Datasets:

haritzpuerto
/

the_pile_00_PubMed_Central

Dataset card Viewer Files Files and versions Community

Split (3)

train · 190k rows

text stringlengths 4 2.52M	meta dict
Introduction ============ A gene tree represents the evolutionary relationships between a set of homologous genes. Gene trees are useful to unveil the molecular evolutionary events that have shaped today\'s genomes. They are traditionally constructed from sequence alignments \[[@B1]\], while recent methods also use the information from species phylogenies through reconciliation \[[@B2]-[@B8]\]. But constructing good gene trees is still challenging: for example, while they yield orthology and paralogy relationships between genes, often alternative or additional information, such as conserved synteny, is used to provide or confirm orthology \[[@B9]\]. The orthology information suggested by gene tree reconciliation may be contradictory with that suggested by an external source, such as conserved synteny \[[@B10],[@B11]\]. We explore a way to reconcile them by performing slight modifications to a given gene tree in order to fit external information on orthology. We propose two kinds of gene tree modification, which consist in computing a gene tree as close as possible to the initial one, satisfying two kinds of constraints. One kind is a set of pairs of genes that should be orthologous but are seen as paralogous in the initial tree. This occurs when orthologs are computed with synteny for example \[[@B11]\]. The other kind is a set of clades that should be rooted by speciation nodes but are rooted by duplication nodes in the initial tree. This occurs when dubious duplications are detected because of the absence of extant support for a duplication, or because of ancestral synteny information \[[@B10]\]. We give polynomial-time algorithms for both problems under the Robinson-Foulds distance, thus proposing several ways to improve gene trees according to external information. There are very few gene tree reconstruction methods including synteny information \[[@B12]\], whereas integrating this information could be valuable \[[@B13]\]. The modifications we propose could be included in a local search framework as other kinds of modifications based on duplications and losses \[[@B14]-[@B17]\]. We assess the validity of the modifications we propose by computing the likelihood ratio between initial and modified trees according to sequence alignments on Ensembl trees \[[@B18]\], showing that often the two trees are statistically equivalent. Different gene tree corrections =============================== Phylogenies ----------- A phylogenyis a rooted binary tree which represents the evolutionary relationships between the nodes. Internal nodes are extinct ancestors, leaves are extant elements and edges represent direct descents between parents and children. Given a node xof a phylogeny T, we call an ancestorof xany node on the path from the root (inclusively) of Tto the parent of x. For a leaf-subset Xof T, lca~T~(X), the lowest common ancestorof X, denotes the farthest node from the root which is an ancestor of all elements of X. We use the notation l(x), and call the cladeof x, the set of leaves which are descendant from an internal node x. We also denote by l(T) the set of leaves, and by V(T) the set of nodes of T. We define two kinds of phylogenies: species trees and gene trees. Species are identified with genomes. For our purpose, genomes are simply sets of genes. Therefore, each gene g, extant or ancestral, belongs to a species s(g). We then have one species tree S, where nodes are identified with species, and many gene trees, where nodes are identified with genes. The set of genes in a gene tree is called a gene family. A reconciliationbetween a gene tree Gand a species tree Sconsists in assigning to each gene gof G(both extant and ancestral) the species s(g) corresponding to the lowest common ancestor in Sof the set {s(l), for all l ∈ l(g)}. Every internal node gof Gis labeled by an event E(g), verifying E(g) = speciation if s(g) is different from s(g~ℓ~) and s(g~r~) where g~ℓ~and g~r~are the two children of g, and E(g) = duplication otherwise. The reconciliation of Gand Sgives all informations about the gene family history. In particular it defines the gene content of an ancestral species at the time of speciation. A reconciliation also implies the orthology and paralogy relationships between genes: two genes gand g\' of Tare said to be orthologousif E(lca~T~(g, g\')) = speciation; gand g\' are paralogousif E(lca~T~(g, g\')) = duplication. For example, Figure [1(1)](#F1){ref-type="fig"} shows a gene tree reconciled with a species tree. In this gene tree a~1~and b~1~are paralogous as their lowest common ancestor is dwhich is a duplication node, while a2 and b2 are orthologous. The number of dots inside big circles represents the number of genes in the corresponding genome (each big circle represents a species). ![Description of the two problems. (1) A gene tree (the \"initial tree\") for the gene family {c, b1, b2, a1, a2} is shown with small red nodes and single thin red edges. It is reconciled with the phylogeny of the three species A, Band Cshown with large green nodes and hollow edges represented by a pair of parallel black lines. Duplication nodes of the reconciled gene tree are squared, while speciation nodes and leaves are dots. (2) The two neighbors of b1 on genome Band of a1 on genome Aare inferred to be orthologous according to their lowest common ancestor in their respective gene trees (not shown). This is an argument for infering orthology between b1 and a1, which is in contradiction with the information provided by the initial tree: their lowest common ancestor is a duplication, and thus they are inferred to be paralogous. (3) A solution to the GOC problem, that is a gene tree of minimum RF distance with the initial tree verifying the constraint of b1 and a1 being orthologous. (4) A solution to the COC problem, that is a reconciled tree in which the clade {b1, b2, a1, a2} of din the initial tree is rather rooted by a speciation node in the corrected tree. This is an example where the optimal solutions to the two problems differ.](1471-2105-14-S15-S5-1){#F1} The Robinson-Fould (RF) distance -------------------------------- The RF distance RF(G, G\') between two phylogenies Gand G\' is the cardinality of the symmetric difference between the clade-sets of the two trees. In other words, denote by c(G, G\') the number of clades that are in Gbut not in G\'. Then RF(G, G\') = c(G, G\') + c(G\', G). In this paper, since we only compare rooted binary trees sharing the same leaf-sets, they always have the same number of internal nodes, and hence the same number of clades. Therefore c(G, G\') = c(G\', G), and RF(G, G\') = 2c(G, G\'). Two correction problems ----------------------- Suppose that in addition to a species tree and a set of reconciled gene trees, we are given additional information of two kinds: • Pairs of genes that we know are orthologous; • Duplication nodes of some gene trees that we suspect to be false. Constraints of orthology on pairs of genes may for example be generated from synteny analysis \[[@B9],[@B11]\]. Some pairs may contradict the information given by the gene tree. Let Pbe a set of pairs (g~1~, g~2~) of orthologous extant genes (verifying s(g~1~) ≠ s(g~2~)). A gene tree Gis said to satisfya set Pif, for any pair (g~1~, g~2~) ∈ P, lca~G~(g~1~, g~2~) is a speciation node. Problem 1 Gene Orthology Correction \[GOC\] Problem *Input: A gene tree G reconciled with a species tree S, and a set P of gene pairs that are required to be orthologous;* *Output: A corrected gene tree G~P~satisfying P, such that RF(G, G~P~) is minimum among all possible solutions. An example is given in Figure [1](#F1){ref-type="fig"}: (1) is the initial tree, and (2) depicts two syntenic regions of size 3 surrounding genes b1 and a1. In general (if we neglect the effect of gene conversion) genes in two syntenic regions should be either all pairwise orthologous or all pairwise paralogous \[[@B11]\]. Consequently, if the two neighbors of b1 on genome Band of a1 on genome Care inferred to be orthologous (according to their lowest common ancestor in their respective gene trees), then an orthology constraint should be imposed on the pair (b1, a1). Figure [1](#F1){ref-type="fig"}. This principle is usually considered as one of the most efficient method to detect orthologies \[[@B9]\]. (3) is a corrected tree. On the other hand, duplication nodes of a gene tree can be considered dubious for different reasons. For example, in Ensembl \[[@B19]\], \"dubious\" is a label assigned to the non-apparent duplication nodes \[[@B20],[@B21]\] pointing to an incongruence between the gene tree and the species tree. Alternatively, inferred ancestral synteny may also point to dubious duplication nodes \[[@B10]\]. Formally, clades corresponding to some duplication nodes may erroneously be considered as sets of paralogous genes, and should rather be considered as orthologous. A gene tree Gis said to satisfya set Cof its clades if E(lca~G~(c)) = speciation for all c ∈ C. Problem 2 Clade Orthology Correction\[COC\] Problem* *Input: A gene tree G reconciled with a species tree S, and a set C of clades of G assigned to duplication nodes;* *Output: A corrected tree GC satisfying C, such that RF(G, G~C~) is minimum among all possible solutions. The two problems are different, as exemplified by Figure [1](#F1){ref-type="fig"}, where (3) is an optimal solution to GOC while (4) is an optimal solution to COC, the latter more distant to the initial tree. In the next two sections, we use Sfor the species tree name, Gfor the reconciled gene tree, and we give efficient solutions to these two problems. The Gene Orthology Correction Problem ===================================== Notice that for any instance of the GOC problem, a corrected tree satisfying Palways exists. Indeed, for any extant species xof S, one can make a tree whose leaf-set is all the extant genes gof Gfor which s(g) = x. Doing this for every species yields a forest whose roots can be reconnected by matching the topology of S, ensuring that any pair of genes not in the same species are orthologous. However, the obtained tree can be very far from the original. Let Pbe a set of gene pairs (which are leaves of G) required to be orthologous. Notice that if (a, b) ∈ P, then we also have (b, a) ∈ P. For any pair (a, b) ∈ P, if lca~G~(a, b) is a duplication in G, then (a, b) is a pair of false paralogs. The set P~f~⊆ Pdenotes the set of all false paralogous pairs of P. Given two distinct leaves aand bof G, we set r~a,b~= lca~G~(a, b), s~a,b~= lca~S~(s(a), s(b)), and define h~a,b~as the highest node (closest to the root) on the path from ato r~a,b~such that s(h~a,b~) is a descendant of s~a,b~. Notice that h~a,b~can be aitself, but not r~a,b~. For instance on Figure [2(1)](#F2){ref-type="fig"}, a~1~, c~2~are false paralogs with $r_{a_{1},c_{2}} = e_{3}$ and $s_{a_{1},c_{2}} = E$. From this, one can deduce that $h_{a_{1},c_{2}} = d_{2}$ and $h_{c_{2},a_{1}} = c_{2}$. We show below that, for any pair (a, b) of false paralogs, h~a,b~is the highest node on the path from ato r~a,b~over which we can move bto make lca~G~(a, b) a speciation node. The reason for moving bas high as possible is to preserve as many clades as possible, allowing a minimum RF distance between the initial and corrected tree. ![GOC Procedure. (1) A gene tree Greconciled with species tree S. Duplication nodes are denoted by a black square. The leaves and internal nodes of Gare labeled with the letter of their corresponding species. Brackets denote the required orthologs given by the input set P= {(a~1~, b~2~), (a~1~, c~1~), (a~1~, c~2~)}. The non-preservable nodes (nodes of H) are depicted by red crosses, while preservable nodes are circled in green. (2) The species tree associated with G. (3) The tree G~P~, a solution to the GOC problem, which preserves every possible clade.](1471-2105-14-S15-S5-2){#F2} Lemma 1*Let(a, b) be a pair of false paralogs in G, and let G\' be a tree in which a and b are orthologous. If x is an ancestor of h~a,b~and a descendant of r~a,b~, then the clade of x is not in G\'. Proof:Suppose otherwise that there is some x\' ∈ V(G\') with the same clade as x(and hence s(x) = s(x\')). Let r\'~a,b~= lca~G\'~(a, b), which should be a speciation. Since bwas not in the clade of x, it cannot be in the clade of x\' either, implying that r\'~a,b~is an ancestor of x\'. Also, since s(x\') = s(x) and xis above h~a,b~in G, we have that s(x\') is s~a,b~or one of its ancestors (otherwise we would have picked xto be h~a,b~). But r\' has x\' in one of its subtrees, and bin the other, implying that r\'~a,b~is a duplication: contradiction. □ We now have a way to identify a set of clades that cannot be in G~P~. For any (a, b) ∈ P~f~, denote by H~a,b~the set of ancestors of h~a,b~that are descendants of r~a,b~. If G~P~satisfies the set P~f~, G~P~cannot contain any clade from the set $H = \cup_{{({a,b})} \in P_{f}}H_{a,b}$. It follows that a minimum of \\|H\\|clades of Gare missing in G~P~. We claim that a solution G~P~to the GOC problem is obtained by modifying exactly c(G, G~P~) = \\|H\\|clades. Theorem 1Let G~P~be a solution to the GOC problem. Then RF(G, G~P~) = 2\\|H\\|. In what follows, we give a constructive proof of Theorem 1 by describing an algorithm for solving the GOC problem. An algorithm for the GOC problem -------------------------------- Call V(G)\\Hthe set of preservable nodesof G(those that we hope to preserve). For example in Figure [2(1)](#F2){ref-type="fig"}, $H = H_{a_{1},c_{2}} \cup H_{c_{2},a_{1}} \cup H_{a_{1},c_{1}} \cup H_{c_{1},a_{1}} \cup H_{a_{1},b_{2}} \cup H_{b_{2},a_{1}} = \left\{ e_{1} \right\} \cup \left\{ e_{2} \right\} \cup \varnothing \cup \varnothing \cup \left\{ d_{1} \right\} \cup \varnothing = \left\{ {e_{1},e_{2},d_{1}} \right\}$.The nodes of Hare represented by red crosses, while the preservable nodes are circled in green. Notice that the root rof Gis preservable, since any solution G~P~to the GOC problem should share the same leaf-set as G. Consider the set $~\mathcal{G}$ of subtrees of Grooted on the highest preservable descendantsof r, i.e. preservable nodes for which ris the unique preservable ancestor. Observe that since any leaf of Gis preservable, we have $\cup_{G_{x} \in ~\mathcal{G}}l\left( G_{x} \right) = l\left( G \right)$. If, for some (g~1~, g~2~) ∈ P, g~1~and g~2~are scattered across two subtrees of G, we call these subtrees required orthologous subtrees(or simply required orthologswhen the context is clear as to whether we are comparing genes or subtrees). For example in the tree Gof Figure [2(1)](#F2){ref-type="fig"}, Gis the set of subtrees rooted at d~2~, c~1~, b~3~and c~2~(the last four restricted to a single leaf), and the subtrees rooted at d~2~and c~1~are required orthologs, as well as those rooted at d~2~and c~2~. However, connecting two subtrees under a speciation might not always be feasible. A definition of possible orthologsfollows. Definition 1 (Possible orthologs)Two subtrees G~1~, G~2~∈ $~\mathcal{G}$rooted at x~1~, x~2~respectively arepossible orthologs if and only if s(x~1~) and s(x~2~) are unrelated, i.e. neither is an ancestor of the other in S. The following lemma ensures that the roots of required orthologous subtrees can actually be joined under a common parent which is a speciation. Lemma 2Let G~1~, G~2~∈ $~\mathcal{G}$be required orthologs. Then G~1~and G~2~are possible orthologs. Proof:Let x~1~, x~2~be the roots of G~1~, G~2~respectively, and let (g~1~, g~2~) ∈ Psuch that g~1~∈ l(G~1~) and g~2~∈ l(G~2~). Let s~ℓ~, s~r~be the left and right children of $s_{g_{1},g_{2}}$, and denote by S~ℓ~and S~r~the subtrees of Srooted at s~ℓ~and s~r~respectively. Suppose without loss of generality that s(g~1~) is in l(S~ℓ~) and s(g~2~) is in l(S~r~). Since x~1~is preservable and on the path between g~1~and $r_{g_{1},g_{2}}$, we have $x_{1} \notin H_{g_{1},g_{2}}$ and thus s(x~1~) ∈ V(S~ℓ~). Similarly, s(x~2~) ∈ V(S~r~). Therefore s(x~1~) and s(x~2~) are unrelated and possible orthologs. The problem, formally defined in the sequel as the maximum orthology tree, consists in joining all trees of $~\mathcal{G}$ into a single tree G\' in a way ensuring that each pair of possible orthologs is joined under a speciation. More precisely, for some possible orthologs $G_{1},G_{2} \in ~\mathcal{G}$ rooted at nodes x~1~, x~2~, we get that lca~G\'~(x1, x~2~) is a speciation, with G~1~, G~2~being unchanged. We begin by giving an overview of the whole algorithm. Algorithm Outline: 1\. Compute the set $H = \cup_{{({a,b})}\; \in \; P_{f}}H_{a,b}$ of internal nodes of Gcorresponding to clades that cannot be in G~P~; 2\. Compute the set $~\mathcal{G}$ of subtrees rooted at the highest preservable descendants of the root of G. If $~\mathcal{G}$ is empty, return Gand terminate; 3\. Construct a tree G\' by joining all trees of $~\mathcal{G}$ in a way ensuring that possible orthologs are joined under speciation. We call G\' the maximum orthology treefor $~\mathcal{G}$; 4\. For every tree $G_{x} \in ~\mathcal{G}$, construct G~x,P~by recursively repeating Steps 2 to 4 with Gbeing G~x~, and replace the G~x~subtree of G\' by G~x,P~. The tree obtained corresponds to the corrected tree G~P~we want. Running this algorithm on the Gtree of Figure [2](#F2){ref-type="fig"} yields the corrected tree G~P~. This algorithm terminates, since we eventually reach all the leaves of G, which correspond to terminal cases in the recursion. Implementing step 1 is straightforward, while step 2 can be done by performing a depth-first search from the root, in which upon visiting a preservable node, we add it to $~\mathcal{G}$ and continue the search without visiting its children. Step 3 is the purpose of the next section, so assume for now that it can be performed correctly as stated. This algorithm can be implemented to run in O(\\|P \\| × \\|V(G)\\|) steps in the worst case, the main bottleneck being the computation of H. The algorithm correctness follows from the two lemmas below. Lemma 3Any preservable node x of G ispreserved in G~P~, meaning that the clade of G rooted at x is a clade of G~P~. Proof:Let xbe a preservable node of Gand G~x~be the subtree rooted at x. It is not hard to see that eventually, steps 2-4 will be run on G~x~and return a tree G~x,P~, which will itself be a subtree of the final corrected tree G~P~. As the algorithm only moves and reconnects subtrees of G~x~, we have that l(G~x~) = l(G~x,P~). Since G~x,P~is a subtree of G~P~, it follows that the clade of xis preserved in G~P~. Lemma 4Let(g~1~, g~2~) ∈ P. Then g~1~and g~2~are orthologs in G~P~. Proof:Denote by G~v~the subtree rooted at v, for some v∈ V(G). Let xbe a preservable node and G~x,P~be the subtree produced after running steps 2-4 on G~x~. Let Dbe the set of highest preservable descendants of x. We say that a gene pair (g~1~,g~2~) is containedin G~x~if g~1~, g~2~∈ l(G~x~). We use induction on the height of the tree to show that all gene pairs in Pthat are contained in G~x~are orthologous in G~x,P~(which proves the lemma since xcan be the root). This is trivially true for leaves as they are preservable and contain no gene pairs. We thus suppose by induction that for any d∈ D, gene pairs in Pthat are contained in G~d~are orthologous in G~d,P~. Let (g~1~, g~2~) ∈ Psuch that (g~1~, g~2~) is contained in G~x~, but there is no d∈ Dsuch that G~d~contains (g~1~, g~2~). What is left to prove is that g~1~and g~2~are orthologous in G~x,P~. We first observe that g~1~, g~2~belong to two different subtrees $G_{d_{1}},G_{d_{2}}$, where d~1~, d~2~∈ D. Otherwise $G_{d_{1}} = G_{d_{2}}$, implying that (g~1~, g~2~) is contained in $G_{d_{1}}$ and we are done. Therefore, $G_{d_{1}}$,$G_{d_{2}}$are required orthologs, and hence possible orthologs. Since we may assume that G~d1~ and G~d2~ are joined under a speciation in G~x,P~, we get that $\textsf{lc}\textsf{a}_{G_{x,P}}\left( {g_{1},g_{2}} \right)$ is a speciation. The result follows from observing that G~x,P~is a subtree of G~P~. Maximum orthology tree ---------------------- We now describe a solution to the maximum orthology tree problem. Formally, given a set of kpossible orthologous subtrees of Grooted on a set of nodes X= {x~1~, . . . , x~k~}, the problem is to construct a tree Fwith l(F) = X, such that for each pair x~i~, x~j~∈ Xthat correspond to roots of possible orthologs, x~i~and x~j~are orthologous in F. Roughly speaking, the algorithm proceeds as follows: start with F~0~being a copy of S. Iterate over ifrom 1 to k, at each step constructing F~i~by grafting x~i~on F~i-1~right above the node v∈ V(F~0~) such that s(v) = s(x~i~). Proceeding this way, we show in Lemma 5 that nodes of V(F~0~) are ensured to remain speciation nodes all over the procedure, and in lemma 6 that the lowest common ancestor of two possible orthologs belongs to V(F~0~), leading to corollary 1 stating that possible orthologs are in fact orthologous in the output tree. Finally remove the leaves artificially introduced by F~0~and standardizethe tree, which means • remove all nodes with no descendant labeled with extant genes; • contract non-root degree 2 nodes, then contract the root if it is of degree one. Starting with F~0~being a copy of Sis a step that might be omitted, but the set of nodes V(F~0~) serves as a skeleton around which we graft our x~i~\'s, making it both easily implementable and provable. Figure [3](#F3){ref-type="fig"} shows how the algorithm proceeds on the set of highest preservable descendants of the root of the tree Gin Figure [2(1)](#F2){ref-type="fig"}. ![The Max Orthology problem. An instance of the max orthology problem, with Xbeing the highest preservable descendants of the root of Gin figure 2. (1) The starting tree F~0~, which is a copy of S. (2) The F~k~tree, which depicts the tree obtained after grafting every node of X. (3) The final tree F, obtained by removing the leaves initially in F~0~and standardizing.](1471-2105-14-S15-S5-3){#F3} Algorithm 1findMaxOrthology(S, X= {x~1~, \... , x~k~}) F~0~←A copy of S V~0~← V(F~0~) L ← l(F~0~) fori= 1 → kdo Find the unique node v∈ V~0~such that s(v) = s(x~i~) F~i~←a copy of F~i\ -\ 1~on which we graft x~i~on the edge linking vto its parent node (or if vis the root of F~i\ -\ 1~, create a new root with children vand x~i~) end for F ←F~k~on which we remove Land stardardize Lemma 5If r∈ V(F~0~) ∩ V(F), then r is a speciation. Proof:Since F~0~is a copy of S, all nodes of V(F~0~) are initially speciation nodes. We show that each grafting operation does not change the event corresponding to these nodes. Say that at iteration i, we graft x~i~on the edge linking vto its parent node p. We first observe that the only nodes that can be transformed from speciation in F~i-1~to duplication in F~i~are on the path from pto the root of F~i-1~. Suppose without loss of generality that vis the left child of pin F~i-1~, and let wbe the newly created node between pand vin F~i~. Thus whas children xiand v, and since s(x~i~) = s(v), we get that s(w) = s(v). It follows that if pwas a speciation in F~i-1~, it remains a speciation in F~i~. Moreover, this implies that s(p) is left unchanged in F~i~, implying in turn that any ancestor of pcannot change from speciation to duplication. Therefore, no grafting operation can affect speciation of any vertex in V(F~i-1~). Finally, we note that removing leaves or deleting degree two nodes in Falso cannot affect speciation nodes. Lemma 6Let x~i~, x~j~∈ X be the roots of possible orthologous subtrees. Then, lca~F~(x~i~, x~j~) ∈ V(F~0~). Proof:First recall that if x~i~,x~j~are the roots of possible ortholog subtrees, then there is some s∈ V(S) such that s(x~i~) and s(x~j~) are in the left and right subtrees of s, respectively. Now, let rbe the unique node in V(F~0~) such that s(r) = s, and let v~i~, v~j~∈ V(F~0~) such that s(v~i~) = s(x~i~) and s(v~j~) = s(x~j~). It is clear that in F~0~, lca(v~i~,v~j~) = r. This also holds for any F~i~by observing that grafting nodes cannot change the lca relationship. Since x~i~is grafted on some edge between v~i~and r, and x~j~between v~j~and r, it follows that lca(x~i~, x~j~) = r∈ V(F~0~). Corollary 1Let x~i~, x~j~∈ X be the roots of possible orthologs. Then they are orthologous in F. The Clade Orthology Correction Problem ====================================== We prove several results characterizing the solutions to the COC problem. Let Cbe a set of clades that has to be satisfied. For a clade c∈ C, we denote by s(c) the value of s(r(c)) where r(c) is the root of c, and by E(c) the value of E(r(c)) that we call the label of c. First, unlike in the GOC problem, a solution to the COC problem does not always exist. Indeed, it is possible that no gene tree has all clades in Clabeled by speciations. We give a necessary and sufficient condition for the existence of a solution. The following lemma is obvious from the definition of reconciliation, and will be used in several proofs. Lemma 7For a reconciled gene tree G, if a node x is an ancestor of a node y and s(x) = s(y) then E(x) = duplication. Theorem 2There is a solution to the COC problem if and only if for every clade c∈ C, s(c) is not a leaf of S, and if for every pair c~1~, c~2~∈ C, either c~1~and c~2~are disjoint sets of leaves, or s(c~1~) ≠ s(c~2~). The necessity of these conditions directly follow from Lemma 7, since s(c~1~), s(c~2~) and the ancestry relationship between c~1~and c~2~remain unchanged in a solution. Their sufficiency will be constructively demonstrated in the sequel. Suppose that the conditions are satisfied. We give a way of finding all optimal solutions according to the RF distance, followed by two ways of finding an optimal one optimizing other criteria in addition. Given a duplication node xof G, pushing x by multifurcationmeans applying the following procedure: • Let s= s(x), and Aand Bbe the two children of sin S. • Let T^A^be the set of maximal subtrees of the subtree of Grooted at x, such that all their leaves lverify that s(l) is a descendant of A(including Aitself). Let G^A^\[x\] be the multifurcated tree obtained by joining all roots of trees in T^A^under a common root. • Let symmetrically T^B^be the set of maximal subtrees of the subtree of Grooted at x, such that all their leaves lverify that s(l) is a descendant of B(including Bitself). Let G^B^\[x\] be the multifurcated tree obtained by joining all roots of trees in T^B^under a common root. • Let G\' be obtained from Gby replacing the clade rooted at xby a new subtree, obtained by joining G^A^\[x\] and G^B^\[x\] under a common root. This rearrangement is described in \[[@B16]\] and applied to dubious duplications as a preprocessing step for ancestral genome reconstruction. A binary resolution G~b~of a multifurcated tree Gis a binary tree in which all the clades of Gare in G^b^. Theorem 3If there is a solution to the COC problem, then a binary gene tree is an optimal solution if and only if it is a binary resolution of the multifurcated tree obtained by pushing the roots of the elements of C by multifurcation (in any order). Proof:It is clear that a binary resolution is a solution, provided that the conditions for the existence of a solution are satisfied. Indeed any clade is preserved through pushing a duplication node, so this operation can be done for all clades in Cindependently. This proves the converse part of Theorem 2. Then it is an optimal solution because by Lemma 7, no clade xwhich is a descendant of the pushed clade csuch that s(c) = s(x) may be conserved if we want cto be a speciation node. And by construction all clades such that s(c) ≠ s(x) are preserved by this operation. Binary resolutions which minimize the number of duplications and losses are studied by \[[@B22]\] and may be applied to provide bona fidephylogenies. We describe an alternative maximizing the number of common triplets. A tripletin a tree Gis a set of three leaves ((a, b), c) of G, such that the LCA of the three is strictly more ancient than the LCA of the first two. Given a species tree S, a reconciled gene tree Gand one of its duplication nodes x, pushing x by tree duplicationmeans applying the following procedure, illustrated in Figure [4](#F4){ref-type="fig"}: ![COC Procedure. An instance of the COC problem. In the gene tree G, where s(a~1~) = s(a~2~) = A, s(b~1~) = s(b~2~) = Band s(c) = C, extract and copy the subtree rooted at × to get the subtrees G^A^(x) and G^B^(x). Remove b~1~and b~2~from G^A^(x) and a~1~and a~2~from G^B^(x). Join G^A^(x) and G^B^(x) under a common root and replace G(x) by the new subtree in the gene tree G\'.](1471-2105-14-S15-S5-4){#F4} • Let s= s(x), and Aand Bbe the two children of sin S. • Let G^A^\[x\] be a tree obtained from the subtree of Grooted at x, by deleting all leaves lwith s(l) being a descendant of A, and standardizing it, which as in the previous sections, means -removing all nodes with no descendant labeled with extant genes; -contracting non-root degree 2 nodes, then contracting the root if it is of degree one. • Let symmetrically G^B^\[x\] be a tree obtained from the subtree of Grooted at x, by deleting all leaves lwith s(l) being a descendant of B, and standardizing it. • Let G\' be obtained from Gby replacing the clade rooted at xby a new subtree, obtained by joining G^A^\[x\] and G^B^\[x\] under a common root. Note that if a clade yis disjoint from xor assigned to a different species, then pushing xby tree duplication does not affect the subtree rooted at y. In consequence, pushing several clades by tree duplications in any order gives a unique solution if the clades satisfy the properties of Lemma 2. Theorem 4If there is a solution to the Clade Orthology Correction problem, the gene tree obtained by successively pushing the roots of the elements of C by tree duplication (in any order) is an optimal solution. Among all optimal solutions, it maximizes the number of common triplets with G. Proof:As already noticed pushing a duplication by multifurcation preserves all clades assigned to species which are different from the species assigned to the pushed node. So it is an optimal solution. Now we have to prove that none of the triplets that are in Gbut not in G\' can be preserved in any other optimal solution. For this we characterize the triplets that can be preserved. For a triplet ((a, b), c) of G, let T~((a,b),c)~be the rooted phylogeny with three leaves and two internal nodes containing the triplet. If the leaves a, b, care in the pushed clade x, then the triplet can be preserved only if in the reconciliation of T~((a,b),c)~, the lowest internal node is not mapped to s(x). Otherwise by Lemma 7, the root node of the triplet cannot be a speciation. Let ((a, b), c) be a triplet such that in the reconciliation of T~((a,b),c)~, the lowest internal node is not mapped to s(x). This triplet is entirely included in G^1^\[x\] or G^2^\[x\]. So it is preserved. In consequence all triplets possibly preserved are indeed preserved by the operation, showing the optimality of the procedure reguarding the number of common triplets. Now if there is no solution to the Clade Orthology problem, we advice to push duplication nodes in Cstarting from the highest ones, without having formalized why we find this solution adequate. Fish gene trees =============== Using synteny as evidence of orthology, we wanted to test the ability of our algorithm designed for the GOC problem to correct gene trees. To this end, we considered the four fish genomes Gasterosteus aculeatus(Stickleback), Oryzias latipes(Medaka), Tetraodon nigroviridis, and Danio rerio(Zebrafish) with human and mouse as outgroups. We used the Ensembl Genome Browserto collect all available gene trees, and filtered each tree to preserve only genes from the taxa of interest. We then reconciled the trees with the known species trees, and identified duplication and speciation nodes. Following our methodology in \[[@B11]\], a region surrounding a gene is defined as the substring containing the gene and both its left and right adjacencies, and two regions are considered syntenic if they contain homologous genes in the same order. We observed in \[[@B11]\] that more than 22% of the 6241 collected gene trees contain at least one false paralogy, that is a pair of genes required from synteny to be orthologous, but the LCA of the corresponding leaves being a duplication rather than a speciation node. For 1000 of the trees containing at least one false paralogy, we applied the correction procedure previously described, and retrieved the gene family alignment from Ensembl. With PhyML \[[@B23]\], we computed the likelihood of the initial and corrected tree, given the alignment. These two likelihood values were compared with Consel \[[@B24]\]. For only 17.7% of the trees, the correction was rejected by the AU test. In other words, the correction algorithm is valid for a vast majority (82.3%) of the tested trees. Moreover, the likelihood of the corrected tree is higher than the original for 44.4% of the trees. Interestingly, 14.8% of the original Ensembl gene trees were rejected when compared to the corrected trees. The correction of the gene tree for the ZNF800gene family, which is related to transcriptional regulation, is given as an example in Figure [5](#F5){ref-type="fig"}. The corrected tree was highly favored by the AU Test, giving it a statistical support advantage with a p-value below 0.001. Furthermore, the non-apparent duplication of G, located at the root of the (m~1~, t~1~, s~1~) subtree, was eliminated, resulting in one less duplication in G~P~. ![An example of corrected fish tree. The tree for the ZNF800 gene family before and after correction, restricted to the species Stickleback (S), Medaka (M), Tetraodon (T) and Zebrafish (Z). (1) The original gene tree Ggiven by Ensembl, using the same notation as in figure 2 for duplications, preservable nodes and required orthologs. Gene region analysis gave us the required orthologs P= {(m~1~, s~1~), (t~1~,d~2~)}. (2) The species tree associated with the four species. (3) The gene tree given by our correction algorithm.](1471-2105-14-S15-S5-5){#F5} Conclusion ========== We give two efficient algorithms for two new gene tree rearrangement problems, related to the correction of a gene tree according to some external information on orthology. The rearrangements are modifications that are as small as possible, given some distance criterion (namely the RF distance), but can be more significant according to other distances such as the usual NNI (nearest neighbor interchange) distance. We show that for fish genomes, the rearrangements we define can be efficient to explore statistically equivalent gene trees when sequence alignement is used to compute likelihood. As corrected trees satisfy synteny contraints, we can be confident enough that they describe the gene family evolution better. Many algorithmic and theoretical problems remain open. For example, is there a similar way for handling paralogy constraints? What about having both orthology and paralogy constraints? It can be shown that there exist sets of constraints with both types that cannot be satisfied. What are the conditions for a set of orthology/paralogy constraints to be satisfiable? These algorithms may be used in a global framework to contruct large gene tree sets which are arguably better than those found in standard databases. The implementation of such a framework is an on-going work. Competing interests =================== None Authors\' contributions ======================= ML, MS, KS, ET, NE modeled the problem, devised the algorithms and wrote the paper. ML implemented the software. Declarations ============ This work is funded by the Natural Sciences and Engineering Reserach Council of Canada (NSERC), Fonds de Recherche Nature et technologies of Quebec, Agence Nationale pour la Recherche and Ancestrome project ANR-10-BINF-01-01. This article has been published as part of BMC BioinformaticsVolume 14 Supplement 15, 2013: Proceedings from the Eleventh Annual Research in Computational Molecular Biology (RECOMB) Satellite Workshop on Comparative Genomics. The full contents of the supplement are available online at <http://www.biomedcentral.com/bmcbioinformatics/supplements/14/S15>.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-polymers-11-01925} =============== Thermoplastic resins, such as polyethylene, polypropylene, polyethylene terephthalate, and polyvinyl chloride, have been widely used in packaging applications, for example, bottles, containers, vessel, labels, foods wraps, and cap seals, which need shrinkable polymer films. The characteristic of shrink film is its ability upon exposure to heat to either shrink or, if restrained, to create shrink tension within the film. This behavior occurs because the film's molecules are oriented by rapidly quenching after being stretched and heated to their orientation temperature range. Upon reheating, these oriented molecular segments in the amorphous region start to relax and return to their original random coil state. Therefore, when the wrapped product is passed through a hot air, it causes the film to shrink around the product producing a tight pack \[[@B1-polymers-11-01925]\]. The manufacture of shrink films usually use multilayer films because they allow a combination of their properties, such as shrink tension, optical clarity, hot seal strength, and shrink temperature range, which are unobtainable in a monolayer film. In addition, film multilayering processes result in high-speed packaging lines \[[@B2-polymers-11-01925]\]. However, these processes are relatively expensive due to the cost of the interpolymer, which is used as a compatabilizing layer \[[@B3-polymers-11-01925],[@B4-polymers-11-01925]\]. Heat shrink films conventionally used for packaging and the like are mainly polyvinyl chloride \[[@B5-polymers-11-01925],[@B6-polymers-11-01925]\], polyethylene terephthalate \[[@B7-polymers-11-01925],[@B8-polymers-11-01925]\] polystyrene \[[@B9-polymers-11-01925]\], or polypropylene \[[@B10-polymers-11-01925]\]. These kinds of plastic products are usually used and disposed immediately after use. Lately, many countries have attempted to identify and reduce, or even ban some single-use plastic items as this take-make-use-dispose is not sustainable. The replacement of fossil-based plastics with renewable-based plastics is one of the sustainability options. The biodegradable polymers have currently been utilized as alternative materials to the existing commodity plastics in some certain applications. Among the biodegradable and compostable plastics, poly(lactic acid) (PLA) is one of the most market-available biodegradable polymer. It possesses good mechanical properties as well as superior optical clarity. However, to be suitable for some applications, it requires an improvement in its elongation at break, impact strength, and film flexibility while maintaining its transparency. In addition, PLA film fabricated by blown film extrusion is very difficult to produce due to its brittleness and low melt strength. One of the efficient methods for the PLA film development is the blending with flexible or soft polymers. Ethylene vinyl acetate (EVA) is a copolymer of ethylene and vinyl acetate (VA). The properties of EVA copolymer usually depend on concentration of VA. EVA copolymer with low VA content (\<20%) are usually employed as thermoplastics, whereas those with high levels of VA are generally considered as oil-resistant elastomers \[[@B11-polymers-11-01925],[@B12-polymers-11-01925]\]. Several works have been carried out for PLA/EVA blends and reported that the incorporation of EVA into PLA significantly improved the elongation at break, toughness, and impact strength of PLA \[[@B13-polymers-11-01925],[@B14-polymers-11-01925],[@B15-polymers-11-01925],[@B16-polymers-11-01925],[@B17-polymers-11-01925]\]. Moreover, the addition of reactive agents could further improve the compatibility between polymers, for instance, glycidyl methacrylate \[[@B18-polymers-11-01925]\], phthalic anhydride \[[@B19-polymers-11-01925]\], and peroxide \[[@B20-polymers-11-01925],[@B21-polymers-11-01925]\]. The addition of reactive agents could promote in situ copolymerization through the free radical reaction using a peroxide initiator. The interfacial adhesion between phases of two polymers can be consequently improved. To the best of our knowledge, the systematic reactive blend of PLA/EVA including the mechanical model of the stretching and shrinking behavior of PLA/EVA reactive blend films have not been reported. Therefore, this work aimed to investigate the mechanical properties and thermal properties of PLA/EVA in reactive blend systems. The addition of EVA could lead to an increase in the blown film extrusion capability of PLA. The heat shrinkable behavior of PLA/EVA reactive blends film was also studied and discussed. 2. Materials and Methods {#sec2-polymers-11-01925} ======================== 2.1. Materials {#sec2dot1-polymers-11-01925} -------------- The grade LX-175 PLA was kindly supplied by Total Corbion PLA (Bangkok, Thailand) Ltd. The melting and the glass transition temperatures were 155 and 55 °C, respectively. The grade TV-1055 EVA, having a vinyl acetate content of 28%, was purchased from TPI Polene (Public) Co., Ltd. (Bangkok, Thailand). The Joncryl^®^ 4368 was obtained from BASF, Bangkok, Thailand. Its molecular weight (M~w~) was 6800 Da with the epoxy equivalent weight of 285 g/mol. Di(tert-butylperoxyisopropyl) benzene or Perkadox14s (peroxide 38.64%) was purchased from AkzoNobel, Bangkok, Thailand. 2.2. Processing of PLA/EVA Film {#sec2dot2-polymers-11-01925} ------------------------------- PLA and EVA pellets were first dried at 60 °C for 6 h before use. The formulations and compositions of PLA/EVA blends were showed in the [Table 1](#polymers-11-01925-t001){ref-type="table"}. The amount of Perkadox (designated as P) was varied from 0.1 to 0.2 phr whereas Joncryl^®^ (referred to J) was mixed as a viscosity modifier with the fixed concentration at 0.5 phr. PLA/EVA and reactive agents were melt-mixed in a twin screw extruder under a temperature profile from the feed zone to the die set of 80/110/150/170/180/180/190/190/190/190 °C with a screw speed of 450 rpm. The extruded PLA/EVA blends were dried, cut, and kept in a plastic zip bag. Neat PLA and PLA/EVA with reactive agent films were fabricated through extrusion blowing under a temperature profile from the feed zone to the die set of 135/165/175/185/190 °C with the screw speed of 45.5 rpm. A film-blowing tower was collected with a calendering nip and take-up rolls. 2.3. Characterization and Testing of Films {#sec2dot3-polymers-11-01925} ------------------------------------------ The melt flow index of all samples was performed on a melt flow indexer operating at 190 °C with a 2.16 kg load according to ASTM D1238. The mechanical properties of the blend films were tested by using a universal testing machine (Instron 5969, Instron, MA, USA) in accordance with ASTM D882. The samples were cut into strips from blown films and divided in two directions; the machine direction (MD) and transverse direction (TD). The strips were tested at a cross-head speed of 50 mm/min. The dynamic mechanical thermal properties of PLA/EVA blend were examined using ANTON PAAR, modular compact rheometer (MCR302, Anton Paar, Graz, Austria) equipped with rectangular fixture (SRF) holders. A temperature sweep was heated from 30 to 150 °C at a heating rate of 3 °C/min under the dynamic mechanical thermal analysis (DMTA) torsion mode at a frequency of 1 Hz. To prepare the DMTA specimen, the extruded PLA/EVA was compressed by compression molding and the sample were laser-cut to dimensions of 10 mm × 1.16 mm × 40 mm. To investigate the reaction between PLA and EVA, the gel content of PLA/EVA film was measured. Sample films were dissolved in chloroform at 60 °C for 24 h. The gel content was computed by the following Equation (1):$${{Gel}\ {content}\ \left( \% \right)\ = \ }\left( \frac{w_{g}}{w_{0}} \right){\ \times \ 100}$$ where w~0~ is the original weight (dry) of the samples, and w~g~ is the weight remaining (dry gel component). To follow the possible chemical reaction of PLA and EVA, Fourier transform infrared spectroscopy (FTIR) was performed at ambient temperature on a Vertex 70 spectrometer (Bruker, MA, USA). Before testing, the dried gel from the gel content experiments were mixed with potassium bromide (KBr) into flakes. The FTIR spectra of the samples were recorded in the wavenumber range of 400--4000 cm^−1^. 2.4. Stretching and Shrinking Tests {#sec2dot4-polymers-11-01925} ----------------------------------- To study the stretching and shrinking behavior of films, the films of PLA/EVA with reactive agents were stretched by a uniaxial stretching machine at a laboratory scale. The film samples were prepared with a size of 5.08 cm × 12.4 cm. The grid lines of a 4 × 8 tables with 1 cm × 1 cm cells were drawn on the films surface for further shrinkage measurement. The films were held and stretched by the uniaxial stretching machine with stretching ratios of 2, 3, and 4 under a temperature of 70 °C. Then the stretched films were cooled immediately to preserve the stretched films' shapes and sizes. After that the stretched films were reheated by introducing to a water bath at 70 °C for 1 min to create the stretched film shrinkage. The dimensions of films after soak in the water bath were measured and the percentage of shrinkage in the machine direction (MD) and the transverse direction (TD) was calculated by Equation (2):$${\%\ {shrinkage}\ = \ }\frac{\left. \left( L \right._{stretched}{- \ L}_{shrunk} \right)}{\left. \left( L \right._{stretched}{- L}_{i} \right)} \times {\ 100}$$ where L~shrunk~ is the length of the sample after it was shrunk, L~stretched~ is the length of the sample after stretching and L~i~ is the initial length of the sample. The morphology of blends and PLA/EVA reactive blend films were investigated by field emission scanning electron microscopy (FESEM) (TESCAN MIRA3 LMH Schottky, Brno, Czech Republic). The samples were fractured under liquid nitrogen. The samples were gold-coated before studying the morphology. 3. Results and Discussion {#sec3-polymers-11-01925} ========================= 3.1. Melt Flow Index of Neat PLA and PLA/EVA Blends {#sec3dot1-polymers-11-01925} --------------------------------------------------- The melt flow index values (MFI) of all samples were listed in [Table 2](#polymers-11-01925-t002){ref-type="table"}. The MFI values of PLA/EVA increased with the EVA contents. The MFI values of Joncyl or Perkadox added to the PLA/EVA blend decreased when compared to that of neat PLA, indicating the higher viscosity and the possible reaction between PLA or EVA and incorporated reactive agents. The addition of Joncryl into the PLA/EVA blend could result in the coupling reaction between end groups of PLA and multifunctional epoxide groups of Joncryl, which consequently led to the higher molecular weight and viscosity of PLA \[[@B22-polymers-11-01925],[@B23-polymers-11-01925]\]. On the other hand, the Perkadox could promote the free radical reaction between PLA and EVA, hence leading to the improved interfacial adhesion between PLA and EVA. The viscosity was thus increased as a result of the higher molecular weight or copolymer. The possible reaction cross-linked PLA and EVA is proposed and discussed later. 3.2. Morphology and Fracture Surface of PLA/EVA Blend {#sec3dot2-polymers-11-01925} ----------------------------------------------------- The morphology of blends and PLA/EVA reactive blend films were investigated by field emission scanning electron microscope (FESEM). From SEM micrograph of PLA/EVA reactive blends in [Figure 1](#polymers-11-01925-f001){ref-type="fig"}, it was shown that the dispersion of EVA particles within the PLA matrix with the dimension of EVA dispersed phase of approximately 2 μm. The voids between PLA matrix and dispersed EVA particles could be observed in non-reactive blend ([Figure 1](#polymers-11-01925-f001){ref-type="fig"}a). It indicated the poor interface of PLA and EVA. In the case of adding Joncryl to the blend (in [Figure 1](#polymers-11-01925-f001){ref-type="fig"}b), the good and fine dispersion with the smaller particle size of EVA could be seen. This implied that the interfacial adhesion between PLA and EVA phases was improved by the reaction between epoxide groups and PLA or EVA chains. For the addition of Perkadox, the SEM micrographs of PLA90 + J0.5/P0.1 and PLA90 + J0.5/P0.2 were showed in [Figure 2](#polymers-11-01925-f002){ref-type="fig"}a) and [Figure 2](#polymers-11-01925-f002){ref-type="fig"}b, respectively. The addition of Perkadox at 0.2 phr exhibited the larger EVA particles and non-uniform size when compared to 0.1 phr. This attributed to the possible increasing of the cross-link reaction and led to the poor interfacial adhesion between PLA and EVA phases. This result agreed with the gel content and DMTA results discussed in the next section. Furthermore, for the tensile properties of PLA/EVA reactive blend, it was expected that the increased Perkadox content from 0.1 to 0.2 phr would increase the elongation at break but the elongation at break of PLA90 + J0.5/P0.2 was lower than expectation. Therefore, the fracture surface of tensile testing was investigated. The result showed that SEM micrograph of fracture surface of PLA90 + J0.5/P0.2 (in [Figure 3](#polymers-11-01925-f003){ref-type="fig"}) exhibited the dent on the surface of film. This might be due to the gel on the film and could act as a stress concentrator and initiate the crack and defect, thus lowering the elongation at break. 3.3. Gel Content {#sec3dot3-polymers-11-01925} ---------------- The gel contents were determined in order to understand the role of the reactive agents on the blend system. The high gel contents imply the change of the undissolved macromolecules, which can be derived from the network structures or larger molecular weights of the polymers. The gel contents of neat PLA and PLA/EVA reactive blend films is summarized in [Table 3](#polymers-11-01925-t003){ref-type="table"}. The results show that the addition of Joncryl in PLA leads to an increase in the gel content from 0.33% to 2.78%. This is because Joncryl acts as a chain extender which reconnects the PLA chains \[[@B23-polymers-11-01925]\]. Consequently, the larger macromolecules and the higher molecular weight of PLA structures were formed. This effect was less when EVA was added for all content. Moreover, the gel contents increased with Perkadox and EVA contents. The higher peroxide concentrations could promote higher degree of the cross-links. The gel residues from the gel content experiments were subjected to the FTIR spectroscopy characterization. The FTIR spectra of gels obtained from neat PLA, PLA/EVA without reactive agents and neat EVA are showed in [Figure 4](#polymers-11-01925-f004){ref-type="fig"}. The peak signal at 1759 cm^−1^ refers to the carbonyl group (C=O) of the ester group of PLA. The peak appearing at 1087 cm^−1^ belongs to the ester bond (C--O--C). The peaks at 2978 and 2947 cm^−1^ are assigned to C--H stretching. The peaks at 1458 and 1385 cm^−1^ are assigned to the bending of --CH~3~. For EVA, the peaks at 2923 and 2852 cm^−1^ are assigned to C--H stretching. The peak at 1740 cm^−1^ is assigned to C=O stretching. The peak located at 1043 cm^−1^ refers to C--O stretching, and the peaks at 1463 and 1372 cm^−1^ correspond to the bending of --CH~3~. Furthermore, we were unable to detect peak shifting in the PLA/EVA without the reactive agents' spectrum. This could mean that there were no reactions between PLA and EVA for non-reactive blend. The comparison of the FTIR spectra between PLA/EVA without reactive agents gel and PLA97 + J0.5/P0.2 gel are shown in [Figure 5](#polymers-11-01925-f005){ref-type="fig"}. The result reveals that the FTIR spectrum of PLA97 + J0.5/P0.2 gel exhibits characteristic peaks both of neat PLA and neat EVA. In addition, it was observed that the peak at 1736 cm^−1^, which refers to the carbonyl group (C=O), is shifted to a lower wavenumber. This suggests evidence of the reaction. The reaction between PLA and EVA has been reported elsewhere \[[@B24-polymers-11-01925],[@B25-polymers-11-01925]\] and could occur through a transesterification reaction in the case of addition with catalyst \[[@B25-polymers-11-01925]\]. In this work, the reaction between PLA and EVA could arise from the free radical reaction. The free radical produced by the decomposition of Perkadox abstracted the hydrogen in both PLA and EVA, leading to free radical generation in PLA and EVA. This free radical via hydrogen abstraction led to the cross-linked structures between PLA and EVA. The possible reactions between PLA and EVA is presented in [Figure 6](#polymers-11-01925-f006){ref-type="fig"}a. In addition, the free radical reaction products could be not only PLA--EVA cross-linking but also EVA--EVA and PLA--PLA cross-linkings \[[@B24-polymers-11-01925],[@B26-polymers-11-01925]\] as shown in [Figure 6](#polymers-11-01925-f006){ref-type="fig"}b. 3.4. Tensile Properties {#sec3dot4-polymers-11-01925} ----------------------- The film samples were prepared using a blow film extrusion. In this case, the fixed amount of multifunctional epoxides was added to all samples in order to improve the blowability and the melt viscosity. The tensile properties of blend films were tested in both machine direction (MD) and transverse direction (TD). The stress-strain plots of both directions were showed in [Figure 7](#polymers-11-01925-f007){ref-type="fig"} and [Figure 8](#polymers-11-01925-f008){ref-type="fig"}, respectively. Since neat PLA film was not smooth and there were a large number of creases on the film due to its low melt strength, it was not suitable for blown film extrusion. Therefore, neat PLA film was not introduced to the test. It was obvious that EVA addition caused the film fabrication enhancement, hence, PLA/EVA blend films can be produced and tested. It can be seen in [Figure 7](#polymers-11-01925-f007){ref-type="fig"} that the elongation at break of PLA/EVA blends are comparable to PLA100 + J0.5, except the PLA97 + J0.5/P0.1 sample exhibited the highest toughness film in the machine direction (MD). Even though the elongation at break of PLA/EVA reactive blend films was insignificantly improved in MD, but more importantly, PLA/EVA reactive blend films were successfully blown. In addition, the tensile results of the blend films with the transverse direction (TD) obviously illustrated the plastic deformation for PLA/EVA blend. This indicated that the addition of EVA could improve the elongation at break of PLA. PLA93 + J0.5/P0.1 showed the highest elongation at break in the transverse direction. Young's modulus, tensile strength, and elongation at break of all blend films are summarized in [Figure 9](#polymers-11-01925-f009){ref-type="fig"}, [Figure 10](#polymers-11-01925-f010){ref-type="fig"} and [Figure 11](#polymers-11-01925-f011){ref-type="fig"}, respectively. Considering the Young's modulus of the blend films (in [Figure 9](#polymers-11-01925-f009){ref-type="fig"}), it could be seen that the Young's modulus of blend films in the MD direction were slightly higher than that of the TD direction. The addition of 5 wt % EVA did not significantly affect the Young's modulus of the blend films. With a further increase of EVA concentration, the Young's modulus slightly decreased because EVA acts as a soft and tough phase. Young's modulus of PLA95 + J0.5/P0.1 which composed of 5 wt % EVA was 2.8 GPa. It decreased to 2.6 GPa for PLA93 + J0.5/P0.1. Moreover, the addition of Perkadox with 0.1 phr could increase the Young's modulus. This could be attributed to the cross-link reaction between PLA and EVA promoted by peroxide. However, when the content of Perkadox increased from 0.1 to 0.2 phr, the Young's modulus of samples tended to decrease. Considering the tensile strength of blend films in MD compared to TD as shown in [Figure 10](#polymers-11-01925-f010){ref-type="fig"}, the tensile strength of the MD films were higher than those of the TD films because the molecules of polymers oriented along the blowing direction or machine direction. When the Perkadox concentration increased from 0.1 to 0.2 phr, the tensile strength of the blend films tended to decrease. PLA90 + J0.5/P0.1 film exhibited a tensile strength in MD of 49.1 MPa. When Perkadox was increased to 0.2 phr, PLA90 + J0.5/P0.2 film, the tensile strength decreased to 46.2 MPa. This might be caused by the higher cross-link degree and the stress concentration in the blends \[[@B27-polymers-11-01925]\]. The elongation at break of the TD films shown in [Figure 11](#polymers-11-01925-f011){ref-type="fig"} was higher than those of the MD films. This could result from the low frost line in the blown film process and cause the partial orientation of the polymer's molecules along the machine direction in blend films while the others could orient in the radial or transverse direction. Furthermore, the elongation at break of the TD films increased with EVA contents. However, when the content of EVA was increased to 5%, the increase of Perkadox content led to the lower elongation at break, in which the decrease was seen from 8.5% to 7.0% for PLA95 + J0.5/P0.1 and PLA95 + J0.5/P0.2 films in TD, respectively. This may be attributed to the higher degree of cross-links evolved during the free radical reaction and may be a result of the reaction between EVA itself, leading to the larger the formation of cross-linked EVA particles. These might, consequently, be attributed to the poor interfacial adhesion between PLA and EVA phases. 3.5. Effect of EVA and Perkadox Contents on Dynamic Mechanical Thermal Properties of PLA/EVA Blend {#sec3dot5-polymers-11-01925} -------------------------------------------------------------------------------------------------- The effect of EVA and Perkadox contents on the dynamic mechanical thermal properties of PLA/EVA blends was investigated. The dependence of the storage modulus on the temperature of neat PLA and PLA/EVA reactive blends were showed in [Figure 12](#polymers-11-01925-f012){ref-type="fig"}a. To clarify the effect of EVA contents on mechanical thermal properties of PLA/EVA blends, the PLA/EVA reactive blends with 3 and 10 wt % of EVA were chosen for comparison. From [Figure 12](#polymers-11-01925-f012){ref-type="fig"}a, it could be observed that neat PLA showed the high storage modulus in the glassy region (25--50 °C) with the storage modulus of around 2200 MPa. The storage modulus of PLA/EVA reactive blends decreased when the EVA contents increased. Further, at temperature above 60 °C, the storage moduli of PLA and blends were significantly declined. At the temperature above T~g~, PLA and EVA are in a rubbery state. In the rubbery region, blending with EVA 10 wt % showed the lower the storage modulus when compared to neat PLA and blending with EVA 3 wt %. This was attributed to the addition of soft and tough EVA phase in the blends, which enhanced the impact load transfer and led to the toughening of the blends \[[@B14-polymers-11-01925]\]. Then, the storage moduli of neat PLA and blends were increased again after 100 °C. This phenomenon was due to the nature of PLA. The increase in the storage modulus is a consequence of the cold crystallization of PLA in the test specimens during the DMTA temperature scan, which was also reported in other studies \[[@B28-polymers-11-01925],[@B29-polymers-11-01925],[@B30-polymers-11-01925]\]. In addition, upon increasing the EVA content to 10 wt %, the glass transition temperature (T~g~) of the PLA/EVA reactive blend was slightly shifted to the lower temperature (as shown in [Figure 12](#polymers-11-01925-f012){ref-type="fig"}b). This was caused by the plasticizing effect of EVA in PLA matrix when the temperature increased. The effect of the Perkadox contents on the dynamic mechanical thermal properties of PLA/EVA blend is shown in [Figure 13](#polymers-11-01925-f013){ref-type="fig"}a. The storage modulus of PLA/EVA reactive blends decreased with the increase in Perkadox content. This result is in agreement with Young's modulus values obtained from the tensile testing. [Figure 13](#polymers-11-01925-f013){ref-type="fig"}b shows a plot of damping factor against the temperature for PLA, PLA90 + J0.5/P0.1, and PLA90 + J0.5/P0.2. The damping peak height of PLA90 + J0.5/P0.1 was slightly higher than that of PLA90 + J0.5/P0.2. This could imply that Perkadox might induce the reaction between PLA and EVA phases and resulted in strong bond interaction in PLA/EVA blends \[[@B24-polymers-11-01925]\]. During the dynamic mechanical thermal testing, the interaction bonds dissociated and reformed, consuming a great deal of energy. Consequently, the loss factor increased and the damping properties improved \[[@B31-polymers-11-01925]\]. This indicated that the interaction between phases of PLA/EVA reactive blends with Perkadox 0.1 phr was better than that of Perkadox 0.2 phr. This result is in agreement with the FESEM images. 3.6. Model of Stretching and Shrinking Behavior of PLA/EVA Reactive Blend {#sec3dot6-polymers-11-01925} ------------------------------------------------------------------------- The polymer which possesses the capability to recover to its original shape after application of an external stimulus (e.g., light, heat, pH, etc.) is called a shape memory polymer (SMP). The primary mechanism of shape memory in polymer is related to the low entropy state during deformation from an original shape above the transition temperature (T~trans~). Shape recovery occurs and returns to a high entropy state when the polymer is reheated above the T~trans~. The T~trans~ can be either a glass transition temperature (T~g~) or a melting temperature (T~m~) of the polymer \[[@B32-polymers-11-01925],[@B33-polymers-11-01925]\]. In addition, the shrinkage of amorphous polymers can be also described by a viscoelastic model. The viscoelastic materials can exhibit both viscous and elastic behavior. The deformation of liquid materials is illustrated by a dashpot. The elasticity of the solid materials is represented by a spring. Therefore, the viscoelastic materials can be described by the combining spring and dashpot. The mechanical model and molecular mechanism to explain stretching and shrinking behavior of PLA/EVA with reactive agents were shown in [Figure 14](#polymers-11-01925-f014){ref-type="fig"}a. The model consisted of a spring (S) and a dashpot (D) connected in parallel. It is noted that the solid circle in molecular mechanism figures indicated the cross-linked points within the blend structure. When the film was stretched at the temperature above its T~g~, the spring and a dashpot were stretched easily because of the low viscosity of the dashpot ([Figure 14](#polymers-11-01925-f014){ref-type="fig"}b). At this temperature, the chain segments were flexible. After that, the stretched film was cooled immediately and created the increase of the dashpot viscosity resulting in the internal stress increase in the extended spring. This process causes the polymer film to memorize its original shape. The flexibility of the entire segment was limited. Furthermore, the temporary entanglement of polymer chains and the cross-link points can be used for the fixation the stretched shape. Then, the stretched film was reheated, causing the decrease of dashpot viscosity. Hence, the spring started to contract to its original track by the force from retained stress ([Figure 14](#polymers-11-01925-f014){ref-type="fig"}c). Thus, the temporary fixed polymer chains recoiled back to its original dimension. However, this model represented only the case of fully shrinking film. The mechanical model of the film where the shrinkage did not reach 100% could be described by using the modeling of polyester \[[@B34-polymers-11-01925]\]. This simple model combines a unit of a spring S~1~ and a dashpot D~1~ connected in parallel, and a unit of the spring S~2~ and the dashpot D~2~ connected in series, as shown in [Figure 14](#polymers-11-01925-f014){ref-type="fig"}d. After re-heating, the dashpot D~2~ started to relax and extend because of the contraction from the S~1~--D~1~ parallel unit. Therefore, the series unit cannot shrink to its original position, thus causing partial film shrinkage. In this part, the stretching of PLA/EVA with reactive agent films was prepared by a uniaxial stretching machine at 70 °C. The film specimens were prepared as shown in [Figure 15](#polymers-11-01925-f015){ref-type="fig"}. The a, b, c and d were the four different zones of film specimen. The percentage of film shrinkage was measured from two dimensions (the length (L) and the width (W)) based on the length difference before and after reheat. The average percentage of the shrinkage of PLA/EVA blend films at different stretching ratios of 2 and 3 for both MD and TD directions are shown in [Table 4](#polymers-11-01925-t004){ref-type="table"} and [Table 5](#polymers-11-01925-t005){ref-type="table"}. The results revealed that the percentage of film shrinkage in the MD films was comparable to the TD films. For the films stretched with stretching ratios of 2 and 3, the shrinkage percentage reached 100%. However, with a stretching ratio of 4, the shrinkage of the film tended to decrease. Considering the effect of EVA contents on the shrinkage behavior, it was found that when EVA contents was increased in blend from 3% to 5% by weight (at a fixed content of reactive agents), those percentages of shrinkage were comparable. Therefore, to clarify the effect of EVA content on the percentage of shrinkage, EVA amounts of 3 and 10 wt % were chosen for the sake of the comparison. When EVA contents increased to 10 wt %, the percentage of shrinkage was higher when compared to the PLA blend with EVA 3 wt %. This result was similar for the stretching ratios of 2 and 3 (as shown in [Table 4](#polymers-11-01925-t004){ref-type="table"} and [Table 5](#polymers-11-01925-t005){ref-type="table"}), except at a stretching ratio of 4. The stretching ratio of 4 led to a strain greater than the yield point of the polymer blend films, hence, it could not return to its original length. From the mechanical model as explained previously, springs were used to describe the reversible deformation of the materials. Hence, the spring modulus could represent the material's modulus. From the tensile results, it could be seen that the addition of EVA in PLA led to a lowering in the Young's modulus of the blends. This could mean that EVA reduced the spring modulus of the blends; therefore, the PLA/EVA film blend easily deformed and returned to its original length. Moreover, the amount of shrinkage depends on the concentration and orientation of the oriented and disoriented amorphous phases which are the predominant mechanisms \[[@B32-polymers-11-01925],[@B35-polymers-11-01925]\]. The increase of EVA content gives the increase of the amorphous phase. Therefore, the percentage of the shrinkage of films increased with EVA content. Furthermore, the increases of Perkadox content caused the percentage of shrinkage to increase for both of MD films and TD films with EVA content of 3, 5, and 7 wt %. The degree of shrinkage was controlled by the concentration and orientation of the amorphous phase \[[@B32-polymers-11-01925]\] as well as the memory points depending upon cross-links \[[@B36-polymers-11-01925],[@B37-polymers-11-01925]\]. Cross-link points help in the shrinking process by serving as a memory in the stretched sample. They revert to their original positions during shrinkage \[[@B38-polymers-11-01925]\]. Consequently, a higher cross-linking reaction between the polymer molecules was of great advantage to the shrinking mechanism. However, the blend film with 10 wt % of EVA exhibited the decreases of percentage of films shrinkage when the Perkadox content increased. The highest of percentage of film shrinkage was observed in PLA90 + J0.5/P0.1. In the case of the stretching ratio of 4, the increase in the percentage of shrinkage with Perkadox content were observed in blends at EVA amount of 3 and 5 wt %. Then it decreased with an increase of EVA higher than 5 wt %. The low percentage of shrinkage could be observed from the formula that exhibited the high gel content (2.72% for PLA90 + J0.5/P0.2). The increase of EVA content led to the increase of gel content. Due to the greater number of tertiary type of carbons, the chance of cross-linking reaction increases \[[@B39-polymers-11-01925]\]. This might be due to the presence of gel in the film which could obstruct the shrinkage of the film. When comparing the percentage of shrinkage in the length and the width directions of the same specimen, it could be seen that the shrinkage in the L direction was higher than the W direction. This is because, when the films were stretched along the L direction, the polymer chains of the films oriented along the stretched direction and had a better return to the original random coil state in the L direction than the W direction while reheating. In addition, the shrinkage of the film was obviously observed at a and d positions because they were the edge of the specimen film. Consequently, they were easier to shrink than the middle position. 3.7. Morphology of Stretched and Shrunk of PLA/EVA Reactive Blend Films {#sec3dot7-polymers-11-01925} ----------------------------------------------------------------------- The morphology of the cross-section of stretched and shrunk PLA/EVA reactive blend films in TD and MD are illustrated in [Figure 16](#polymers-11-01925-f016){ref-type="fig"} and [Figure 17](#polymers-11-01925-f017){ref-type="fig"}, respectively. It can be seen in [Figure 16](#polymers-11-01925-f016){ref-type="fig"} that TD films displayed a stack morphology. This characteristic is from the perpendicularly film preparation direction when film was blown. In addition, the PLA90 + J0.5/P0.1 stretched film ([Figure 16](#polymers-11-01925-f016){ref-type="fig"}a) showed larger gap between layers than the PLA90 + J0.5/P0.2 stretched film ([Figure 16](#polymers-11-01925-f016){ref-type="fig"}b). However, the PLA90 + J0.5/P0.1 film exhibited a better stacking pack than the PLA90 + J0.5/P0.2 film after films shrinking. The shrink film morphology of PLA90 + J0.5/P0.2 sample also showed some restraint between layers that might cause the smaller percentage of shrinkage than the PLA90 + J0.5/P0.1 sample. The morphology of MD sections ([Figure 17](#polymers-11-01925-f017){ref-type="fig"}) revealed the occurrence of pinholes in the PLA90 + J0.5/P0.2 film. The holes were larger and still located in the film after shrinking. Moreover, the surface roughness was roughly greater in the PLA90 + J0.5/P0.2 sample than in the PLA90 + J0.5/P0.1 sample. 4. Conclusions {#sec4-polymers-11-01925} ============== The brittleness of PLA was improved when EVA was introduced. The addition of EVA led to an increase of blown film extrusion capability of PLA. In addition, the PLA/EVA film with 0.1 phr of Perkadox was observed as the optimal compatibilizer concentration which led to the maximum toughness at PLA/EVA 93/7 and highest modulus at 95/5. The high loading of peroxide could cause the high cross-link degree and phase separation. The percentage of shrinkage of PLA/EVA reactive blends films were 100%, when the films were stretched two and three times of their original length. However, the shrink recovery was less than 100% when the films were stretched up to four times of its original length. The heat-shrinkable behavior of the PLA/EVA reactive blend could be described by the shape memory effect. It can be illustrated as the spring and dashpot connected in parallel regarding to the fully shrinkage film. However, the film where the shrinkage had not reached 100% could be demonstrated by the addition of series connections between the spring and dashpot to the parallel spring-dashpot unit. The authors would like to thank the Department of Materials Engineering, Faculty of Engineering, Rajamangala University of Technology Rattanakosin, Institute of Research and Development Rajamangala University of Technology Rattanakosin for funding support. The authors gratefully acknowledge the Department of Materials Science and Engineering, Faculty of Engineering and Industrial Technology Silpakorn University, as well as Department of Physics, Faculty of Science Silpakorn University. The authors also would like to acknowledge Total Corbion PLA (Thailand) Ltd. for PLA supply. Conceptualization: S.S.; data curation: R.K.; formal analysis: R.K., S.S. and M.S.; funding acquisition: S.S.; investigation: R.K., T.P., P.P., and R.B.; methodology: R.K. and S.S.; supervision: R.K. and S.S.; writing---original draft: R.K.; writing---review and editing, R.K. and S.S. This research was funded by Biodiversity-Based Economy Development Office (Public Organization) (BEDO) and National Research Council of Thailand (NRCT) fiscal year 2015. The authors declare no conflict of interest. ![Morphology of (a) PLA95 and (b) PLA95 + J0.5.](polymers-11-01925-g001){#polymers-11-01925-f001} ![Morphology of (a) PLA90 + J0.5/P0.1 and (b) PLA90 + J0.5/P0.2.](polymers-11-01925-g002){#polymers-11-01925-f002} ![Fracture surface of PLA90+J0.5/P0.2.](polymers-11-01925-g003){#polymers-11-01925-f003} ![FTIR spectra of gel of neat PLA, PLA/EVA, and neat EVA.](polymers-11-01925-g004){#polymers-11-01925-f004} ![FTIR spectra of gel of PLA/EVA compared to gel of PLA97 + J0.5/P0.2.](polymers-11-01925-g005){#polymers-11-01925-f005} ![(a) Possible reaction between PLA and EVA and (b) the reaction scheme for PLA/EVA reactive blend.](polymers-11-01925-g006){#polymers-11-01925-f006} ![Stress-strain curves of PLA/EVA blend film in machine direction (MD).](polymers-11-01925-g007){#polymers-11-01925-f007} ![Stress-strain curves of PLA/EVA blend film in transverse direction (TD).](polymers-11-01925-g008){#polymers-11-01925-f008} ![Young's modulus of PLA/EVA blends film in MD and TD direction.](polymers-11-01925-g009){#polymers-11-01925-f009} ![Tensile strength of PLA/EVA blends film in MD and TD direction.](polymers-11-01925-g010){#polymers-11-01925-f010} ![Elongation at break of PLA/EVA blends film in MD and TD direction.](polymers-11-01925-g011){#polymers-11-01925-f011} ![The effect of EVA content on dynamic mechanical thermal properties of PLA/EVA blend. (a) Storage modulus and (b) damping factor with an enlarged inset at T~g~.](polymers-11-01925-g012){#polymers-11-01925-f012} ![The effect of Perkadox content on dynamic mechanical thermal properties of PLA/EVA blend. (a) Storage modulus and (b) damping factor with an enlarged inset at T~g~.](polymers-11-01925-g013){#polymers-11-01925-f013} ![The proposed mechanical model with molecular mechanism for stretching and shrinking behavior of PLA/EVA reactive blend. The red line, blue line and black dot were PLA chains, EVA chains and crosslink point between PLA and EVA chains, respectively. (a) The proposed mechanical model; (b) heating-stretching-cooling; (c) reheating and shrinking; and (d) the partial shrinkage mechanical model.](polymers-11-01925-g014){#polymers-11-01925-f014} ![Film specimen for heat shrinkage testing and stretching direction.](polymers-11-01925-g015){#polymers-11-01925-f015} ![The morphology of the cross-section in TD of (a) stretched PLA90 + J0.5/P0.1 film, (b) stretched PLA90 + J0.5/P0.2 film, (c) shrunk PLA90 + J0.5/P0.1 film, and (d) shrunk PLA90 + J0.5/P0.1 film.](polymers-11-01925-g016){#polymers-11-01925-f016} ![The morphology of the cross-section in MD of (a) stretched PLA90 + J0.5/P0.1 film, (b) stretched PLA90 + J0.5/P0.2 film, (c) shrunk PLA90 + J0.5/P0.1 film, and (d) shrunk PLA90 + J0.5/P0.1 film.](polymers-11-01925-g017){#polymers-11-01925-f017} polymers-11-01925-t001_Table 1 ###### Sample formulations and blend compositions. Formula Compositions ---- ------------------- -------------- ---- ----- ----- 1 Neat PLA 100 \- \- \- 2 PLA100 + J0.5 100 \- 0.5 \- 3 PLA97 + J0.5/P0.1 97 3 0.5 0.1 4 PLA97 + J0.5/P0.2 97 3 0.5 0.2 5 PLA95 95 5 \- \- 6 PLA95 + J0.5 95 5 0.5 \- 7 PLA95 + J0.5/P0.1 95 5 0.5 0.1 8 PLA95 + J0.5/P0.2 95 5 0.5 0.2 9 PLA93 + J0.5/P0.1 93 7 0.5 0.1 10 PLA93 + J0.5/P0.2 93 7 0.5 0.2 11 PLA90 + J0.5/P0.1 90 10 0.5 0.1 12 PLA90 + J0.5/P0.2 90 10 0.5 0.2 polymers-11-01925-t002_Table 2 ###### Melt flow index (MFI) of samples. Sample MFI (g/10min) ------------------- --------------- Neat PLA 18.7 ± 1.5 PLA100 + J0.5 5.0 ± 0.1 PLA97 + J0.5/P0.1 3.9 ± 0.3 PLA97 + J0.5/P0.2 3.9 ± 0.2 PLA95 7.7 ± 0.5 PLA95 + J0.5 3.3 ± 0.3 PLA95 + J0.5/P0.1 6.9 ± 0.3 PLA95 + J0.5/P0.2 7.2 ± 0.3 PLA93 + J0.5/P0.1 5.5 ± 0.4 PLA93 + J0.5/P0.2 5.1 ± 0.1 PLA90 + J0.5/P0.1 2.6 ± 0.3 PLA90 + J0.5/P0.2 3.6 ± 0.3 polymers-11-01925-t003_Table 3 ###### Gel content of PLA/EVA reactive blend film. Formula Gel Content (%) ------------------- ----------------- Neat PLA 0.33 ± 0.02 PLA100 + J0.5 2.78 ± 0.54 PLA97 + J0.5/P0.1 0.10 ± 0.14 PLA97 + J0.5/P0.2 0.47 ± 0.01 PLA95 + J0.5 2.30 ± 0.03 PLA95 + J0.5/P0.1 0.49 ± 0.04 PLA95 + J0.5/P0.2 1.23 ± 0.09 PLA93 + J0.5/P0.1 0.55 ± 0.13 PLA93 + J0.5/P0.2 1.78 ± 0.52 PLA90 + J0.5/P0.1 0.62 ± 0.44 PLA90 + J0.5/P0.2 2.72 ± 0.20 polymers-11-01925-t004_Table 4 ###### The average percentage of shrinkage of film at a stretching ratio of 2 in MD and TD. Direction Position Shrinkage (%) ----------- ------------- --------------- -------------- -------------- -------------- ------------- MD a W 90.0 ± 23.2 93.3 ± 14.9 106.7 ± 18.3 100.0 ± 0.2 L 105.4 ± 6.3 92.0 ±8.4 96.7 ± 7.5 95.1 ± 7.4 b W 92.7 ± 23.6 95.0 ± 11.2 107.0 ± 18.0 100.0 ± 0.1 L 99.0 ± 0.1 93.8 ± 5.7 100.0 ± 0.1 93.2 ± 3.9 c W 93.0 ± 1.8 100.0 ± 0.1 106.7 ± 18.3 100.0 ± 0.1 L 97.0 ± 10.9 98.0 ± 11.5 100.0 ± 0.1 96.5 ± 4.8 d W 100.0 ± 14.9 80.0 ± 44.7 107.0 ± 18.0 95.0 ± 11.2 L 99.2 ± 6.7 98.3 ± 11.9 100.0 ±0.2 100.0 ± 9.1 TD a W 100.0 ± 13.8 56.0 ± 33.4 92.7 ± 1.9 81.3 ± 27.2 L 100.0 ± 7.1 89.9 ±10.2 87.9 ± 3.5 86.7 ±10.9 b W 100.0± 14.9 51.7 ± 30.3 92.7 ± 1.5 75.3 ± 15.2 L 100.0 ± 7.7 89.3 ± 7.9 89.3 ± 0.6 78.4 ± 26.2 c W 100.0 ± 14.2 45.0 ± 27.4 92.0 ± 1.8 80.3 ±18.7 L 100.0 ± 9.0 89.3 ± 7.9 89.3 ± 0.8 81.8 ± 19.1 d W 98.3 ± 19.1 62.0 ±17.2 92.7 ± 1.8 76.7 ± 13.7 L 106.9 ± 4.4 84.7± 12.8 90.2 ±5.4 89.9 ± 11.5 polymers-11-01925-t005_Table 5 ###### The average percentage of shrinkage of film at a stretching ratio of 3 in MD and TD. Direction Position Shrinkage (%) ----------- ------------- --------------- ------------- -------------- ------------- ------------- MD a W 63.3 ± 32.4 88.0 ± 26.8 92.9 ± 18.6 80.0 ± 28.3 L 104.3 ± 6.1 87.5 ± 22.7 97.9 ± 2.8 80.1 ± 28.1 b W 82.3 ± 24.6 87.0 ± 18.6 108.7 ± 25.9 80.0 ± 24.5 L 104.4 ± 6.1 83.5 ± 24.2 100.0 ± 0.1 83.7 ± 18.6 c W 79.1 ± 33.9 66.0 ± 29.0 117.0 ± 23.1 84.0 ± 26.1 L 86.2 ± 4.2 67.9 ±23.2 82.5 ± 2.1 72.7 ± 13.7 d W 70.5 ± 39.1 84.0 ± 26.1 109.1 ± 18.6 79.0 ± 14.3 L 103.4 ± 5.6 85.4 ± 24.3 99.0 ± 2.1 90.2 ± 10.2 TD a W 77.3 ± 24.4 62.0 ± 24.6 96.9 ± 3.0 80.0 ± 34.6 L 90.4 ± 23.0 77.3 ± 17.0 95.1 ± 0.1 84.1 ± 30.3 b W 91.2 ± 32.2 83.0 ± 9.7 97.4 ± 2.9 75.0 ± 32.8 L 92.8 ±17.1 91.0 ± 7.4 95.0 ± 0.2 82.1 ± 35.0 c W 82.4 ± 40.1 79.0 ± 2.2 97.4 ± 2.4 71.0 ± 30.1 L 73.4 ± 43.9 77.6 ± 38.1 80.0 ± 2.2 69.2 ± 35.3 d W 91.3 ± 34.3 78.0 ± 2.7 97.6 ± 2.2 75.0 ± 32.8 L 89.1 ± 34.1 94.0 ± 4.2 95.9 ±4.4 82.2 ± 40.2	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#SEC1} ============ Pathway diagrams are the roadmaps for molecular biology. Just as roadmaps show the connections between villages, towns, and cities, pathway diagrams illustrate the connections between genes, proteins and metabolites. A well-illustrated pathway diagram provides biological context to complex molecular processes in an easily understood and highly visual manner. In this regard, pathway diagrams provide a remarkably useful means for scientists to share, integrate, interpret and visualize 'omics data and 'omics measurements. Even now, compilations of pathway diagrams (i.e. pathway databases) are among the most frequently used and oft-referenced resources in the entire field of bioinformatics. This is because pathway databases are widely called upon to interpret genomics, metagenomics, transcriptomics, proteomics and/or metabolomics data. Furthermore, they are frequently used to facilitate more complex analytical tasks such as cellular modeling ([@B1]), flux balance analysis ([@B2]), gene-set enrichment analysis ([@B3]), metabolite-set enrichment analysis ([@B4]) and over-representation analysis ([@B5]). A wide variety of pathway databases exist, including both commercial and open-access resources. Some of the best-known online pathway databases are also among the oldest. These include KEGG ([@B6]) and the 'Cyc' databases ([@B7]), both of which emerged in the mid 1990s. Both of these resources are superb collections, providing extensive, easily searched, web-accessible pathways and annotations that are widely used by thousands of scientists each day. More recently, a number of other pathway databases have emerged such as Reactome ([@B8]) and Wikipathways ([@B9]), which cover pathways for a selected (small) number of model organisms. Additionally, Wikipathways employs a community annotation model and exploits a number of continuing advances in web visualization and interactivity to offer a somewhat richer user/contributor experience. There are also a variety of specialized pathway databases such as BioCarta (<https://cgap.nci.nih.gov/Pathways/BioCarta_Pathways>) ([@B10]), InnateDB ([@B11]), PharmGKB ([@B12]) and the Pathway Interaction Database ([@B13]), which either focus on specific pathway types, specific (single) organisms or provide a different, often visually more impressive way of exploring pathways than their comprehensive pathway database cousins. In addition to these freely available or open-access pathway databases, there are also several well-known commercial pathway databases such as GeneGo (<https://portal.genego.com/>), Protein Lounge (<http://www.proteinlounge.com/>) or IPA (Ingenuity Pathway Analysis) (<https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/>). In particular, the Ingenuity Pathway Analysis system by Qiagen is perhaps the most popular and most extensive commercial resource in terms of pathway coverage and pathway analysis tools. While there are many excellent features in each of the pathway databases just mentioned, there are also some notable limitations as well. For instance, even though KEGG and the Cyc databases are explicitly small molecule pathway databases, they do not provide many pathways covering lipid synthesis, metabolite signaling, small molecule hormone signaling or small molecule drug action. While the KEGG database does provide some protein signaling, developmental and disease process pathways, the CycDBs do not. Neither KEGG nor the CycDBs provide pathways relating to cellular responses, and KEGG does not provide information on cellular locations or biological context. Being the oldest of today\'s pathway databases, KEGG and the Cyc databases also use somewhat dated approaches to web-based pathway visualization. While the Reactome and Wikipathways databases do provide somewhat more modern web interfaces and cover many aspects of protein and cellular signaling, they are also much more limited in their coverage of metabolism than KEGG or the Cyc databases. Likewise, Reactome and Wikipathways largely ignore metabolite signaling, hormone signaling, drug action and many common disease pathways. Even with their modern pathway visualization tools, the Reactome and Wikipathway pathway diagrams still lack the visual appeal or biological context of the pathway diagrams found in other pathway databases such as BioCarta ([@B10]), DrugBank ([@B14]) or SMPDB ([@B15]). Ideally what is needed is a pathway database that combines the comprehensive metabolic pathway coverage found in the KEGG or the Cyc databases, with the rich protein/cellular signaling pathway coverage found in Reactome, Wikipathways or IPA, and the visualization quality and biological context found in BioCarta or SMPDB---into a single, freely available resource. Herein we describe just such a database---called PathBank. PathBank is a comprehensive, visually rich pathway database containing more than 110 000 machine-readable pathways found in 10 model organisms (including humans, rodents, plants, insects and microbes). PathBank provides detailed, fully searchable, hyperlinked diagrams of metabolic, protein/metabolite signaling, disease, drug and physiological pathways. Its pathway diagrams include information on the relevant organs, organelles, subcellular compartments, locations and structures to provide important biological context. All molecular/protein images are hyperlinked to databases with rich descriptions and all pathways are viewable using a modern 'Google Maps'-like or AJAX-based visual navigation system. All of PathBank\'s pathways have detailed descriptions and all of its images, image maps, descriptions and reactions are freely downloadable in several common data exchange formats. The entire database may be browsed, searched or annotated using a variety of easy-to-use tools. Likewise, community or crowd-sourced annotations and submissions to PathBank are supported through the web-based PathWhiz ([@B16]) pathway illustrator. PathBank DESCRIPTION AND CONTENT {#SEC2} ================================ PathBank (version 1.0) contains 110 234 richly annotated, fully colored, machine-readable pathways for 10 model organisms (Homo sapiens, Bos taurus, Rattus norvegicus, Mus musculus, Drosophila melanogaster, Caenorhabditis elegans, Arabidopsis thaliana, Saccharomyces cerevisiae, Escherichia coli and Pseudomonas aeruginosa). The database currently contains 78 488 different compounds (including metabolites, drugs and other xenobiotics), 8993 different proteins and describes more than 176 535 different reactions and interactions. PathBank pathways are divided into two broad categories or classes: (i) Metabolite/Compound pathways and (ii) Protein pathways. Metabolite/Compound pathways contain a majority (\>50%) of small molecules compared to proteins as pathway entities. Protein pathways contain a vast majority of proteins (typically \>80%) compared to metabolites as pathway entities. In the current version of PathBank there are 109 836 Metabolite/Compound pathways and 398 Protein pathways. Within PathBank there are six subcategories of Metabolite/Compound pathways: (i) Metabolic (catabolism/anabolism); (ii) Physiological/Endocrinological (primarily involving metabolites); (iii) Metabolite Signaling; (iv) Drug Metabolism; (v) Drug Action; and (vi) Disease (primarily involving small molecule metabolism). In contrast to the Metabolite/Compound pathways there are 15 subcategories of Protein pathways: (i) Immunological; (ii) Cellular Response; (iii) Gene Regulatory; (iv) Growth Factor; (v) Cytokine Signaling; (vi) Protein/Peptide Hormone-Mediated; (vii) Neurological Signaling; (viii) Developmental Signaling; (ix) Kinase Signaling; (x) Apoptosis Signaling; (xi) Stress Activated Signaling; (xii) Pathogen-Activated Signaling; (xiii) Transport/Degradation; (xiv) Cytoskeletal Signaling; and (xv) Disease (primarily involving protein dysregulation). All pathways have detailed text descriptions that provide summaries of the pathway or the processes being depicted. Averaging a length of 342 words, these original descriptions highlight relevant background information, biological context and key reactions. Similarly, all compounds in PathBank have detailed text descriptions averaging 305 words. Further, all compounds have manually curated abbreviations for use in the easier-to-read large font pathway visualizations. Most of these compound names, abbreviations and descriptions were obtained from our own locally maintained databases such as HMDB ([@B17]), DrugBank ([@B14]), YMDB ([@B18]) and others. All protein names and descriptions are obtained from, and linked to, UniProt. Links and citation data for more than 12 958 different references (textbooks and journals) used to help create PathBank\'s pathways are also provided. PathBank supports a wide variety of searching and browsing functions, including pathway browsing/searching (by name), compound browsing/searching (by name), protein browsing/searching (by name), sequence searching, compound similarity searching, molecular weight searching and advanced (Boolean) text searching. Users may also search PathBank by using lists of compound names, protein names and/or gene names or their corresponding identifiers (HMDB ID, Affymetrix ID, GenBank ID, UniProt ID, etc.). All of the pathways, proteins, and metabolites in PathBank are hyperlinked to other resources and all of the pathways are navigable through a modern visualization interface modeled after Google Maps (<https://www.google.com/maps>). PathBank also has a variety of tools to facilitate the display of specific pathway components and to map their relative concentrations. Similarly, different types of pathway displays are possible to facilitate printing or the preparation of slides and/or images. All of PathBank\'s pathway images, image maps, descriptions and reactions are freely downloadable in several common machine readable or data exchange formats including BioPAX, SBML, PWML, SBGN, RXN, PNG and SVG. PathBank was implemented using a Ruby on Rails (<http://rubyonrails.org>, version 4.2.0) web framework incorporating a MySQL relational database (<https://www.mysql.com>, version 5.1.50) to manage all of the pathway data, including entity relationships, external references, descriptions, visualization specifications and chemical structures. PathBank uses the model--view--controller architecture, in which internal data logic is separated from user input and data presentation. The raw information stored in the database is dynamically extracted and rendered into web pages by PathBank\'s HTML interface responder. PathBank is hosted on a Digital Ocean server equipped with 4 CPUs, 50 GB of disk space, 8 GB of RAM and an additional 200 GB storage space on an Amazon S3 storage facility. PathBank offers a number of features not found in other pathway databases. These features are highlighted in Table [1](#tbl1){ref-type="table"}. In terms of the number of pathways per organism, PathBank offers the largest and possibly the most comprehensive collection of any public database. In terms of the total number of pathways, it is second only to KEGG (which has 653 007 pathways from 6076 organisms compared to the 10 organisms in PathBank). As shown in Table [1](#tbl1){ref-type="table"}, PathBank also appears to provide much more visual detail, more biological context and more pathway-related data (text descriptions) than almost any other pathway database. In terms of biological context and detail, PathBank pathways illustrate organs, tissues, cells and organelles as well as physiological consequences (especially for drug action pathways). In terms of pathway-related data, PathBank provides almost twice as much text for its pathway descriptions as any other database, significantly more references than most other databases as well as uniquely providing both protein and metabolite descriptions (see Table [1](#tbl1){ref-type="table"}). PathBank is particularly unique in its collection of lipid synthesis, disease, drug action, drug metabolism and metabolite signaling pathways. A key objective of PathBank is to provide pathway resources for the metabolomics and lipidomics communities. Technology now exists to detect and quantify thousands of individual lipids. Some of these lipids are unique to specific diseases, conditions, tissues or organisms. However, pathways describing the biosynthesis or degradation of these unique lipids are generally lacking in all current pathway databases (except PathBank). Indeed, because of these lipid pathways (as well as PathBank\'s many disease and drug action pathways) we estimate that \>95% of the pathways found in PathBank are not found in any other pathway database. For instance, if pathway name similarity mapping is used to compare database overlap, \<3% of PathBank\'s pathways are found in KEGG, \<2% are found in Reactome, \<1% are found in Wikipathways and \<1% are found in BioCarta. Alternately, if one uses the average number of pathways/organism in each database and if one assumes all pathways in other databases are mostly covered in PathBank (a reasonable assumption) then only 1% of PathBank\'s pathways are found in KEGG, 11% are found in Reactome, 1% are found in Wikipathways and 3% are found in BioCarta. ###### A comparison between PathBank and other pathway databases PathBank KEGG Reactome Wiki-Pathways BioCarta ----------------------------------- ---------- --------- ---------------- --------------- ---------- No. of organisms 10 6076 16 31 2 No. of pathways 110 234 653 007 20 882 2785 590 Ave. no. pathways/organism 1 1 023 107 1305 90 295 Protein pathways? Yes Yes Yes Yes Yes Metabolic pathways? Yes Yes Yes Yes No No. of pathways (Human) 48 701 335 2227 1,102 314 No. of pathways (E. coli) 1178 119 0 12 0 No. of pathways (A. thaliana) 3356 138 0 93 0 No. of pathways (Mouse) 12 651 331 1649 225 276 Hyperlinked proteins/genes Yes Yes Yes Yes No Hyperlinked metabolites Yes Yes Yes Yes No No. of disease pathways 20 692 91 411 118 5 No. of signaling pathways 64 65 718 211 39 No. of drug action pathways 404 75 46 39 0 Compound descriptions Yes No No No No Protein descriptions Yes No No No No Pathway summaries Yes Yes Yes Yes No Ave. word count per summary 342 68 179 143 0 Ave. ref. count per pathway 12 14 3 8 0 Rich visual display Yes No Yes No Yes Displays protein 4° structure Yes No Yes No No Displays metabolite structure Yes Yes Yes No No Displays cellular organelles Yes No Yes Yes Yes Displays cellular locations Yes No Yes Yes Yes Displays organs/tissues Yes No Yes (in EHLDs) No No Multicolored pathways Yes No Yes Yes Yes Multiple pathway views Yes No No No No No. pathway image formats 2 1 6 7 1 XML compatible Yes KGML Yes Yes No SBML compatible Yes No Yes Yes No BioPAX compatible Yes No Yes Yes No 'Google Maps' interactivity Yes No Yes No No Community editable Yes No No Yes No Open access/Free download Yes No Yes Yes Yes Integrated with other DBs Yes Yes Yes Yes No Integrated with other Apps Yes Yes Yes Yes No PathBank\'s collection consists of a large number of lipid synthesis pathways. A major goal of PathBank is to have a pathway for every protein and a map for every metabolite. Lipids constitute more than 70% of the nameable metabolite entities in most organisms but unfortunately there is a tendency for many biochemists and pathway databases to ignore lipids or to treat them as generic molecules. From the perspective of classical metabolism, the use of generic classes is helpful. However, from the perspective of metabolomics and lipidomics, this approach is very limiting. Indeed, the primary data outputs of lipidomics and metabolomics experiments are individual compound identities and concentrations for hundreds to thousands of unique (but often similar) molecules. Converting these data into 'generics' or compound classes leads to a significant loss of information. Furthermore, recent work in lipidomics is highlighting how important individual lipid molecules are for certain biological functions or disease processes ([@B19]). This is why there has been a strong push within the metabolomics and lipidomics community to develop a database like PathBank where the goal is to have 'a pathway for every molecular entity'. PathBank LAYOUT AND NAVIGATION {#SEC3} ============================== A screenshot montage of the PathBank interface and its search and browsing tools is shown in Figure [1](#F1){ref-type="fig"}. PathBank has a landing/home page very similar to that of its smaller cousin, SMPDB ([@B15]). This includes a short textual description of the database, a circulating 'carousel' of selected PathBank images and a series of tabs at the top of the page for navigation. An empty search box is located at the top left of each PathBank page. PathBank has five major tabs: (i) Browse; (ii) Search; (iii) About; (iv) Downloads; and (v) Contact Us. Under the Browse tab users may browse the website using four different pages: a) Pathways; b) Table of Primary Pathways (TOPP); c) Compounds (metabolites or drugs); or d) Proteins. All pathway search and browse results are filterable by species (10 model species available) and by pathway type (Metabolite/Compound and Protein) using two dropdown filters. Additionally, selecting a pathway type will render visible a third dropdown menu to allow filtering of the results by pathway sub-categories (6 Metabolite/Compound pathway sub-categories and 15 Protein pathway sub-categories). ![A screenshot montage of different browsing and searching screens taken from PathBank. A more detailed description of the different functions and capabilities of the various browse and search tools in PathBank is given in the text.](gkz861fig1){#F1} If users choose to browse by Pathway, a multi-page table is presented that contains four columns: (i) the PathBank identifier (with the thumbnail image); (ii) the pathway details, which includes the pathway name, species, description and download links; (iii) the compounds (which are hyperlinked); and (iv) the proteins (which are also hyperlinked). Clicking on the thumbnail image of a given pathway will direct users to the full-size pathway image. Likewise, clicking on the metabolite/compound names (column iii) or protein names (column iv) will open a page with detailed compound descriptions (from the appropriate organism-specific metabolome database such as HMDB or from a specialized organism-specific database) or protein descriptions (from UniProt). If the TOPP option is selected, a multi-page table is presented showing three columns: (i) the PathBank identifier; (ii) the pathway name including the species name (if filtered by all species); and (iii) the link to the pathway view. In the TOPP, similarly named (replicated) pathways are grouped together to facilitate browsing of largely unique pathways. Groupings are represented by a single primary pathway, hence the table name, which has a clickable pathway name displayed in blue text. These hyperlinked pathways may be selected by the user to navigate to a popover view containing all of the related pathways which can be similarly browsed. In the case of lipid pathways, only the generic version of the pathway is listed in the main table; the pathways for the specific lipids are browsable in the generic lipid\'s popover view. Users can browse PathBank\'s compounds and proteins in much the same manner. Multi-page tables present compounds and proteins alphabetically in three columns: (i) the compound/protein\'s identifier along with the corresponding View buttons; (ii) the compound/protein\'s name (with a short description); and (iii) the associated pathways (which are hyperlinked). Clicking on a compound\'s View button will take a user to the appropriate 'Compound Card', corresponding to the database most appropriate for the chosen (or filtered) organism. The default is HMDB if no organism has been selected. Clicking on a protein\'s corresponding View button will take a user to the appropriate 'UniProt Card'. Unlike the pathway browse results, compounds and proteins are filterable only by species using the dropdown menu provided. An additional filter in the form of a Search box at the top right of the table allows users to filter the page by compound, protein or pathway name. PathBank places a strong emphasis on providing users with high quality, artistically pleasing pathway diagrams that are not only correct and informative but also colorful, interactive and richly detailed. All of the pathway diagrams in PathBank use scalable vector graphics (SVG) and a web interface technology inspired by Google Maps. This allows rapid and continuous zooming using a mouse pad or a mouse scroll wheel or through simply clicking on-screen zoom icons. It also allows facile navigation around zoomed-in pathway diagrams through a simple click-and-drag operation or through clicking on-screen up/down or left/right arrows located near the on-screen zoom functions. A full-screen view of each PathBank pathway diagram is also available, which can be toggled off and on by clicking the full-screen icon located between the navigation arrows. On the right side of each pathway diagram is a pathway display panel with five tabs (Description, Highlight, Analyze, Downloads, Settings/Display). The default view is the Description panel, which describes the pathway in detail and provides one or more literature references. The Highlight panel (viewed by clicking the 'Highlight' tab) allows users to select and color different proteins and metabolites for display purposes. The Analyze panel (viewed by clicking the 'Analyze' tab) allows users to enter concentration (relative or absolute) data on proteins/transcripts and/or metabolites and to have these colored on the pathway diagram. It is highly recommended that users switch to the black and white version of the pathway when using the highlight and analyze tools in order to clearly see the elements being colored. The download panel (viewable by clicking the 'Download' tab) allows users to select the type of file format that they wish to have their pathway image (annotated or untouched) saved as. Options are available for downloading images in BioPAX ([@B22]) image format (full color SVG, grayscale SVG, simple SVG, large font SVG, and simple large font SVG), BioPAX only format, SBGN ([@B23]), SBML ([@B24]), PWML (a custom PathWhiz format), PNG or all of the above. At last, the Display panel (viewed by clicking the 'Display' tab) allows users to change the pathway display to suit a particular taste or application. Users may toggle between simple (thick line) and complex (individual lipid) membranes; between a dark blue (aqueous) background or a white background; between the full color, highly detailed version and the simplified KEGG-like (black and white) representation; and finally between full-names of pathway components and their large font abbreviations. There is also an additional fifth option to render the simplified (KEGG-like) pathway representation using the easier-to-read large font abbreviations. In addition to these browsing functions, PathBank also offers extensive search functions. These include a general text search (available at the top of every PathBank web page) as well as several, more specific, search functions listed under the 'Search' tab. The text search supports Boolean logic (AND, OR and NOT operations) along with field-specific searches covering compound names, protein names, compound/protein identifiers, pathway names, and pathway descriptions. Instructions for the text search are accessible using the Search dropdown list. There are four additional searches available: (i) the Path-MAP Advanced Search; (ii) the ChemQuery Structure Search; (iii) the Molecular Weight Search; and (iv) the Sequence Search. The Path-MAP search allows users to enter long lists of compound names, protein names, compound identifiers, protein identifiers or gene identifiers to search for organism-specific pathways enriched with these entities. The result is a list of pathways with enrichment scores calculated using the frequency of matches and the number of pathway entities as scaled using a hypergeometric function. It is also possible to enter similar lists with Path-MAP but with concentrations or relative concentrations, as might be obtained from a typical proteomics, transcriptomics or metabolomics experiment. The result of this type of search is a pathway annotated and colored with the corresponding metabolite/protein concentrations according to a yellow (low)-red (high) concentration gradient. Other searches such as the ChemQuery and Molecular Weight searches are intended for metabolite or compound searching against PathBank\'s chemical database of 78 271 compounds and are identical to those used by HMDB, SMPDB and many other databases. The Sequence search uses BLAST ([@B25]) to find sequence matches or sequence similarities to the protein sequences within PathBank\'s sequence database of 8973 proteins. PathBank\'s 'About' tab provides additional information about PathBank, its associated release notes, the required citations, database statistics, the PathBank style guide, links to other Pathway Databases, as well as images (via a Pathway legend) for the different components (organelles, organs, tissues) seen in PathBank pathways. PathBank\'s 'Download' tab allows users to download pathways, metabolite names and protein names (in CSV files) as well as all of its pathways in BioPAX, SBGN, SBML and PWML format. A subset of PathBank\'s pathways (i.e. the primary pathways browsable through the TOPP) are also downloadable in SVG and PNG format. Due to the extremely large file size, users who wish to download all pathway images will need to submit a request using PathBank\'s Contact form. All of PathBank\'s sequences (both gene and protein) are available in FASTA text format, all of its chemical structures are available in SDF format and all of its reactions are available in RXN format. PathBank ASSEMBLY, QUALITY CONTROL AND CURATION {#SEC4} =============================================== PathBank was assembled using the PathWhiz pathway illustrator ([@B16]) using a team of pathway curators who have been manually researching, illustrating and refining pathway diagrams for more than 5 years. PathBank was initially built from a number of small molecule (metabolite) pathway diagrams that were originally assembled for SMPDB ([@B15]), HMDB ([@B17]), DrugBank ([@B14]), YMDB ([@B18]) and ECMDB ([@B26]). These pathway databases are restricted to depicting the metabolic or metabolite signaling pathways associated with humans (SMPDB, HMDB, DrugBank), yeast (YMDB) and E. coli (ECMDB). PathBank extends these existing pathway collections to cover other model organisms and to include disease, physiological, protein and cellular signaling pathways. The PathBank curation team took advantage of a number of recent improvements to the PathWhiz pathway illustrator ([@B16],[@B27]). In particular, the latest version of PathWhiz has been enhanced to permit near-automated pathway propagation (to create homologous pathways for other species) and pathway replication (to create similar pathways within a species). PathWhiz has also been modified to support protein signaling, protein--protein interactions, protein--DNA interactions, transport and transcriptional events. An example of a PathBank pathway diagram generated via PathWhiz (illustrating the Warburg effect) is shown in Figure [2](#F2){ref-type="fig"}. ![A screenshot from PathBank\'s 'Warburg Effect' pathway. First identified in the 1920's, this is a pathway commonly activated in cancer and it is classified as a 'Metabolite Disease' pathway. Curiously, no other commonly used pathway database appears to illustrate this very well-known and important pathway.](gkz861fig2){#F2} As described previously ([@B15],[@B16]), all pathways constructed via PathWhiz are built according to standard operating protocols (SOPs), checklists and pathway illustration style guides to ensure general uniformity and consistency across pathways and across species. The comprehensive PathBank style guide, complete with numerous visual examples, is posted on the web page under the 'About' tab. In selecting which pathways would be drawn for a given species, members of the curation team would first perform an extensive literature search to identify unique metabolite classes or metabolic processes for the organism of interest. Second, a literature review would be done to search for unique proteins or unique protein/cellular signaling processes. Next, a detailed review of existing pathway diagrams (both metabolic and protein signaling) from multiple public pathway databases for that organism would be performed. These results would then be compiled and compared against existing internal reference pathway databases corresponding to the closest species (such as the HMDB for other mammals) to that organism. Further discussions and team meetings would be held wherein curators would discuss their findings with PathBank\'s lead curator (CL) and its principal investigator (DSW). Through this process new pathways for a given organism would be identified, selected (or rejected), assessed for their similarity to existing pathways, sketched (by hand) and finally manually drawn via PathWhiz (if unique) or replicated via PathWhiz and modified (if similar to another species) by members of the PathBank curation team. After a given pathway was illustrated using PathWhiz, it would be evaluated independently by a second pathway curator (often the lead curator or the PI) to ensure it adhered to the PathWhiz style guides and SOPs. If the pathway failed to meet style guide requirements or failed to follow the SOPs, it would be corrected and re-evaluated until it met those requirements. Only after a pathway diagram passed secondary inspection would it be allowed to be propagated (to other species) or replicated (for related pathways within a species). This quality control protocol was applied to all hand-drawn primary pathways in PathBank which serve as template pathways that can be automatically replicated and/or propagated via PathWhiz. Spot checks (of ∼5% of the pathways) for any automatically replicated/propagated pathways were manually done by the lead curator(s) to ensure logical and biological consistency. Global or system-wide checks were also done periodically to ensure that organelle or organ/tissue images were appropriate for a given organism. Many of the primary pathways (especially for metabolism) that have been online for several years have undergone heavy remediations in order to comply with the guidelines outlined in PathBank\'s detailed style guide. As a result of this huge effort, these redrawn pathways have a standard, more aesthetically pleasing appearance, greater biological context (e.g. transporters across cell and organelle membranes) and increased connectivity between pathway entities. Many pathways have been updated with new descriptions and icons that have been drawn for PathBank\'s growing image collection. User feedback has also been used to further refine and improve these pathways. All members of the PathBank curation team were required to have at least an undergraduate degree in bioinformatics or molecular biology. This ensured that they had sufficient biological and/or biochemical knowledge to understand the scientific literature and the biological processes being illustrated. All curation team members were also given extensive training by the lead curator(s) in pathway illustration via hands-on mentoring, text instructions, peer support, and video tutorials. However, it is important to note that given the inherently artistic nature of pathway illustration, the differences in artistic/illustrative skills among the curation team and the complexity of some biological processes, it is impossible to guarantee that all pathways in PathBank exhibit the same level of visual or conceptual quality. Improvements and updates to PathBank\'s pathways are an ongoing process. Minor corrections or small improvements in image or pathway quality or pathway descriptions will be done without a formal update announcement. However, significant changes, additions or improvements to an individual pathway diagram (\>10% in the number of pathway components), will be listed in the pathway description and the last updated date will be modified to reflect any such changes. As this is only version 1.0 of the PathBank database, all pathway diagrams are dated with August 2019 as the last update date. Large-scale updates and improvements to the database in the future will be given database version numbers (2.0, 3.0, etc.) and suitable database update dates. They will also be described in detail as publications or online update descriptions as appropriate. PathBank is made available under the Open Database License in which users are free to contribute to the database using its PathWhiz pathway drawing tool. Indeed, one of the authors listed on this manuscript (MRG) contributed \>50 primary pathways to PathBank as an external 'community' member. Of course, not all community contributions are of sufficient quality to merit inclusion to PathBank. Any decision to include/exclude community or crowd-sourced pathway submissions depends on the quality/quantity of the contribution and their adherence to PathBank\'s SOPs and style guide. The final decision for any external contribution is ultimately made by PathBank\'s lead curator. The PathBank curation team will closely work with, and will train, external pathway illustrators who are serious about contributing to this resource. PathBank LIMITATIONS AND FUTURE PLANS {#SEC5} ===================================== Like all pathway databases, PathBank is a work in progress. Currently both the quantity and level of coverage of PathBank\'s protein pathways falls far short of what is currently available for PathBank\'s metabolite/compound pathways. Indeed, only 398 protein pathways have been illustrated to date, compared to 109 836 metabolite pathways. This imbalance reflects the past focus of the PathBank team\'s curation effort (which was primarily in metabolism), the extremely large number of lipid pathways (\>90 000) and the challenges that exist in depicting the somewhat more complex protein signaling pathways. As PathBank\'s collection expands, so too will its genome coverage (currently ranging between 1.9 and 25.1%) for each model organism (viewable on PathBank\'s statistics page). Of course, even with this major annotation effort, not every known pathway will appear in PathBank---now or in the near future. Likewise, it is expected that a number of users will notice that some PathBank pathways do not depict all expected or relevant entities in the pathway diagrams. In these situations, we certainly encourage users to contact the PathBank curation team (via the 'Contact' tab) so that those corrections can be made or those PathBank depictions can be more fully explained. Another clear limitation with PathBank is the relatively small number of model organisms that are currently covered. Currently, only 10 organisms are included in PathBank\'s 'zoo' with several well-known model organisms, including a few key vertebrates, several parasites and a number of important bacteria, not being included. Over the coming two years, it is expected that at least nine other model organisms will be added to the database, including the zebrafish (Danio rerio), the African clawed frog (Xenopus laevis), the chicken (Gallus gallus), rice (Oryza sativa), the malaria parasite (Plasmodium falciparum), the sleeping sickness parasite (Trypanosoma brucei) and several bacteria, including Bacillus subtilis, Staphylococcus aureus, and Thermotoga maritime. Additional organisms or species may be added depending on user feedback. PathBank is also somewhat limited in the analytical tools that it offers. This limited offering was a deliberate decision. Because all of PathBank\'s data are freely downloadable, we believe it would be far better for PathBank to serve as a data resource hub rather than an analytical hub. By making PathBank more of a data resource hub, we hope to encourage bioinformatics specialists to download or use the PathBank data to develop and implement their own, much more sophisticated analytical tools and web servers. Indeed, this has already happened. For instance, PathBank\'s data is now being used to support a very popular analytical web server called MetaboAnalyst 4.0 ([@B28]). PathBank was originally designed to facilitate interactive, web-based visualization and pathway navigation. However, this has led to the creation of pathways that are not particularly 'printer friendly'. Indeed, many PathBank pathways have entity labels that are generally too small to be seen in conventional slides or manuscript figures. To improve the situation, PathBank now offers two alternative pathway displays (available through the Settings tab of the Display panel) that allows users to toggle between 'regular text' and 'large text' for both the full-color and simplified visualizations so that the image labels are more visible and printer-friendly. In order to accommodate the larger font size, manually curated abbreviations and gene symbols adhering to a strict seven-character limit are used in place of the longer full names of compounds and proteins. Additional efforts are ongoing to create even smaller or more compact pathway visualizations with less empty space and more grid-like entity layouts. CONCLUSION {#SEC6} ========== PathBank represents a new type of pathway database. It has been designed to take some of the best features of earlier pathway databases and to combine them with newer, more modern kinds of web visualization tools and more advanced web interfaces to create a fully open access, visually rich and highly comprehensive pathway resource. PathBank also brings a number of unique features into the Pathway database field including exceptional pathway coverage, richly detailed pathway and entity descriptions, a number of new pathway types or categories (metabolite signaling, drug action, drug metabolism, disease-specific pathways), new kinds of pathway or entity visualization details (cellular location, organelles, organs, protein quaternary structure) and a much wider variety of pathway viewing and data exchange/storage options. Being the first release of this particular pathway database, we expect that users will note some missing elements or inconsistences with certain pathways or pathway data. However, we anticipate that through user feedback, continued community engagement and ongoing curation efforts, the next release of PathBank will be even more comprehensive and potentially much more useful to the community. We would like to thank Aidin Foroutan and Noah A. Wishart for their contributions to the earliest versions of PathBank. FUNDING {#SEC7} ======= Canadian Institutes for Health Research (CIHR); Canada Foundation for Innovation, Major Science Initiatives Fund (CFI-MSI); Genome Alberta (a division of Genome Canada). Funding for open access charges: Genome Alberta. Conflict of interest statement. None declared.	{ "pile_set_name": "PubMed Central" }
The authors confirm that all data underlying the findings are fully available without restriction. Data are available from Figshare: <http://dx.doi.org/10.6084/m9.figshare.1232203>. Introduction {#s1} ============ Work is a part of a process of social integration that directly affects the physical and mental health of individuals. Sickness absence is an indicator of the state of health of workers [@pone.0115885-Vahtera1]. Sociodemographic characteristics [@pone.0115885-Krokstad1]--[@pone.0115885-Markussen1], health habits and lifestyles [@pone.0115885-Krokstad1]--[@pone.0115885-North1], environmental work conditions [@pone.0115885-Krokstad1], [@pone.0115885-North1]--[@pone.0115885-Fischer1], psychosocial factors at work [@pone.0115885-North1] and the individuals\' state of health [@pone.0115885-Ferreira1] are some of the factors associated with sickness absence. Sick leave due to mental disorders is a global cause of concern that deserves attention as a function of its associated costs [@pone.0115885-Dewa1]. In Brazil, sick leaves lasting up to 15 days are paid by the worker\'s employer, while workers on longer leaves must claim sickness benefits from the public social security system (National Social Security Institute - INSS). Applicants are evaluated by medical experts to establish whether the sick leave benefits should be granted. In recent years, mental and behavioral disorders became the third leading cause of temporary sickness benefits. About 203,000 new sick leave benefits due to mental problems were granted from 2008 to 2011, costing more than US\$ 90 million to the Brazilian public pension system [@pone.0115885-SilvaJunior1]. An earlier joint publication by the International Labour Office and the World Health Organization (ILO/WHO) called the attention to the relevance of the psychosocial factors at work for the employees\' state of health. Long-term exposure to such stressors is associated with psychosomatic complaints, psychiatric symptoms and changes in wellbeing [@pone.0115885-ILO1]. Theoretical models were developed to define and explain the effects of occupational stress on the workers\' health, two among which were given special attention, to wit, the Demand-Control-Support (DCS) [@pone.0115885-Karasek1] model, which is assessed by means of the Swedish Demand-Control-Support Questionnaire [@pone.0115885-Sanne1], and the Effort-Reward Imbalance (ERI) model [@pone.0115885-Siegrist1], which is also assessed by means of a specific questionnaire [@pone.0115885-Siegrist2]. According to the literature, there is overlap between those questionnaires dimensions [@pone.0115885-Calnan1]. Analyses combining both models provide better estimates of the effects of stressful experiences at the workplace [@pone.0115885-Belkic1]. Recent studies on sick leave applied the DCS and ERI models to the investigation of psychosocial factors at work, either alone or in combination [@pone.0115885-Griep1]--[@pone.0115885-Wang1]. However, those models have not yet been used in combination to the investigation of such factors and LTSA-MD. A first spell of sickness absence due to mental disorders is associated with risk for recurrence of further sick leaves [@pone.0115885-Koopmans1]. For that reason, an accurate understanding of the factors that are related with sickness absence might increase the efficacy of preventive actions and point to strategies to facilitate return to work [@pone.0115885-Cornelius1]. The aim of the present study was to compare factors associated with long-term sickness absence between workers who claimed social benefits due to mental disorders or by other causes. We investigated individual features and occupational characteristics. In addition, we evaluated psychosocial factors at work assessed by the DCS and ERI models, and whether they were associated with the outcome variable. Material and Methods {#s2} ==================== The present case-control study was performed at the largest public social security branch in the city of São Paulo, Brazil, in 2011. The group of patients assigned to each medical expert was considered to represent a cluster. Every day along the study period, and at each work shift, clusters were randomly selected to compose the study population. After the completion of the social security medical evaluation, the participants from the randomly selected clusters who met the inclusion criteria were referred for interview with one of the investigators. The participants must have worked in formal employment and requested sick leave benefits after having being off from work for more than 15 days. The ones who had dual employment or were out of work for more than 12 months were excluded. A total of 438 eligible individuals were invited to participate in the study, being that 53 refused. Difference in sex and medical diagnosis was not found among the individuals who agreed or refused to participate, however, the former where younger than the latter (p\<0.05), being that their average age was 34.8 (±9.5) and 38.0 (±9.44) years old, respectively. The final study population comprised 385 participants. As cases were defined 160 volunteers who requested sick leave benefits after having being diagnosed with one of the "Mental and behavioral disorders" listed in International Statistical Classification of Diseases and Related Health Problems - 10^th^ Revision (ICD-10) Chapter 5. Most of the sample was characterized as having: "Mood \[affective\] disorders - F30-F39" (53.7%) and "Neurotic, stress-related and somatoform disorders -- F40- F49" (32.6%). Other mental and behavioral disorders found were: "Organic, including symptomatic, mental disorders F00-F09) - (1.9%), "Mental and behavioral disorders due to psychoactive substance use - F10-F19" (5%), "Schizophrenia, schizotypal and delusional disorders - F20-F29" (4.9%) and "Disorders of adult personality and behavior - F60-F69" (1.9%). Controls were 225 participants who applied for sickness benefits due to diseases listed in any other ICD10 chapter, being their distribution as follows: chapter XIX - Injury, poisoning and certain other consequences of external causes (36.9%), XIII - Diseases of the musculoskeletal system and connective tissue (19.6%), XI - Diseases of the digestive system (7.6%), IX - Diseases of the circulatory system (5.3%), II - Neoplasms (5.3%), and others (25.3%). All the participants signed an informed consent form in compliance with the Declaration of Helsinki. The study was authorized by INSS and approved by the ethics committee of School of Public Health, University of São Paulo, Brazil. Interviews were conducted to fill the forms used to collect data on covariates that could show association with sickness absence [@pone.0115885-Krokstad1]--[@pone.0115885-Ferreira1]; [@pone.0115885-Griep1]--[@pone.0115885-Wang1]; [@pone.0115885-Allebeck1]. The investigated individual features were sociodemographic data (sex, age, self-reported skin color, marital status, education), health habits and lifestyle (tobacco use, alcohol intake, physical activity), and health conditions (comorbidities, body mass index). The versions validated for the Brazilian Portuguese language of the following questionnaires were used to evaluate the participants\' health habits and lifestyle: Fagerström Tolerance Questionnaire for tobacco use [@pone.0115885-Halty1], Alcohol Use Disorders Identification Test (AUDIT) for level of alcohol intake [@pone.0115885-Figlie1] and International Physical Activity Questionnaire (IPAQ) [@pone.0115885-Matsudo1]. Morbidities that required medical care in the year previous to data collection were investigated to establish the participants\' health conditions. The volunteers\' height and weight were measured at the time of the interview to calculate the body mass index (BMI). The occupational data assessed were: nature of work and employment status, current job position and violence at the workplace. The psychosocial factors at the workplace assessed by the DCS model were investigated using the version validated for the Brazilian Portuguese language [@pone.0115885-Alves1] of the "Swedish Demand-Control-Support Questionnaire" (DCSQ). A short version of an ERI Brazilian Portuguese questionnaire was used to assess dimensions of the ERI model [@pone.0115885-Chor1]. The diagnosis of the sick leave-related illnesses was obtained from the reports issued by the medical experts from the INSS branch where the study was conducted. The allocation of participants reporting more than one disease to the group of cases or controls was decided based on their main diagnosis, as established by the INSS examiner. Univariate logistic regression models were built including the outcome and the independent variables. Variables with p-value equal to or lower than 0.20 on univariate regression were selected for the stepwise forward process used to fit multiple logistic models. A correlation matrix was elaborated to assess the potential overlap effect of variables. P-value less than 0.05 was considered to be significant in the final model. Four final regression models were built to evaluate the psychosocial factors at work and their association with LTSA-MD. Regression analysis (A) evaluated the association of LTSA-MD with the interaction of three psychosocial dimensions assessed by the DCS model. Regression analysis (B) investigated the association of the outcome with the interaction of three dimensions assessed by the ERI model. Regression analysis (C) included both models separately as covariates associated with the outcome. Finally, regression analysis (D) assessed the interaction of both models and their association with LTSA-MD. Software Epidata version 3.1 was used for data collection. The data were tabulated and analyzed using software Epi-Info version 3.5.2 and SPSS version 19. Results {#s3} ======= The study population was mostly composed of women (56.6%), with mean age of 34.8 (±9.5) years old. Most participants self-reported to be white (49.1%), live with someone (married/stable relationship) (51.4%), and had attended 11 years or more of formal education (67.7%) ([Table 1](#pone-0115885-t001){ref-type="table"}). 10.1371/journal.pone.0115885.t001 ###### Distribution of participants according individual features, São Paulo, Brazil, 2011 (N = 385). ![](pone.0115885.t001){#pone-0115885-t001-1} Case (N = 160) \% Mean (SD) Control (N = 225) \% Mean (SD) p-value ------------------------------------------- ---------------- ------ -------------- ------------------- ------ --------------- --------- Sex Male 48 28.7 119 71.3 Female 112 51.4 106 48.6 \<0.001 Age 34.21 (8.13) 35.17 (10.37) 0.325 Self-reported skin color White 92 48.7 97 51.3 Black 46 32.4 96 67.6 Other 22 40.7 32 59.3 0.053 Marital status Single 54 40.0 81 60.0 Married/stable relationship 83 41.9 115 58.1 Separated/divorced/widower 21 27 0.625 Education (formal schooling in years) \<11 34 27.4 90 72.6 \> = 11 126 48.3 135 51.7 0.001 Tobacco use None 130 41.5 183 58.5 Low 13 26.5 36 73.5 High 17 73.9 6 26.1 0.1617 Alcohol intake None 143 45.1 174 54.9 Low 12 19.7 49 80.3 High 5 71.4 2 28.6 0.036 Physical activity None 36 36.0 64 64.0 Yes 124 43.5 161 56,5 0.191 Comorbidities 0 or 1 17 15.5 93 84.5 \> = 2 143 52.0 132 48.0 \<0.001 Body mass index Low weight 7 58.3 5 41.7 Normal weight 58 38.2 94 61.8 Overweight 51 39.8 77 60.2 Obesity 44 47.3 49 52.7 0.427 A total of 81.3% participants were nonsmokers, and 82.3% reported being abstainer. Physical activity in the previous month was reported by 74% of the participants ([Table 1](#pone-0115885-t001){ref-type="table"}). Most participants worked in the private sector (82.6%), being that "service provider" was the most prevalent professional group (61.6%) ([Table 2](#pone-0115885-t002){ref-type="table"}). 10.1371/journal.pone.0115885.t002 ###### Distribution of participants according to occupational characteristics, São Paulo, Brazil, 2011 (N = 385). ![](pone.0115885.t002){#pone-0115885-t002-2} Case (N = 160) \% Control (N = 225) \% p-value ------------------------------------- ---------------- ------ ------------------- ------ --------- Employment and status Private sector (outsourced) 19 38.0 31 62.0 Private sector (regular) 129 40.6 189 59.4 Public sector\* 12 70.6 5 29.4 0.103 Occupational groups Operational-manufacturing 11 17.7 51 82.3 Management positions\\ 7 33.3 14 66.7 Technical-administrative\* 30 46.2 35 53.8 Service providers\* 112 47.3 125 52.7 0.046 Violence at workplace None 9 12.0 66 88.0 Yes\* 151 48.7 159 51.3 \<0.001 Demand-Control quadrant High strain\* 92 53.2 81 46.8 Low strain 7 19.4 29 80.6 Active work\\ 55 37.9 90 62.1 Passive work 6 19.4 25 80.6 \<0.001 Social support Low\* 86 69.9 37 30.1 High 74 28.2 188 71.8 \<0.001 DCS model No exposure 99 32.7 204 67.3 High strain + low social support 61 74.4 21 25.6 \<0.001 Effort-Reward ratio \<1.00 64 26.8 175 73.2 \> = 1.00 \* 96 65.8 50 34.2 \<0.001 Overcommitment Low 21 15.2 117 84.8 High\* 139 56.3 108 43.7 \<0.001 ERI model No exposure 41 18.2 184 81.8 High ER ratio + high overcommitment 93 58.1 67 41.9 \<0.001 Combined DCS + ERI models No exposure 110 34.0 213 66.0 Exposure 50 80.6 12 19.4 \<0.001 Note: DCS = demand-control-support; ERI = effort-reward imbalance; ER = effort-reward Violence at the workplace was mentioned by 80.5% of the participants ([Table 2](#pone-0115885-t002){ref-type="table"}), corresponding to: verbal abuse (61.8%), harassment (53.8%), be the target of jokes and discrimination (48.6%), threats of aggression (35.1%), assault at work or on the way to work (29.3%), sexual harassment (12.2%), physical aggression (9.3%) and sexual violence (1.0%). The results of Cronbach\'s alpha for the DCSQ dimensions were: demand, 0.73; control: 0.56; and social support, 0.84. High job strain was the most prevalent quadrant in the DCS model (44.9%). Social support was reported to be high by 68.1% of the participants. The Cronbach\'s alpha for the ERI questionnaire dimensions was: effort, 0.79; reward, 0.86; and overcommitment, 0.85. The ERI condition was reported by 37.9% of the participants. High overcommitment to work was reported by 64.2% of the participants ([Table 2](#pone-0115885-t002){ref-type="table"}). Most of the participants (91%) reported having sought medical care in the previous year. The main health problems motivating those medical visits were: emotional disturbances (48.3%) and back pain (35.6%). The average BMI of the participants was 28.84 kg/m^2^ (±5.58 kg/m^2^) representing a tendency to overweight ([Table 2](#pone-0115885-t002){ref-type="table"}). The individual features selected for multiple modeling were: sex, self-reported skin color, education, level of tobacco use, level of alcohol intake and physical activity and number of morbidities reported in the previous year ([Table 1](#pone-0115885-t001){ref-type="table"}). The selected variables corresponding to occupational factors were: employment status, occupational group, violence at the workplace, DCS model, ERI model, combined DCS and ERI models ([Table 2](#pone-0115885-t002){ref-type="table"}). Although variable participants\' age did not showed a p-value equal to or lower than 0.20, it was kept in the final model because it might modify the outcome variable. All the regression analyses showed that LTSA-MD was associated with the female sex, self-reported white skin color, higher educational levels, high tobacco consumption, high alcohol intake, two or more comorbidities reported in the previous year, exposure to violence at work ([Table 3](#pone-0115885-t003){ref-type="table"}). 10.1371/journal.pone.0115885.t003 ###### Multiple logistic regression analyses and factors associated with long-term sickness absence due to mental disorders^a^, São Paulo, Brazil, 2011 (N = 385). ![](pone.0115885.t003){#pone-0115885-t003-3} Regression A OR (95%CI) Regression B OR (95%CI) Regression C OR (95%CI) Regression D OR (95%CI) ------------------------------------------------------------------ -------------------------- ------------------------- ------------------------- -------------------------- Sex Male 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) Female 2.22 (1.27--3.87)\\ 2.09 (1.21--3.62)\\ 2.07 (1.18--3.64)\* 2.30 (1.32--4.01)\\ Self-reported skin color White 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) Other 0.50 (0.29--0.84)\\ 0.51 (0.30--0.86)\* 0.49 (0.29--0.84)\\ 0.49 (0.29--0.83)\\ Education (formal schooling in years) \<11 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) \> = 11 2.61 (1.45--4.70)\\ 2.41 (1.34--4.32)\\ 2.51 (1.39--4.56)\\ 2.43 (1.35--4.37)\\ Tobacco use None 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) Low 0.66 (0.29--1.51) 0.65 (0.28--1.50) 0.70 (0.30--1.62) 0.66 (0.29--1.49) High 8.85 (2.44--32.10)\\\* 7.16 (2.01--25.48)\\ 8.66 (2.33--32.11)\\ 8.08 (2.24--29.19)\\ Alcohol intake None 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) Low 0.29 (0.12--0.69)\\ 0.36 (0.16--0.83)\* 0.31 (0.13--0.74)\\ 0.30 (0.13--0.69)\\ High 10.04 (1.18--85.63)\* 11.05 (1.42--85.90)\* 9.98 (1.16--86.11)\* 9.07 (1.05--78.12)\* Comorbidities 0 or 1 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 2 or more 5.55 (2.81--10.95)\\\* 4.64 (2.36--9.13)\\\* 4.77 (2.38--9.56)\\\* 5.35 (2.73--10.49)\\\* Violence at workplace No 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) Yes 4.43 (1.96--9.99)\\\* 3.60 (1.59--8.14)\\ 3.47 (1.51--7.97)\\ 4.36 (1.94--9.79)\\\* DCS model No exposure 1.0 (Ref) 1.0 (Ref) High job strain + Low social support 4.43 (2.33--8.45)\\\* 3.37 (1.71--6.66)\\\* ERI Model No exposure 1.0 (Ref) 1.0 (Ref) ERI + high overcommitment 3.04 (1.77--5.24)\\\* 2.15 (1.21--3.84)\\ Interaction of DCS + ERI models No exposure 1.0 (Ref) High job strain + low social support + ERI + high overcommitment 5.20 (2.40--11.25)\\\* Note:^a^ adjusted for age; \* p \<0.05; \\ p\<0.01; \\\* p\<0.001; OR = odds ratio; 95% CI = 95% confidence interval; DCS = demand-control-support; ERI = effort-reward imbalance. Perceived psychosocial factors at work were associated with LTSA-MD. The analyses showed that the both DCS and ERI models were associated with the outcome. Statistical difference was not found between separate or combined analysis of those models dimensions ([Table 3](#pone-0115885-t003){ref-type="table"}). Discussion {#s4} ========== The case-control design selected for the present study allowed detecting that individual features and psychosocial work conditions were significantly associated with LTSA-MD compared to other causes of sick leave. In addition, dimensions of occupational stress assessed by the DCS and ERI models were significantly associated with LTSA-MD. LTSA-MD and Individual Features {#s4a} ------------------------------- The tendency of women to show greater concern about their health might account for the fact that sex was one of the sociodemographic variables associated with LTSA-MD. This female cultural trait might be reflected in an earlier search for medical help, because women deem reporting psychiatric symptoms more acceptable than men do [@pone.0115885-Stansfeld1]. In addition, the women\'s frequent assumption of the "double burden" of juggling work and family is also considered to be a factor strongly associated with sick leave episodes [@pone.0115885-Allebeck1]. The association of LTSA-MD with self-reported white skin color and high educational levels might be accounted for by the greater access these groups of individuals have to health information in Brazil. More informed workers might detect more easily situations that pose a threat to their mental health [@pone.0115885-Silva1], and thus seek appropriate actions that might demand sick leave. Those facts notwithstanding, the relevance of the educational level is controversial. A recent review found that high educational levels were protective against, but also a risk factor for longer sickness absence due to mental disorders [@pone.0115885-Cornelius1]. Sick leave due to psychiatric disorders is described in male alcohol drinkers and female smokers [@pone.0115885-Foss1]. The present study showed that both, high alcohol intake and high tobacco consumption, were associated with LTSA-MD regardless of sex. However, as a function of the study design, we were not able to establish the temporal precedence between these habits and mental distress. Multiple health complaints [@pone.0115885-Roelen1] and self-rated poor health status [@pone.0115885-Ferreira1] are considered risk factors for sick leave, and our results corroborate this hypothesis. Reporting two or more comorbidities in the previous year was associated with LTSA-MD. Self-perceived deterioration of the general state of health probably represents one further psychological stressor that increases the workers\' mental fatigue. This situation might eventually predispose workers toward absenteeism. LTSA-MD and Psychosocial work conditions {#s4b} ---------------------------------------- Perception of the work conditions as characterized by low control, high demands, low social support, high effort, low reward and high overcommitment exhibited significant association with LTSA-MD. Altogether, those stressors might denote cognitive overload, resulting from excessive demands at work, the effort associated with multitasking, and lack of perspectives for professional growth. Previous studies showed that all-cause LTSA is associated with high job strain in middle-aged workers [@pone.0115885-Wang1]. LTSA was also associated with dimensions assessed by the combination of DCS and ERI models among nursing professionals [@pone.0115885-Griep1]. A recent study conducted in Canada showed that only ERI was independently associated with sick leave due to mental health problems [@pone.0115885-Ndjabou1]. Job control at work is a part of the theoretical construction of the two models used herein to evaluate occupational stress. This type of psychosocial factor at work combines task performance (micro aspect) with broader issues related with the rewards, such as salary and career progression (macro aspect) [@pone.0115885-Bosma1]. Low control and/or low reward might make mental suffering and fatigue worse. A recent study conducted in Belgium showed that only low reward at work was independently associated with LTSA-MD, although other dimensions of the DCS and ERI models had also been included in analysis [@pone.0115885-Janssens1]. In the ERI model, effort might behave as an extrinsic (like, e.g., work activities) and also as an intrinsic, or personal component (as, e.g., high overcommitment) liable to affect the workers\' mental health. Combination of low effort and high demand is significantly associated with medically certified absence for mental health problems [@pone.0115885-Ndjabou1]. High job strain is an important factor associated with LTSA [@pone.0115885-Wang1] and this finding agrees with our results relative to sick leave due to mental disorders. However, the results of the Belstress III study did not find association of LTSA-MD with high job strain or social support at work [@pone.0115885-Ndjabou1]. Both low social support and excessive personal commitment were associated with LTSA-MD in the present study. Workers might sometimes feel themselves alone, while having to behave in a resolute manner vis-à-vis excessive demands relative to which they have little decision-making power [@pone.0115885-Vzina1]. Thus being, one might hypothesize that cases of this study with LTSA-MD could be triggered by the attempts made by workers to excel in an unstable and or competitive job employment [@pone.0115885-Siegrist3]--[@pone.0115885-Hasselhorn1]. Further studies might investigate the effect of subjective assessments of productivity or lack of recognition by peers and superiors on absenteeism due to mental sickness. Exposure to violence at the workplace stood out as an important factor associated with sickness absence, as also an European study found [@pone.0115885-Niedhammer1]. Bullying, more particularly, is associated with LTSA-MD [@pone.0115885-Janssens1]. This fact is probably due to deterioration of both the vertical (supervisor-subordinate) and the horizontal (between coworkers) work relationships. Exposure to violence at the workplace and low social support at work were associated with LTSA-MD in the present study. Combined analyses of different scales assessing psychosocial factors at work have the advantage of decreasing the limitations inherent to each individual model [@pone.0115885-Silva1]--[@pone.0115885-Foss1], [@pone.0115885-Rugulies1]--[@pone.0115885-Griep2]. In the present study, the model that exhibited the highest odd ratio values was the one that included interactions among dimensions from the two theoretical models of stress used. However, the results of the comparison of the four regression analyses did not indicate statistically significant differences. To the best of our knowledge, the present study was the first to use the aforementioned model analyses to evaluate psychosocial factors at work associated with long-term mental sick leave in workers applying for social security benefits. There are not legal threshold limit values available for occupational exposure to psychosocial factors at work in spite of their acknowledged negative consequences on the mental health. Those factors should be discussed by managers and subordinates when planning return-to-work actions, in conjunction with preventive measures. Studies on the relationships between work conditions and recurrence of LTSA-MD are needed [@pone.0115885-Koopmans1]. Although self-reported data have been used in the investigation of factors associated with absenteeism and illness [@pone.0115885-Griep1], studies on long-term sickness absence ought to give the preference to data from official registries [@pone.0115885-Svedberg1] or confirmed by a physician [@pone.0115885-Marmot1]. One of the strengths of the present study is that both cases and controls were selected based on medical diagnoses confirmed by a highly qualified social security medical expert. The questionnaires used to investigate the participants\' exposure to psychosocial factors at work exhibited satisfactory internal validity, as shown by the Cronbachs alpha values. However, one might question whether interviewing individuals with poor mental health conditions might have not interfered with the results. To be sure, some of the participants might have judged their work conditions as less favorable than they would have done when they were active at work [@pone.0115885-Foss1], resulting in overestimation of the perceived unfavorable psychosocial factors at work. Although the study sample consisted of an urban working population holding formal jobs at a major Brazilian city, the results are similar to the ones of other studies conducted in different parts of the world [@pone.0115885-Allebeck1], [@pone.0115885-Foss1], [@pone.0115885-Rugulies1]--[@pone.0115885-Lu1]. Conclusions {#s5} =========== The present study showed that individual features and psychosocial work conditions were associated with long-term sickness absence due to mental disorders. Combined analysis of stress models showed that the psychosocial factors at work were significantly associated with LTSA-MD. The high job strain condition, low social support, effort-reward imbalance and high overcommitment might represent a threat to the mental health of workers. Resourceful use of this information may contribute to the implementation of preventive actions and strategies to facilitate return to work targeting the populations most susceptible to mental disorders. The authors would like to thank: the National Social Security Institute (INSS) of Brazilian Social Security Ministry, for granting permission for data collection; the Laboratory of Population Studies (LEP) and the Public Health Graduate Program, School of Public Health, University of São Paulo, Brazil, for technical assistance and funding, respectively. [^1]: Competing Interests:The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: JSSJ FMF. Performed the experiments: JSSJ. Analyzed the data: JSSJ FMF. Contributed reagents/materials/analysis tools: JSSJ. Wrote the paper: JSSJ FMF.	{ "pile_set_name": "PubMed Central" }
Introduction {#sec1-1} ============ Extracorporeal membrane oxygenation (ECMO) was introduced into clinical practice over 40 years ago, to support patients with advanced but potentially reversible cardiopulmonary failure. Early adoption as a widely accepted support modality occurred in severe neonatal respiratory failure following the publication of prospective clinical trials\[[@ref1][@ref2][@ref3]\] and in pediatric cardiac failure, particularly in the perioperative environment.\[[@ref4][@ref5]\] It subsequently became increasingly adopted in pediatric respiratory failure\[[@ref6]\] and adult cardiac support with apparent success but without prospective studies. Use in adults for respiratory failure was not widely adopted until recently, following publication of the CESAR trial\[[@ref7]\] and reports from the H1N1 epidemic suggesting a survival benefit.\[[@ref8][@ref9]\] At the present time, the most rapidly growing application of ECMO is to provide circulatory support to patients in cardiac arrest following failure of conventional cardiopulmonary resuscitation (CPR) (extracorporeal CPR \[ECPR\]) to provide a return of spontaneous circulation (ROSC).\[[@ref10]\] The higher level of cardiac output support provided by ECPR compared to conventional CPR provides several potential advantages, including a higher rate of successful ROSC, support of postresuscitation cardiogenic shock while arranging and performing coronary interventions, and maintaining organ perfusion during recovery of native cardiac output. The Extracorporeal Life Support Organization (ELSO) maintains a registry of extracorporeal life support (ECLS) cases reported by its member of health-care organizations, which currently number over 300 and represent a global cross-section of ECLS. The number of ECPR cases reported annually to the registry are given in [Figure 1](#F1){ref-type="fig"} and the growth in this area of ECLS represents the most rapidly growing segment. The number of actual cases worldwide is likely much higher since the ELSO participation is voluntary and the registry does not capture all cases performed. ![Growth of extracorporeal cardiopulmonary resuscitation cases reported annually to the Extracorporeal Life Support Organization registry from 1992 to 2015](ACA-20-4-g001){#F1} The term ECPR has been used to refer to the general use of ECLS in the periarrest period, including the use of postarrest cardiogenic shock following conventional resuscitation. ELSO defines ECPR as the use of ECLS in the patient in sustained cardiac arrest without ROSC, or in patients in ROSC may be transient but not sustained. This review adopts the narrower ELSO definition in which ECLS is used before sustained ROSC, in patients experiencing either out-of-hospital or in-hospital cardiac arrest. Rationale for Extracorporeal Cardiopulmonary Resuscitation {#sec1-2} ========================================================== Survival is low for both in-hospital and out-of-hospital cardiac arrest managed with conventional CPR, on the order of 10% for out-of-hospital and 15%--20% for in-hospital.\[[@ref11][@ref12][@ref13][@ref14][@ref15]\] Conventional CPR even when optimally performed provides only a fraction of normal cardiac output, in particular cerebral and myocardial blood flow.\[[@ref16][@ref17]\] Several factors are related to outcome from cardiac arrest.\[[@ref18]\] An initial rhythm of ventricular fibrillation (VF) or ventricular tachycardia (VT) (shockable rhythm) is associated with improved outcome,\[[@ref19]\] most likely representing a rhythm disturbance in a previously healthy myocardium as opposed to myocardial failure from underlying chronic cardiac disease. Even with VF as a presenting rhythm, preexisting comorbidities decrease the chance of survival.\[[@ref20]\] Witnessed cardiac arrest with bystander CPR results in earlier initiation of CPR and is associated with improved outcomes.\[[@ref19]\] The time to initial defibrillation in a shockable rhythm\[[@ref21]\] or administration of epinephrine in a nonshockable rhythm\[[@ref22]\] is inversely related to survival, suggesting that a sufficient duration of ischemia results in diminished response to resuscitation efforts or the irreversible loss of cellular energetics. Properly performed CPR results in improved cardiac output and appears to increase the time before irreversibility sets in, increasing the chance of survival.\[[@ref23][@ref24]\] Management of refractory cardiac arrest with ECPR has the potential to alter some of the above factors. ECPR eliminates the need to achieve ROSC to provide vital organ perfusion. Once the patient is cannulated and extracorporeal flow initiated, perfusion of the brain, myocardium (in nonoccluded coronary arteries), and splanchnic organs is immediately restored, and the postresuscitation low-flow state is eliminated. Improvement in myocardial blood flow enhances the chance of myocardial recovery and return of a spontaneous rhythm or a shockable rhythm more responsive to defibrillation. Introduction of interventions becomes easier. With earlier achievement of adequate systemic perfusion, the risk of neurologic injury may be reduced despite delayed return of spontaneous rhythm. Therapeutic hypothermia can be rapidly achieved and sustained. Coronary interventions are facilitated even if the patient remains in VF since stable perfusion has been achieved. Restoration of coronary perfusion may permit defibrillation or cardioversion of otherwise refractory VF or VT. The success of ECPR relies on early initiation in a patient with the potential for myocardial recovery. Indications {#sec1-3} =========== There are no prospective studies of ECPR, and no clearly identified selection and initiation criteria for its use in cardiac arrest.\[[@ref25]\] The American Heart Association recognizes that at the present time, there is insufficient evidence to recommend routine use of ECPR in cardiac arrest but indicates that it may be considered in settings where ECPR is readily available, blood flow interruption following arrest is brief, and the underlying condition leading to arrest is reversible.\[[@ref26]\] The challenges are identifying those patients with reversibility and initiating support sufficiently early. Despite the lack of clear-cut indications, there is a body of experience that can provide guidance for patient selection. A major determinant of outcome following cardiac arrest is the time to return of adequate perfusion, which in conventional CPR is considered to be time to ROSC, and in ECPR is time to initiation of extracorporeal flow. There are observational studies on survival versus time to ROSC following CPR but none following time to ECPR. In in-hospital cardiac arrest, nearly half of patients who had ROSC did so within 10 min of CPR and about 75% within 20 min.\[[@ref27]\] Eighty-one percent of those having ROSC within 15 min had good functional recovery. In out-of-hospital cardiac arrest, ROSC occurred within 16 min in nearly 90% of those patients with good functional recovery. After 15 min, good functional recovery was only 2%. These data suggest that CPR should be considered refractory at 15 min. Implementation of ECPR should take place as soon as possible after this time. The target to complete the transition to ECPR is 30 min when possible and no more than 60 min. The determination of reversibility is more challenging. Having a witnessed arrest is important in identifying the exact time of arrest and documentation of time to CPR, which should be within 5 min. A no-flow period of 10 min or longer should preclude ECPR. Once CPR is initiated, it should have been continuous and of good quality, preferably with a mechanical compression device. The etiology of arrest is a clear determinant of reversibility. In out-of-hospital cardiac arrest, the etiology and comorbidities are often unknown. A shockable rhythm suggests an ischemic event with retained myocardial viability and provides strong support for ECPR. A patient in pulseless electrical activity (PEA) poses the most challenge. Patients with known major comorbidity are generally not candidates. Younger patients with a respiratory etiology for arrest and patients with suspected massive pulmonary embolism may be considered. Inpatients usually have a well-characterized diagnosis and known comorbidities. Patients experiencing in-hospital cardiac arrest with known or suspected acute coronary syndrome, suspected pulmonary embolism, or arrhythmia associated with acute ventricular failure who may be ventricular assist device or transplant candidates are candidates. Other special circumstance warrants consideration for ECPR. Deep hypothermic (\<20°C) cardiac arrest is successfully managed with ECLS,\[[@ref28]\] for which it is considered a standard of care.\[[@ref29]\] Prolonged times to institution of ECPR can be considered, with high survival and durations of up to 226 min resulting in survival.\[[@ref30]\] Cardiac arrest or refractory nonperfusing rhythms due to severe cardiovascular drug intoxication or myocardial toxin are other applications of ECPR with reported success since these are typically reversible by metabolic elimination.\[[@ref31][@ref32][@ref33]\] Technique and Application {#sec1-4} ========================= The mode of support for ECPR is VA since other modes (venovenous, arteriovenous) provide gas exchange but not circulatory support, whereas VA support provides both. VA support involves cannulation of the right atrium for venous drainage and the aorta or a large artery (femoral or carotid) for return to the arterial system. The heart and lungs are bypassed during ECPR, and the extracorporeal circuit provides full cardiopulmonary support. On ROSC, systemic blood flow is the sum of native cardiac output and extracorporeal flow. Cannulation {#sec2-1} ----------- Several cannulation approaches are available, with the choice dictated by the circumstances surrounding cardiac arrest. Cannulation can be central or peripheral. Central cannulation is performed through a median sternotomy, with venous drainage directly from the right atrium or atrial appendage and return to the aortic root. Central cannulation is usually reserved for patients in the postoperative period following a prior median sternotomy for cardiac surgery. Outside of this setting, peripheral cannulation is standard practice. Peripheral cannulation is usually performed through the femoral vessels in adults and adolescents. The femoral site is easily accessible during CPR. A single operator can cannulate both the common femoral artery and femoral vein on the same side. For faster cannulation, one vessel on each side can be cannulated by two operators. Either percutaneous or surgical cannulation can be performed. In pediatric patients, the femoral vessels are of inadequate size for cannulation, and the carotid artery and internal jugular vein are accessed through surgical cutdown. The choice between percutaneous and surgical approaches for femoral access is usually based on the experience and preferences of the operator. Traditionally, a surgical approach was used since percutaneous access in the absence of pulses made a percutaneous approach prone to complications. Bedside vascular ultrasound allows the identification of vessel anatomy as well as sizing of the vessels to assure success with percutaneous access. As a result, percutaneous access is the preferred approach since it is faster, is associated with less bleeding complications, can be performed by nonsurgeons (intensivists, emergency physicians, interventional cardiologists, and interventional radiologists), and does not require ligation of the artery.\[[@ref34][@ref35][@ref36]\] A two-stage technique can be employed that may facilitate decision-making and cannulation. If there is no response to conventional resuscitation within 10 min, then cannulation of the femoral vessels using small vascular catheters can be performed. If further resuscitation attempts remain unsuccessful and ECPR is elected, then cannulation is facilitated by placing guidewires in the existing catheters.\[[@ref36]\] Cannulas used in ECPR must be capable of sufficient flow to provide full circulatory support (at least 4 L/min in the adult). The femoral venous cannula used in adults for percutaneous access is 24--28 Fr and is long enough to reach the right atrium from the femoral insertion site. Adult femoral arterial cannulas are short and range from 16 to 18 Fr. Surgical cannulation of the femoral artery in all cases and percutaneous in many cases will result in limb ischemia, necessitating placement of an antegrade cannula into the superficial femoral artery that can support over 200 mL/min, usually 6--8 Fr in size. When surgical placement is chosen, the cannula size is often chosen based on the observed vessel size. Central cannulation is performed with cannulas used for cardiopulmonary bypass, typically 32 Fr atrial and 24 Fr aortic, with appropriately smaller sizes for pediatric patients. Decannulation following successful return of adequate circulation depends on the cannulation approach. Surgically placed cannulas are removed by surgical cutdown and repair of the vessels. Percutaneous venous cannulas are simply withdrawn, with a horizontal mattress suture straddling the cannula to provide hemostasis. Percutaneous arterial cannula usually requires vessel repair, but withdrawal and manual pressure or a vascular occlusion device used for sizes up to 16 Fr. Decannulation of the centrally cannulated patient requires an operative approach with closure of the sternotomy incision. If long-term support is needed, the patient can be converted to peripheral cannulation to reduce risk of infection and bleeding associated with central cannulation. Extracorporeal circuit {#sec2-2} ---------------------- Extracorporeal circuits incorporating a centrifugal pump and hollow-fiber membrane lung have become the most widely adopted design for extracorporeal support, including ECPR. This circuit design has several advantages. The systems are rapidly primed. The centrifugal pump can provide a controlled negative inlet pressure to facilitate venous drainage. Being afterload dependent, these pumps will not generate very high pressure in the event of an occlusion. Hollow-fiber membrane lungs have low pressure drop and have highly efficient gas exchange, supporting up to 400 mL/min or more of oxygen transfer. Adult circuits incorporate ⅜" inner diameter tubing between all of the components and the cannulas. Centrifugal pumps and adult membrane lungs have ⅜" connectors. Pediatric circuits use ¼" inner diameter tubing. Pediatric membrane lungs have ¼" connectors, but centrifugal pumps are only available in ⅜" size, necessitating step-up connectors. Adult hollow-fiber membrane lungs have rated flows on the order of 7 L/min, more than adequate for most applications. Pediatric membrane lungs have rated flows on the order of 3 L/min. The extracorporeal circuit is primed with isotonic-balanced electrolyte solution for adults and pediatric patients in whom the extracorporeal circuit volume is no more than 30% of the patient\'s blood volume. Albumin is added by some centers to reduce osmotic shifts on initiation of support. Pediatric patients have higher extracorporeal circuit volumes relative to their blood volume and may require a blood prime due to hemodilution or immediate transfusion to restore hematocrit. Circuits preprimed with crystalloid solution only (no albumin or blood products) may be prepared ahead of time and stored for ready availability for ECPR. Preprimed circuits have been shown to be safe for at least 30 days.\[[@ref37][@ref38]\] Complications {#sec2-3} ------------- Bleeding is the most common complication associated with all forms of ECLS, including ECPR. The most common site of bleeding is at the cannulation site and is greater following surgical placement than for percutaneous placement. Cannula site bleeding can usually be managed with topical thrombotic agents but may require surgical exploration. Systemic anticoagulation increases the risk of bleeding at other sites, including intracranial and gastrointestinal. Vascular injury during percutaneous insertion can result in failure to cannulate, vascular injury, loss of distal flow, arteriovenous fistulas, and retroperitoneal hematomas. Since ECPR is deployed under time-sensitive circumstances, it could be expected to have a higher complication rate. Failure to achieve adequate extracorporeal flow can be due to inadequate intravascular volume, inadequately sized cannulas (venous in particular), or mispositioning of the cannulas. Ultrasound guidance during percutaneous insertion helps assure proper intravascular placement. Fluoroscopic imaging is helpful for assuring proper positioning but is generally not available during ECPR initiation, but may be helpful once ECPR is initiated. Postreturn of Spontaneous Circulation Management {#sec1-5} ================================================ Successful initiation of extracorporeal support may or may not be associated with early return of a stable perfusing cardiac rhythm. Persistence of or development of VF or VT warrants continued aggressive attempts at arrhythmia control with pharmacologic agents, cardioversion/defibrillation, and coronary reperfusion in the case of myocardial ischemia. Persistent PEA is approached with correcting underlying electrolyte abnormalities and acidosis (e.g., with renal replacement therapies). Restoration of myocardial contractility is critical since prolonged stasis of blood in the ventricles and aortic root can lead to clotting of these chambers. Targeted temperature management for postresuscitation hypothermia is easily performed with the extracorporeal circuit. The remainder of post-ROSC management including institution and maintenance of systemic anticoagulation, management of peripheral vascular tone, and management of mechanical ventilation, fluid balance, sedation, etc., is standard for all ECLS and is not detailed here. Outcomes Following Extracorporeal Cardiopulmonary Resuscitation {#sec1-6} =============================================================== There are no prospective controlled studies that compare outcomes from ECPR with conventional CPR. Evaluation of outcomes is currently limited to reports from observational studies with and without matching and meta-analyses. The ELSO registry reports a survival to discharge following ECPR of 29% of 2885 adults, 41% of 2996 children, and 41% of 1336 neonates.\[[@ref10]\] The difference between age groups is likely due to differences in demographics as well as age-related physiology and comorbidities. Neonatal and pediatric cases are more likely to be patients with congenital heart disease with in-hospital arrest in the perioperative period under close observation. It is also possible that these registry cases may include support of circulatory shock immediately following successful resuscitation with conventional CPR. A more detailed analysis of the ELSO registry data in 492 infants and children with heart disease receiving ECPR, a 42% survival, was reported. In this cohort, predictors for nonsurvival included single ventricle physiology, prior Stage 1-type procedure, and extreme metabolic acidosis.\[[@ref39]\] Observational pediatric studies from single institutions report outcomes similar to the ELSO registry. In a cohort of 80 children with primarily cardiac diagnoses receiving ECPR, survival with favorable neurologic outcome was reported as 30%.\[[@ref40]\] In a cohort of 32 inpatient pediatric patients, 82% of whom had a cardiac diagnosis, a survival of 73% with grossly normal neurologic function was reported.\[[@ref41]\] A similar cohort of 31 pediatric patients with predominantly cardiac disease, survival with good neurologic outcome was reported in 23%.\[[@ref42]\] Interestingly, this group found no relationship between the duration of conventional CPR before institution of ECPR and outcome. In a retrospective analysis of 54 pediatric patients with mixed cardiac and noncardiac causes of in-hospital cardiac arrest, an overall survival of 46% was reported, with approximately equal survival between the cardiac and noncardiac causes.\[[@ref43]\] Survivals in adults reported in literature are similar. A meta-analysis of eleven observational studies in adult patients receiving ECPR following witnessed in-hospital cardiac arrest and failure of conventional CPR found an overall survival to discharge of 40%.\[[@ref44]\] The most common diagnosis was acute myocardial infarction. This cohort, however, included patients receiving extracorporeal support for cardiogenic shock immediately following ROSC. With the stricter definition of ECPR applied before ROSC, the survival was higher at 48%, with a mean time of resuscitation of 40 min before initiation of ECPR. This may be due to the earlier institution of ECLS in the pre-ROSC group. A shorter pre-ECPR time (\<30 min) was found to be associated with a 1.9 odds of survival compared with a longer time. A prospective observational study using propensity matching comparing ECPR with conventional CPR in 172 adult patients with in-hospital cardiac arrest revealed a higher survival to discharge and higher 30-day and 1-year survival.\[[@ref45]\] In another study of 353 patients using propensity matching, ECPR was associated with better short- and long-term survival, with the only significant and independent predictor of survival being the use of ECPR (hazard ratio 0.57) as opposed to conventional CPR.\[[@ref46]\] Prognosis following cardiac arrest associated with deep hypothermia managed with ECPR with rewarming appears to have a better prognosis than standard extracorporeal rewarming. In a series of 59 patients, 25 patients were treated with ECPR and the rest with standard therapy. Sixty-four percent of these in the standard group died, primarily of severe pulmonary edema, but none in the ECPR group died. A multivariate analysis yielded a 6.6-fold higher chance of survival when using ECPR for deep hypothermic resuscitation.\[[@ref47]\] Summary {#sec1-7} ======= ECPR is an emerging technology for the management of patients with cardiac arrest refractory to conventional CPR and resuscitative approaches. ECPR is defined as the application of VA ECMO in patients in cardiac arrest who fail to achieve ROSC, or fail to retain ROSC, while undergoing resuscitation. Observational studies suggest that survival with good neurologic recovery may be better with ECPR than with conventional CPR but prospective controlled studies are lacking. Consideration for ECPR is given to individuals with in-hospital or out-of-hospital witnessed cardiac arrest receiving good quality sustained external chest compressions within 5 min of arrest, and in whom ROSC does not return within 15--30 min. Cannulation for ECPR should ideally occur within 30 min of arrest, with up to 60 min considered. A potentially reversible cause of cardiac arrest should be identified, such as a shockable rhythm for out-of-hospital cardiac arrest or a known reversible cause for in-hospital arrest. Both children and adults are candidates for ECPR. Further studies are needed to better define its role and application; however, at the present time, ECPR should be considered in appropriate patients in hospitals with existing ECLS programs. Financial support and sponsorship {#sec2-4} --------------------------------- Nil. Conflicts of interest {#sec2-5} --------------------- There are no conflicts of interest.	{ "pile_set_name": "PubMed Central" }
Introduction ============ People with severe mental illness (SMI), including schizophrenia, bipolar disorder, and other nonorganic psychotic illnesses, are known to have shorter life expectancy, by \~10--20 years, than the general population.[@b1-clep-10-421]--[@b6-clep-10-421] Their premature death is attributed not only to suicide and accident but also to physical illnesses -- especially cardiovascular diseases such as coronary heart disease and stroke.[@b5-clep-10-421]--[@b9-clep-10-421] The high rate of cardiovascular deaths in patients with SMI has been attributed to the high prevalence of lifestyle-related conditions (eg, smoking and obesity),[@b10-clep-10-421]--[@b13-clep-10-421] suboptimal screening/assessment of cardiovascular risks,[@b14-clep-10-421],[@b15-clep-10-421] and suboptimal management of underlying diseases (eg, hypertension, diabetes, dyslipidemia).[@b16-clep-10-421]--[@b18-clep-10-421] Chronic kidney disease (CKD) -- the continuous presence of kidney damage and/or decreased level of kidney function[@b19-clep-10-421] -- is independently associated with increased risk of death, cardiovascular events, and all-cause hospitalization.[@b20-clep-10-421] CKD can progress to end-stage renal disease (ESRD) requiring renal replacement therapy (RRT), a substantial burden for both -- quality of life of patients and health budgets. The UK National Institute for Health and Care Excellence guidance for CKD recommends early identification and appropriate management of CKD.[@b21-clep-10-421] People with SMI have an increased prevalence of several risk factors for CKD, including smoking and diabetes.[@b13-clep-10-421],[@b16-clep-10-421],[@b22-clep-10-421] Moreover, patients with SMI may be treated with lithium, which is associated with the development of CKD.[@b23-clep-10-421],[@b24-clep-10-421] However, there has been limited research investigating the prevalence of, and risk factors for, CKD in people with SMI.[@b13-clep-10-421],[@b25-clep-10-421],[@b26-clep-10-421] In the UK, primary care plays an important role in the management of physical health conditions in people with SMI. As part of the Quality and Outcomes Framework (QOF), there have been recent financial incentives for general practitioners (GPs) to undertake physical health checks in people with SMI.[@b27-clep-10-421] However, it is unclear whether CKD in patients with SMI is appropriately recognized and managed by GPs. Therefore, we used a UK primary care database to: 1) compare the prevalence of CKD and RRT among patients with SMI (with and without a history of lithium use) and those without SMI; 2) investigate whether there is an independent association between SMI and CKD after adjusting for known CKD risk factors; and 3) compare the recognition and management of CKD between CKD patients with and without SMI. Materials and methods ===================== Data source ----------- The Clinical Practice Research Datalink (CPRD) is an observational data and interventional research service provided by the National Health Service. Around 98% of the UK population are registered with a primary care practice. Currently, \>650 general practices contribute data conforming to quality control standards to the CPRD, including \~ 7% of the UK population. The included practices are representative of the UK general population in terms of age and sex.[@b28-clep-10-421] The database includes the following data: patient demographics, diagnoses based on Read codes (a hierarchical coding system), prescriptions based on British National Formulary codes, laboratory test results, and referrals made by GPs. Ethical approval for this study was obtained from the Independent Scientific Advisory Committee of CPRD (protocol no. 16_055). Informed consent was waived because data are anonymized for research purposes. Study population ---------------- Our study population included all people, aged 25--74 years, registered in the CPRD for at least 1 year as on March 31, 2014 (ie, the end of a financial or QOF year). We selected the 25--74 age range because our preliminary analysis suggested that CKD and/or SMI were rare in people outside these parameters, thereby limiting the power for comparative prevalence analyses. To ensure that we had reliable measures of morbidity (to allow time for the recording of past medical history in newly registered patients), we required that all participants had at least 1 year of continuous registration in the CPRD. Disease definition ------------------ We identified SMI using Read morbidity codes that have been validated in UK primary care data, with high sensitivity (91%), specificity (99.9%), and positive predictive value (91%).[@b29-clep-10-421] Patients with SMI were further classified as those with and without any record of lithium prescription in the period between the CPRD registration and March 31, 2014. We excluded patients who had been prescribed lithium without recorded SMI from the general population comparison group because the indication for treatment was not known. Our definition of CKD was based on two measurements of estimated glomerular filtration rate \<60 mL/min/1.73 m^2^ separated by 3 months or longer,[@b19-clep-10-421] calculated from serum creatinine records in the CPRD in the past 5 years, using the Chronic Kidney Disease Epidemiology Collaboration equation.[@b30-clep-10-421] RRT was characterized on the basis of diagnosis codes suggesting hemodialysis, peritoneal dialysis, or kidney transplantation. Previous research has shown that the prevalence of estimated glomerular filtration rate \<60 mL/min/1.73 m^2^ and RRT in the CPRD is similar to that estimated in a nationally representative population survey (Health Survey for England) and disease registry (UK Renal Registry), suggesting that most cases of CKD and RRT are captured in the CPRD.[@b31-clep-10-421] Covariate definition -------------------- We defined baseline characteristics of patients with and without SMI: age, sex, ethnicity, socioeconomic status (SES), country of the UK (ie, England, Northern Ireland, Scotland, and Wales) body mass index (BMI), smoking status, chronic diseases that are associated with CKD[@b21-clep-10-421] -- diabetes, hypertension, cardiovascular disease (ie, myocardial infarction, chronic heart failure, peripheral arterial disease, and stroke), urological disease (ie, vesicoureteral reflux, renal tract stone, and prostatic hypertrophy), systematic lupus erythematosus, and polycystic kidney disease -- based on relevant diagnosis codes. Based on previous studies using UK primary care data,[@b32-clep-10-421],[@b33-clep-10-421] we grouped patients with no record of ethnicity into those of white ethnicity. SES was assigned at the general practice level using quintiles of the Index of Multiple Deprivation in each country of the UK.[@b34-clep-10-421] BMI and smoking status were based on the most recent records prior to the date of study inclusion (ie, March 31, 2014). For chronic diseases, if a diagnosis code was recorded for a patient during the period between CPRD registration and the study date, then that disease was regarded as being present. Data analysis ------------- We compared baseline characteristics and overall prevalence of CKD (including patients on RRT) among the three groups: patients with SMI with and without a history of lithium prescription as well as those without SMI, using chi-squared tests. We stratified CKD prevalence by age (using 10-year age bands) and sex. We estimated the prevalence of RRT among the three groups and compared it overall by chi-squared tests. Moreover, we compared the distribution of the most recent category of RRT modality (hemodialysis, peritoneal dialysis, or kidney transplantation) between patients with and without SMI using chi-squared tests. Next, we carried out logistic regression analyses to examine the association between SMI (with and without a history of lithium use) and CKD. To understand which factors are more likely to explain the association between SMI and CKD, we adjusted step-by-step for age and sex, ethnicity, SES, BMI, smoking status, and chronic diseases associated with decline in kidney function (ie, diabetes, hypertension, cardiovascular disease, urological disease, systemic lupus erythematosus, and polycystic kidney disease).[@b21-clep-10-421] Patients with missing data of BMI and smoking status were excluded from the analysis. We took into account variations in coding and testing practices by different general practices through clustering by general practice in the logistic regression models. Finally, we restricted analysis to patients with CKD who were not on RRT and compared the recognition and management of CKD between patients with and without SMI. We used the QOF CKD indicators (QOF version 2013/14) as markers of the recognition and management of CKD in the UK primary care.[@b27-clep-10-421] We determined the proportion of CKD patients with and without SMI who had: 1) a record of a diagnostic code for CKD, using CKD codes listed in the QOF version 2013/14;[@b35-clep-10-421] 2) the most recent blood pressure measure ≤140/90 mmHg (recorded in the year before the date of the study); 3) urine test for proteinuria/albuminuria (including dipstick test) in the year before the date of the study; and 4) for those with hypertension and proteinuria (recorded any time between CPRD registration and study date), at least one prescription for a renin--angiotensin system antagonist in the past 3 months. In addition, because the National Institute for Health and Care Excellence guidance for cardiovascular disease recommends statins as primary prevention for all CKD patients regardless of serum cholesterol levels,[@b36-clep-10-421] we compared the proportion of CKD patients with and without SMI who had at least one prescription for a statin in the past 3 months. All statistical analyses were conducted using Stata 14 (Stata Corp, College Junction, TX, USA). Results ======= Of the 2,418,730 people aged 25--74 years who were registered in the CPRD for \>1 year on March 31, 2014, we identified 28,396 (1.17%) patients with SMI, including 4,295 patients with a history of lithium prescription in the CPRD and 24,101 without ([Figure 1](#f1-clep-10-421){ref-type="fig"}). In patients with SMI (both with and without a previous lithium prescription) compared to those without, the SES was generally lower and the proportions of black ethnicity, obesity (BMI ≥30 kg/m^2^), and current smokers were higher. Diabetes, hypertension, and cardiovascular diseases were more common in patients with SMI, whereas the prevalence of other diseases (urological disease, systematic lupus erythematosus, and polycystic kidney disease) was similar between patients with and without SMI ([Table 1](#t1-clep-10-421){ref-type="table"}). Overall, the prevalence of CKD was 14.64% (629/4,295) in patients with SMI and history of lithium prescription, 3.34% (805/24,101) in patients with SMI and no history of lithium prescription, and 2.09% (49,870/2,387,988) in patients without SMI (P \< 0.001). The absolute difference in CKD prevalence among the three groups increased with age in both men and women ([Figure 2](#f2-clep-10-421){ref-type="fig"}). Moreover, there was evidence (P = 0.012) that the overall prevalence of RRT was higher in patients with SMI (0.23% \[10/4, 295\] and 0.15% \[36/24, 101\] in those with and without a history of lithium prescription, respectively) than those without SMI (0.11% \[2,645/2,387,988\]). There was strong evidence that the distribution of RRT modalities was different between patients with and without SMI (P = 0.001). In patients with SMI, 50% (23/46) were receiving hemodialysis and 2% (1/46) peritoneal dialysis, and 48% (22/46) underwent kidney transplantation. However, in patients without SMI, the corresponding figures were 26% (700/2,645), 7% (194/2,645), and 66% (1,751/2,645). The age--sex-adjusted odds ratio (OR) for CKD was 7.13 (95% confidence interval \[CI\] 6.47--7.85) for patients with SMI and history of lithium use, and 1.69 (1.56--1.83) for those with SMI and no history of lithium use, compared to those without SMI. After adjusting for SES, BMI, smoking, and chronic diseases associated with CKD, the adjusted ORs decreased to 6.49 (5.84--7.21) and 1.45 (1.34--1.58) but were still significant in patients with SMI with and without previous lithium prescription, respectively ([Table 2](#t2-clep-10-421){ref-type="table"}). Among patients with biochemically defined CKD (not on RRT), a higher proportion of people with SMI had a recorded diagnosis of CKD (SMI 70.3%, without SMI 65.5%) and a recorded blood pressure ≤140/90 mmHg (SMI 80.1%, without SMI 75.6%; [Table 3](#t3-clep-10-421){ref-type="table"}). However, a lower proportion of patients with SMI had recently been prescribed statins (SMI 50.5%, without SMI 59.0%). Discussion ========== Main findings ------------- We found that, in UK primary care, patients with SMI had a greater prevalence of CKD compared to the general population. This was pronounced (6.5-fold increase) in patients with a history of lithium prescription, but there was a 1.5-fold increase in odds of CKD even among patients never known to be prescribed lithium and after adjustment for differences in known risk factors for CKD. In addition, patients with SMI had an increased prevalence of RRT and were more likely to be receiving hemodialysis as their modality than patients without SMI on RRT. Strengths and limitations ========================= There are several strengths of our study. There is likely to be a good recording of the diagnoses examined because GPs manage patients with SMI for their physical and mental health -- with or without the support of psychiatrists in secondary care. In addition, the QOF -- the reward and incentive program for GPs -- included both SMI and CKD during this time period.[@b27-clep-10-421] Moreover, the diagnosis of SMI has been validated at the individual level,[@b29-clep-10-421] and the prevalence estimates of CKD and RRT have been validated at the population level.[@b31-clep-10-421] However, we need to acknowledge several limitations. First, this is a cross-sectional study, in which temporal relationships cannot always be clarified. The majority of patients with SMI have developed symptoms by early adulthood,[@b37-clep-10-421] whereas the prevalence of CKD starts to increase after age 55.[@b31-clep-10-421] Therefore, it is unlikely that CKD precedes the onset of SMI. However, the interpretation of potential risk factors in the association between SMI and CKD requires caution. For example, cessation of smoking and lithium may be as a result of the development of CKD. Therefore, we did not differentiate previous and current users in the regression models. Second, a greater prevalence of CKD among patients with SMI may, in part, be influenced by surveillance or ascertainment bias. Patients with SMI take medications, such as lithium and other psychotropic drugs, which need regular monitoring. In addition, in 2013--2014, GPs were incentivized to monitor blood glucose and cholesterol levels for people with SMI; therefore, many patients would have had concurrent testing of renal function.[@b38-clep-10-421] In contrast, in the general population, creatinine testing in primary care is not universal: in 2013--2014, this was recommended and incentivized only for people with known CKD risk factors.[@b21-clep-10-421],[@b27-clep-10-421] However, we have previously demonstrated that the prevalence of CKD identified in the CPRD was very similar to that seen in the Health Survey for England, a nationally representative survey of the general population.[@b31-clep-10-421] This suggests that most patients with CKD are captured by the current testing strategy in primary care, and the proportion of patients with unmeasured CKD in CPRD is small. Therefore, underascertainment of CKD in people without SMI is unlikely to be a substantial contributor to our results. Third, RRT is a rare outcome: the incidence of RRT in the UK is \~100 per million population.[@b39-clep-10-421] Combined with the cross-sectional design, this meant that, despite the large sample size, the number of patients with SMI and RRT was limited (n = 46). Although there was a statistically significant difference in the prevalence of RRT between patients with and without SMI, more detailed comparison (eg, stratification by age and sex) or analysis of risk factors for RRT was not possible with these numbers. Finally, our adjusted analyses suggested that differences in the prevalence of risk factors between patients with and without SMI did not completely explain the association between SMI and CKD. This finding may be due to additional unconsidered risk factors, differences in patient management, or residual confounding. For example, antipsychotics used for SMI have been suggested to cause acute kidney injury,[@b40-clep-10-421],[@b41-clep-10-421] which is a known risk factor for subsequent CKD.[@b21-clep-10-421] Furthermore, insufficient control -- or later initiation -- of treatment for the CKD risk factors (eg, high blood pressure, diabetes, and obesity) may lead to higher incidence of CKD in the SMI population; however, this information could not be adequately captured in this cross-sectional study. Using recorded CPRD diagnoses, moreover, means that the misclassification or underidentification of disease status (eg, diabetes and heart failure) is possible. We did not have any information on the use of lithium before the CPRD registration and, therefore, some patients may have been wrongly classified as not having used lithium. This also meant that we were unable to examine the association between the length or cumulative dose of lithium prescription and CKD. Comparison with other studies ----------------------------- Few studies have examined the prevalence of CKD in the population with SMI. A small cross-sectional study in London, UK, found a similar relative difference in CKD prevalence between those with and without SMI despite a less accurate CKD definition than in the present study.[@b13-clep-10-421] A Taiwanese cohort study showed that people with schizophrenia are more likely to develop CKD,[@b25-clep-10-421] whereas a cross-sectional analysis of Scottish primary care demonstrated that people with bipolar disorder had a higher prevalence of CKD than people without the condition.[@b18-clep-10-421] Several studies have focused on the prevalence or incidence of CKD in patients using lithium. A Swedish cohort study showed that the prevalence of CKD (defined as serum creatinine level \>150 μmol/L) and ESRD requiring RRT was higher in people ever exposed to lithium than in the general population,[@b42-clep-10-421] whereas a Danish cohort study also showed that lithium prescription was associated with an increased rate of CKD diagnosis.[@b26-clep-10-421] The relative risk of lithium use for the prevalence of CKD in our study (the fully adjusted OR of 6.5) was larger than those estimated in previous studies: the adjusted hazard ratio by lithium for CKD was nearly 2 in a cohort study in Oxford, UK,[@b43-clep-10-421] and around 3 in the Danish cohort study.[@b26-clep-10-421] These differences can be explained by the different nature of the study population in each study. We compared people with lithium prescription for SMI and those without SMI in the general population, whereas the UK cohort study compared those with and without lithium use among people with at least two blood samplings in hospitals in Oxford,[@b43-clep-10-421] and the Danish study estimated the risk of lithium prescription in the population with a single manic episode or bipolar disorder.[@b26-clep-10-421] Therefore, our comparison group is healthier than in the other studies, although again, ascertainment bias may play a role to some extent in our study. The management of CKD includes prevention of cardiovascular events. Several previous studies have focused on inequalities in medication use for cardiovascular risk between patients with and without SMI. Studies from the UK showed that statin prescribing is lower in patients with schizophrenia and bipolar disorder than in the general population, although their cardiovascular risks are higher.[@b16-clep-10-421],[@b18-clep-10-421] A US study suggests that prescribing of renin--angiotensin system antagonists and statins is suboptimal in patients with SMI and type 2 diabetes.[@b17-clep-10-421] Our finding of lower prescribing of statins among patients with SMI and CKD, compared to the general population, are consistent with these earlier studies. Explanation of findings and clinical relevance ---------------------------------------------- CKD is strongly and independently associated with mortality and cardiovascular risk.[@b20-clep-10-421] Therefore, the higher prevalence of CKD we have established may contribute to the known shorter life expectancy in people with SMI. Moreover, we have shown that the difference in CKD prevalence between patients with and without SMI increased with age. It is possible that progressive accumulation and biological effect of CKD risk factors in patients with SMI (eg, obesity, smoking, and diabetes) leads to an increased incidence of CKD at an older age. We have confirmed a higher burden of CKD risk factors among patients with SMI, and adjustment for these partially explained the association between SMI and CKD. However, many questions about the cause of the higher prevalence of CKD remain unanswered. Our snapshot of GP's management of patients with SMI suggests that CKD is more commonly coded as a diagnosis than in the general population. The extent to which this is driven by incentivized management schemes and regular testing of renal function for patients with SMI -- particularly those prescribed lithium and other psychotropic medications -- is unknown. Whereas blood pressure appeared better controlled among SMI patients, this may reflect different underlying renal pathologies associated with lower rates of hypertension (eg, interstitial nephritis related to lithium). Other aspects of management -- proteinuria testing as well as prescription of statins and renin--angiotensin system antagonists -- were lower among patients with SMI. Of greatest concern is our finding of the substantial increase in the prevalence of RRT among patients with SMI. The development of ESRD results in markedly reduced quality of life and psychological stress. Renal transplantation is associated with better quality of life and, possibly, longer survival.[@b44-clep-10-421] However, our results showed that the proportion of patients receiving kidney transplantation was substantially lower in patients with SMI than those without. It is important to understand why this difference arises and ensure that there are no inappropriate barriers to the consideration for renal transplantation for people with SMI. Conclusion ========== We found that the prevalence of CKD and RRT was substantially higher in patients with SMI as compared to the general population. There was a greater burden of risk factors among patients with SMI, but these did not fully explain the increased prevalence of CKD. Further information about the management of CKD in SMI patients such as referral to specialist care and management of comorbidities is needed to identify opportunities for prevention of CKD and its progression. Masao Iwagami is supported by the Honjo International Scholarship Foundation. Laurie A Tomlinson is funded by a Wellcome Intermediate Clinical Fellowship (WT101143MA). Liam Smeeth is supported by a Wellcome Senior Research Fellowship in Clinical Science (grant number 098504/Z/12/Z). David PJ Osborn is supported by the UCLH NIHR Biomedical Research Centre and he was also in part supported by the National Institute for Health Research (NIHR) Collaboration for Leadership in Applied Health Research and Care (CLAHRC) North Thames at Bart's Health NHS Trust. The funders had no role in the execution of this study or in the interpretation of results. Disclosure The authors report no conflicts of interest in this work. ![Flow chart showing identification of study participants.\ Abbreviations: CPRD, Clinical Practice Research Datalink; SMI, severe mental illness.](clep-10-421Fig1){#f1-clep-10-421} ![Prevalence of CKD in patients with and without SMI by age in men and women.\ Abbreviations: CKD, chronic kidney disease; SMI, severe mental illness.](clep-10-421Fig2){#f2-clep-10-421} ###### Characteristics of patients by SMI diagnosis and history of lithium use Characteristic Patients without SMI diagnosis N = 2,387,988 Patients with SMI diagnosis (N = 28,396) -------------------------------------------------------------- ---------------------------------------------- ------------------------------------------ -------------- Type of SMI diagnosis Schizophrenia -- 9,134 (37.9) 448 (10.4) Bipolar disorder -- 5,881 (24.4) 3,374 (78.6) Other nonorganic psychosis -- 9,086 (37.7) 473 (11.0) Age category (years) 25--34 468,602 (19.6) 3,654 (15.2) 242 (5.6) 35--44 503,405 (21.1) 5,475 (22.7) 688 (16.0) 45--54 571,325 (23.9) 6,361 (26.4) 1,187 (27.6) 55--64 456,457 (19.1) 4,844 (20.1) 1,209 (28.2) 65--74 388,199 (16.3) 3,767 (15.6) 969 (22.6) Sex (male) 1,193,218 (50.0) 13,183 (54.7) 1,761 (41.0) Ethnicity White/unrecorded[\](#tfn1-clep-10-421){ref-type="table-fn"} 2,232,403 (93.5) 22,051 (91.5) 4,184 (97.4) Black 42,339 (1.8) 836 (3.5) 35 (0.8) South Asian 72,190 (3.0) 734 (3.1) 42 (1.0) Other 41,056 (1.7) 480 (2.0) 34 (0.8) Country* England 1,739,874 (72.9) 17,507 (72.6) 2,815 (65.5) North Ireland 92,714 (3.9) 931 (3.9) 234 (5.5) Scotland 307,936 (12.9) 3,084 (12.8) 765 (17.8) Wales 247,464 (10.4) 2,579 (10.7) 481 (11.2) Socioeconomic status 1 (least deprived) 468,867 (19.6) 3,463 (14.4) 856 (19.9) 2 496,229 (20.8) 4,295 (17.8) 864 (20.1) 3 483,617 (20.3) 4,706 (19.5) 833 (19.4) 4 508,424 (21.3) 6,084 (25.2) 948 (22.1) 5 (most deprived) 430,851 (18.0) 5,553 (23.0) 794 (18.5) Body mass index (kg/m^2^) \<18.5 37,526 (1.6) 497 (2.1) 56 (1.3) 18.5 to \<25 772,208 (32.3) 6,868 (28.5) 988 (23.0) ≥25 to \<30 716,303 (30.0) 7,480 (31.0) 1,482 (34.5) ≥30 530,065 (22.2) 8,281 (34.4) 1,724 (40.1) Missing data 331,886 (13.9) 975 (4.1) 45 (1.1) Smoking status Nonsmoker 962,061 (40.3) 5,578 (23.1) 1,123 (26.2) Current smoker 516,109 (21.6) 10,677 (44.3) 1,485 (34.6) Ex-smoker 861,414 (36.1) 7,773 (32.3) 1,685 (39.2) Missing data 48,404 (2.0) 73 (0.3) \<5 (\<0.1) Chronic diseases associated with CKD Diabetes mellitus 166,917 (7.0) 3,421 (14.2) 718 (16.7) Hypertension 682,868 (28.6) 8,826 (36.6) 2,138 (49.8) Cardiovascular disease Myocardial infarction 31,659 (1.3) 390 (1.6) 81 (1.9) Chronic heart failure 10,997 (0.5) 192 (0.8) 38 (0.9) Peripheral arterial disease 15,383 (0.6) 232 (1.0) 47 (1.1) Stroke 24,097 (1.0) 419 (1.7) 96 (2.2) Urological disease Vesicoureteral reflux 1,676 (0.1) 16 (0.1) 5 (0.1) Renal tract stone 30,665 (1.3) 266 (1.1) 41 (1.0) Prostatic hypertrophy 40,855 (1.7) 379 (1.6) 105 (2.4) Other Systemic lupus erythematosus 3,423 (0.1) 47 (0.2) 5 (0.1) Polycystic kidney disease 1,980 (0.1) 22 (0.1) \<5 (\<0.1) Note: Number of patients without recorded ethnicity was 1,141,478 (47.8%), 9,687 (40.2%), and 2,045 (47.6%), respectively. Abbreviations: CKD, chronic kidney disease; SMI, severe mental illness. ###### Adjusted logistic regression analyses of the association between SMI and chronic kidney disease Model Adjusted odds ratio (95% CI) for chronic kidney disease --------------------------------------------------------------------------------------------------- --------------------------------------------------------- ------------------- ------------------- Model 1: adjusted by age and sex 1 (ref) 1.69 (1.56--1.83) 7.13 (6.47--7.85) Model 2: Model 1 + adjusted by ethnicity, country, and socioeconomic status 1 (ref) 1.65 (1.52--1.79) 7.13 (6.48--7.86) Model 3: Model 2 + adjusted by smoking status 1 (ref) 1.60 (1.48--1.74) 6.91 (6.27--7.61) Model 4: Model 3 + adjusted by body mass index 1 (ref) 1.53 (1.42--1.66) 6.44 (5.84--7.09) Model 5: Model 4 + adjusted by diabetes 1 (ref) 1.42 (1.31--1.54) 6.23 (5.64--6.88) Model 6: Model 5 + adjusted by other chronic diseases[\](#tfn3-clep-10-421){ref-type="table-fn"} 1 (ref) 1.45 (1.34--1.58) 6.49 (5.84--7.21) Note:* Hypertension, cardiovascular disease (myocardial infarction, chronic heart failure, peripheral arterial disease, and stroke), urological disease (vesicoureteral reflux, renal tract stone, and prostatic hypertrophy), systematic lupus erythematosus, and polycystic kidney disease. Abbreviations: CI, confidence interval; ref, reference; SMI, severe mental illness. ###### Recognition and management of CKD in CKD patients with and without SMI Indicator of CKD recognition/management Patients with biochemically defined CKD (not on RRT) --------------------------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------ -------------------- 1\) Recorded diagnosis of CKD 976 (70.3) 30,922 (65.5) 2\) Most recent blood pressure ≤140/90 mmHg 1,112 (80.1) 35,705 (75.6) 3\) Prescription of renin--angiotensin system antagonists (among patients with hypertension and proteinuria diagnoses), numerator/denominator (%) 52/66 (78.8) 2,905/3,428 (84.7) 4\) Proteinuria/albuminuria testing in the past year 421 (30.3) 15,263 (32.3) 5\) Prescription of statins 701 (50.5) 27,853 (59.0) Abbreviations: CKD, chronic kidney disease; RRT, renal replacement therapy; SMI, severe mental illness.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION ============ The National Resident Matching Program (NRMP) is a private, not-for-profit corporation established in 1952 that aims to provide an impartial venue for matching applicants' and programs' preferences for each other. In 2007 the NRMP enrolled 127 emergency medicine (EM) residency programs in the match, which offered 1,288 EM positions and filled 1,282 of those positions. Of those, 1,027 were filled with United States graduates and 234 with independent applicants. A total of 1,489 applicants applied to an EM residency program. Of those, 1,105 were graduates of accredited U.S. medical schools and 384 were independent applicants.[@b1-wjem11_5p474] In its Statement on Professionalism, the NRMP outlines an expectation that all match participants conduct their affairs in an ethical and professionally responsible manner. While there are specific guidelines for certain explicit violations of match ethics listed in the Match Participation Agreement, other potential violations are less well-defined and subject to interpretation. One such prohibition concerns misleading communications. While the NRMP "permits program directors and applicants to express a high degree of interest in each other," it "prohibits statements implying a commitment." The distinction between what is an expression of interest and what implies a commitment is left up to interpretation by the program and applicant. Although statements such as "we plan to rank you very high on our list" and "we hope to have the opportunity to work with you in the coming year" are noted to be non-binding, the NRMP reports that these statements are frequently misinterpreted. Applicants are advised to not rely on them when creating rank order lists, and program directors are advised to avoid making misleading statements in their interactions with applicants.[@b2-wjem11_5p474] Previous publications suggest that violations of professional behavior in the match process may be common.[@b3-wjem11_5p474]--[@b11-wjem11_5p474] However, the frequency of misleading communications is unknown. In our experience advising students applying to EM residency programs, we have heard that there is wide variation in program practices regarding contacting applicants after the interview day but before rank lists are submitted. While some students report only rare contact by any means, other students report frequent communication by e-mail or phone. Both the variation in program practices, and the nature of communication (which often is interpreted as at least an "expression of interest"), can be confusing and anxiety-provoking for the applicant. To our knowledge, there are no published descriptions of EM residency programs' practices regarding contacting applicants after the interview day, a communication practice with the potential to be interpreted as misleading and impact applicants' decisions in creating rank order lists. We sought to describe the experiences of our applicants regarding being contacted after interview day by EM residency programs, and characterize their perceptions regarding these communications. METHODS ======= Study Design ------------ This cross-sectional study surveyed all applicants to our EM residency program in the 2006--2007 NRMP cycle. Study Setting and Population ---------------------------- This study was conducted at a three-year EM residency program that offered nine post-graduate year (PGY)-1 positions in the 2006--2007 cycle. During the study period, our program's policy was to not contact interviewees after the interview day. Invitations to participate in the study were offered by e-mail to all 706 individuals who had applied to our program. Of these applicants, 62.5% were male, and 37.5% were female. One hundred twelve applicants (16%) had interviewed at our program. To reassure applicants of anonymity and emphasize we were interested in their overall interview experience rather than their experience at our program, the survey did not ask respondents if they interviewed at our program. The geographic distribution of the 706 eligible subjects included 174 (24.6%) from the Midwest, 125 (17.7%) from the Northwest, 96 (13.6%) from the Southeast, 193 (27.3%) from the West and 118 (16.7%) foreign medical graduates. Study Protocol -------------- All applicants to our residency program were invited to participate in an anonymous and voluntary Web-based survey. A commercially available survey software program was used to administer the survey in February 2007, after rank lists were submitted but before match results were announced. Two additional invitations to participate were sent over the next month. This study was submitted to our Institutional Review Board and was exempt from formal review. Measurements ------------ The survey used a combination of multiple-choice and interval-scale type items to assess applicants' perceptions about whether they were contacted by EM residency programs after the interview day, followed by items that characterized the nature of the contact they received and their feelings about being contacted. Applicants who responded that they were not contacted were asked analogous questions assessing how they felt about not being contacted. Both groups were asked a question pertaining to their understanding of the "rules of the match." The survey also asked applicants to provide the following information: number of EM programs applied to and interviewed at; number of EM programs ranked; participation in the couples match; and ranking programs outside of EM. The survey was piloted on 10 current residents and reviewed before the study was initiated. Minor formatting and wording edits were made to avoid ambiguity and improve understandability. No significant problems were identified. Data Analysis ------------- The survey software reported descriptive statistics for data. The response rate was calculated as the number of applicants who responded to the question that referred to our primary research objective ("Were you contacted, in any way, by a member of a residency program after your interview but before rank lists were due?") divided by the number of applicants who were invited to participate. Other items were not made mandatory so that applicants who were uncomfortable or unsure answering a certain question could opt out of that question but still participate in the survey. All responses collected are reported. Percentages of respondents selecting each answer were calculated and compared between the contacted and not contacted groups. RESULTS ======= Of 706 applicants invited to participate, 240 (34%) completed the survey. Not all respondents completed each item, and therefore all percentages are reported for the total number of respondents for each item. [Table 1](#t1-wjem11_5p474){ref-type="table"} describes the type and number of programs respondents applied to, interviewed at, and ranked. Fifty-two percent (123/237) of respondents answered that they expected to be contacted by a residency program after their interview but before rank lists were due; 89% (214/240) of respondents reported actually being contacted. Fifty-six percent were contacted by 3--5 programs, with a third of applicants being contacted by more than five programs and the remainder of respondents being contacted by fewer than three. Out of those who were contacted, 91% report being contacted by e-mail; 67% by mail; and 55% by phone (some applicants were contacted by more than one method, therefore numbers do not add up to 100%). Fifty-one percent reported that being contacted changed the order of their rank list in at least one case. A majority of contacted applicants felt "happy" or "excited" about being contacted, but significant numbers reported feeling "put on the spot" (21%), or "uncomfortable" (17%). Of the 25 respondents who were not contacted after the interview day, 48% reported feeling "nervous" and 48% reported feeling "disappointed" about their lack of contact. A majority of applicants in both groups (80% of contacted applicants, 73% of not contacted applicants) believed that it is appropriate for programs to contact applicants after interview day but before the rank lists are in, but 39% of contacted subjects responded that contact by phone is either "always inappropriate" or "usually inappropriate." [Table 2](#t2-wjem11_5p474){ref-type="table"} demonstrates responses regarding being contacted. Understanding of the "rules of the match" is depicted for both groups in [Table 3](#t3-wjem11_5p474){ref-type="table"}. DISCUSSION ========== During the annual residency match, program directors are faced with the challenge of adhering to the highest professional standards while competing with other programs for the most qualified applicants. Although the NRMP does provide a framework for this process, many decisions, ranging from how to select invitees, how to rank applicants, and how to communicate with top candidates after the interview day, are left up to the individual programs. Even if the nature of communication does not violate the rules of the match, program directors may risk offending applicants with contact that is either too direct, or by not providing communication that applicants expect because they are receiving it from other programs. To avoid misleading communication, one pediatric residency program recently published their "no call policy" between interview day and the day rank lists are due, and reported that over a four-year period, 10.3% of respondents reported that a recruiting call would have caused them to rank those programs more favorably.[@b12-wjem11_5p474] Although communication after interview day is not meant to imply commitment, these data suggest that it does influence rank order lists for a subset of applicants. To our knowledge, this is the first cross-sectional investigation of EM applicants' perceptions about being contacted after the interview day but before the match. While over half of applicants reported that they expected to be contacted during this period, we were surprised to find that 89% actually were contacted. Although this represents a sample of the global pool of applicants, program directors can conclude that it is likely that a majority of their top candidates are being contacted by at least one other program after interview day. We have seen that the method and nature of this contact varies, but that in just over half of respondents, this communication did affect the placement of the program on their rank order list. This begs several follow-up questions: Are applicants flattered, and therefore ranking programs higher? Are they offended by the contact, and ranking them lower? Or do they perceive that they are not competitive because they were not contacted and thus rank the program lower in favor of other more persuasive programs? While further studies are needed to answer these questions we did find that the contact that is presently being initiated by program directors is eliciting a variety of feelings in applicants, who report feeling everything from "happy" to "uncomfortable" due to this communication. Some programs directors may opt to have a "no contact" policy, but it is important to note that nearly half of the applicants who were not contacted felt "nervous" or "disappointed." Programs may choose to address the issue directly with applicants by presenting the program policy and philosophy regarding this practice during or before the interview day. For those programs that continue to contact applicants after the interview day, decisions must also be made about who should do the contacting, and how it should be done. While most respondents felt comfortable with e-mail contact, over a third deemed phone contact inappropriate -- notable since 55% of contacted respondents reported receiving telephone communication. Similarly, few respondents rated being contacted by an interviewer or the program director as inappropriate, while a majority responded that being contacted by a resident or faculty who did not interview them is inappropriate. In light of the NRMP's prohibition of statements that imply a commitment, either on behalf of the program or applicant, we found applicants' responses regarding the rules of the match intriguing. Over three-quarters of applicants felt it was appropriate to tell programs where they would be ranked, and over one-third felt it was appropriate for programs to notify applicants of their place on the program's rank list. Although these practices may be common, they are not in compliance with the written rules of the match. Because they are not binding, these communications may also be misleading. If these results are applicable to a broader pool of EM applicants, and especially to applicants across specialties, they will be of interest to NRMP and medical school personnel charged with educating applicants about the rules of the match. Our results suggest that a decision whether to contact top applicants after the interview day, and how to go about it, is indeed a decision that may cause reactions in applicants and even affect the order of their rank list. Further studies will be helpful in determining whether these results have external validity across a greater sample of applicants, and the most appropriate and ethical means by which to conduct communication with top residency candidates. This information is needed from a practical standpoint to best use the time and resources of academic departments and from a professional standpoint to ensure that our specialty adheres to highest ethical standards during a selection process that can at times lead us to push the boundaries of our professionalism. LIMITATIONS =========== Our candidate pool may differ from the general pool of EM applicants in several ways. For example, an interviewee at our program may not be representative of candidates who consider other areas of the country or prefer a four-year program. Although our general applicant pool does represent a geographically and gender diverse population, in order to reassure applicants of their anonymity we did not collect this demographic information on respondents and therefore can not determine the ways that respondents differed from non-respondents. In addition, we opted to survey all applicants rather than only our interviewees to broaden our sample, as we only are only able to interview approximately the top 15% of our applicants. We recognized a priori that this may negatively impact our response rate, because applicants who were not offered an interview may be less interested in participating in a voluntary, anonymous study administered by our program months after they were not invited to interview. However, given the lack of any data on this topic in our field, we concluded that a more representative sample was more important than a traditionally high sample size, and our resulting relatively low sample size is a limitation of the study. CONCLUSION ========== Most EM residency applicants report being contacted by programs after the interview day but before rank lists are submitted. Applicant perceptions regarding this communication vary widely depending on the method of communication and who they are being contacted by. Although applicants think this practice is appropriate in general, over a third of subjects believe that contact by phone is inappropriate. Furthermore, half of our respondents reported that being contacted by a residency program changed that program's position on their rank list. These findings suggest that the decisions that residency programs make regarding their communication practices with applicants after the interview day are both important and may have a significant impact on the outcome of their match. Reprints available through open access at <http://escholarship.org/uc/uciem_westjem> Conflicts of Interest: By the WestJEM article submission agreement, all authors are required to disclose all affiliations, funding sources and financial or management relationships that could be perceived as potential sources of bias. The authors disclosed none. ###### Application characteristics of respondents[Δ](#tfn1-wjem11_5p474){ref-type="table-fn"} 1--5 6--10 11--15 16--20 \>20 -------------------------------------------- ------- ------- -------- -------- ------- How many EM programs did you apply to? 1.7% 4.8% 9.6% 20.0% 63.9% How many programs EM did you interview at? 10.4% 39.1% 40.0% 9.1% 1.3% How many EM programs did you rank? 14.3% 46.5% 31.3% 7.8% 0.0% YES NO ------------------------------------------------------------------------------------------------------------------------------------------------ ------ ------- Did you participate in the couples match? 7.0% 93.0% Did you rank programs in a field other than emergency medicine? (Not including combined programs)[\](#tfn2-wjem11_5p474){ref-type="table-fn"} 5.2% 94.8% Percentages were rounded to the first decimal place, and therefore not all numbers add up to 100%. This item received 229 responses. All other items in table received 230 responses. EM,* emergency medicine ###### Applicant perceptions regarding method and initiator of post-interview communication Percentages of respondents who answered that each scenario is "always" or "usually" inappropriate --------------------------------------------------------------------------------------------------- -------------- ------------- Phone contact 81/206 (39%) 9/24 (38%) E-mail contact 12/206 (6%) 6/23 (26%) Mail contact 19/206 (9%) 5/23 (22%) Being contacted by the program director 10/204 (5%) 4/23 (17%) Being contacted by someone who interviewed you, but not the program director 19/204 (9%) 3/23 (13%) Being contacted by a resident who did not interview you 93/204 (5%) 17/23 (74%) Being contacted by a faculty member who did not interview you 113/204 (6%) 17/23 (74%) Being contacted by the residency coordinator 54/203 (27%) 7/23 (30%) For this item, there were a total of 206 respondents in the "contacted" group and 24 respondents in the "not contacted" group. However, not all respondents completed each subitem. Percentages are calculated from only those respondents who completed each subitem. ###### Applicants' understanding regarding the "rules of the match" Percentages of applicants answering "true" to the following statements: ------------------------------------------------------------------------- --------------- ------------- It is acceptable for me to tell a program where I will rank them 165/206 (80%) 18/24 (75%) It is acceptable for a program to tell me where they will rank me 85/206 (41%) 8/24 (33%) It is acceptable for me to ask a program where I will be ranked 13/206 (6%) 5/24 (21%) It is acceptable for a program to ask me where I plan on ranking them 10/205 (5%) 2/24 (8%) For this item, there were a total of 206 respondents in the "contacted" group and 24 respondents in the "not contacted" group. However, not all respondents completed each subitem. Percentages are calculated from only those respondents who completed each subitem. [^1]: Supervising Section Editor: Michael Epter, DO	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-nanomaterials-10-00157} =============== The first application of solid polymer electrolytes in lithium-based batteries was pioneered by Armand and co-workers \[[@B1-nanomaterials-10-00157]\], which has inspired a series of studies in this field. Due to the safety issues of classical Li-ion batteries, solid-state lithium batteries are being revived \[[@B2-nanomaterials-10-00157],[@B3-nanomaterials-10-00157],[@B4-nanomaterials-10-00157]\]. In addition, in these solid-state batteries, Li metal can be employed as the anode because of the non-flammable nature of the solid electrolytes \[[@B5-nanomaterials-10-00157],[@B6-nanomaterials-10-00157]\]. The renewed interest in lithium solid electrolytes reflects high demands of safe and high-energy battery technologies \[[@B7-nanomaterials-10-00157],[@B8-nanomaterials-10-00157]\]. In particular, solid polymer electrolytes consisting of lithium salts and poly (ethylene oxide) (PEO) represent a unique category of lithium solid electrolytes, which are compatible with Li metal \[[@B9-nanomaterials-10-00157],[@B10-nanomaterials-10-00157]\] and suitable for building cells and strongly relaxed lithium dendrites \[[@B11-nanomaterials-10-00157],[@B12-nanomaterials-10-00157]\]. Moreover, these materials enable the construction of flexible, stretchable, compact and laminated batteries \[[@B13-nanomaterials-10-00157],[@B14-nanomaterials-10-00157],[@B15-nanomaterials-10-00157]\]. Despite their advantages, the application of PEO-based polymer electrolytes is drastically hindered by their low ionic conductivities. The PEO chains are coordinated by the lithium ions in PEO-LiX electrolytes, thus dividing Li^+^ cations from X^−^ anions \[[@B16-nanomaterials-10-00157],[@B17-nanomaterials-10-00157]\]. During the breaking/forming processes of the Li--O bonds, lithium ions transport via inter/intrachain hooping in the PEO-LiX electrolytes \[[@B18-nanomaterials-10-00157],[@B19-nanomaterials-10-00157]\]. Lithium ion transport in PEO-LiX polymer electrolytes thereby requires local relaxation and segmental rearrangement of the PEO chains, which can only be realized when the PEO-LiX polymer electrolytes are in an amorphous condition, viz. above the melting point, \~60 °C. Therefore, the PEO-LiX polymer electrolytes generally exhibit inferior ionic conductivities below 60 °C, especially at room temperature. The room-temperature conductivities of PEO-LiX polymer electrolytes are typically in a range of 10^−8^\~10^−6^ S cm^−1^ \[[@B20-nanomaterials-10-00157],[@B21-nanomaterials-10-00157],[@B22-nanomaterials-10-00157]\]. In addition, the recrystallization processes of PEO-LiX polymer electrolytes lead to a stepwise decrease of conductivities at room temperature, which gradually increases the internal resistance of the rechargeable cells and causes capacity decay. Scrosati et al. reported that the room-temperature conductivity of the PEO~8~-LiClO~4~ polymer electrolyte was decreased from \~10^−6^ S cm^−1^ to \~10^−7^ S cm^−1^ after 20 days \[[@B23-nanomaterials-10-00157]\]. Maranas et al. determined that the recrystallization process of the PEO~6~-LiClO~4~ phase needs at least three days at room temperature, which decreases the conductivities by more than two orders of magnitude \[[@B24-nanomaterials-10-00157]\]. Therefore, tremendous efforts have been devoted to suppressing the crystallization of PEO-LiX and improving and stabilizing their conductivities at ambient temperature. One common method is plasticizing the PEO-LiX via adding liquid plasticizers, i.e., organic solvents \[[@B25-nanomaterials-10-00157],[@B26-nanomaterials-10-00157]\]. The conductivities can be significantly enhanced by the addition of liquid plasticizers. Unfortunately, the mechanical performance of the polymer electrolytes will be destroyed. It should be noted that the PEO-LiX polymer electrolytes intrinsically possess inferior mechanical performance. The destroyed inferior mechanical performance of polymer electrolytes changes some of the intrinsic features. In addition, the liquid plasticizers in the PEO-LiX matrix may lead to incompatibility with the lithium metal anode, which will terminate the most important advantage of the PEO-based electrolytes. It is thereby interesting to improve the ionic conductivities of PEO-LiX polymer electrolytes, but without compromising the mechanical performance and stability with Li metal. Nanoscale oxides were found to meet this merit criterion by the pioneering work of Scrosati and co-workers \[[@B27-nanomaterials-10-00157]\]. The TiO~2~ and Al~2~O~3~ nanoparticles with a respective size of 13 nm and 5.8 nm have been used as fillers in the PEO-LiClO~4~ matrix, which enhance the conductivities from 10^−8^ S cm^−1^ to \~1.7 × 10^−5^ S cm^−1^ at ambient temperature. The recrystallization kinetics of the PEO chains after cooling from the amorphous state to room temperature is inhibited by the dispersed and large surface-area ceramic fillers, thus increasing the conductivities. Meanwhile, the mechanical performance of the polymer matrix is also improved. The effects of various ceramic nanofillers, such as SiO~2~, ZnO, ZrO~2~, LiAlO~2~, BaTiO~3~ and clays etc. in the electrical performance of polymer electrolytes have been investigated \[[@B28-nanomaterials-10-00157],[@B29-nanomaterials-10-00157],[@B30-nanomaterials-10-00157],[@B31-nanomaterials-10-00157],[@B32-nanomaterials-10-00157]\]. Silicon dioxide, as a ceramic nanofiller, plays a vital role in optimizing the property of materials \[[@B33-nanomaterials-10-00157]\]. In our previous work, we observed that 10 wt.% SiO~2~ (5--10 nm particle size) increases the room-temperature conductivity of PEO-LiClO~4~ to \~10^−5^ S cm^−1^. However, a stepwise decrease of conductivity to \~10^−6^ S cm^−1^ has been observed after one week, indicating that the suppression on the recrystallization process is incomplete \[[@B34-nanomaterials-10-00157]\]. Besides inert fillers, ceramic electrolytes including garnet-type Li~7~La~3~Zr~2~O~12~, nasicon-type Li~1.5~Al~0.5~Ge~1.5~(PO~4~)~3~, perovskite-type Li~0.33~La~0.55~TiO~3~ and sulfide Li~10~GeP~2~S~12~ have been used as active fillers for the PEO-LiX polymer electrolytes to enhance their conductivities \[[@B35-nanomaterials-10-00157],[@B36-nanomaterials-10-00157],[@B37-nanomaterials-10-00157],[@B38-nanomaterials-10-00157],[@B39-nanomaterials-10-00157]\]. The inert and active ceramic nanoparticle fillers added through preformed ceramics may result in agglomeration of nanoparticles, heterogeneous dispersion of fillers, and weak interactions between PEO chains and ceramics, hindering further enhancement of the conductivities of PEO-LiX polymer electrolytes. Cui et al. reported an aqueous hydrolysis preparation of the PEO-LiClO~4~-SiO~2~ polymer electrolyte with an enhanced conductivity of 4.4 × 10^−5^ S cm^−1^ at 30 °C \[[@B40-nanomaterials-10-00157]\]. Herein, we propose an alternative route to process the nanocomposite polymer electrolyte membranes based on the in situ non-hydrolytic sol gel reaction. Homogeneous dispersion of SiO~2~ nanoparticles and robust chemical interaction between the SiO~2~ nanoparticles and PEO are demonstrated. The in situ synthesized nanocomposite polymer electrolyte membrane achieves an improved conductivity of 1.1 × 10^−4^ S cm^−1^ at 30 °C, which is two orders of magnitude higher than that of the preformed synthesized composite polymer electrolyte. In addition, an extended electrochemical stability voltage window of up to 5 V vs. Li/Li^+^ is achieved. By using this reliable nanocomposite polymer electrolyte membrane, the Li/nanocomposite polymer electrolyte/Li symmetric cell demonstrated a stable long-term cycling performance of over 700 h at 0.01--0.1 mA cm^−2^ without short circuiting, and the high-energy solid lithium cell with Li metal paired with LiFePO~4~ cathode exhibited a reversible capacity of 123.5 mAh g^−1^ at 0.1 C and 55 °C. 2. Experiment {#sec2-nanomaterials-10-00157} ============= 2.1. Chemical Reagents and Materials {#sec2dot1-nanomaterials-10-00157} ------------------------------------ Polyethylene oxide (PEO, M~n~ = 1,000,000 g mol^−1^, Aladdin, Shanghai, China) was dehydrated at 60 °C for 24 h in vacuum. Lithium perchlorate (LiClO~4~, anhydrous, Aladdin, Shanghai, China) and lithium bis(trifluoromethanesulfonyl)imide (LiTFSI, Aladdin, Shanghai, China) were dried at 120 °C in vacuum for 24 h. The as-dried PEO, LiClO~4~, and LiTFSI were stored in an Ar-filled glove box before use. Tetraethoxysilane (TEOS, Aladdin, Shanghai, China), formic acid (FA, Aladdin, Shanghai, China), N,N-dimethylformamide (DMF, Aladdin, Shanghai, China), and acetonitrile (anhydrous, Aladdin, Shanghai, China) were used as-received. Lithium iron phosphate (LiFePO~4~, MTI Co. Ltd., Shenzhen, China) and carbon black (Super P, MTI Co. Ltd., Shenzhen, China) were dried at 120 °C in vacuum for 24 h before use. Lithium metal strips (China Energy Lithium Co. Ltd., Tianjin, China) were stored in an Ar-filled glove box and used without any surface treatment or modification. 2.2. Preparation of Electrolyte Membranes {#sec2dot2-nanomaterials-10-00157} ----------------------------------------- PEO and LiClO~4~ were mixed at the ether-oxygen-to-lithium ratio of 10:1, and dissolved in DMF to \~10% weight at 60 °C. A measured amount of TEOS, based on the complete conversion of TEOS to SiO~2~, was then added to the PEO-LiClO~4~ solution (weight of SiO~2~ equaled to 10 wt.% of total weight of PEO and LiClO~4~). FA was slowly added with vigorous stirring to the above clear solution (the mole ratio FA/TEOS = 7.8/1) to initiate the gelation. Afterwards, the solution was stirred for 12 h, outgassed under partial vacuum, cast onto a PTFE substrate, and dried firstly in air and subsequently in vacuum at 60 °C for 24 h. The processes yielded free-standing, transparent and flexible nanocomposite polymer electrolyte membranes. The membranes were kept in the inert glovebox for at least one week before characterization. 2.3. Characterizations of Electrolyte Membranes {#sec2dot3-nanomaterials-10-00157} ----------------------------------------------- The microstructure of the nanocomposite polymer electrolyte membrane was examined on the surfaces of the membranes via scanning electron microscope (SEM) (Hitachi, Su8020, Tokyo, Japan). The distribution of the elements was detected via energy-dispersive spectroscopy (EDS) (HORIBA, EX250, Kyoto, Japan). The interactions of the in situ composite polymer membrane were evaluated via X-ray photoelectron spectroscopy (XPS) (Axis Ultra DLD, Shimadzu, Kyoto, Japan). The ionic conductivities of the membranes were determined via electrochemical impedance spectroscopy (EIS) on an Autolab PGSTAT302N (Metrohm, Herisau, Switzerland) under frequencies from 1 MHz to 0.1 Hz and amplitude of 10 mV at room temperature in a CR2025 coin cell consisting of the given membrane and stainless steel blocking electrodes. The conductivity--temperature curves for the membranes were obtained at the temperature range from room temperature to 100 °C. The voltage of the membranes was determined by linear sweep voltammetry (LSV, Autolab PGSTAT302N, Herisau, Switzerland) at a sweep rate of 0.1 mV s^−1^. The membranes were sandwiched by a lithium strip and stainless steel in a CR2025 coin cell. The Li plating/stripping process was studied in a Li/polymer membrane/Li symmetric cell. Li plating/stripping was performed at variable current densities from 0.01 to 0.1 mA cm^−2^ (2 h for each plating/stripping cycle) at 55 °C. To build the solid-state lithium cell, Li metal anode and LiFePO~4~ cathode were paired with the nanocomposite polymer electrolyte membrane. The LiFePO~4~ cathode was prepared in a similar manner as the previous solid-state Li metal battery \[[@B35-nanomaterials-10-00157]\]. Briefly, LiFePO~4~, Super P, PEO and LiTFSI were dispersed in acetonitrile and ball milled at 300 rpm for 24 h at a weight ratio of LiFePO~4~:Super P:PEO:LiTFSI = 60:10:20:10. The composite cathode with a LiFePO~4~ loading of \~1.0 mg cm^−2^ was obtained after casting the resulting slurry on Al foil followed by vacuum drying at 60 °C for 24 h. The charge-discharge characterization was performed in a voltage range of 2--4 V under the constant current modes of 0.1 C and 0.2 C (1 C = 170 mA g^−1^) at 55 °C. 3. Results and Discussion {#sec3-nanomaterials-10-00157} ========================= 3.1. Physicochemical Properties of the Nanocomposite Polymer Electrolyte Membranes {#sec3dot1-nanomaterials-10-00157} ---------------------------------------------------------------------------------- [Figure 1](#nanomaterials-10-00157-f001){ref-type="fig"} schematically shows the in situ synthesis of SiO~2~ nanoparticle-filled PEO-LiClO~4~ nanocomposite polymer electrolyte membranes. We prepared the nanocomposite polymer electrolytes through a classical non-hydrolytic sol gel reaction \[[@B41-nanomaterials-10-00157]\]. This reaction scheme was first used with a simple silicon alkoxide, i.e., TEOS. FA was used as a catalyst to create the oxide nanofillers. Unlike other hydrolysis reactions, this method is well-suited for the synthesis of oxide nanoparticles without changing the pH of the solution. The reactive by-products only included trace ethanol and water, but not deionized water, which eliminates the need for the time-consuming and energy-intensive drying process \[[@B40-nanomaterials-10-00157],[@B42-nanomaterials-10-00157]\]. Moreover, TEOS, FA, LiClO~4~ and PEO are soluble in a single solvent, such as DMF, at elevated temperatures above the melting point of PEO, such as 60 °C, thereby leading to the infinite miscibility of the precursor fillers in the PEO network and homogeneous distribution of the nanofillers in the PEO matrix. At this point, a semi-transparent and stable resin was obtained after sufficient stirring, and cast on a PTFE sheet. A free-standing, transparent and flexible nanocomposite polymer electrolyte membrane was obtained after a careful drying process. It should be noted that the as-prepared membranes were stored in an inert glovebox for at least one week before any characterization to remove trace water and complete the recrystallization dynamics. Liu and Lin et al. previously described similar materials, viz. PEO-LiTf/LiClO~4~-SiO~2~ composites \[[@B40-nanomaterials-10-00157],[@B43-nanomaterials-10-00157]\]. PEO has many advantages, such as a high capability of dissolving lithium salts, good stability towards lithium metal anode, easy processability, low toxicity, and low cost etc. These advantages make PEO a popular material system. Our work is different from their work. Liu et al. developed the PEO-SiO~2~ composite polymer electrolyte through simultaneous formation of the polymer matrix and the inorganic particles. Their work emphasized the polymer matrix formed by ultraviolet irradiation of a PEO macromer concurrent with addition of SiO~2~ nanoparticles. The synthesis process was relatively complicated, the SiO~2~ nanoparticles were not fully in situ addition, and the polymerization reaction was also not easy to control. By contrast, in our work, the SiO~2~ nanoparticles were in situ formed in the PEO matrix, and high-molecular-weight PEO (1,000,000 g mol^−1^) was selected. In both Liu and Lin's work, the formation of SiO~2~ nanoparticles were via a hydrolysis reaction (addition of TEOS to deionized water). On the contrary, we prepared the nanocomposite polymer electrolytes through a non-hydrolytic sol gel reaction. The reactive by-products only included trace ethanol and water, but not deionized water, which eliminates the need for the time-consuming and energy-intensive drying process. We think these are the advantages of our approach over other in situ approaches for SiO~2~ in PEO. [Figure 2](#nanomaterials-10-00157-f002){ref-type="fig"}a shows the SEM image of the nanocomposite polymer electrolyte membrane. The membrane displays a glass-like morphology, which is different from the spherulitic morphologies of typical PEO-LiX polymer electrolytes \[[@B44-nanomaterials-10-00157]\]. The surface of the present membrane is smooth compared with the rough surfaces of pure PEO-LiX polymer electrolytes, indicating the membrane has an amorphous structure \[[@B45-nanomaterials-10-00157]\]. SiO~2~ nanoparticles were not observed in the SEM image, probably because the SiO~2~ nanoparticles are embedded in the PEO matrix and the resolution of the SiO~2~ nanoparticles is obstructed by the polymer. To study whether the SiO~2~ nanoparticles are distributed homogeneously in the PEO matrix, EDS analysis was conducted on elements C, O, Si and Cl, which represent PEO, SiO~2~ nanoparticles and LiClO~4~, respectively. As shown in [Figure 2](#nanomaterials-10-00157-f002){ref-type="fig"}b, the EDS spectra confirm the homogeneous distribution of SiO~2~ nanoparticles in the PEO matrix. By contrast, the membrane prepared by directly mixing preformed SiO~2~ nanoparticles exhibited heterogeneous morphologies with aggregation and inhomogeneous distribution of SiO~2~ nanoparticles \[[@B34-nanomaterials-10-00157],[@B46-nanomaterials-10-00157],[@B47-nanomaterials-10-00157]\]. 3.2. Ionic Conductivity and Electrochemical Stability Window {#sec3dot2-nanomaterials-10-00157} ------------------------------------------------------------ As shown in [Figure 3](#nanomaterials-10-00157-f003){ref-type="fig"}a, the EIS of the nanocomposite polymer electrolyte membrane (in situ membrane) was acquired to determine the ionic conductivity at 30 °C. The ionic conductivities of the composite polymer electrolyte (ex situ membrane) and ceramic-free polymer electrolyte (ceramic-free membrane) membranes were provided as a comparison. It was found in [Figure 3](#nanomaterials-10-00157-f003){ref-type="fig"}a that the EIS of the membranes was similar. The EIS consists of an incomplete high-frequency semicircle, which represents the ionic impedance of the polymer electrolytes, and a straight-line in the low-frequency region, which is ascribed to the bulk effect of the blocking electrodes. The ionic conductivity of the in situ membrane was around \~1.1 × 10^−4^ S cm^−1^ at 30 °C; whereas, the ex situ membrane and ceramic-free membrane exhibited ionic conductivities of 9.1 × 10^−7^ S cm^−1^ and 7.1 × 10^−8^ S cm^−1^ at 30 °C, respectively, which agrees well with previous results \[[@B40-nanomaterials-10-00157]\]. In particular, the in situ non-hydrolytic sol gel reaction significantly enhanced the ionic conductivity by approximately two orders of magnitude compared with that of the ex situ membrane and three orders of magnitude compared with that of the ceramic-free membrane. [Figure 3](#nanomaterials-10-00157-f003){ref-type="fig"}b compares the Arrhenius plot of the in situ membrane with those of the ex situ and ceramic-free membranes from 30 °C to 110 °C. The heating procedure of the ceramic-free membrane displayed a break at about 60 °C, demonstrating the typical phase transition of PEO from crystalline to the amorphous form. On the other hand, the ex situ membrane showed a slightly different trend for the heating procedure. The conductivity break was also observed but decreased to 50 °C, demonstrating that PEO crystallization was alleviated by the addition of the ceramic filler. Interestingly, the break was almost disappeared during the heating procedure of the in situ membrane, demonstrating the PEO crystallization was almost completely suppressed by the in situ non-hydrolytic sol gel reaction. In addition, the cooling cycle exhibited a distinct trend compared with the heating cycle. The conductivity trend of the ceramic-free and ex situ membranes was reproduced in the cooling cycle only at above the break temperature. The conductivities of the ceramic-free and ex situ membranes were around 10^−5^ S cm^−1^ when the membranes were cooled to 30 °C, while the conductivities were 10^−7^ S cm^−1^ and 10^−6^ S cm^−1^ for the starting stage, respectively. This is because the crystalline PEO is changed to the amorphous state when the temperature is above the melting point of PEO. When cooled below the melting point, the amorphous PEO does not have enough time to recrystallize due to the slow recrystallization kinetics, resulting in conductivity changes. For the in situ membrane, since the in situ non-hydrolytic sol gel reaction can markedly prevent the crystallization of PEO, the heating and cooling procedures had little effect on the phase transition of PEO. Therefore, the conductivity of the in situ membrane does not have significant changes during the heating and cooling cycles. Furthermore, the conductivity of the in situ membrane at ambient temperature stayed stable (approximately 7 × 10^−5^ S cm^−1^), as shown in [Figure 3](#nanomaterials-10-00157-f003){ref-type="fig"}c, where the conductivity is plotted as a function of time at ambient temperature. It is well-known that the recrystallization process of PEO-LiX polymer electrolytes can lead to a significant decrease of conductivities, i.e., one to two orders of magnitude, restricting the application of PEO for solid-state batteries, especially at ambient temperature \[[@B34-nanomaterials-10-00157]\]. The inhibitory effect of the in situ non-hydrolytic sol gel reaction on the recrystallization dynamics of PEO provides a promising application for PEO-based polymer electrolytes at ambient temperature. High-energy batteries require the electrolyte to possess a wide electrochemical stability voltage window. We showed that the in situ non-hydrolytic sol gel reaction could produce high-voltage electrolytes. The electrochemical stability window of the in situ, ex situ and ceramic-free membranes was determined via LSV from 2 V to 6.5 V at 1 mV s^−1^. The stainless steel and lithium metal serve as working and reference electrodes, respectively. As shown in [Figure 3](#nanomaterials-10-00157-f003){ref-type="fig"}d, a low current was observed until 3.5 V vs. Li/Li^+^ for the ceramic-free membrane, which represents the oxidation process and decomposition of PEO \[[@B48-nanomaterials-10-00157]\]. A stable current could be extended to 4.5 V vs. Li/Li^+^ for the ex situ membrane, implying the positive effect of nanofillers on the electrochemical stability voltage window of polymer electrolytes. Notably, the in situ reaction intensified this positive effect with an enhanced voltage window of up to 5 V vs. Li/Li^+^, which is comparable with those previously reported high-voltage solid polymer electrolytes. 3.3. Interactions Between Polymer and Nanofiller {#sec3dot3-nanomaterials-10-00157} ------------------------------------------------ This work presents a nanocomposite polymer electrolyte membrane obtained from the in situ non-hydrolytic sol gel reaction of TEOS in the PEO matrix which is believed to trigger the interactions between the polymer and nanofillers. To support this hypothesis, we characterized the nanocomposite polymer electrolyte membrane via XPS. As seen in [Figure 4](#nanomaterials-10-00157-f004){ref-type="fig"}, the XPS spectra can be properly deconvoluted into corresponding components according to different energy states of these elements. It is notable that the spectra of the sample differ from those of the individual elements. For example, the C 1s spectrum shows binding energies of carbon around 284.9 and 286.5 eV, which are related to the C atoms of PEO. However, they are different from those of the individual PEO \[[@B49-nanomaterials-10-00157],[@B50-nanomaterials-10-00157]\]. In addition, the binding energies of carbon at approximately 288.4 and 289.2 eV correspond to the carbonyl group, indicating the reactions between formic acid and PEO or SiO~2~ \[[@B51-nanomaterials-10-00157],[@B52-nanomaterials-10-00157]\]. More importantly, the Si 2p spectrum shows the binding energy of Si around 103.9 and 102.1 eV, which can be assigned to the typical Si of SiO~2~ and siloxanes, respectively \[[@B53-nanomaterials-10-00157],[@B54-nanomaterials-10-00157]\]. These results indicate that the hydroxyl groups of the PEO chains are chemically binding with the SiO~2~ nanoparticles during the non-hydrolytic process. The Li 1s, Cl 2p and O 1s spectra show peaks with binding energies of 55.8, 207.8, 209.4 and 532.7 eV, corresponding to the lithium and chlorine atoms in LiClO~4~ and oxygen atoms in PEO, respectively \[[@B49-nanomaterials-10-00157]\]. Therefore, the results reveal the combined structures of the nanocomposite polymer electrolyte under discussion. 3.4. Electrochemical Properties {#sec3dot4-nanomaterials-10-00157} ------------------------------- Obstruction of lithium dendrite growth is a key requirement for the development of lithium metal batteries. The stability of the in situ membrane towards lithium metal was demonstrated by cycling the Li/in situ membrane/Li symmetric cell at 55 °C. [Figure 5](#nanomaterials-10-00157-f005){ref-type="fig"} shows the galvanostatic cycling. The current density was first increased from 0.01 to 0.05 mA cm^−2^, then reduced back to 0.01 mA cm^−2^ with three loops, and subsequently boosted to 0.1 mA cm^−2^. Notably, the voltage overpotential was reproduced with varying current densities, that is 10 mV versus 0.01 mA cm^−2^, 50 mV versus 0.05 mA cm^−2^, 10 mV versus 0.01 mA cm^−2^, 50 mV versus 0.05 mA cm^−2^, 10 mV versus 0.01 mA cm^−2^, and 50 mV versus 0.05 mA cm^−2^. The results indicate the interface contact was stable. The long cycling property was further demonstrated at 0.1 mA cm^−2^ for 400 h, indicating that a good compatibility is achieved between the nanocomposite polymer electrolyte and Li metal anode. To test the nanocomposite polymer electrolyte membrane, a solid battery was built with LiFePO~4~ as the cathode, the in situ membrane as the electrolyte, and the Li metal as the anode. The battery was charged to 4 V and discharged to 2 V at 55 °C. As shown in [Figure 6](#nanomaterials-10-00157-f006){ref-type="fig"}a, the cell exhibited a typical potential plateau of 3.38 V and 3.47 V, representing the discharge/charge potential plateau of LiFePO~4~ at 0.1 C, respectively. The initial discharge capacity was 123.5 mAh g^−1^, which is 72.6% of the theoretical capacity (170 mAh g^−1^). The relatively low reversible capacity may be a result of the low conductivity of the electrolyte membrane and the large interfacial resistances. As shown in [Figure 6](#nanomaterials-10-00157-f006){ref-type="fig"}b, the specific capacity of Li/in situ membrane/LiFePO~4~ cells increased slightly during the initial 10--20 cycles, due mainly to the improved contact and conductivity between PEO, filler and active material after repetitive Li-ion diffusion \[[@B55-nanomaterials-10-00157],[@B56-nanomaterials-10-00157]\]. In the subsequent cycles at 0.2 C, the Li/in situ membrane/LiFePO~4~ cell suffered from a progressive decrease in the specific capacity. The discharge capacity was 81 mAh g^−1^ after 90 cycles at 0.2 C, corresponding to 70% of the initial capacity at 0.2 C. Other groups also reported capacity decays upon cycling for the Li/LiFePO~4~ cells using PEO-based composite polymer electrolytes, probably owing to the large volume change and destructive interfacial contact \[[@B38-nanomaterials-10-00157]\]. The Coulombic efficiency of the battery was kept at \>99.0% throughout the cycling. 4. Conclusions {#sec4-nanomaterials-10-00157} ============== In summary, we developed a new method for the in situ non-hydrolytic sol gel synthesis method for the preparation of nanocomposite polymer electrolytes. Highly uniformly dispersed SiO~2~ nanoparticles in the PEO matrix were obtained because of the infinite miscibility of all precursors. Moreover, robust chemical interactions between the SiO~2~ nanoparticles and PEO chains were triggered by the in situ non-hydrolytic sol gel reaction. As a consequence, the developed nanocomposite polymer electrolyte membrane showed an enhanced ionic conductivity of \~1.1 × 10^−4^ S cm^−1^ at 30 °C, which is two orders of magnitude higher than that of the preformed synthesized composite polymer electrolyte. Moreover, the electrochemical window was distinctly extended to 5 V from 3.5 V vs. Li/Li^+^. The Li/nanocomposite polymer electrolyte/Li symmetric cell demonstrated a stable long-term cycling performance of over 700 h at 0.01--0.1 mA cm^−2^ without short circuiting. The all-solid-state battery consisting of the nanocomposite polymer electrolyte, Li metal anode and LiFePO~4~ cathode possessed a high discharge capacity of 123.5 mAh g^−1^, a Coulombic efficiency of above 99%, a good capacity retention of 70% after 10 cycles at 0.1 C and 90 cycles at 0.2 C. The in situ synthesized nanocomposite polymer electrolyte membrane with greatly enhanced electrochemical performance provides a promising solution for developing safe and high energy-density lithium metal batteries. S.S. and N.H. conceived the ideas. S.V.S. performed and analysed the characterization of XPS. X.T. performed the synthesis and characterization of materials and batteries. Y.W., W.T., J.Y., Z.W., L.L. and J.M. analysed the electrochemical performance of the materials and batteries. X.T., S.S. and N.H. analysed the data and wrote the paper. All authors have read and agree to the published version of the manuscript. This research was funded by National Natural Science Foundation of China (No. 51702030, No. 11632004, and No. U1864208), Shanghai Aerospace Science and Technology Innovation Foundation (No. SAST2017-137), Key Research and Development Project of Chongqing (No. cstc2017zdcy-zdyfX0060), Key Program for International Science and Technology Cooperation Projects of the Chinese Ministry of Science and Technology (No. 2016YFE0125900), and National Science and Technology Major Project (2017-VII-0011-0106). The authors declare no conflict of interest. ![Schematic of the in situ synthesis of the nanocomposite polymer electrolyte membrane.](nanomaterials-10-00157-g001){#nanomaterials-10-00157-f001} ![(a) Scanning electron microscope (SEM) image. (b) Energy-dispersive spectroscopy (EDS) elemental mapping images of the nanocomposite polymer electrolyte membrane. (C, O, Si, Cl) EDS elemental map of the membrane showing distribution of C, O, Si, Cl, respectively.](nanomaterials-10-00157-g002){#nanomaterials-10-00157-f002} ![(a) Alternating current (AC) impedance plots of the nanocomposite polymer electrolyte membranes. The inset is an enlarged image. (b) Conductivity as a function of temperature. (c) Conductivities with time evolution plots. (d) Current--potential curve.](nanomaterials-10-00157-g003){#nanomaterials-10-00157-f003} ![X-ray photoelectron spectroscopy (XPS) spectra of the in situ synthesized nanocomposite polymer electrolyte membrane: (a) XPS spectra, (b) C 1s spectrum, (c) O 1s spectrum, (d) Li 1s spectrum, (e) Cl 2p spectrum, and (f) Si 2p spectrum.](nanomaterials-10-00157-g004){#nanomaterials-10-00157-f004} ![Galvanostatic cycling measurements of Li/in situ membrane/Li symmetrical cells at various current densities and 55 °C.](nanomaterials-10-00157-g005){#nanomaterials-10-00157-f005} ![Electrochemical performance of Li metal battery Li/in situ membrane/LiFePO~4~ at 0.1 C/0.2 C and 55 °C: (a) The corresponding charge/discharge profiles at the 1st, 3rd and 5th cycles. (b) Cycling performance.](nanomaterials-10-00157-g006){#nanomaterials-10-00157-f006}	{ "pile_set_name": "PubMed Central" }
The University of Tennessee (UT), the Oak Ridge National Laboratory (ORNL), and the Kentucky Biomedical Research Infrastructure Network (KBRIN), have collaborated over the past decade to share research and educational expertise in bioinformatics. One result of this collaboration is the joint sponsorship of an annual regional summit to bring together researchers, educators and students who are interested in bioinformatics from a variety of research and educational institutions. This summit provides unique opportunities for collaboration and forging links between members of the various institutions. This year, the Ninth Annual UT-ORNL-KBRIN Bioinformatics Summit was held at Lake Barkley State Park in western Kentucky, from March 19 to 21, 2010. A total of 269 participants pre-registered for the summit, with 150 from various Tennessee institutions and 93 from various Kentucky institutions. A number of additional participants came from universities and research institutions from other states and countries, e.g. Emory University, University of British Columbia, University of Cincinnati, Iowa State University, etc. Eighty-one registrants were faculty, with an additional 94 students, 63 staff, and 31 postdoctoral participants. The conference program consisted of three days of presentations. The first day included a pre-summit of talks by researchers from Kentucky institutions, two workshops on the use of Bioconductor for DNA microarray data analysis, and a session on Next Generation Sequencing. The next two days were dedicated to scientific presentations divided into two primary sessions on Epigenetics and Medical Informatics. Pre-Summit Kentucky bioinformatics session ========================================== Dr. Nigel Cooper, the project director of KBRIN, started the pre-summit Kentucky Bioinformatics session with a summary of Bioinformatics in the state of Kentucky. Dr. Cooper reviewed the history, development and current state of bioinformatics programs in Kentucky institutions, as well as plans for the future. The presentation highlighted the many sources of funding providing support for the various bioinformatics initiatives, the undergraduate, graduate, and post-graduate programs currently underway, as well as physical resources available to bioinformatics researchers in Kentucky. Dr. Cooper also emphasized the need to have institutional support in the grant application process both for continuing grant support and in applying for larger grants to allow bioinformatics research to move forward in the future. The remaining talks of the pre-summit session highlighted current bioinformatics research from faculty at various Kentucky institutions. The subjects covered included sequence analysis, next generation sequencing, transcriptomics, proteomics and metabolomics. In the first research talk, Dr. Eric Rouchka of the University of Louisville, used sequence analysis of the 5\' untranslated region (UTR) of 18000 mouse transcripts to determine those that may be translationally regulated by CPEB1 \[[@B1]\]. Regulation of gene expression at the translational level is important because it allows the cell to create a pool of mRNA transcript that can be quickly translated to protein without the need to start transcription. This work required a model of the CPEB1 binding site, and a scoring method that considered how many binding sites the 5\' UTR contained in addition to how well the sites matched the consensus sequence. Of the 18000 transcripts examined, 1200 were found with a high score and three binding sites. Ontological analysis of those genes determined a subset that appears to be functionally related. The next talk, by Dr. Arnold Stromberg from Kentucky University, highlighted a rather unusual situation in DNA microarray analysis, having only one sample available for each treatment under investigation. In DNA microarray studies, with few exceptions, current practice is to have multiple samples for each treatment under investigation. However, in this study multiple samples were not possible. To answer the biological questions being considered, namely which samples were most similar and which genes show differential expression between two of the samples, Dr. Stromberg examined the various comparisons between pairs of samples, as well as the expression patterns across all four samples. Two important considerations emerged in this analysis, that of false positives, and the determination of genes that are truly \"unexpressed\". Dr. Ted Kalbfleisch from the University of Louisville presented his work on assessing the prevalence of the LINE-1 retrotransposon using Next Generation Sequence (NGS) data. The long interspersed nuclear element 1 (LINE-1) is a repeat sequence that has been found pervasively throughout the genome. The LINE-1 element is important because it has the ability to re-insert copies of itself into the genome, thereby causing changes in gene regulation, overall expression, or generating splice isoforms of a gene. This action has been previously documented as the causative agent in some diseases. However, the true prevalence of LINE-1 in the human genome is unknown, and until the advent of NGS was very difficult to assess. However, even with the availability of NGS data, determining the frequency of this element in the genome is difficult due to its repeatability, the very thing that is normally removed from consideration during sequence assembly. Using short portions of the known LINE-1 sequence, Dr. Kalbfleisch discovered at least 22 insertions that are currently being verified in the lab. Future work includes developing an assay to determine a comprehensive LINE-1 insertion profile, as well as the screening of large populations to fully characterize the variation present in the human population. Dr. Susmita Datta presented her work regarding an improved automated method for detecting monoisotopic peaks in MS data. Automated peak detection in MS data analysis is important because of the continued use of various MS methods in both metabolomics and proteomics, however the amount of data generated in any given MS run is overwhelming to be able to analyze by hand, and current peak detection methods mistakenly identify peaks due to many different factors, including the presence of molecules with overlapping mass profiles. To allow detection of the peaks, Dr. Datta modelled the isotopic distribution of the peptides using a mixture of location shifted Poisson distributions, and an EM algorithm to determine the weighting parameter of the distributions. Statistical analysis and comparison with the LIMPIC method showed improvement in the number of detected peaks and their identification. Dr. Ryan Gill from the University of Louisville presented his work on the determination of differences in gene networks between two biological samples. Traditional DNA microarray analysis focuses on generating lists of differentially expressed genes; however there may be greater significance in identifying how gene association networks change between conditions. To determine if networks have changed between conditions, however; one must be able to statistically differentiate between two networks. Dr. Gill focused on the methods used to determine differential connectivity of a single gene or class of genes, and differences in modular structure \[[@B2]\]. Particularly important in this work was the determination of the sensitivity of the differential score to the network construction parameters, as these will influence the final networks obtained from the underlying data. Using data from an experiment designed to determine the underlying genetic changes between fat and lean mice, Dr. Gill demonstrated the results and effectiveness of the differential measure developed. Dr. Xiang Zhang of the University of Louisville presented work on developing a metabolic pipeline geared towards solving three challenges in metabolomics bioinformatics: metabolite identification, metabolite quantification, and the visualization of metabolite networks. In contrast to many other methods, Dr. Zhang\'s pipeline uses an in-silicomethod to predict the gas chromatographic retention times of the possible metabolites. This is followed by a peak alignment method that considers both dimensions of separation and the spectra of the molecules to allow peak alignment over different temperature ranges in the GC column. Using a combination of significance scores from different statistical tests, those molecules that are regulated by the process under investigation are then determined. The inter-molecular correlations can then be visualized using the SysNet \[[@B3]\] software. This workflow has provided Dr. Zhang and colleagues with many interesting insights into various biological problems. The final talk of the pre-summit session was presented by Dr. Guy Brock on the biological impact of missing value imputation on the downstream analysis of DNA microarray data \[[@B4]\]. DNA microarrays frequently have missing expression values, and a common solution to this problem is to impute values using different methods. Although many different studies have examined the problem of missing value imputation, this study was unique in that it examined the ability of the imputation methods to recreate the original data, performance measures used to determine which method should be used, impact of the methods on the down-stream analysis using three different biological impact measures. Bioconductor workshops ====================== Dr. Guy Brock began the official summit program with two workshops on the use of R and Bioconductor \[[@B5]\] for bioinformatics data analysis. The first workshop started by providing a basic introduction to the R environment, as well as an overview of the philosophy and capabilities of R and the Bioconductor suite. This was followed by an introduction to the ExpressionSet data structure and methods used to interact with the underlying data. Using the ALL expression data set available in Bioconductor, attendees were walked through a typical data analysis workflow to determine which genes are differentially expressed in a DNA microarray experiment. The second workshop continued by introducing some of the tools available to the researcher who has a list of differentially expressed genes and is searching for biological commonalities among them. These include examining gene annotations such as gene names and the Gene Ontology. Dr. Brock also reviewed the use of clValid \[[@B6]\], a recently released R package for cluster validation, which can be used to determine the validity of gene clustering results based on various measures. Finally, the use of Gene Set Enrichment Analysis (GSEA) to determine the presence of over-represented biological functions or pathways was reviewed. Next generation sequencing ========================== The Next Generation Sequencing (NGS) session was kicked off by Dr. Ted Kalbfleisch from the University of Louisville with a thorough introduction to this rapidly developing field. Dr. Kalbfleisch reviewed DNA sequencing methods from the early days of Maxam-Gilbert sequencing through the current slate of Next Generation sequencers such as the 454 and SoLiD platforms. He also reviewed the various software tools used to perform alignment and raw assembly with commentary on some of the pros and cons associated with each method, in addition to the many applications for which NGS is being used. His presentation ended with a discussion of a particular examination of genetic variation of the Line-1 retrotransposon from NGS data that is publicly available. This project required aligning known Line-1 insertion patterns to whole genome NGS data for identification of novel insertion sites. Dr. David Sexton of Vanderbilt University followed up with an examination of the resources required to manage the massive amounts of data produced by NGS based upon his experience as the Director of the Computational Genomics Core in the Center for Human Genetics Research \[[@B7]\]. Dr. Sexton continually emphasized the need to be able to handle the incredible volumes of data that these sequencers generate, in terms of storage, backup, transport, and analysis. For every consideration, he described the currently implemented solutions at Vanderbilt, giving attendees a glimpse into the day-to-day operations of the NGS center. The last talk of the day was by Dr. Steven Jones from the University of British Columbia. Dr. Jones focussed on the use of NGS to characterize individual mutations in cancer cells, and thereby offer tailored therapies, or individualized genomics. Using NGS, they were able to determine which gene pathways were modified in individual cancers, and suggest appropriate treatments for the individual. This stands in marked contrast to the previously accepted method of isolating one particular strain of cancer cells and treating it as a representative for other cancers in the same bodily tissue \[[@B8]\]. One particularly interesting example involved one individual cancer where the mutations were characterized over time in response to treatment, with examination of the various gene pathways that were disrupted as the cancer underwent further mutation in response to the treatments. Epigenetics =========== The second day of talks began with the session on Epigenetics. Dr. James Cheverud of Washington University presented his research on epigenetic sources of variation in mice obesity. This work involved the crossing of very large (overweight, LG) females with very small (lean, SM) males, and correlating the resultant phenotypes with various loci, and determining how the interactions between loci lead to a particular phenotype \[[@B9]-[@B11]\]. Particular emphasis was placed on examining combinations of gene-effects to provide more explanatory power than examining single gene effects in isolation. Dr. Rosanna Weksberg from the Hospital for Sick Children at the University of Toronto presented research on the genetic imprinting effect in Beckwith-Wiedemann syndrome (BWS) \[[@B12],[@B13]\] which has both genetic causes and epigenetic phenotypes. Two groups of genes have been shown to have changed methylation patterns in connection with BWS, leading to changes in their expression. Dr. Weksberg also discussed Prader Willi Syndrome, and Russell-Silver Syndrome, which are additional human imprinting disorders associated with growth. Rounding out the epigenetics session was Dr. Robert Lane from the University of Utah Department of Pediatrics and his examination of genetic imprinting via histone methylation as a mechanism of adaptation. It has been known for some time that the methylation state of genes affects their expression, however there are still many questions regarding the causes and effects of particular histone methylation patterns. The primary system examined by Dr. Lane was intrauterine growth restriction (IUGR), a condition of restricted growth in the embryo that leads to altered expression of insulin-like growth factor 1 (IGF-1) in the adult, with subsequent effects on growth and development \[[@B14],[@B15]\]. A rat model of IUGR combined with examination of the methylation patterns of the histones found reproducible changes to the methylation of particular histone residues that were also gene position dependent. This resulted in a change of the ratio of particular IGF-1 transcripts, thereby resulting in altered phenotypes. Similar examination of pups of hyperglycemic rats (simulating diabetic pregnancies) revealed analogous changes in methylation of the histones. Incredibly, they were able to prevent the phenotypic changes and reprogram the methylation via supplementation with essential nutrients, thereby re-establishing IGF-1 expression to control levels. Following the poster session and evening buffet, there was an overview of Computable Genomix\'s Gene Indexer program. Gene Indexer is an automated latent semantic indexing engine that extracts both explicit and implicit relationships from the literature. It is efficient and uses an investigator\'s questions, not predetermined pathways or ontologies, to interrogate the scientific literature and determines the functional cohesiveness of any groups of genes. Dr. Robert Hettich of the Oak Ridge National Laboratory presented work on the use of -omics to examine human gut microbial communities and their involvement in Crohn\'s disease \[[@B16],[@B17]\]. This work required examining the metaproteomes and metagenomes, or set of all proteins and genes from a heterogeneous population of bacteria in twins where one, both or neither suffered from Crohn\'s disease (CD) or Ulcerative Colitis (UC). Dr. Hettich\'s group specifically performed the proteomics work to identify as many expressed proteins as possible in the stool samples based on the gene sequences that had been previously determined. They were able to identify changes in the composition of the microbiome and the metaproteome that correlated with the presence or absence of disease, providing rich data sets that will require extensive follow-up. Day two\'s keynote address was given by Dr. Mike Hawrylycz of the Allen Institute for Brain Science on the informatics of large scale digital aliasing in neuroscience \[[@B18],[@B19]\]. As an example, Dr. Hawrylycz provided an in-depth examination of the information required to create the Developing Mouse Brain Atlas. This resource combines 2 D slices of brain tissue, 3 D models of the brain changing during development \[[@B20]\], with in-situhybridization gene expression levels for \~2000 genes. This publicly available resource allows anyone to examine the expression levels of these genes over the developmental stages and determine those genes that have correlated expression either temporally or spatially. A variety of analysis tools and metrics are built into the Atlas, providing a wealth of information regarding gene expression in the brain to the interested researcher. Medical informatics =================== Only one talk was presented in the Medical Informatics session, by Dr. Myriam Fornage from the University of Texas at Houston, on the search for genes responsible for vascular diseases in the brain. Although studies have shown that by far the greatest indicators for vascular disease are age, sex, and high blood pressure, it may be that particular genetic factors predispose individuals to ischemic injuries in the brain. This has been borne out by previous studies that indicate many individuals suffer from pre-clinical lesions, and that true ischemic stroke tends to be \"the tip of the iceberg\", in that it is a clinical symptom of a long underlying process. Dr. Fornage reported on a series of genome wide association studies performed to try and determine genes that may make it more likely that an individual will suffer from brain vascular disease (BVD). A variety of SNP\'s have been identified and confirmed in follow up studies. Two genes in particular were found to be associated with a higher risk of BVD, Ninj2 and Wnk1 \[[@B21]\]. Ninj2 is a cell surface adhesion molecule that is induced after nerve injury, and is theorized to affect how the brain tolerates/recovers from ischemic insults. Wnk1 is a kinase that has been previously associated with hypertension, and is theorized to promote injury to the brain. Using a mouse ischemia model, the expression of both genes was followed over time following injury to the brain, and the expression of both was shown to be concordant with their known roles. Although GWAS analyses are identifying new genetic variants involved in BVD, the effect of many of them is small, thereby requiring large sample sizes to allow their detection. A question that does result though, is how small an effect is too small to be clinically relevant and not worthy of further investigation or drug development? In addition, there is the potential for more work on examining gene-gene interactions, as it is likely that variants found in conjunction will have a larger effect. Posters & short talks ===================== During the two-hour poster session on Saturday afternoon, fifty-two posters (eighteen more than the previous year) were presented. The abstracts were divided into the general groupings of Bioimaging, Bioinformatics of Health and Disease, Bioinformatics Infrastructure, Comparative Genomics, Databases, Functional Genomics, Gene Regulation, Genome Annotation, Genomics, Machine Learning/Algorithms, Microarrays, Neuroscience, Proteomics, Sequence Analysis, and Structure and Function Prediction. Two sessions (one each on Saturday and Sunday) featured presentations from the poster abstracts. They included \"Inferring Gene Coexpression Networks for Low Dose Ionizing Radiation using Graph Theoretical Algorithms and Systems Genetics\" (Gary L. Rogers), \"Integrating metagenomics with metaproteomics for characterization of the molecular activities of the human distal gut microbiome in healthy and Crohn\'s Disease\" (Alison Erickson), \"Discovering Disease-specific Biomarker Genes for Cancer Diagnosis and Prognosis\" (Zhongming Zhao), \"Analysis of equine protein-coding gene structure and expression by RNA-sequencing\" (Steven Coleman), \"A systematic pathway-based analysis of GWAS data revealed susceptibility pathways to schizophrenia\" (Peilin Jia), \"Association of Genomewide Newborn DNA Methylation Patterns with Maternal Diet, Birth Weight and SNP Variation\" (Ronald Adkins), \"High-throughput sequencing of the DBA/2J mouse genome\" (Rob Williams), and \"Subsystems-based servers for rapid annotation of genomes and metagenomes\" (Rami Aziz). Future plans ============ The 2011 Bioinformatics summit will return to the state of Tennessee in the spring of 2011. Potential focus areas include current technological trends in molecular biology, applications of next-generation sequencing, and systems biology. The format will likely be expanded to include more local talks and hands-on workshops at introductory and advanced levels. Acknowledgements ================ We would like to thank the additional Conference Program Committee members Nigel Cooper (University of Louisville), Dan Goldowitz (University of British Columbia), Ramin Homayouni (University of Memphis), Julia Krushkal (University of Tennessee-Memphis), Mike Langston (University of Tennessee-Knoxville), Terry Mark-Major (University of Tennessee-Memphis), Cynthia Peterson (University of Tennessee-Knoxville), Arnold Stromberg (University of Kentucky), Rob Williams (University of Tennessee-Memphis) and Zhongming Zhao (Vanderbilt University) for organizing an outstanding scientific program. In addition, we wish to thank Stephanie Dearing, Terry Mark-Major, Michelle Padgett, Donna Schulte and Jane Thornton for all of their efforts in dealing with the conference organization details. We would also like to thank Kerry Allen and the wonderful folks at Lake Barkley State Park, who helped make our stay as pleasant as possible. Funding for the UT-ORNL-KBRIN Summit is provided in part by the Kentucky Biomedical Research Infrastructure Network (KBRIN), University of Tennessee Center for Integrative and Translational Genomics, University of Tennessee Molecular Resource Center, UT-ORNL Science Alliance, and NIH grant P20RR16481.	{ "pile_set_name": "PubMed Central" }
![](glasgowmedj76317-0023){#sp1 .1} ![](glasgowmedj76317-0024){#sp2 .2} ![](glasgowmedj76317-0025){#sp3 .3} ![](glasgowmedj76317-0026){#sp4 .4} ![](glasgowmedj76317-0027){#sp5 .5} ![](glasgowmedj76317-0028){#sp6 .6} ![](glasgowmedj76317-0029){#sp7 .7} ![](glasgowmedj76317-0030){#sp8 .8} ![](glasgowmedj76317-0031){#sp9 .9} ![](glasgowmedj76317-0032){#sp10 .10} ![](glasgowmedj76317-0033){#sp11 .11} ![](glasgowmedj76317-0034){#sp12 .12} ![](glasgowmedj76317-0035){#sp13 .13} ![](glasgowmedj76317-0036){#sp14 .14} ![](glasgowmedj76317-0037){#sp15 .15} ![](glasgowmedj76317-0038){#sp16 .16} ![](glasgowmedj76317-0039){#sp17 .17} ![](glasgowmedj76317-0040){#sp18 .18} ![](glasgowmedj76317-0041){#sp19 .19} ![](glasgowmedj76317-0042){#sp20 .20} ![](glasgowmedj76317-0043){#sp21 .21} ![](glasgowmedj76317-0044){#sp22 .22} ![](glasgowmedj76317-0045){#sp23 .23} ![](glasgowmedj76317-0046){#sp24 .24} ![](glasgowmedj76317-0047){#sp25 .25} ![](glasgowmedj76317-0048){#sp26 .26} ![](glasgowmedj76317-0049){#sp27 .27} ![](glasgowmedj76317-0050){#sp28 .28} ![](glasgowmedj76317-0032-a){#f1 .10} ![](glasgowmedj76317-0032-b){#f2 .10} ![](glasgowmedj76317-0034-a){#f3 .12} ![](glasgowmedj76317-0036-a){#f4 .14} ![](glasgowmedj76317-0036-b){#f5 .14} ![](glasgowmedj76317-0038-a){#f6 .16} ![](glasgowmedj76317-0038-b){#f7 .16} ![](glasgowmedj76317-0039-a){#f8 .17} ![](glasgowmedj76317-0039-b){#f9 .17}	{ "pile_set_name": "PubMed Central" }
Introduction ============ An abdominal pseudocyst is a rare complication of a ventriculo-peritoneal shunt. Such cysts may cause diagnostic problems in regions such as the Middle East, where echinococcosis disease of the liver is endemic, due to similarities in clinical presentation and radiological appearance. Case presentation ================= An 18-year-old Caucasian male patient presented with a 10-day history of generalized tonic-clonic seizures. His past medical history included right ventriculo-peritoneal (VP) shunt insertion at two weeks of age for bacterial meningitis complicated by hydrocephalus. Four years prior to his current admission he had presented with abdominal pain and a computed tomography (CT) scan of his abdomen at that time was interpreted as being consistent with a right hepatic hydatid cyst (8 × 6 cm). Serology workup was negative for hydatid disease at the time of the CT scan, however, due to the characteristic radiological findings, antihelminthic treatment (albendazole) was commenced and he was subsequently lost to follow-up. During his current admission, he underwent a CT scan of his brain and abdomen, which revealed an increase in the size of the cerebral ventricles and an increase in the size of the liver cyst(11 × 9 cm), revealing the presence of the VP shunt tip inside the cyst (Figure [1](#F1){ref-type="fig"}). Exploratory laparotomy was performed and the tip of the shunt was found inside the cyst, which was opened and drained. The hepatic cyst was found to contain cerebrospinal fluid with no evidence of hydatid disease. The VP shunt was repositioned in his pelvis; our patient made an excellent postoperative recovery and was discharged home after four days. A follow-up CT scan showed regression of the dilated cerebral ventricles (Figure [2a, b](#F2){ref-type="fig"}). ![Abdominal CT scan showing 11 × 9 cm pseudocyst of his right hepatic lobe, with peripheral calcifications and the tip of the VP shunt going inside the cyst (arrow).](1752-1947-5-475-1){#F1} ![CT scan of the patient\'s brain. (A) Preoperative ventricular dilatation. (B) Postoperative decompression of the ventricle](1752-1947-5-475-2){#F2} Discussion ========== VP shunts are foreign bodies that may cause intra-abdominal complications. Major intra-abdominal complications include ascites, peritoneal infections, intestinal obstructions and perforations, pseudocyst, abscess formation and inguinal hernia. Migration of the distal catheter and metastases of brain tumor have also been reported \[[@B1]-[@B3]\]. The incidence of intra-abdominal cerebrospinal fluid (CSF) pseudocyst varies between 1% and 3% in different studies. Hepatic pseudocyst secondary to a shunt is extremely rare \[[@B4],[@B5]\]. The formation of an abdominal CSF pseudocyst was first described by Harsh in 1954 \[[@B6]\]. Non-specific clinical presentations may cause diagnostic and therapeutic difficulties; physicians should be aware of this complication, especially in unconscious patients \[[@B7]\]. The precise etiology for abdominal pseudocyst formation is still unknown. An inflammatory process, either sterile or infectious, is generally regarded as the main causative factor \[[@B8]-[@B10]\]. Other predisposing factors have been postulated such as peritonitis, prior surgical peritoneal adhesion, a history of central nervous system (CNS) infections and CNS tumors, distal shunt migration, multiple shunt revisions, malabsorption of CSF and allergic reaction \[[@B10],[@B11]\]. The time between last VP shunt operation and development of an abdominal pseudocyst has been reported from three weeks to 10 years \[[@B12]\]. Hepatic pseudocysts secondary to VP shunts are classified as intra-axially or extra-axially growing pseudocysts when penetrating the Glisson capsule; the shunt tube can cause extra-axial subcapsular pseudocyst formation \[[@B13]\]. Alternatively, the tip of the shunt can be lodged in the liver parenchyma and cause formation of an intra-axially growing pseudocyst deep within the parenchyma \[[@B14]\]. The most important factors causing hepatic pseudocyst are migration of the peritoneal tip of the shunt to the liver surface and its chronic irritation \[[@B7]\]. Consequently, the oncotic pressure of the cystic fluid increases, interstitial fluid passes into the cyst, and the cyst increases in size \[[@B2]\]. Hydatid disease is endemic in the Middle East. The combination of imaging and serology are usually used to make the diagnosis. In this case, the initial diagnosis was misled by the radiological findings of peripheral calcifications that may represent the common appearance of hydatid cyst. Imaging findings of echinococcosis reflect a spectrum depending on the developing stages of the parasitic cyst in the human tissue, ranging from a single unilocular cyst, to multiple daughter cyst formation, and then gradually to a solid and calcified cyst \[[@B15]\]. The fact that the hydatid indirect hemagglutinin (IHA) test was negative does not rule out hydatid disease however, in retrospect, would make a peritoneal cyst or pseudocyst as likely. The sensitivity and specificity of the IHA are 86.7% and 95% respectively \[[@B16]\]. Only a positive hydatid serology is valuable; a negative serologic test does not exclude the diagnosis \[[@B17]\]. The suspicion of an abdominal pseudocyst is often made at the time of physical examination and on the basis of conventional radiology \[[@B7]\]. Visualization of the distal tip of the VP shunt within a homogeneous intraperitoneal collection is the principal diagnostic sign of an abdominal CSF pseudocyst on ultrasound and CT. Ultrasonography is the method of choice because it is fast and reliable \[[@B8]\]. However, a CT scan of the abdomen provides a more accurate diagnosis \[[@B18]\]. For the diagnosis of an extra-axially growing hepatic pseudocyst, abdominal CT images are typical. The pseudocyst is surrounded by an annulus showing continuity with hepatic tissue \[[@B13]\]. The standard treatment of a hepatic pseudocyst secondary to catheter tip migration, in cases with no infection or prominent inflammatory reaction in the peritoneal cavity, should be simple repositioning of the peritoneal catheter in the abdominal cavity. This procedure may be combined with percutaneous or open drainage in resistant cases \[[@B3]\]. However, there are numerous therapeutic approaches for the management of shunt-related abdominal pseudocysts reported in the literature \[[@B19]\]; simple aspiration of the cyst under CT or ultrasound guidance \[[@B5]\]; removal of the shunt and installation of a new one once the cyst is resolved; if it is not resolved, shunt revision following cyst aspiration; or draining the cyst fluid through a explorative laparotomy, unroofing the cyst wall, then shunt revision or repositioning. In recent years, laparoscopic approaches have been advocated \[[@B20]\]. Conclusion ========== We report a case of a hepatic CSF pseudocyst secondary to the migration of a VP shunt, mistaken for hepatic hydatid disease with serious sequelae. The formation of a pseudocyst should always be considered in patients with VP shunts in situand can be easily treated by simple repositioning. Abbreviations ============= CNS: central nervous system; CSF: cerebrospinal fluid; CT: computed tomography; IHA: indirect hemagglutinin; VP: ventriculo-peritoneal. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= WF drafted the manuscript; HHA and DM participated in the design of the study; MK participated in the design and coordination of the study. All authors read and approved the final manuscript. Consent ======= Written informed consent was obtained from the patient for publication of this manuscript and any accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal.	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ Stroke is a major cause of death and long-term disability worldwide. For ischemic strokes, clots in the brain can be dissolved with recombinant tissue plasminogen activator (tPA) [@pone.0054203-Zivin1]. Timing of tPA application is critically important. The sooner patients receive tPA and reperfuse, the better the odds ratio for improved outcomes. However, in current clinical practice, performing a brain scan is considered indispensable prior to tPA therapy in order to rule out patients with intracerebral hemorrhage (ICH) [@pone.0054203-Adams1]. This induces a substantial time-to-treatment delay, as tPA cannot be given at the patients' home or in the ambulance as is the case for myocardial infarction [@pone.0054203-Morrison1]. Of course, tPA is not completely benign. Many experimental studies now show that excessive tPA can amplify excitotoxic neuronal death and promote blood brain barrier injury [@pone.0054203-Armstead1], [@pone.0054203-Cheng1]. But it is important to remember that, even after factoring in rates of complications and side effects, tPA is still clinically effective when given to the right patients at the right time. Yet, tPA usage is still limited to less than 5% of all ischemic strokes today, more than 10 years after FDA approval. From a clinical and practical perspective, the fear of inadvertently administering tPA in ICH is a major factor that limits the use of this important therapy. The widespread assumption that tPA therapy would worsen ICH seems intuitive, but lacks scientific validation. Here, we tested the effects of intravenous tPA therapy in different experimental models of ICH in mice. Results {#s2} ======= An in vitro activity assay confirmed that the recombinant human tPA dosing used in our experiments was able to convert mouse plasminogen into active plasmin, the enzyme responsible for clot lysis ( [Fig. 1A](#pone-0054203-g001){ref-type="fig"} ) [@pone.0054203-Ishii1]. Enzyme activity was further confirmed in vivo using a standard rat model of thromboembolic focal cerebral ischemia [@pone.0054203-Zhu1]. Homologous blood clots were intraluminally placed into the middle cerebral artery, and then rats were treated with either saline or 10 mg/kg of tPA at 1 hr post occlusion. Laser Doppler flowmetry confirmed that tPA effectively restored cerebral blood flow ( [Fig 1B](#pone-0054203-g001){ref-type="fig"} ). ![tPA activity measures.\ (A) Human tPA but not saline activates mouse plasminogen. In this assay, tPA converts plasminogen to plasmin, then plasmin converts D-Val-Leu-Lys-7-Amino-4-Methylcoumarin to a fluorescent product, which was measured over 1 hr at 5 minute intervals using a fluorescence plate reader. Data are provided in arbitrary fluorescence units (A.U.). Data are mean±SEM, 3 measures per group and time point. (B) Thrombolytic activity of tPA was proven in-vivo in a standard model of thromboembolic middle cerebral artery occlusion in rats. tPA restored regional cerebral blood flow (rCBF) more rapidly than saline. Data are mean±SEM, 4 animals per group.](pone.0054203.g001){#pone-0054203-g001} The first model of ICH involved the standard and widely-used stereotactic injection of collagenase type VII-S (0.2 U) into mouse striatum to provoke ICH. Consistent with previous work [@pone.0054203-Foerch1], [@pone.0054203-Rosenberg1], ICH began within 30 min after collagenase injection, and hematoma development was well underway by 1 hr (see Methods and [Fig. 2A](#pone-0054203-g002){ref-type="fig"} ). At 30 min after ICH induction, mice were blindly and randomly assigned to one of 3 treatment groups: saline controls (500 µl, n = 15), tPA (10 mg/kg in 500 µl saline, n = 15), or the anticoagulant heparin (used as a positive control, 100 U/kg in 500 µl saline, n = 4). Treatments were infused over 30 min via a jugular vein catheter. Twenty-four hrs after ICH induction, hematoma volumes were assessed using a photometric assay. Surprisingly, hematoma volumes were not different between saline controls (mean±SD 7.5±3.4 µl) and tPA-treated mice (7.6±3.5 µl), but heparin significantly worsened hemorrhage (19.8±8.8 µl, one-way ANOVA between group differences p\<0.001, post-hoc saline vs. tPA p = 1.000, saline vs. heparin p\<0.001, tPA vs. heparin p\<0.001, [Fig. 2B](#pone-0054203-g002){ref-type="fig"}). Mortality rate was 0/15 in saline mice, 2/15 in tPA mice, and 2/4 in heparin mice. The two tPA-treated mice that died had pronounced bleeding at the surgical areas (head, neck), but ICH volume was not increased (2.6 and 7.3 µl, respectively). Most likely, death resulted from extracerebral bleeding complications. In contrast, the dead heparin mice had extensive ICH volumes (33.0 and 14.7 µl). The functional impact of ICH, assessed by means of a standard hanging wire test, was not different between saline- and tPA-treated mice ([Fig. S1](#pone.0054203.s001){ref-type="supplementary-material"}). Because this result was somewhat surprising, a second independent study was initiated to confirm these findings. Using different batches of tPA and collagenase, 24 ICH mice were randomized to saline controls (500 µl, n = 8), tPA (10 mg/kg in 500 µl saline, n = 8), or heparin (100 U/kg in 500 µl saline, n = 8). Once again, treatments were infused over 30 min via a jugular vein catheter. Three mice had to be excluded due to catheter complications (rupture of jugular vein). At 24 hrs, hematoma volumes were generally larger than in the first experimental series, likely due to variations of enzyme activity between different collagenase batches. However, the same pattern of results was obtained between groups. Controls (mean±SD 16.4±7.4 µl) and tPA-treated mice (18.5±4.3 µl) had similar hematoma volumes, whereas heparin-treated mice had larger bleeds (27.4±5.1 µl, one-way ANOVA between group differences p = 0.004, post-hoc saline vs. tPA p = 1.000, saline vs. heparin p = 0.005, tPA vs. heparin p = 0.033, [Fig. 2C](#pone-0054203-g002){ref-type="fig"}). Mortality rate was 2/7 in saline mice, 4/6 in tPA mice, and 7/8 in heparin mice. ![Collagenase-induced ICH volumes are not increased by administration of i.v. tPA.\ (A) Brain sections showing hematoma at 30 min, 60 min and 24 hrs after ICH induction. ICH starts to occur within 30 min and has largely developed by 1 hr. (B) Hematoma volumes at 24 hrs after ICH induction (mean±SD). tPA did not alter hematoma sizes, but heparin significantly worsened ICH as compared to saline and tPA treatment. (C) Results of the reconfirmation study. No difference was observed between saline and tPA mice, but heparin significantly increased ICH volume (mean±SD) as compared to saline and tPA treatment.](pone.0054203.g002){#pone-0054203-g002} To further confirm these findings, we also tested tPA injection at a later time point (i.e., between 4 and 4.5 hrs after ICH induction when more clot formation in the extravasated blood has occurred). Following the same experimental protocol as above, no difference was found between mice treated with tPA (mean±SD 8.2±5.8 µl) and saline (7.9±2.7 µl, n = 8 per group, t-test p = 0.882), suggesting that no re-bleeding was induced by means of the tPA injection at this later time point after the initiation of ICH. To rule out the possibility that our findings were caused by artifacts specific to the collagenase model of ICH, we repeated these experiments using an independent ICH model, involving laser-mediated rupture of cerebral blood vessels under in vivo imaging (see Methods for details). Cranial windows were implanted in mice (n = 11). Five days later, bleeding was induced in one to four penetrating arterioles per mouse by injuring the targeted endothelium with tightly-focused femtosecond laser pulses [@pone.0054203-Nishimura1]. This technique produces hematoma volumes about 20,000 times smaller than the ICH model, but this model provides a powerful way to image, in real time, the development of vascular rupture and bleeding in vivo. During the production of these hemorrhages, animals received intravenous saline (n = 4), tPA (n = 4), or heparin (n = 3), as described above. Using in vivo two-photon excited fluorescence microscopy, we found no difference in the size of the red blood cell (RBC)-filled microhemorrhage core or the surrounding blood plasma-filled region between saline and tPA-treated mice ( [Fig. 3A--C](#pone-0054203-g003){ref-type="fig"} ). In contrast, hemorrhages produced while infusing heparin were significantly larger than saline controls (ANOVA, post hoc; RBC diameter: p\<0.001; plasma extravasation diameter: p = 0.019). We identified 18 of the femtosecond laser-induced intracerebral microhemorrhages in post-mortem tissue sections. Across all treatments, we found that, in addition to making a spherical hematoma, RBCs spread vertically along the perivascular space surrounding the targeted penetrating arteriole ([Fig. S2a--c](#pone.0054203.s002){ref-type="supplementary-material"}). Blood plasma, however, was able to diffuse through the parenchyma tissue and exhibited a more extended spatial pattern ([Fig. S2d--f](#pone.0054203.s002){ref-type="supplementary-material"}). We consistently observed that heparin treatment led to a larger spherical RBC-filled volume compared to saline and tPA treatments ([Fig. S2c and S2f](#pone.0054203.s002){ref-type="supplementary-material"}). ![Femtosecond laser-induced cortical microhemorrhage volumes are not increased by i.v. administration of tPA.\ (A) Two-photon excited fluorescence in vivo image projections of fluorescently-labeled blood plasma spanning a 20 µm depth centered at the hemorrhage origin. Image stacks are shown before and after rupturing the wall of a single penetrating arteriole using tightly focused femtosecond laser pulses. Representative examples from the three different intravenous infusion groups (saline, tPA, heparin) are shown. Extravasated plasma is visualized by diffuse fluorescence and can be seen in the post hemorrhage images in a halo surrounding the target vessel. The dark core immediately adjacent to the target vessel is filled with red blood cells. Post-hemorrhage (B) red blood cell (RBC) and (C) blood plasma extravasation diameters (mean±SEM) are shown for the three treatments: saline (n = 9 microhemorrhages), tPA (n = 11), and heparin (n = 10). No difference was found between the saline and tPA group, but heparin significantly increased both red blood cell and blood plasma extravasation diameters.](pone.0054203.g003){#pone-0054203-g003} Our results showed that tPA did not worsen bleeding in two independent animal models of ICH. In order to further assess the specificity of our findings, we next tested tPA in a pathophysiologically different type of brain injury, subarachnoid hemorrhage (SAH). Following a standard model, SAH was induced in mice by perforating the left anterior cerebral artery using a blunted nylon suture [@pone.0054203-Sozen1]. Thirty min later, saline or tPA was infused, as described above. Subarachnoid blood volume was quantified at 24 hrs using the same photometric assay as before. SAH blood volumes were significantly larger in tPA-treated animals compared to saline controls (mean±SD 20.0±10.8 vs. 8.3±3.5 µl, p = 0.010, t-test) ( [Fig. 4](#pone-0054203-g004){ref-type="fig"} ). SAH severity was clearly worsened by tPA as 3 out of 6 tPA-treated mice died within 24 hrs, compared to 0 out of 7 saline animals. Functional outcome assessed on an ordinal scale was also significantly worsened in tPA-treated SAH mice (p = 0.014, Mann Whitney U-test). ![Treatment with tPA worsened subarachnoid hemorrhage.\ Blood volumes at 24 hrs after SAH induction (mean±SEM) are provided. tPA significantly increased SAH blood volume (tPA, n = 6; saline, n = 7).](pone.0054203.g004){#pone-0054203-g004} It is known that tPA can exacerbate neurovascular injury during cerebral ischemia. Indeed, many labs have shown that administration of tPA worsens blood brain barrier damage during reperfusion injury in ischemic stroke, in part by amplifying matrix metalloproteinases (MMPs) [@pone.0054203-Wang1], [@pone.0054203-Yepes1]. Hence, we hypothesized that the different outcomes after tPA application in focal cerebral ischemia versus ICH may be accompanied by different responses in MMP activity. Mice were subjected to either transient focal cerebral ischemia or collagenase-induced ICH and randomized into saline or tPA groups. For the ischemia group, the middle cerebral arteries were occluded with intraluminal sutures for 3 hrs, then mice were treated with saline or 10 mg/kg tPA intravenously, followed by reperfusion via withdrawal of the occluding sutures. Brains were extracted and MMP levels were measured after 3 hrs of reperfusion. For the ICH group, mice were treated with saline or 10 mg/kg tPA intravenously at 3 hrs after collagenase injection, then brains were extracted for MMP measurements at 6 hrs. As expected, MMP-2 and MMP-9 levels were elevated in injured ipsilateral brain tissue in all ischemia and hemorrhage mice. In focal cerebral ischemia, MMP levels were significantly higher in tPA-treated mice compared to saline-treated mice. However, tPA did not amplify MMPs in brains from ICH mice ( [Fig. 5A--B](#pone-0054203-g005){ref-type="fig"} , [Fig. S3](#pone.0054203.s003){ref-type="supplementary-material"}). ![Gelatin zymography of MMP responses.\ (A) Representative zymogram gels showing MMP-2 and MMP-9 levels in brain homogenates derived from mice subjected to focal cerebral ischemia (induced by middle cerebral artery occlusion, MCAO) or primary ICH mice. (B) Quantified densitometry of the zymogram results (arbitrary units normalized to constant loading of MMP standards). After 3 hrs of focal cerebral ischemia followed by 3 hrs reperfusion, tPA significantly amplified MMP-2 and MMP-9 levels. In contrast, there were no differences in MMPs in saline versus tPA-treated mice after ICH.](pone.0054203.g005){#pone-0054203-g005} Discussion {#s3} ========== Our results in two different mouse models of ICH suggest that intravenous tPA application during the phase of hematoma formation does not increase intracerebral hemorrhage volume. Our findings stand in sharp contrast to the widespread -- but so far unproven -- assumption that tPA would enlarge hematomas if inadvertently given to patients with acute ICH. Unlike tPA, the anticoagulant heparin significantly increased intracerebral hemorrhage volume. This positive control demonstrates that both of our ICH models are able to detect hematoma enlargement if it occurs. And the results are consistent with our previous findings that another anticoagulant, warfarin, exacerbates hemorrhage in ICH mice as compared to anticoagulation-naïve controls [@pone.0054203-Foerch1]. Although we did not quantify absolute fibrinolytic activity in comparison to reference standards, we showed that our tPA preparations were thrombolytically active, both in-vivo and in-vitro. From a mechanism standpoint, the massive hematoma enlargement after heparin therapy suggests that drugs were administered prior to definitive clot formation. Unlike warfarin and heparin, tPA has no effect on primary hemostasis and does not delay initial coagulation. But as an activator of the fibrinolytic cascade, tPA increases the risk of re-bleeding from previously hemostatic stable areas [@pone.0054203-Murray1]. So tPA could theoretically worsen intracerebral bleeding. On the other hand, the fibrinolytic activity of tPA is highly dependent on parameters such as clot structure. For example, platelet-rich thrombi are much less susceptible to tPA dissolution compared with erythrocyte-rich thrombi, mainly due to the release of platelet-derived plasminogen activator inhibitor [@pone.0054203-Jang1]. Intracerebral hematomas consist of whole blood containing large amounts of platelets so it is possible that these structures may be more tPA-resistant. Another possible explanation for our finding is that the ICH clots surrounding ruptured brain blood vessels are heterogeneous and larger than singular intra-arterial thrombi in ischemic stroke [@pone.0054203-Rosenberg1]. Standard doses of tPA may not be able to dissolve significant parts of the coagulated hematoma in order to induce re-bleeding. Furthermore, ICH may be stabilized by the counterpressure of surrounding brain tissue, which limits hematoma expansion. Elevated pressure in the brain areas near the hematoma may even prevent systemic intravenously administered tPA from reaching the initial sites of vessel rupture. Our finding that tPA does not amplify MMP-2 and MMP-9 in ICH potentially suggests that this drug is not activated by a parenchymal (i.e. extra-arterial) clot formation in the same way as it is activated by an intraluminal thrombus in ischemic stroke [@pone.0054203-Kahles1]. However, there might be other members of the MMP family involved in the pathophysiologic cascades following ICH that are activated by tPA treatment [@pone.0054203-Xue1]. Brain edema was not measured in our study. But tPA was found to aggravate brain edema if injected into the ventricles in the case of ICH with ventricular extension (in this setting, tPA injection is intended to restore the drainage of the cerebrospinal fluid) [@pone.0054203-Ducruet1], [@pone.0054203-Thiex1]. Nevertheless, it is unlikely that differences in edema formation between tPA- and saline-treated mice have influenced our main outcome parameter (i.e., ICH volume), as hematoma expansion and edema formation occur sequentially rather than simultaneously. In contrast to ICH, our results clearly show that tPA dramatically increases blood volume in SAH. Pathophysiologically, SAH results from the rupture of a larger intracranial artery in the subarachnoid space outside the brain. Blood extravasation is immediately followed by the induction of the coagulation cascade, leading to the formation of a fibrin-rich clot in the vessel wall [@pone.0054203-Ishikawa1]. In ischemic stroke, tPA is able to dissolve thrombi in arteries of similar diameter [@pone.0054203-Zivin1]. Thus, it is conceivable that tPA applied in acute SAH is able to reach the site of the initial vessel rupture, thereby dissolving newly-formed arterial clots. Any re-bleeding would then easily expand into the subarachnoid space. What is extremely important is that our results must be distinguished from the fact that tPA therapy increases the risk of hemorrhagic transformation after ischemic stroke [@pone.0054203-NINDSStudyGroup1]. The mechanisms involved in primary ICH are different from those that underlie hemorrhagic transformation and reperfusion injury after ischemia. Hemorrhagic transformation during reperfusion injury is now known to be related to excitotoxic effects of tPA and neurovascular upregulation of MMPs [@pone.0054203-Cheng1], [@pone.0054203-Wang1], [@pone.0054203-Yepes1], rather than the re-bleeding of a previously terminated hematoma. tPA may not worsen ICH, but this is different from potential deleterious effects of tPA during cerebral ischemia and reperfusion. Our present data support this idea. tPA seems to amplify MMPs during cerebral ischemia-reperfusion but not during ICH formation. We performed an extensive literature research, but did not find clinical or experimental data targeted to explore the effects of tPA on hematoma volume in the acute phase of ICH. However, a recent study on memantine as an adjunctive therapy to tPA in ischemic stroke also analyzed a group of mice subjected to collagenase ICH. [@pone.0054203-Montagne1] Whereas a bolus injection of tPA (10 mg/kg) 30 min after ICH induction increased hematoma size on MRI scans, the combined administration of tPA and memantine did not change ICH as compared to controls. As memantine is a NMDA-antagonist which does not reduce the thrombolytic activity of tPA itself, these findings are worthwhile to be further explored in the context of the present study. In addition, we found one more study that explored tPA administration in a guinea pig model of collagenase ICH. [@pone.0054203-Mihara1] In this work, a dose-dependent increase of hematoma volumes was observed. The results of the higher dose groups stand in apparent contrast to our findings and need further evaluation`.` One should be cautious when transferring our experimental findings to the clinical setting. Despite testing our hypothesis in two pathophysiologically independent models of ICH, significant species differences and model limitations may play a role. In particular, the collagenase model represents "deep" hemorrhage formation occurring in or nearby the basal ganglia, while the laser-rupture model more closely represents cortical microhemorrhages, with a diameter of only 100 µm. It is difficult to extrapolate from these models to, for example, large cortical hemorrhages in humans. The latter frequently result from the rupture of pathological vessels, for example arteriovenous malformations or vessels with cerebral amyloid angiopathy, and our models of bleeding from normal vessels cannot rule out differences in the effects of tPA due to pre-existing vessel pathology. Furthermore, the coagulation system of mice and humans are not entirely comparable, as some murine factors have different roles than their human counterparts [@pone.0054203-Hogan1]. Of course, if it would be true that intravenous tPA does not enlarge hematomas or worsen outcomes in ICH, then tPA could be administered at much earlier time points to patients exhibiting signs of stroke. This could vastly increase the effectiveness of this drug in patients with ischemic stroke, while not worsening outcome for the patients with ICH. However, based on clinical findings alone, it is not always possible to distinguish patients with SAH from those with ICH. Some SAH may mimic ICH, due to a parenchymal extension of the primary bleeding into the lobes that surround the aneurysm [@pone.0054203-Fisher1]. According to our findings, treating a SAH patient with tPA would have negative sequelae. In summary, our experimental data do not support the assumption that intravenous tPA application significantly increases hematoma volume in acute ICH. Due to species differences and technical limitations of the models used, our findings cannot be immediately transferred to a clinical setting. More clinical and experimental studies are warranted. Methods {#s4} ======= All experiments were performed under protocols approved by the Massachusetts General Hospital Institutional Animal Care and Use Committees (IACUC, protocol N-00087) and the Cornell Institutional Animal Care and Use Committees (IACUC, protocol 2009--0043) following the NIH Guide for Use and Care of Laboratory Animals. The ARRIVE guidelines were followed (please see checklist below). In-vitro tPA Activity Assay and In-vivo Reperfusion Studies {#s4a} ----------------------------------------------------------- A 96-well plate was pre-loaded with mouse plasminogen (10 µg/ml) and the fluorogenic plasmin substrate D-Val-Leu-Lys-7-Amino-4-Methylcoumarin (200 µM) in a final volume of 50 µl phosphate buffered saline (PBS). Thereafter, 0.125 µl of human recombinant tPA stock-solution (4 mg/ml) or saline was added. In this assay, tPA converts plasminogen to plasmin, then plasmin converts D-Val-Leu-Lys-7-Amino-4-Methylcoumarin to a fluorescent product, which was measured over 1 hr at 5 min intervals using a pre-warmed fluorescence plate reader (37°), with excitation set at 360 nm and emission at 460 nm. Three measures were performed per group and time point [@pone.0054203-Ishii1]. In-vivo thrombolytic activity of tPA was demonstrated by a standard model of ischemic stroke based on a thromboembolic occlusion of the middle cerebral artery in rats (see [@pone.0054203-Zhu1] for details). In brief, one hour after initiation of ischemia, 4 animals per group were treated intravenously with either saline or tPA (10 mg/kg). Laser-Doppler flowmetry was used to monitor regional cerebral blood flow for up to 150 min after treatment. Mouse Model of Collagenase-induced Intracerebral Hemorrhage {#s4b} ----------------------------------------------------------- ### Experimental procedures {#s4b1} Experiments were performed in male CD-1 mice (12--16 weeks) under isoflurane anesthesia, following an institutionally approved protocol (Massachusetts General Hospital N-00087) and in accordance with the National Institute of Health's Guide for the Care and Use of Laboratory Animals. Under isoflurane anesthesia (1.5% to 2%) with spontaneous respiration in a nitrous oxide--oxygen mixture, the right jugular vein was prepared. After ligating the distal part of the vessel, a PE-10 tube was inserted into the jugular vein and moved forward in the proximal direction. After properly fixing the tube and closing the suture, the animals were placed in a stereotactic frame. Under ongoing anesthesia, a small borehole was drilled, and a 32-gauge 0.5 µl microinjection needle (Hamilton, 7000 series, Hamilton, Reno, NV, USA) was slowly lowered into the right striatum at the following stereotactic coordinates from bregma: 0.0 mm anterior, 2.0 mm lateral, and 3.5 mm depth. Over a period of 5 min, 0.5 µl of saline containing 0.2 U of collagenase VII-S (Sigma-Aldrich, St Louis, MO, USA) was injected [@pone.0054203-Foerch1], [@pone.0054203-Rosenberg1]. The needle was left in place for 10 min, and then slowly removed over a period of 5 min. Thereafter, the borehole was sealed with bone wax and the scalp closed. The mice were then allowed to wake up and were put in a mouse restrainer under light restraint. Beginning 30 min or 4 hrs after hemorrhage induction, respectively, saline (500 µl), recombinant human tPA (10 mg/kg, diluted in 500 µl saline) or heparin (100 U/kg, diluted in 500 µl saline) were continuously injected through the jugular vein catheter for 30 min using an injection pump (10% bolus injection over the first minute). The correct intravenous position of the catheter was double-checked by drawing some blood into the tube after the injection. The catheter was then ligated and cut near the skin and the mice were returned to their cages. ### Functional outcome and ICH volume determination {#s4b2} Twenty-four hrs after hemorrhage induction, mortality rate was determined and functional outcome was assessed using a standard hanging wire test [@pone.0054203-Foerch1]. Afterwards, mice were perfused with PBS and sacrificed under deep anesthesia. Brains were removed, separated into left and right hemispheres, and placed in glass tubes containing 3 ml PBS. After 30 sec of homogenization, ultrasound was applied for 1 min. After centrifugation for 30 min (13 000 rpm, 4°C), 250 µl supernatant was added to 1 ml Drabkins reagent. With use of a photometer, absorption rates were determined at 540 nm, and hematoma volumes were calculated for the entire brain (both hemispheres) on the basis of a standard curve [@pone.0054203-Foerch1]. Mice that died within the first 24 hrs after ICH induction could not undergo perfusion. In these cases, we subtracted 2.1 µl from the total blood volume derived from the hemoglobin assay. This volume was found to be the mean difference in blood content between perfused and non-perfused brains [@pone.0054203-Foerch1]. ### Determining the kinetics of hematoma development {#s4b3} In order to assure that the medication is administered during hematoma formation, we sacrificed untreated mice under deep anesthesia at 30 min (n = 3), 60 min (n = 4) and 24 hrs (n = 1) after hemorrhage induction. Brains were removed and cut in slices using a 1-mm matrix to visually identify the location and size of the hematoma ( [Fig. 2A](#pone-0054203-g002){ref-type="fig"} ). All mice showed bleeding in the right striatum. Together with previous studies using the collagenase model [@pone.0054203-Foerch1], these data suggested that the time interval of treatment administration (i.e., 30 to 60 min after hemorrhage induction) well matches the kinetics of hematoma formation. Mouse Model of Laser-induced Parenchymal Microhemorrhage {#s4c} -------------------------------------------------------- ### Experimental procedures {#s4c1} All animal procedures were approved by Cornell University Institutional Animal Care and Use Committee, following an institutionally approved protocol (2009--43) in accordance with the National Institute of Health's Guide for the Care and Use of Laboratory Animals. Eleven C57BL/6J mice (The Jackson Laboratory; 5 female, 6 male) ranging from 21 to 32 g in mass were used to image the size of femtosecond laser induced hemorrhages. To prepare a chronic cranial window for imaging, mice were anesthetized using 5% isoflurane (VetOne) and maintained at 1.5--2%. Body temperature was kept at a constant temperature of 37.5°C with a heating blanket and thermometer (50-7053P; Harvard Apparatus). Glycopyrrolate (0.005 mg/kg mouse (Baxter Healthcare Corp.)) was administered intramuscularly while ketoprofen (5 mg/kg mouse (Fort Dodge, Inc.)) and dexamethasone sodium phosphate (0.2 mg/kg mouse (American Regent, Inc.)) were both administered subcutaneously prior to surgery. A 6 mm diameter bilateral craniotomy was performed over the parietal cortex. An 8 mm glass cover slip (Deckglaser; WPI) was then glued over the exposed brain using cyanoacrylate and dental cement (Lang Dental Mfg Co.). Animals were administered 5% (wt/vol) glucose in physiological saline (0.1 ml/100 g mice) subcutaneously and gradually transitioned off isoflurane anesthesia. Mice were then administered ketoprofen (5 mg/kg) every 24 hrs for 72 hrs and allowed a minimum of five days recovery. ### Two-photon excited fluorescence microscopy {#s4c2} Before imaging sessions, mice were anesthetized with isoflurane and the jugular vein was cannulated, as described above. Mice were intravenously injected with 0.1 ml of 5% (wt/vol) lysine-fixable Texas-red dextran (70 kDa) fluorescent dye (D1864; Invitrogen) in physiological saline to label the vasculature. In-vivo imaging was conducted with a locally-designed two-photon excited fluorescence (2PEF) microscope using a 1045-nm, 1-MHz, 350-fs pulse train from a Yb-fiber oscillator/amplifier system (µJewel FCPA, IMRA America, Inc.) as the excitation source. Fluorescence emission was collected with a 65 nm wide filter centered at 645 nm (Chroma). 2PEF imaging revealed surface blood vessels and sub-surface capillaries ([Fig. 3A](#pone-0054203-g003){ref-type="fig"}). In addition to visualizing vascular topology, we measured the diameter and blood flow speed and direction in individual vessels, as described previously [@pone.0054203-Schaffer1]. For hemorrhages, we targeted cortical penetrating arterioles (PA), i.e. arterioles that branch from the surface arteriole network and dive into the brain to feed capillary beds [@pone.0054203-Nishimura2]. These vessels were identified by tracing through the vascular network from readily-identifiable surface arterioles and we confirmed they were penetrating arterioles by checking that the blood flow direction was into the brain. ### Penetrating arteriole microhemorrhage by femtosecond laser ablation {#s4c3} Each mouse was subjected to one to four microhemorrhages with a minimum of 1 mm spacing between hemorrhaged vessels. 2PEF image stacks of each microhemorrhage site ([Fig. 3A](#pone-0054203-g003){ref-type="fig"}) as well as diameter and blood flow speed measurements of the target PAs were collected before microhemorrhages were induced. A 50 µl bolus injection of saline (9 microhemorrhages across 4 animals), tPA (11 microhemorrhages across 4 animals), or heparin (10 microhemorrhages across 3 animals) was infused through the jugular vein catheter, then after five minutes a constant infusion (18 µl/min for 25 min) was set using a syringe pump (PHD2000; Harvard Apparatus), as described above. During the period of steady infusion, the targeted PAs were once again located and microhemorrhages were produced in the descending segment of the PA, about 50--100 µm below the surface of the brain. To induce a microhemorrhage, 50 fs duration laser pulses were tightly focused (1.0 numerical aperture) on the wall of the targeted PA, as described by Nishimura, et al [@pone.0054203-Nishimura1]. Briefly, a laser energy of about 700 nJ was initially applied for a 10 pulse burst (1-kHz repetition rate). If the vessel did not rupture, the laser energy was increased by about 50% and the vessel was irradiated again. This process was repeated until extravasation of red blood cells (RBC) and blood plasma into the parenchyma of the brain was observed. The variation of required laser energy to trigger a microhemorrhage depends mainly on the depth of the target vessel beneath the brain surface and the presence or absence of large blood vessels on the brain surface above the target vessel. We use the minimum laser energy required to rupture the vessel wall of the targeted PA and initiate bleeding into the brain. 2PEF image stacks were taken immediately and 30 min after each PA hemorrhage ([Fig. 3A](#pone-0054203-g003){ref-type="fig"}). ### Red blood cell and plasma diameter measurements {#s4c4} During the microhemorrhage, RBCs and plasma exit the lumen of the targeted PA. Since the fluorescent marker only labels blood plasma, under 2PEF imaging the RBCs are visualized as dark patches in a sea of fluorescently-labeled blood plasma leaving the vessel. After the hemorrhage, the region immediately surrounding the targeted PA is filled with RBCs [@pone.0054203-Nishimura1]. This region appears dark in the 2PEF image because the RBCs are densely packed and largely exclude the fluorescently-labeled blood plasma ([Fig. 3A](#pone-0054203-g003){ref-type="fig"}). Blood plasma exiting the lumen of the target vessel moves further into the brain parenchyma than the RBCs and can be visualized by a ring of increased fluorescence surrounding the RBC filled hemorrhage core. RBC hemorrhage and plasma extravasation diameters were calculated from a maximal projection of 20 µm of the 2PEF image stack (ImageJ) taken 30 min after the microhemorrhage, with the stack centered on the depth where the microhemorrhage was created. The researcher making the diameter measurements was blinded to the treatment group. ### Post-mortem histology {#s4c5} At the end of each imaging session, animals were transcardially perfused with 30 ml of phosphate buffered saline (PBS) (Sigma-Aldrich) and 100 ml of 4% (wt/vol) paraformaldehyde (Fisher Scientific) in PBS. The brain was extracted from the skull and cryoprotected by immersion in 30% (wt/vol) sucrose in PBS for 24 hrs and then in 60% sucrose in PBS. Fiducial marks were made in the left corners of the craniotomy window by vertically inserting a 25-gauge needle 2 mm into the brain. The needle was externally tinted with black ink (Parker). The fiducial marks were placed in known locations relative to the microhemorrhage sites. The brain was then frozen and cut into 50 µm thick coronal sections on a cryostat. The sections were mounted onto microscope slides (Superfrost Plus, Fisher Sci.) and incubated for 12 min with diaminobenzadine (DAB) (Peroxidase Substrate kit, Vector labs, No. SK-4100) to stain endogenous peroxidase in RBCs and then rinsed with deionized water. The sections were photographed under brightfield and fluorescence microscopy. Fiducials were mapped and used to identify the slices and location within slices where microhemorrhages were located. Microhemorrhages were identified on the basis of both DAB stained RBCs and Texas red-dextran fluorescence in the parenchymal space ([Fig. S2A--C](#pone.0054203.s002){ref-type="supplementary-material"}). ### Penetrating arteriole volume blood flux and laser energy used to make hemorrhage do not influence hemorrhage sizes {#s4c6} We tested whether there was any correlation between the measured RBC and plasma extravasation diameters and the volume blood flux of the targeted PA or the laser energy used to trigger the microhemorrhage. Within the range of laser energies and PA blood fluxes studied here, there were no significant correlations between the RBC and plasma diameters with penetrating arteriole flux rates or deposited energy across all three treatment groups (Pearson product-moment correlation coefficient for RBC hemorrhage diameter, Spearman's rank correlation coefficient for plasma diameter). Mouse Model of Subarachnoid Hemorrhage {#s4d} -------------------------------------- ### Experimental procedures {#s4d1} In male CD-1 mice, isoflurane anesthesia (1.5% to 2%) was induced with spontaneous respiration in a nitrous oxide--oxygen mixture. A heating pad was used to maintain the body temperature at 37.0±0.5°C. A blunted 4--0 monofilament nylon suture was introduced into the external carotid artery and advanced through the internal carotid artery to the left anterior cerebral artery (ACA) near the anterior communicating artery, where resistance was encountered and a severe decrease was seen in cerebral blood flow. The filament was advanced 3 to 5 mm further to perforate the ACA, then immediately withdrawn. SAH was confirmed by the persistence of reduction in CBF [@pone.0054203-Sozen1]. Then, the right jugular vein was prepared and the catheter for drug infusion was inserted as described above. Thirty min after SAH induction, saline or tPA was infused as described above. ### Functional outcome and quantification of subarachnoid blood volume {#s4d2} Twenty-four hrs after SAH induction, mortality and functional outcome was assessed using a modified 18-point scoring system in a blinded fashion [@pone.0054203-Sugawara1]. Subarachnoid blood volume was also assessed by means of a previously published photometric hemoglobin assay method, with slight modifications [@pone.0054203-Sugawara1], [@pone.0054203-Lee1]. In brief, mice were perfused using PBS and sacrificed under deep anesthesia. Brains were dissected and removed with the dura intact to avoid the loss of any subarachnoid blood, and whole brain placed in glass tubes containing 3 ml PBS. Hemoglobin content was measured as described above. Mice that died within the first 24 hrs after SAH induction could not undergo perfusion. In these cases, we subtracted 2.1 µl from the total blood volume derived from the hemoglobin assay, as described above. SDS-PAGE Gelatin Zymography {#s4e} --------------------------- Gelatin zymograms were used to measure the levels of MMP-2 and MMP-9 in brain homogenates from ICH mice and from mice that underwent focal cerebral ischemia as described elsewhere in detail [@pone.0054203-Asahi1], [@pone.0054203-Hayakawa1]. For the ICH group, mice were treated with saline (n = 5) or 10 mg/kg tPA (n = 5) intravenously at 3 hrs after collagenase injection. Following transcardial perfusion with ice-cold PBS under deep anesthesia, brains were extracted at 6 hrs after ICH induction. For the ischemia group, the middle cerebral arteries were occluded with intraluminal sutures for 3 hrs. At the time point of reperfusion, mice were treated with saline (n = 5) or 10 mg/kg tPA (n = 5) intravenously. Brains were extracted after 3 hrs of reperfusion. Brains were divided into ipsilateral and contralateral hemispheres, then frozen immediately in liquid nitrogen and stored at −80°C. Samples were homogenized in lysis buffer including protease inhibitors on ice. After centrifugation, supernatant was collected, and total protein concentrations were determined using the Bradford assay (Bio-Rad). Prepared protein samples were loaded and separated by 10% Tris-glycine gel with 0.1% gelatin as substrate. MMP activity was quantified via standard densitometry. MMP indexes were calculated based on purified MMP-2 and -9 standards. Statistical Analysis {#s4f} -------------------- Statistical analysis was performed using SPSS 19.0 (IBM, Armonk, NY) and StatsDirect v.2.7.8 (Cheshire, UK). Data were tested for normality and variance homogeneity. We used one-way ANOVA to determine if there were differences across the three treatment groups. If statistically significant differences were found by ANOVA, we performed pair-wise comparisons between groups using the Bonferroni multiple-comparisons correction. Depending on scale level and data distribution, t-test or Mann Whitney U-test were used to compare means between two groups, respectively. p-values less than 0.05 were considered statistically significant. Supporting Information {#s5} ====================== ###### Functional outcome 24 hrs after ICH induction was assessed by means of a standard hanging wire test (time to fall-off, maximum 60 sec, three attempts per mouse). No difference was observed between saline- and tPA-treated mice. (TIF) ###### Click here for additional data file. ###### Representative coronal post-mortem tissue sections of microhemorrhages from animals treated with saline (a, d), tPA (b, e) or heparin (c, f). Panels (a-c) show white-light transmission images of DAB treated sections, which show the RBCs that have hemorrhaged into the tissue as black. Panels (d-f) show fluorescence images of extravasated Texas Red-dextran for the same sections as panels (a-c). (TIF) ###### Click here for additional data file. ###### Gelatin zymography of MMP responses. Representative zymogram gels showing MMP-2 and MMP-9 levels in brain homogenates derived from middle cerebral artery occlusion (MCAO) (A) or primary ICH (B) mice. In both MCAO and ICH models, MMP-2/−9 levels on the contralateral side were not affected by tPA treatment, indicating that tPA treatment does not change MMP-2/−9 baseline levels. In contrast, in MCAO mice, MMP levels on the ipsilateral side were increased by tPA treatment (see [Fig. 5](#pone-0054203-g005){ref-type="fig"} for more details). (TIF) ###### Click here for additional data file. We thank IMRA America, Inc. for the loan of laser equipment. [^1]: Competing Interests:The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: CF CS EL. Performed the experiments: NR FC FS AL KY KvL EM KA XF. Analyzed the data: NR FC KY KvL EM KA CS XW. Wrote the paper: CF NR HS CS EL.	{ "pile_set_name": "PubMed Central" }
27.1 Introduction {#Sec1_28} ================= Effective communicable disease control relies on high- quality disease surveillance, which is the systematic and regular collection of information on the occurrence, distribution, and trends of an event on an ongoing basis with sufficient accuracy and completeness to provide the basis for action. Such a system therefore provides information for planning, implementation, monitoring, and evaluation of public health programs. It includes case detection and registration, case confirmation, data reporting, data analysis, outbreak investigation, response and preparedness activities, feedback, and communication. Health authorities must also provide appropriate supervision, training, and resources for the surveillance system to operate properly. The Georgian National Health Policy, adopted in 1999, declares the reduction of communicable and socially dangerous diseases to be a major priority for maintaining and improving the health of the Georgian population over the next decade. 27.2 Surveillance System in Georgia {#Sec2_28} =================================== Ideally, a surveillance system is sensitive enough to correctly identify all cases of a particular disease occurring in the community, especially in the case of dangerous pathogens, when even a single case must be detected. All clinically diagnosed or laboratory confirmed cases of communicable diseases that come to health facilities for treatment or consultation, regardless of whether they are reported urgently or once a month, must be registered. The usefulness of public health surveillance data depends on its uniformity, simplicity, and timeliness. State and local public health officials use the information about occurrence of diseases to accurately monitor trends, plan and make decisions, and evaluate the effectiveness of interventions. Established surveillance systems should be regularly reviewed based on the explicit criteria of usefulness, cost, and quality, and existing systems should be modified as a result of such reviews. Attributes of quality include: (1) sensitivity; (2) specificity; (3) representativeness; (4) timeliness; (5) simplicity; (6) flexibility; and (7) acceptability. The sensitivity of a surveillance system is its ability to detect health events (completeness of reporting). Its specificity is inversely proportional to the number of false-positive reports. Representativeness can be measured by comparing surveillance data covering part of the population to either nationwide data, where available, or to random samplesurvey data. Simplicity in a system means it is easy to understand and implement, and is usually expected to be relatively cheap and flexible. A flexible system can easily be adapted by adding new disease entities, conditions, and/or by extending surveillance to additional population groups. Acceptability depends on the perceived public health importance of the event under surveillance, recognition of individual contributions, and the time required for reports. Internationally regulated dangerous infections, such as plague, cholera, yellow fever, poliomyelitis, viral hemorrhagic fevers, tularemia, anthrax, rabies, SARS, smallpox, tick-borne encephalitis (TBE), and influenza caused by a new virus subtype, must be reported to officials immediately. Urgent notification must also be done for groups of cases of any infectious disease, excluding acute respiratory infections and influenza. 27.3 The National Center for Disease Control and Medical Statistics (NCDC) of Georgia {#Sec3_28} ===================================================================================== The NCDC was founded in 1996 to succeed the Georgian Station for Plague Control. An integral part of the Georgian Public Health system, the NCDC reports to the Ministry of Labor, Health and Social Affairs of Georgia. The main responsibilities of NCDC include conducting surveillance on communicable and non- communicable diseases; controlling important diseases; carrying out preventive measurements; promoting healthy lifestyles; and gathering and processing medical statistical data. In addition, NCDC houses the Georgian National collection of especially dangerous pathogens. The NCDC network comprises 11 regional public health centers (CPH) and 66 district (rayon) CPH. Medical facilities and physicians are also part of the network. Surveillance on communicable and non-communicable diseases, including especially dangerous pathogens, continues to be one of the main responsibilities of the NCDC. All institutions and providers delivering healthcare services to the population regardless of their ownership, including laboratories and private care providers, must notify the local public health service whenever they diagnose, suspect, or even receive positive laboratory results that might indicate a disease of interest. The NCDC determines and annually updates the list of notifiable and reportable diseases based on the current epidemiological situation. Following the collapse of Soviet Union, the Republic of Georgia found itself in a very difficult economic situation. The healthcare system seemed nearly non-existent, and only international healthcare agencies were operating with the support of international organizations by conducting surveillance on communicable and non-communicable diseases. With the help of these agencies, NCDC succeeded in control and prevention of diseases important to the public health through health promotion programs, gathering and processing medical statistical data, running immunization programs, and so forth. During the 1990s, there were numerous achievements, like situation analysis and public health strategy design and piloting. Communicable disease surveillance guidelines for CPHs and healthcare providers, job aids for district-level CPHs and facility workers, etc., were established; human resource capacity was strengthened; district-level training and continuous supervision and support were carried out; and nationwide policy (endorsed by MoLHSA Decrees) was developed and implemented. Standard case definitions were determined for AFP/polio, measles, diphtheria, mumps, rubella/CRS, pertussis, tetanus, acute viral hepatitis, rabies, shigellosis, salmonellosis, cholera, bacterial meningitis, and influenza H5N1. Case definitions for anthrax, plague, tularemia, brucellosis, TBE, hemorrhagic fevers (HFRS, CCHF), and other diseases are forthcoming. The NCDC has research potential as well. About 60% of the staff members are specialists with university educations, and 32% of them are doctors of science or are candidates for scientific doctorates. The list of implemented and ongoing projects jointly funded by the United States and other international agencies demonstrates the diversity of their expertise:Internaional Training and Research in Emerging Infectious Diseases (1997--2002), with Fogarty International Center, National Institutes of HealthEstablishing Epidemiological Network on the Territory of Georgia (1997), with the "Open Society Georgia" FoundationImprovement of Epidemiological Network in Georgia (1998), with the "Open Society Georgia" FoundationReproductive Health Survey (1999--2000), with United Nations Population Fund (UNFPA), United Nations Children\'s Fund (UNICEF), United States Agency for International Development (USAID), UN High Commissioner for Refugees (UNHCR), American International Health Alliance (AIHA), and the Centers for Disease Control and Prevention (CDC)Nutritional Status of Children Under Five Years of Age in Six Regions of Georgia (2000--2001), with the USAID/Save the Children-US, Georgia Field OfficeProvision of Epidemiological Survey Services on Baku--- Tbilisi---Ceyhan Pipeline Route (2003), with and the British Petroleum Company, International Training and Research in Emerging Infectious Diseases (ITREID), the Fogarty International Center and NCDCReproductive Health Survey 1999--2000, conducted jointly by the NCDC (Tbilisi, Georgia) and CDC (Atlanta, GA), funding provided by UNFPA, UNICEF, USAID, UNHCR, and AIHA; Nutritional Status of Children Under Five Years of Age in Six Regions of Georgia: 2000--2001; c onducted jointly by the NCDC (Tbilisi, Georgia), with technical assistance provided by Irwin Shorr (private consultant) and the CDC (Atlanta, GA), and managed by the Georgia Field Office of "Save the Children"-USA; funding provided by USAIDEnhanced Epidemiologic and Laboratory Diagnostic Capacity for the Control of Botulinum Intoxication in Georgia. Partner: U.S. Department of Health and Human Services (DHHS)/Biotechology Engagement Program (BTEP); Collaborators: CDC (Atlanta, GA)Prevention of Amebiasis and Creation of Diagnostic Test-Systems for E. histolytica Strains Isolated in Georgia. Partner: U.S. DHHS/BTEP; Collaborators: University of Virginia Health System, Charlottesville, VA; Molecular Epidemiology and Antibiotic-Resistance of Bacterial Infections in Georgia. Partner: U.S. DHHS/BTEP. Collaborators: University of Maryland School of Medicine, Department of Epidemiology and Preventive Medicine, Baltimore, MDClinical and Molecular Epidemiology of Drug-Resistant Tuberculosis in the Republic of Georgia and the Caucasus. Partner: U.S. DHHS/BTEP; Collaborators: Emory University, Atlanta, GAEpidemiology, Molecular Characteristics and Clinical Course of HCV Infection in Georgia. Partner: U.S. DHHS/BTEP; Collaborators: Johns Hopkins University, Baltimore, MDEcology, Genetic Clustering, and Virulence of Yersinia pestis Strains Isolated from Natural Foci of Plague in Georgia. Funding Agency: DTRA, etc. The U.S. Defense Threat Reduction Agency (DTRA) has begun implementation of a project aiming to improve the surveillance system through standardized and repeatable disease monitoring systems, mobile epidemiological response teams, and secure transportation of infectious agents; systems of communications and information technology, including electronic communicable disease reporting system; biosafety and physical security of central reference laboratories, safe transportation of pathogens, and to provide help in promulgating new national rules and regulations, and their relations to BWPPP. All these elements are supplied with verifiable training in their field. A central Epidemiological Monitoring Station (EMS) was arranged at NCDC. It is equipped with modern sophisticated equipment, like light cyclers, RT- PCR, and so forth. EMS carries out surveillance on especially dangerous infections. Here, the first avian flu case in Georgia was identified. The Republic of Georgia is benefiting greatly from these projects. Biosecurity and biosafety at biological facilities will increase, and the disease surveillance infrastructure and capabilities with state-of-the-art technology will be improved, leading to detection and response to outbreaks or epidemics in a timely manner and improvement of early warning systems and data collection necessary for action linked to control measures. This sustainable disease surveillance system will continue to benefit the Republic of Georgia through reduced risk of disease proliferation risk. The laboratory capacity of NCDC reference laboratories has been improved, promoting opportunities to carry out various collaborative research endeavors.	{ "pile_set_name": "PubMed Central" }
1. Prostate Development {#sec1-jdb-04-00030} ======================= The prostate is a male sex-accessory gland that contributes secretions to the ejaculate. The mouse prostate is comprised of a highly branched ductal network in a loose fibro-vascular stroma arranged into anterior, dorsolateral and ventral lobes. The overall appearance and ratio of epithelium and stroma varies significantly between species but the basic mechanisms controlling prostate development appear to be conserved as evidenced by tissue recombination studies using mouse, rat, rabbit, and human prostate mesenchyme and epithelia \[[@B1-jdb-04-00030]\]. The most widely used animal model for prostate development and morphogenesis is the mouse. The prostate develops from the urogenital sinus (UGS), a derivative of endodermal tissue. The first morphological event in prostate development is budding of urogenital sinus epithelium (UGE) into the surrounding urogenital sinus mesenchyme (UGM). This occurs in the mouse at embryonic day 16.5 (E16.5). During prenatal development, these epithelial buds elongate and form solid, unbranched ducts. Postnatal, the solid epithelial ducts elongate, canalize, and undergo branching morphogenesis. Extensive branching occurs by postnatal day 15 (P15) and is largely completed by P30 \[[@B2-jdb-04-00030]\]. This morphogenetic process results in a mature prostate with an intricate branched ductal system that consists of three paired lobes: the ventral prostate (VP), the anterior prostate (AP), and the dorsolateral prostate (DLP). Each prostatic lobe has a distinct branching pattern \[[@B2-jdb-04-00030]\]. 2. Hedgehog Signaling in the Developing Prostate {#sec2-jdb-04-00030} ================================================ In the developing prostate, Shh and Ihh are expressed by the UGS epithelium and induce Hh target gene expression in the adjacent mesenchyme ([Figure 1](#jdb-04-00030-f001){ref-type="fig"}; \[[@B3-jdb-04-00030],[@B4-jdb-04-00030],[@B5-jdb-04-00030],[@B6-jdb-04-00030],[@B7-jdb-04-00030]\]). Mesenchymal target genes include the conserved Hh targets Ptc, Gli1, Ptc2 and Hip1 \[[@B5-jdb-04-00030]\] but also include a number of specific genes such as IGFBP-6, Angiopoetin4 (Agpt4), IGFBP-3, Inhibin beta-B and TIMP3 \[[@B8-jdb-04-00030],[@B9-jdb-04-00030]\]. Several investigators noted minor expression of Ptc and Gli1 at the tips of the growing ducts suggesting a minor component of autocrine signaling \[[@B10-jdb-04-00030],[@B11-jdb-04-00030]\]. 3. Requirement for Prostate Development {#sec3-jdb-04-00030} ======================================= Shh is the most abundantly expressed Hh ligand in the developing prostate \[[@B12-jdb-04-00030]\]. Shh expression in the UGS epithelium increases with the onset of ductal budding. Abundant gene expression is maintained until the beginning of ductal branching at which point Shh begins to steadily decrease and eventually taper off to a low but detectable level in the adult \[[@B3-jdb-04-00030],[@B4-jdb-04-00030]\]. Shh expression is localized to the epithelium at the distal tip as the ducts elongate and branch. During this time Ptc1 and Gli1 are localized to the mesenchyme directly around the emerging buds \[[@B4-jdb-04-00030]\]. Several laboratories have studied the role of Hh signaling in prostate development. The earliest studies used a polyclonal antibody to neutralize Shh signaling and noted inhibition of prostate morphogenesis in the grafted UGS \[[@B3-jdb-04-00030]\]. Subsequent studies showed prostate development was not dependent on Shh signaling since the UGS from a Shh null embryo exhibited normal prostate morphogenesis when grafted \[[@B5-jdb-04-00030],[@B6-jdb-04-00030],[@B7-jdb-04-00030]\]. However; we then observed that Ihh expression was increased in the Shh null UGS and provided functional redundancy that rescued Hh signaling \[[@B12-jdb-04-00030]\]. Whether Hh signaling is strictly required for prostate morphogenesis remains to be determined. 4. Effects on Growth and Ductal Morphogenesis {#sec4-jdb-04-00030} ============================================= Initially it was reported that chemical inhibition of Hh signaling in the cultured UGS resulted in decreased epithelial proliferation and branching \[[@B4-jdb-04-00030]\]. However, later studies reported a myriad of effects on branching during experimental approaches to increase or decrease Hh signaling in the cultured postnatal prostate\[[@B5-jdb-04-00030],[@B6-jdb-04-00030],[@B7-jdb-04-00030]\]. Wang reported inhibition of Hh signaling increased overall epithelial proliferation \[[@B5-jdb-04-00030]\]. However, the effects were regionalized along the duct. In fact, epithelial proliferation was significantly and selectively decreased in the distal, less differentiated part of the duct \[[@B5-jdb-04-00030],[@B6-jdb-04-00030]\]. We postulated that inconstant findings reported by different laboratories might be attributed to the studies being done at different stages of prostate development \[[@B5-jdb-04-00030]\]. This was subsequently shown to be the case. Using a combination of chemical inhibition of Hh signaling and transgenic activation of Hh signaling, it was shown that paracrine Hh signaling promotes epithelial proliferation and budding prenatally while inhibiting epithelial proliferation and branching postnatally. These differential effects were mirrored by stage-specific differences in Hh target gene regulation \[[@B8-jdb-04-00030]\]. These studies established paracrine Hh signaling as a primary regulator of ductal morphogenesis and epithelial proliferation. Stage dependent effects are associated with an evolving palette of target gene regulation and reveal that the effects of paracrine Hh signaling are dependent upon the mesenchymal microenvironment. Beachy and colleagues performed elegant studies localizing Hh ligand expression and pathway activity during branching morphogenesis in the regenerating prostate \[[@B13-jdb-04-00030]\]. They found both Shh and Ihh to be expressed at significant levels but observed focally decreased Hh expression and pathway activity at sites of ductal branching. Manipulations to decrease Hh pathway activity increased ductal branching and this was associated with increased stromal expression of hepatocyte growth factor (HFG) a positive regulator of ductal branching. Ongoing studies are yielding additional evidence of cross-talk between Hh and other regulators of prostate development. Shh expression is diminished in Sox9 genetically deficient mice \[[@B14-jdb-04-00030]\] but prolonged in FoxA1 deficient mice \[[@B10-jdb-04-00030]\]. Pu et al. \[[@B11-jdb-04-00030]\] demonstrated that exogenous Shh decreased FGF10 expression in cultured prostate tissues. 5. Autocrine Signaling {#sec5-jdb-04-00030} ====================== We performed preliminary studies to examine the role of autocrine Hh signaling in the prostate. Using Ptc1^lacZ^ and Gli1^lacZ^ reporter mice to localize Hh pathway activation, we observed robust expression of Ptc1^lacZ^ and Gli1^lacZ^ in the urogenital mesenchyme. Most epithelium was unstained, but we observed strong lacZ staining in a small number of epithelial cells in the nascent prostate ducts at P1 and P5 ([Figure 2](#jdb-04-00030-f002){ref-type="fig"}). Staining of the adult prostate showed relatively infrequent and scattered epithelial staining (data not shown). These observations suggest that autocrine Hh pathway activation occurs in a small number of epithelial cells specifically during prostate development. To examine the role of autocrine signaling specifically, we abrogated epithelial Hh signaling by conditional deletion of the essential pathway component Smoothened (Smo) using a Shh^cre^ construct B6.Cg-Shh^tm1(EGFP/cre)Cjt^/J (stock 005622). We have previously shown that Shh mRNA expression in the developing prostate is restricted to the epithelium \[[@B3-jdb-04-00030],[@B4-jdb-04-00030]\]. When the Shh^cre^ mouse was bred to the lacZ reporter mouse Rosa26 (B6.129S4-Gt(ROSA)26Sor\<tm1Sor\>/J 003474) X-gal staining of the UGS from Shh^cre^;Rosa26 mice exhibited robust positive staining restricted to the epithelium ([Figure 3](#jdb-04-00030-f003){ref-type="fig"}) and Mehta et al. \[[@B15-jdb-04-00030]\]. Shh^cre^ was bred to Smo^c/c^ (Smo^c/c^: STOCK Smo^tm2Amc/J^ 004526) to selectively delete Smo from all cells expressing Shh, including the UGS epithelium. Shh^cre^;Smo^c/c^ mutant pups were small with abnormalities in limb, digits and tail and died as newborns (data not shown). The P1 UGS from Shh^cre^;Smo^c/c^ mutants and controls were grafted under the renal capsule of adult nude males \[[@B12-jdb-04-00030],[@B16-jdb-04-00030]\]. Histologic examination after eight weeks showed well-developed ductal structures and differentiated epithelium and stroma in both groups ([Figure 4](#jdb-04-00030-f004){ref-type="fig"}, top panels). 6. Prostate Regeneration {#sec6-jdb-04-00030} ======================== Development and maintenance of the prostate is exquisitely dependent on androgen. Castration of the adult results in widespread epithelial apoptosis and glandular involution. Regeneration of the prostate occurs with administration of exogenous testosterone and involves robust cell proliferation, branching morphogenesis and complete reconstitution prostate architecture and differentiation. The prostate's unique regenerative capacity has stimulated interest in the role of progenitor cells in this process and the study of prostate stem cells has been an active area of investigation. We performed studies showing that castration induces a selective expansion of slow-cycling progenitor cells. The implication of this finding is that these are "androgen-independent" progenitor cells that could play an important role in prostate regeneration. Interestingly, we found that both Shh and Gli expression are increased significantly in response to castration \[[@B17-jdb-04-00030]\] and observed robust Hh signaling when progenitor cells are grown in anchorage independent culture \[[@B18-jdb-04-00030]\]. These findings suggested a role for Hh signaling in progenitor cell expansion and the capacity for prostate regeneration. A role for Hh signaling in prostate regeneration has been previously recognized \[[@B19-jdb-04-00030]\] demonstrated that inhibition of Hh signaling abrogated prostate regeneration but did not distinguish between autocrine and paracrine signaling. Using grafted P1 UGS from Shh^cre^;Smo^c/c^ mutants and controls, we characterized the effect of disrupted autocrine signaling on the response to castration and testosterone supplement. Eight weeks after grafting, mice were castrated and examined 14 days later. We observed a truly striking difference. Whereas the control grafts exhibited well-formed ducts with cuboidal epithelium, the mutant grafts lacked well defined cuboidal epithelial cells. Grafted tissues castrated x 14 days and then testosterone supplemented x 14 days revealed an equally exciting disparity in re-growth. Whereas control grafts recovered normal morphology comparable to the intact tissue, the mutant grafts exhibited an atrophic appearance with fewer, dilated ductal lumen and minimal epithelial re-growth ([Figure 4](#jdb-04-00030-f004){ref-type="fig"}, bottom panels). These previously unpublished observations indicate that disruption of autocrine signaling may not perturb prostate development but exaggerates the response to castration and impairs regeneration in response to testosterone. A proposed model for the complex roles of autocrine and paracrine Hh signaling in prostate development and capacity for regeneration is shown ([Figure 5](#jdb-04-00030-f005){ref-type="fig"}). 7. Prostate Cancer {#sec7-jdb-04-00030} ================== Hh signaling is active in human prostate cancer and evidence for both autocrine and paracrine signaling in human prostate cancer has been presented. This appears to involve both ligand-dependent signaling as well as non-canonical pathway activation. Paracrine signaling has been modeled using xenograft studies and shown to promote tumor growth \[[@B20-jdb-04-00030]\]. Interestingly, a correlation analysis of Shh target gene expression in human prostate cancer revealed dichotomous expression of prenatal and postnatal target genes in the tumors. In those tumors with a myofibroblastic stroma resembling the mesenchyme of the developing prostate, Shh and Gli1 expression correlated with expression of prenatal target genes. In those with a more mature, myofibroblast poor stroma, there was correlation with expression of postnatal target genes. These findings suggest that the effect of paracrine signaling in prostate cancer as in development, is dependent upon the stromal context and microenvironment in which it occurs. The role of autocrine signaling in prostate cancer is debated. Several studies appeared to show autocrine signaling in prostate cancer cell lines cultured in vitro \[[@B19-jdb-04-00030],[@B21-jdb-04-00030]\] but this has been contested by two other studies \[[@B22-jdb-04-00030],[@B23-jdb-04-00030]\]. Hh signaling appears to be consistently up-regulated in metastatic and hormone refractory prostate cancer and several studies suggest that Hh pathway activation may be a critical step in tumor metastasis and progression to androgen independence \[[@B19-jdb-04-00030],[@B21-jdb-04-00030],[@B24-jdb-04-00030]\]. A proposed model for the roles of autocrine and paracrine Hh signaling in prostate cancer is shown ([Figure 6](#jdb-04-00030-f006){ref-type="fig"}). 8. Summary and Conclusions {#sec8-jdb-04-00030} ========================== Shh is the dominantly expressed ligand in the developing prostate. In absence of Shh, Ihh expression is up-regulated and provides functional compensation. In the regenerating adult prostate, both Shh and Ihh contribute to pathway activity and inhibit ductal branching by restricting the expression of HGF. Paracrine signaling in the developing prostate plays an important role in regulating growth and ductal development. The growth effects and target gene regulation are stage-dependent. There is evidence for autocrine signaling in the developing prostate but its role is unclear. Preliminary studies suggest it may have an important role in the capacity to survive castration and enable ductal regeneration. The role of Hh signaling in development and regeneration is recapitulated in cancer. There is strong evidence for paracrine signaling in prostate cancer. The effect of paracrine signaling may be critically dependent on the phenotype of the tumor stroma. This has major implications for the design of clinical trials of Hh inhibitors in prostate cancer. The role of autocrine signaling remains ill-defined. Our studies suggest a link between autocrine signaling and the capacity of epithelial cells to survive castration and there is a suggestion of pathway activation in high grade, androgen independent prostate cancer. It is as yet unclear if this is ligand-dependent or due to non-canonical activation mechanisms. I wish to gratefully acknowledge Min Yu's provision of the data presented in [Figure 2](#jdb-04-00030-f002){ref-type="fig"} and [Figure 3](#jdb-04-00030-f003){ref-type="fig"}. The author declares no conflict of interest. ![Whole mount in situ hybridization and sections of the P1 urogenital sinus (UGS). (m) mesenchyme (e) epithelium \[[@B4-jdb-04-00030]\].](jdb-04-00030-g001){#jdb-04-00030-f001} ![X-gal/PanCK staining of P1/P5 prostate from Gli1^lacZ^ and Ptc1^lacZ^ mice showed strong positive beta-gal activity (blue) in the mesenchyme generally and in a limited number of epithelial cells (brown) in the nascent ducts (arrowheads). N = 3, scale bar = 40 µm.](jdb-04-00030-g002){#jdb-04-00030-f002} ![Staining for beta-gal in E16 UGS of Shh^cre^;Rosa26R mice. Note efficient recombination through the UGS epithelium (UGE) and bladder epithelium. UGM: UGS mesenchyme; n = 3.](jdb-04-00030-g003){#jdb-04-00030-f003} ![In vivo subcapsular grafts of P1 UGS from control and Shh^cre^;Smo^c/c^ mutants. Histologic appearance after eight weeks growth (top), two weeks after castration (middle) and two weeks after testosterone supplement. n = 3.](jdb-04-00030-g004){#jdb-04-00030-f004} ![Hh signaling in the developing prostate involves both autocrine and paracrine signaling. Autocrine signaling drives proliferation of (androgen independent) progenitor cells at the bud tip. Paracrine signaling elicits a complex transcriptional response, determined by the stage of development, that exerts a variety of effects on epithelial proliferation and differentiation. In the prenatal prostate, Hh pathway activity drives epithelial proliferation and ductal growth. In the postnatal prostate, Hh pathway activity inhibits epithelial proliferation and ductal growth. Androgen independent progenitor cells generated by autocrine Hh signaling during development are able to survive castration and enable prostate regeneration in response to testosterone supplement.](jdb-04-00030-g005){#jdb-04-00030-f005} ![The target genes activated by paracrine Hh signaling in the adult prostate and the effect on epithelial/tumor cell proliferation is determined by the phenotype of the stroma. In the normal adult prostate and prostate cancers with a normal fibroblastic stroma the target gene profile resembles that of the postnatal prostate and the effect of paracrine Hh signaling is homeostatic. In prostate cancers with a so-called "reactive" (myofibrobastic) stroma, the target gene profile resembles that of the prenatal prostate and the effect of paracrine signaling is to promote epithelial/tumor cell proliferation. Autocrine signaling in tumor stem cells may enable these cells to survive castration and enable androgen independent tumor growth.](jdb-04-00030-g006){#jdb-04-00030-f006}	{ "pile_set_name": "PubMed Central" }
Introduction {#section1-2192568218765400} ============ The number of cervical spine surgeries performed in the United States has increased over the past several decades.^[@bibr1-2192568218765400]^ One of the goals of surgical correction is restoration of normal anatomic alignment; however, age-based normative values for cervical parameters have not been well defined. Prior studies have reported radiographic parameters in asymptomatic subjects, but these studies dichotomized groups into \<30 versus \>60 years old,^[@bibr2-2192568218765400]^ but failed to provide a comprehensive analysis of cervical parameters, including overall (C0-C7) lordosis, horizontal gaze parameters, and cervical sagittal vertical axis,^[@bibr2-2192568218765400][@bibr3-2192568218765400]--[@bibr4-2192568218765400]^ and did not obtain baseline health-related quality of life (HRQOL) measurements.^[@bibr2-2192568218765400],[@bibr3-2192568218765400],[@bibr5-2192568218765400][@bibr6-2192568218765400]--[@bibr7-2192568218765400]^ Additionally, studies have been limited by the inclusion of mixed populations of asymptomatic and surgical patients,^[@bibr5-2192568218765400],[@bibr7-2192568218765400]^ focused on occipitocervical, rather than cervical, relationships^[@bibr8-2192568218765400]^ and reporting of patients from other countries,^[@bibr2-2192568218765400],[@bibr6-2192568218765400],[@bibr8-2192568218765400]^ which may differ from a North American population based on variability of spinal parameters between races.^[@bibr9-2192568218765400],[@bibr10-2192568218765400]^ Furthermore, cervical alignment correlates with HRQOL scores, as demonstrated by Tang et al,^[@bibr5-2192568218765400]^ and Villavicencio et al^[@bibr11-2192568218765400]^ who reported more favorable clinical outcomes in patients with cervical sagittal vertical axis (SVA) less than 40 mm and greater cervical lordosis after cervical spine surgery. Recent studies have described age-based changes among spinopelvic parameters^[@bibr12-2192568218765400]^ as well as reciprocal changes in cervical sagittal alignment secondary to spinopelvic parameters.^[@bibr13-2192568218765400]^ Consequently, it is likely that cervical parameters change with age. Therefore, identifying age-adjusted normative cervical parameters will provide radiographic guidelines for the assessment and management of cervical spine patients. Additionally, the inclusion of thoracic and spinopelvic parameters on full-length standing radiographs will provide additional insight into the relationships between spinal segments. The purpose of this study was to investigate cervical alignment parameters in North American subjects without symptomatic spinal pathology. Our primary hypothesis was that sagittal cervical radiographic parameters vary with age. A secondary hypothesis is that the changes in spinopelvic parameters will parallel those of previous studies,^[@bibr12-2192568218765400]^ and will demonstrate relationships with the cervical spine. The identification of age-adjusted cervical parameters may provide benchmarks for both nonsurgical and surgical treatment. Methods {#section2-2192568218765400} ======= This study is a single-center retrospective review of a prospective database of 118 asymptomatic adult subjects who underwent biplanar imaging with 3-dimensional capabilities (EOS Imaging, Paris, France) between October 2014 and October 2015 to determine age-adjusted alignment parameters of the cervical spine. Institutional review board approval was obtained for the study. Patients aged 18 to 79 years were included if they did not have a history of back or neck pain and if they passed the exclusion criteria. Exclusion criteria were the following: coronal deformities (Cobb angle \>10°); a history of previous spine surgery; history of hip or knee arthroplasty or any other realignment surgery of the lower extremities; complaints of back pain or neck pain that resulted in missed work, affected activities of daily living, participation in recreational activities, or required narcotic pain medication; degenerative or pathologic condition of the spine that necessitated physician intervention; nonambulatory patients; history of neuromuscular disorders, inflammatory arthritis, or congenital anomalies; and pregnancy. Solicitation of volunteers was performed via advertising and providing a gift card with a value of \$25. Demographic data (age and body mass index \[BMI\]) and HRQOL data (Oswestry Disability Index \[ODI\] and Neck Disability Index \[NDI\] scores) were collected prior to imaging. Patients were divided into the following age groups: \<35 years, 35-44 years, 45-54 years, 55-64 years, and \>65 years. Radiographic Analysis {#section3-2192568218765400} --------------------- Measurements of cervical, thoracic, and spinopelvic parameters were collected on sagittal EOS imaging via validated and dedicated software (Spineview, ENSAM Laboratory of Biomechanics, Paris, France).^[@bibr14-2192568218765400]^ Measurements were obtained by demarcating the margins of the superior and inferior endplates. Spinal parameters were generated using Matlab software (Matlab 2015b, MathWorks, Natick, MA). Standing, full-body EOS images were obtained with the patient standing in a relaxed position of comfort with their fingers resting on their clavicles and without any specific recommendation regarding horizontal gaze. Radiographic measurements were performed by one assessor followed by subsequent verification by a second assessor. The cervical parameters included in this study were C0-C2 (angle between base of skull and inferior endplate of C2), C2-C7 (angle between inferior endplate of C2 and inferior endplate of C7), and C0-C7 lordosis (angle between foramen magnum and inferior endplate of C7); C2-C7 sagittal vertical axis (SVA; distance between a plumb line drawn from the midbody of C2 and the posterosuperior corner of C7); apex of cervical lordosis; T1 slope minus cervical lordosis (TS-CL); chin brow vertical angle (CBVA; angle subtended by a line tangential to the chin and supraorbital ridge, and a vertical reference line); slope of line of sight (SLS; angle between the horizontal and a line from the infraorbital rim and through the inner ear); McGregor slope (McGS; the angle subtended by a line extending from the hard palate and opisthion and a horizontal reference line); and the contribution of the upper cervical angle to total cervical lordosis (UpperC%), defined as C0-C2 divided by C0-C7 ([Figure 1a and b](#fig1-2192568218765400){ref-type="fig"}). ![Illustrative explanation of horizontal gaze (a) and sagittal cervical parameters (b). SLS, slope of line of sight; McGs, McGregor slope; CBVA, chin brow vertical angle; C0-C2, occiput-C2 angle; C2-C7 angle; cSVA, cervical sagittal vertical axis; T1 S, T1 slope.](10.1177_2192568218765400-fig1){#fig1-2192568218765400} Measurements of the thoracic spine included the T1 pelvic angle (TPA; angle subtended by a line drawn from the centroid of T1 to the femoral head axis and a line from the midpoint of the S1 endplate to the femoral head axis), T1 slope (T1 S; angle between a horizontal reference line and the superior endplate of T1), T4-12 kyphosis, and location of the apex of kyphosis. Spinopelvic measurements were performed as in previous studies^[@bibr15-2192568218765400]^ and included sacral slope (SS), pelvic tilt (PT), pelvic incidence (PI), pelvic incidence--lumbar lordosis (PI-LL) mismatch, L1-S1 lumbar lordosis (LL), C7-S1 SVA, and location of the apex of lumbar lordosis. Statistical Analysis {#section4-2192568218765400} -------------------- The following data was collected for each of the 5 age groups: mean, range, standard deviation (SD), standard error (SE), and 95% confidence interval (CI). Statistical analyses were performed using SPSS 20.0 (IBM Corp, Armonk, NY). The normal distribution of parameters was assessed followed by analysis of variance for comparison of variance between age groups. Post hoc analysis were then completed to determine if the differences between specific age groups were statistically significant, as defined by P ≤ .05. Pearson's correlations were calculated to examine the relationship between age and cervical parameters. Correlation coefficients (r) were interpreted as follows: 0.00-0.29 no correlation, 0.30-0.49 weak correlation, 0.50-0.69 moderate correlation, 0.70-0.89 strong correlation, and 0.90-1.00 very strong correlation. Results {#section5-2192568218765400} ======= Patient Population {#section6-2192568218765400} ------------------ A total of 118 subjects were enrolled ([Table 1](#table1-2192568218765400){ref-type="table"}). The average age of the cohort was 50.5 ± 16.9 years and the average BMI measured 28 kg/m^2^. A total of 31.4% of patients were male. The mean NDI and ODI scores were 3.39 ± 4.5 and 1.64 ± 4.80, respectively. There were no statistically significant differences between groups with regard to NDI (P = .749) and ODI (P = .583) scores. ###### Stratification of Groups Based on Age (N = 118). ![](10.1177_2192568218765400-table1) n Percent ODI NDI ------------------ ----- --------- ----------- ----------- Gender Male 37 31.4 Female 81 68.6 Age group, years \<35 30 25.4 0.9 ± 3.0 3.2 ± 3.1 35-44 18 15.3 3.0 ± 7.2 4.6 ± 5.3 45-54 17 14.4 1.1 ± 3.4 3.9 ± 4.2 55-64 21 17.8 1.3 ± 4.0 3.0 ± 3.4 ≥65 32 27.1 2.1 ± 5.7 3.0 ± 5.9 Total 118 100 Abbreviations: NDI, Neck disability Index; ODI, Oswestry Disability index. Cervical Parameters {#section7-2192568218765400} ------------------- The values of cervical parameters are shown in [Tables 2](#table2-2192568218765400){ref-type="table"} and [3](#table3-2192568218765400){ref-type="table"}. C2-C7, C0-C7, and T1 S demonstrated significant increases with age. C0-C7 lordosis was not significantly different among patients \<55 years. However, patients \<35 years had less lordosis than those aged 55 to 64 (P \< .001) and ≥65 (P \< .001) years. C2-C7 lordosis was significantly greater in patients ≥65 years than in those \<45 years (P ≤ .03). Patients \<35 years were also found to be significantly less lordotic than patients aged 55 to 64 years (P = .001). Significant differences in T1 S were identified in patients \<35 versus ≥65 years (P \< .001), 35-44 versus ≥65 years (P = .005), and 45-54 versus ≥ 65 years (P = .009). T1 S demonstrated a moderate, positive correlation with age (r = 0.446, P \< .001) as shown in [Table 2](#table2-2192568218765400){ref-type="table"}. McGS, SLS, and CBVA did not change linearly with age despite statistical significance, and mean averages of all age groups were within 10° of one another. C0-C2 lordosis and ratio of upper cervical to total cervical angle (UpperC%) were not significantly different among groups. Cervical kyphosis was present in approximately half of subjects in the \<35, 35- to 44-, and 45- to 54-year age groups (56.7%, 50.0%, and 47.1%, respectively) compared with 9.5% of subjects between 55 and 64 years and 12.5% of those ≥65 years. Younger patients had a significantly higher rate of cervical kyphosis compared with older patients (P = .005). T1S-CL and cSVA were not significantly different between groups and failed to show a relationship with age. The mean apex of cervical lordosis was located at C5 in all groups. ###### Cervical Sagittal Radiographic Parameters. ![](10.1177_2192568218765400-table2) Age (Years) C0-C2 (deg) C2-C7 (deg) C0-C7 (deg) T1S-CL (deg) T1 S (deg) cSVA (mm) ------------- ------------- ------------- ------------- -------------- ------------- ------------- \<35 19.1 ± 9.7 −2.2 ± 13.8 17.0 ± 12.7 24.2 ± 10.1 22.0 ± 8.1 28.4 ± 16.7 35-44 23.3 ± 7.9 1.6 ± 7.8 24.8 ± 6.8 21.8 ± 8.0 23.4 ± 5.6 23.4 ± 14.1 45-54 21.6 ± 9.3 3.6 ± 11.1 25.2 ± 10.2 19.9 ± 8.8 23.5 ± 7.4 26.2 ± 26.2 55-64 19.7 ± 7.7 10.7 ± 9.9 30.6 ± 9.1 17.6 ± 8.9 28.4 ± 6.9 22.7 ± 15.6 ≥65 21.0 ± 7.6 11.8 ± 12.1 32.7 ± 11.0 20.6 ± 10.3 32.5 ± 11.5 30.1 ± 15.3 P .526 \<.001\* \<.001\* .170 \<.001\* .381 Abbreviations: CL, cervical lordosis; cSvA, cSVA, cervical sagittal vertical axis; T1 S, T1 slope. \Statistically significant. ###### Horizontal Gaze Radiographic Parameters. ![](10.1177_2192568218765400-table3) Age (Years) CBVA (deg) SLS (deg) McGS (deg) ------------- ------------ ------------ ------------ \<35 6.0 ± 7.1 5.4 ± 7.8 7.8 ± 8.1 35-44 −1.8 ± 6.8 −2.7 ± 5.6 0.8 ± 4.9 45-54 0.7 ± 9.6 −1.6 ± 9.3 1.6 ± 7.3 55-64 1.3 ± 8.7 −1.7 ± 9.0 0.9 ± 9.5 ≥65 1.7 ± 1.7 0.1 ± 7.7 2.2 ± 7.9 P* .033\* .004\* .007\* Abbreviations: CBVA, chin brow vertical angle; McGS, McGregor slope; SLS, slope of line of sight. \Statistically significant. Thoracic Parameters {#section8-2192568218765400} ------------------- T1 pelvic angle was similar among all age groups and failed to correlate with age (mean, 2.9° ± 1.0°). The difference in thoracic kyphosis from T4-12 was not statistically significant among groups. Mean kyphosis measured between 37° and 45° in all cohorts and the overall average thoracic kyphosis for the entire group was 41.5°. The mean apex of thoracic kyphosis was located at T7 in all groups. Spinopelvic Parameters {#section9-2192568218765400} ---------------------- Significant between-group differences were identified for PT, PI-LL mismatch, and C7-S1 SVA ([Table 4](#table4-2192568218765400){ref-type="table"}). Specifically, PT was found to be significantly greater in subjects aged 55-64 years (P* = .036) and ≥65 years (P \< .001) compared with those \<35 years. PI-LL mismatch was found to differ significantly between subjects ≥65 compared with those \<35 years (P = .001); C7-S1 SVA was not significantly different in patients \<65 years, but significant differences were observed in subjects \<35 (P \< .001), 35-44 (P = .010), 45-54 (P = .001), and 55-64 (P = .004) compared with patients ≥65 years. Pelvic incidence, sacral slope, lumbar lordosis did not vary with age. The average apex of lumbar lordosis was located at L3 in all groups. ###### Spinopelvic Parameters. ![](10.1177_2192568218765400-table4) Age (Years) SS (deg) PT (deg) PI (deg) PI-LL (deg) L1-S1 (deg) C7-S1 SVA (mm) T4-12 Kyphosis (deg) ------------- ------------ ------------ ------------- ------------- ------------- ---------------- ---------------------- \<35 39.1 ± 8.3 9.3 ± 6.7 48.4 ± 10.9 12.6 ± 11.5 61.0 ± 13.2 −35.5 ± 34.4 37.8 ± 11.7 35-44 39.8 ± 8.4 13.9 ± 7.5 53.7 ± 10.5 5 ± 11.8 58.7 ± 12.4 −11.4 ± 32.0 39.2 ± 11.9 45-54 37.2 ± 7.7 13.5 ± 6.0 50.8 ± 8.5 6.5 ± 9.2 57.4 ± 10.7 −21.5 ± 27.2 38.2 ± 10.4 55-64 39.1 ± 8.7 15.3 ± 8.2 54.4 ± 14.7 7.2 ± 12.4 61.6 ± 11.6 −12.6 ± 34.1 45.3 ± 11.9 ≥65 35.9 ± 9.9 17.3 ± 6.9 53.3 ± 11.3 12.5 ± 2.2 53.8 ± 14.4 27.2 ± 51.3 45.3 ± 14.5 P .493 .001\* .312 .003\* .151 \<.001\* .057 Abbreviations: LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; SS, sacral slope; SVA, sagittal vertical axis. \Statistically significant. Discussion {#section10-2192568218765400} ========== Previous studies have not comprehensively analyzed changes in cervical alignment with increasing age in an asymptomatic North American population. In our study, we found significant increases in C2-C7 and C0-C7 lordosis with age, without a reciprocal change in C0-C2 angle. Additionally, at least 50% of patients \<45 years are kyphotic from C0-C7. T1 S, which represents the angle of T1 to a horizontal reference line, is also significantly greater in older than in younger patients. Horizontal gaze parameters (CBVA, SLS, McGS) did not demonstrate a meaningful correlation with age. PT, PI-LL, and C7-S1 SVA are significantly different between certain age groups, which is in agreement with prior research.^[@bibr12-2192568218765400]^ C2-C7, C0-C7, and T1 S angles significantly increase with age in an asymptomatic North American cohort. Specifically, differences in C0-C7 and C2-C7 angles are present in patients \<35 compared with those ≥55 years. A mean difference in C0-C7 lordosis of approximately 15° was found between patients \<35 and ≥55 years; a similar difference was observed for C2-C7 lordosis. C0-C2 and the contribution of the upper cervical angle to total cervical lordosis (UpperC%) did not change with age, suggesting that the increase in C0-C7 lordosis is driven by an increase in C2-C7 angle without any change in C0-C2. C0-C2 lordosis may not have changed significantly because the majority of flexion-extension occurs at the occiput-C1 articulation. An increase in C0-C2 lordosis with age would theoretically reduce the amount of extension reserve, thereby negatively affecting the ability to navigate a 3-dimensional environment. Horizontal gaze parameters did not vary linearly with age and meaningful trends were not observed between groups. Mean values for McGS, CBVA, and SLS were within a 10° range between all age groups, reflecting a relatively constant position over time. The unchanged position of the head (and therefore the eyes) likely reflects a requirement of humans to visualize a constant degree of environment; changing these parameters would result in either a more downward or upward gaze, which would not be advantageous. T1 S is significantly greater in patents ≥65 compared with patients \<55 years old, which likely represents a response to increasing C2-C7 SVA and PI-LL mismatch (or decreasing lumbar lordosis) in the setting of fixed thoracic kyphosis. These changes in alignment would drive an increase in T1 S and thus greater cervical lordosis as a compensatory measure. C2-C7 SVA, a measure of cervical offset, did not increase in age as has been found with C7-S1 SVA in our study as well as previous research.^[@bibr12-2192568218765400]^ Rather, C2-C7 became more lordotic possibly to compensate for increasing C7-S1 SVA, PI-LL mismatch, and T1 S. Failure to increase cervical lordosis with age would result in an impairment of horizontal gaze in addition to greater cervical offset and presumably worsening HRQOL as C2-C7 SVA increases beyond 40 mm.^[@bibr5-2192568218765400]^ In our study, significant changes in thoracic kyphosis were not observed despite alterations in cervical and spinopelvic parameters; it appears that the cervical and lumbar spine may change with age because they are less rigid than the thoracic region. Increasing sagittal offset has been shown to negatively affect SF-36 (Short Form--36 health survey) and NDI scores in patients who underwent posterior cervical fusion^[@bibr5-2192568218765400]^ and other studies have shown worse HRQOL scores with abnormal cervical alignment.^[@bibr16-2192568218765400],[@bibr17-2192568218765400]^ However, age-related normative cervical parameters in a North American population have not been established. Hardacker et al^[@bibr4-2192568218765400]^ reported on overall cervical lordosis in asymptomatic subjects but failed to include a comprehensive analysis of other cervical parameters and did not stratify subjects by age. Park et al^[@bibr2-2192568218765400]^ compared cervical parameters in asymptomatic subjects who were less than 30 years of age and those who were at least 60 years, thus excluding patients between 30 and 60 years. In our study, we stratified patients into 5 age groups ranging from \<35 to \>65 years, and therefore included adult subjects of all ages. Additionally, radiographs in our study were obtained with biplanar imaging with 3-dimensional capabilities, which is in contrast to previous studies that used standard radiography.^[@bibr2-2192568218765400][@bibr3-2192568218765400]--[@bibr4-2192568218765400]^ Computer-assisted radiographic measurement, as used in our study, has been shown to have less standard measurement error than manual measurement for the determination of Cobb angles.^[@bibr18-2192568218765400]^ In our study, we found that patients aged \>55 years had, on average, approximately 15 more degrees of C0-C7 lordosis than those \<35 years. However, C0-C2 was not found to be different between groups; consequently, the C2-C7 segment became significantly more lordotic with age as shown by the greater C2-C7 lordosis in patients \>55 compared with those \<35 years. These findings are in agreement with Park et al,^[@bibr2-2192568218765400]^ who reported no difference in C0-C2 angle with age but did find greater C2-C7 lordosis in subjects \>60 compared with those \<30 years. Although we found a similar trend of greater lordosis in older subjects, the average alignment in the \<35 age group measured −2.2° with an average difference of 12.9° and 14° from patients in the 55-64 and ≥65 age groups, respectively. This is greater than the 4.7° difference between younger and older subjects reported by Park et al.^[@bibr2-2192568218765400]^ Yukawa et al^[@bibr3-2192568218765400]^ studied 1230 asymptomatic Japanese subjects and similarly concluded that C2-C7 lordosis increased with age. Their results stratified age groups by decade and direct comparisons are not possible. However, subjects in their third and fourth decades exhibited greater C2-C7 lordosis than patients \<35 years in our study (mean, 8.4° vs −2.2°) and those in their seventh and eighth decades were more lordotic than patients in our study who were ≥65 years (mean, 18.7° vs 11.8°). The differences between the studies may reflect variations in C2-C7 lordosis between Japanese and North American populations. Le Huec et al^[@bibr8-2192568218765400]^ found that C2-C7 lordosis averaged 4.8° in 106 asymptomatic volunteers and had a moderate, positive correlation with C7 slope (r* = 0.516, P \< .001). In our study, we measured T1 S rather than C7 slope, but also found that T1 S increased with C2-C7 lordosis. That T1 S changes with age has not been reported. Cervical parameters were not stratified by age in the study by Le Huec et al,^[@bibr8-2192568218765400]^ though 33.96% of patients presented with cervical kyphosis. In our study, the overall rate of cervical kyphosis was 33.9%, which is identical to that of Le Huec et al.^[@bibr8-2192568218765400]^ As expected, cervical kyphosis occurred more commonly in younger patients (≥50% of patients \<45 years) than in older patients (≤12.5% of patients ≥55 years). The kyphotic cervical alignments may be secondary to subjects' postures during EOS imaging rather than fixed kyphosis as seen in advanced degenerative conditions. Thus, kyphosis or straightening of the cervical spine in younger patients likely represents a flexible reversal of cervical lordosis rather than true kyphosis. In contrast, Been et al^[@bibr19-2192568218765400]^ found that total cervical lordosis, defined as the angle between the foramen magnum and inferior endplate of C7, did not vary between subjects \<20 years of age compared with those between 20 and 50 years old. Additionally, kyphosis was only observed in less than 10% of all subjects. This study is a radiographic assessment of asymptomatic patients and therefore has limited applicability to patients undergoing complex spinal realignment or other fusion procedures for symptomatic, degenerative conditions. Likewise, we cannot conclude that patients with symptomatic spinal pathology, such as cervical radiculopathy or stenosis, demonstrate similar radiographic parameters compared with the asymptomatic population in this study. However, our results provide a benchmark of several radiographic parameters that may assist in the evaluation of patients. An additional shortcoming is the limited sample size in our study and future research is aimed at analyzing larger cohorts. Another limitation includes the inability to determine the relationship between gender and cervical parameters. We acknowledge that gender may affect cervical alignment; the aim of our study was to focus on the effect of age. A retrospective study of the effect of gender on cervical lordosis did not find any difference in total cervical lordosis, but males were found to have greater C3-C7 lordosis and females demonstrated more lordosis from the foramen magnum to C3.^[@bibr19-2192568218765400]^ Interrater reliability for the radiographic measurements was not reported in this study. However, the reliability of computer-assisted sagittal plane radiographs has been previously published.^[@bibr20-2192568218765400]^ Finally, asymptomatic subjects were enrolled but we did not control for pelvic incidence. It is possible that patients with a low PI have less cervical lordosis than patients with a high incidence, for example. Cervical alignment correlates with HRQOL after cervical spine surgery, but a comprehensive analysis of cervical and horizontal gaze parameters has not been described. In our study, we identified several cervical and horizontal gaze parameters that remain relatively constant with age and established parameters that change with age. The evaluation of cervical parameters in patients with cervical spine disease and performing age-based comparisons of these results to an asymptomatic population may benefit future research and surgical plans. Declaration of Conflicting Interests: The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: LGL reports personal fees from Medtronic, grants and personal fees from DePuy-Synthes Spine, personal fees from K2 M, nonfinancial support from Broadwater, nonfinancial support from Seattle Science Foundation, grants and nonfinancial support from Scoliosis Research Society, nonfinancial support from Stryker Spine, nonfinancial support from The Spinal Research Foundation, grants from EOS, grants from Setting Scoliosis Straight Foundation, personal fees from Fox Rothschild, LLC, personal fees from Quality Medical Publishing, other from Evans Family Donation, other from Fox Family Foundation, grants and nonfinancial support from AOSpine, outside the submitted work. VL reports grants from SRS, grants from NuVasive, grants from DePuy Spine, grants from Stryker, grants from K2 M, personal fees from DePuy Spine, personal fees from AO, personal fees from MSD, personal fees from NuVasive, personal fees from Nemaris INC, outside the submitted work. HK reports personal fees from Zimmerbiomet, personal fees from K2 M, grants from ISSGF, other from Stryker, other from AOSPINE, grants from CSRS, outside the submitted work. MG reports personal fees and other from DePuy, personal fees from Orthofix, personal fees from Medtronic, other from Johnson & Johnson, other from Proctor & Gamble, other from Board of Directors, SRS, other from Treasurer, FOSA, outside the submitted work. Funding: The author(s) received no financial support for the research, authorship, and/or publication of this article.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION ============ Type II odontoid fracture is the most common injury of the axis, especially in elderly patients, and it is usually treated by anterior odontoid screw fixation[@b2-jkns-60-5-498],[@b6-jkns-60-5-498]. There is much controversy regarding risk factors of nonunion following anterior odontoid screw fixation. However fracture gap consistently identified as a significant risk factor[@b6-jkns-60-5-498],[@b10-jkns-60-5-498],[@b13-jkns-60-5-498],[@b21-jkns-60-5-498],[@b23-jkns-60-5-498], and we have recently reported that a fracture gap of ≥2 mm resulted in a 21-fold increase fusion failure rates[@b11-jkns-60-5-498]. Inter-fragmentary compression and stable fixation is important for fracture union, and it is generally believed that greater inter-fragmentary compression promotes more predictable fracture healing[@b2-jkns-60-5-498],[@b8-jkns-60-5-498],[@b11-jkns-60-5-498],[@b13-jkns-60-5-498],[@b21-jkns-60-5-498],[@b23-jkns-60-5-498]. Cannulated lag screws and headless compression screws are frequently used for the anterior odontoid fixation, because they can produce a lag effect across the fracture site, resulting in reduction of the fracture gap and increased the stability of the fracture[@b10-jkns-60-5-498],[@b11-jkns-60-5-498],[@b15-jkns-60-5-498],[@b20-jkns-60-5-498],[@b22-jkns-60-5-498]. Many authors recommended surgical techniques that include perforation of the dens tip by the screw and correct sizing of the implant to achieve bicortical purchase in order to enhance the lag effect during inter-fragmentary compression[@b2-jkns-60-5-498],[@b3-jkns-60-5-498],[@b8-jkns-60-5-498],[@b25-jkns-60-5-498]. We originally used cannulated lag screws for anterior odontoid screw fixation at our institute, but recently, in accordance with the technique introduced by Chang et al.[@b8-jkns-60-5-498] and Cho and Sung[@b10-jkns-60-5-498],[@b11-jkns-60-5-498], we have adopted the 4.5-mm Herbert screw (Zimmer, Warsaw, IN, USA) for anterior odontoid screw fixation The Herbert screw was initially developed for internal fixation of displaced scaphoid fractures[@b18-jkns-60-5-498]. Because it is a headless compression screw, it can be inserted through articular cartilage and embedded below the bone surface, which reduces surrounding soft tissue irritation. Herbert screw is unique in that it is double threaded with different pitches on the proximal and distal threads for inter-fragmentary compression[@b11-jkns-60-5-498],[@b17-jkns-60-5-498],[@b18-jkns-60-5-498]. A number of biomechanical studies have been performed to investigate various compression screws, mainly in scaphoid fracture models[@b4-jkns-60-5-498],[@b7-jkns-60-5-498],[@b16-jkns-60-5-498]--[@b18-jkns-60-5-498],[@b20-jkns-60-5-498]. Although the Herbert screw is theoretically known to reduce the fracture gap and compress the fractured fragment as the screw is inserted[@b8-jkns-60-5-498],[@b18-jkns-60-5-498], to our knowledge, there is no study reporting biomechanical comparison of inter-fragmentary compression pressures between lag screws and headless compression Herbert screws for anterior odontoid screw fixation. Accordingly, the purpose of the present biomechanical study was to compare the inter-fragmentary compression pressures in simulated type II odontoid fractures after fixation with 4.5-mm headless compression Herbert screws and 4.0-mm half threaded cannulated lag screws. MATERIALS AND METHODS ===================== We compared the inter-fragmentary compression pressures between 4.5-mm Herbert screw and 4.0-mm cannulated lag screws, both of which have been used for anterior odontoid screw fixation in our institute ([Fig. 1](#f1-jkns-60-5-498){ref-type="fig"}). The cannulated lag screw (Solco Biomedical Co., Pyeongtaek, Korea) had a shaft diameter of 2.6 mm, a thread length of 14 mm, and a head diameter of 5.0 mm. The double-threaded headless compression Herbert screw (Zimmer) has different pitches on the leading and trailing threads, which are separated by a long unthreaded shaft. The distal threads are coarser than proximal threads, and are intended to advance rapidly to create inter-fragmentary compression[@b18-jkns-60-5-498],[@b24-jkns-60-5-498]. The proximal thread length of the Herbert screw was 6.4 mm, diameter of 5.8 mm, and distal thread length was 12 mm, diameter 4.5 mm. The Herbert screw had a 1.42 mm flush reduction potential, which indicated the degree of fracture reduction potential when the screw was tightened 1 mm below the bone surface. Fresh human cadaveric bone has been regarded as the most suitable medium for in vitro biomechanical bone screw fixation studies, but recent studies have shown that the density of cancellous bone can be highly variable[@b1-jkns-60-5-498],[@b4-jkns-60-5-498],[@b5-jkns-60-5-498]. In order to eliminate the variability in geometry and mechanical properties associated with cadaveric specimens, solid rigid polyurethane foam, which is comparable to human osteoporotic cancellous bone, can be used as an alternative test material for comparative evaluation of medical devices[@b7-jkns-60-5-498],[@b16-jkns-60-5-498],[@b17-jkns-60-5-498],[@b20-jkns-60-5-498]. Therefore, in the present study, in order to have uniform samples for comparison, rigid polyurethane foam blocks (Sawbones, Vashon, WA, USA) were used as the test materials for the insertion of the screws. The foam had density of 15 per cubic foot (PCF) (0.16 g/mL), which has been shown to represent the material properties of osteopenic human cancellous bone[@b4-jkns-60-5-498],[@b24-jkns-60-5-498]. Each 40×40×40 mm synthetic bone model was cut precisely into two 20 mm thick segments by cutting machine (JSG-520MB; JIN SAN T&G, Incheon, Korea), and a 1-mm layer of denser foam (0.64 g/mL) was laminated to the top and bottom surfaces to represent cortical bone. [Fig. 2](#f2-jkns-60-5-498){ref-type="fig"} shows the test setup. A custom testing jig was used to hold the blocks and prevent rotation during screw insertion. An ultra-thin load cell sensor was employed to minimize the displacement of the simulated fractures, and a 10-mm hole between 2 thin metal washers was placed at the center of the sensing head in order to accommodate the screw. A washer load cell (FlexiForce A201-100; Tekscan, Boston, MA, USA) was placed between the two foam block segments to measure the compression force and was then connected to a data acquisition system (MSO4034; Tektronix, Beaverton, OR, USA). Each screw was prepared with two different lengths, 40-mm and 45-mm. We compared inter-fragmentary compression pressures between 40- and 45-mm long 4.5-mm Herbert screws (n=8 and n=9, respectively) and 40- and 45-mm long 4.0-mm cannulated lag screws (n=7 and n=10, respectively) after insertion into rigid polyurethane foam test blocks. Because the total length of the foam block was 42 mm, we inserted the 40-mm screw within the cancellous foam block, but were able to penetrate the denser bottom cortical foam with the 45-mm screws. This allowed us to compare inter-fragmentary compression pressures as they are affected by penetration of the apical dens tip by the screws. All of the screws were placed according to the respective manufacturer guidelines and the specimen foam was drilled with drill bits specified by each screw's manufacturer. The screws were inserted at a controlled speed of 10 rpm until the threads were totally embedded in the foam. The test specimens were then mounted onto an universal testing machine (Instron3000; Instron, Norwood, MA, USA) ([Fig. 3](#f3-jkns-60-5-498){ref-type="fig"}) and inter-fragmentary compression was produced by tightening the head of the screw to the surface of the first segment. Output from the washer load cells was measured during insertion of the screws, and the measurement continued after a steady state had been reached. Statistical analysis -------------------- All of the statistical analyses were performed using Prism 7 software (GraphPad, La Jolla, CA, USA). The data are expressed as mean±standard error of mean. Student t-test was for comparison between groups. Values of p\<0.05 were considered statistically significant. RESULTS ======= We compared the inter-fragmentary compression pressures between 4.5-mm diameter Herbert screws and 4.0-mm diameter cannulated lag screws with different lengths (40 mm and 45 mm). Mean compression pressure of the 40- and 45-mm cannulated lag screws were 50.48±1.20 N and 53.88±1.02 N, respectively, which did not reach statistical significance (p=0.0551). Mean peak compression pressure of the 40-mm long Herbert screw was 52.82±2.17, which was not statistically significant compared with the 40-mm and 45-mm long cannulated lag screw (p=0.3679 and p=0.2473, respectively). However the 45-mm long Herbert screw had significantly greater mean compression pressure (60.68±2.03 N) than both the 45-mm long cannulated lag screw and the 40-mm Herbert screw (p=0.0049 and p=0.0246, respectively) ([Fig. 4](#f4-jkns-60-5-498){ref-type="fig"}). DISCUSSION ========== The cannulated lag screws with a short thread are most frequently used for anterior odontoid fracture fixation to produce a lag effect across the fracture site[@b2-jkns-60-5-498],[@b6-jkns-60-5-498],[@b9-jkns-60-5-498],[@b13-jkns-60-5-498],[@b20-jkns-60-5-498],[@b21-jkns-60-5-498]. Theoretically, after the cannulated lag screw has crossed the fracture line, the threads engage the fragment and the lag effect of the screw reduces the fracture. Further tightening of the lag screws pulls the odontoid in a caudal direction, compressing the fracture site and enhancing fracture stability, which results in improved healing of the fracture[@b3-jkns-60-5-498],[@b11-jkns-60-5-498],[@b15-jkns-60-5-498],[@b20-jkns-60-5-498],[@b25-jkns-60-5-498]. A single piece non-cannulated headless compression Herbert screw with variable pitches of the proximal and distal threaded portions, was first designed by Herbert and Fisher[@b18-jkns-60-5-498] in 1984, to provide internal compression and stability of scaphoid fracture while avoiding any protrusion of metal on the joint surface of the bone. And thereafter Whipple modified the Herbert screw by developing a cannulated version to allow for more accurate screw placement. Because of its security of fixation and lack of a protrusive head, the Herbert screw would appear to have significant advantages over standard implants for the fixation of small cancellous bone fragments. In 1994, Chang et al.[@b8-jkns-60-5-498] first described the Herbert-Whipple screw fixation of type II odontoid fracture. They suggested that close reduction and compressive osteosynthesis by double-threaded compression screw is an optimal method of treatment of displaced type II odontoid fractures. At our institute, we originally used the cannulated lag screw for anterior odontoid screw fixation, but recently, in accordance with the method introduced by Chang et al.[@b8-jkns-60-5-498], we have adopted the 4.5-mm Herbert screw (Zimmer)[@b10-jkns-60-5-498],[@b11-jkns-60-5-498],[@b19-jkns-60-5-498]. In our previous study using the Herbert screw for anterior odontoid fixation, the risk of fusion failure was 21 times greater in patients with a fracture gap of more than 2 mm[@b11-jkns-60-5-498]. We believed that in order to facilitate maximal reduction of the fracture gap using the Herbert screw, it is essential for the screw to penetrate the apical dens tip, which has generally been the recommended surgical technique[@b2-jkns-60-5-498],[@b3-jkns-60-5-498]. However, in fact, cortical purchase by the screw may be technically difficult and very stressful to the surgeon due to the risk of injury of adjacent basilar artery and pons[@b3-jkns-60-5-498],[@b25-jkns-60-5-498]. To date, there have been a number of many biomechanical studies that have evaluated different compression screws for bone fractures[@b4-jkns-60-5-498],[@b5-jkns-60-5-498],[@b16-jkns-60-5-498],[@b17-jkns-60-5-498],[@b24-jkns-60-5-498]. Burkhart et al.[@b7-jkns-60-5-498] reported that there were no significant differences in the stability provided by a 3.0-mm headless compression screw and a 2.0-mm cortical screw in radial head fractures in fresh frozen human cadaveric bone specimen. A recent cadaveric scaphoid study found that the mean inter-fragmentary compression generated by an Acutrak 2 Standard screw was significantly greater than that of a Synthes headless compression screw[@b16-jkns-60-5-498]. However, there are few published biomechanical studies for odontoid screws. Magee et al.[@b20-jkns-60-5-498] have reported a biomechanical comparison of a fully threaded, variable pitch screw and a partially threaded lag screw for internal fixation of type II dens fractures in human cadaveric specimens. They found that stiffness and load to failure were greater for the headless, fully threaded variable pitch screw compared with the partially threaded lag screw. However, they did not measure the inter-fragmentary compression pressure between the fracture fragments of the odontoid process. Previous studies have shown that the density and elastic modulus of cancellous bone (e.g., scaphoid bone) are highly variable, and the mechanical properties of trabecular bone from multiple locations have tremendous variations, which can affect the maximum achievable compression force[@b1-jkns-60-5-498],[@b4-jkns-60-5-498],[@b14-jkns-60-5-498]. Rigid polyurethane foam is been regarded as an ideal material for in vitro evaluation of compression screws. The uniform and consistent properties of rigid polyurethane test blocks in comparison to animal and human bone has been well established. Foam densities of 10, 15, and 20 pounds PCF have been shown to have the material properties of osteoporotic, osteopenic, and normal bone, respectively[@b4-jkns-60-5-498],[@b17-jkns-60-5-498],[@b24-jkns-60-5-498]. Because type II odontoid fractures are the most common type of axis fracture in elderly patients, in the present study, we used 15 PCF foam (0.16 g/mL) in order to represent the material properties of osteopenic human cancellous bone. In the present study for the 40 mm-long Herbert screw and the 40-mm long cannulated lag screw, there was no difference in inter-fragmentary compression pressures. But, in our comparison of 45-mm long screws, which were long enough to penetrate the dense cortical foam of the opposite test block, the 45-mm Herbert screw produced significantly higher inter-fragmentary compression pressure than the 40-mm Herbert screw. The 45-mm cannulated lag screws also provided greater inter-fragmentary compression pressure. Although this did not reach statistical significance (p=0.0551), we surmised that the number of test screws (n=7) was too small to achieve a statistical difference. These results support the recommendations of previous studies that, in order to facilitate maximal reduction of the fracture gap using lag screws, it is essential to penetrate the apical dens tip with the screw[@b2-jkns-60-5-498],[@b3-jkns-60-5-498],[@b11-jkns-60-5-498]. When we compared the 45-mm Herbert screw and the 45-mm cannulated lag screw, the Herbert screw produced significantly larger peak compression pressures than the cannulated lag screws (p=0.0049). This result suggests that the specially designed Herbert screw may provide greater reduction of the fracture gap in the odontoid process than the cannulated lag screws when the screws penetrate the apical cortex of the dens. This finding further emphasizes the importance of screw-tip penetration of the apical dens tip during anterior odontoid screw fixation. We have some limitations in this study. First, we did not check other biomechanical parameters, including bending strength or pullout strength. Inter-fragmentary compression pressure alone cannot guarantee the biomechanical superiority of the Herbert screw compared with the cannulated lag screw. Second, clinically, greater inter-fragmentary compression pressure may not mean greater reduction of the fracture gap in odontoid process fixation, because, in humans, the dens is attached by many strong ligaments including apical and alar ligaments[@b12-jkns-60-5-498],[@b26-jkns-60-5-498]. Furthermore, because there are so many risk factors of fusion failure of odontoid process fixation[@b2-jkns-60-5-498],[@b3-jkns-60-5-498],[@b11-jkns-60-5-498],[@b15-jkns-60-5-498], we must consider other factors for higher bone fusion rates after odontoid screw fixation. Third, in human, due to the heterogeneous fracture line and orientation of type II odontoid fracture, odontoid screw insertion generally have a little angled path against the fracture line during the operation. But, in this biomechanical study, we inserted the screws perpendicularly into the test blocks, which may not reflect the real situation of anterior odontoid screw fixation. CONCLUSION ========== Our results showed that inter-fragmentary compression pressures of the Herbert screw were significantly increased when the screw tip penetrated the opposite dens cortical foam. This can support the generally recommended surgical technique that, in order to facilitate maximal reduction of the fracture gap using anterior odontoid screws, it is essential to penetrate the apical dens tip with the screw. This work was supported by Biomedical Research Institute grant, Kyungpook National University Hospital (2013). ![Showing the studied implants; the 4.0-mm cannulated cortical screw (left) and the 4.5-mm headless compression Herbert screw (right).](jkns-60-5-498f1){#f1-jkns-60-5-498} ![Illustration for test setup. PCF: per cubic foot.](jkns-60-5-498f2){#f2-jkns-60-5-498} ![Showing the test specimens which was mounted onto a Testometric M 500 Universal testing machine.](jkns-60-5-498f3){#f3-jkns-60-5-498} ![Mean compression pressures of cannulated lag screws and headless compression Herbert screws (n=7--10/groups).](jkns-60-5-498f4){#f4-jkns-60-5-498}	{ "pile_set_name": "PubMed Central" }
Findings ======== Spread of antimicrobial-resistant bacteria from companion animals to humans causes concern but the role of companion animals as reservoirs of antimicrobial resistant bacteria requires further investigation \[[@B1]\]. Escherichia coliare commonly found in the intestinal tract of animals, including dogs, and constitute a reservoir of resistant genes for potentially pathogenic bacteria \[[@B1]\]. Dogs may also be a reservoir of E. colistrains that cause extraintestinal infections in humans \[[@B2]\]. Antimicrobial resistance of canine E. colihas previously been investigated \[[@B3]-[@B5]\]. Antimicrobial-resistant E. colihave been isolated more frequently in kennel dogs than in individually owned dogs \[[@B3]\]. Breeding of multiple dogs at one location may increase the risk of spreading antimicrobial resistant clones in the population similar to livestock on the same premises \[[@B6],[@B7]\]. The purpose of the present study was to compare phenotypic and genetic characteristics of antimicrobial resistant E. coliisolated from the faeces of pups in kennels, and to investigate genetic relatedness among these isolates as the epidemiology of antimicrobial resistant bacteria in dog populations has not been extensively studied. Faecal samples were obtained from 43 apparently healthy pups not above two months of age from two kennels (A and B). Twenty-five pups were from eight litters in kennel A (A-a to A-h), and 18 were from five litters in kennel B (B-a to B-e; Table [1](#T1){ref-type="table"}). There was no history of antimicrobial use in the pups but the dams may have been administered lincomycin for postpartum infection prophylaxis. Briefly, faecal swabs were plated onto desoxycholate-hydrogen sulphide-lactose agar (Eiken Chemical, Tochigi, Japan) and incubated overnight at 37°C in an aerobic atmosphere. Subsequently, two lactose fermenting colonies with typical E. colimorphology were picked and subjected to confirmatory Gram staining, indole, methyl red, Voges-Proskauer, and Simmons\' Citrate tests. ###### Details of surveyed pups in this study. Kennel Litter Breed No. of pups Pup identity Date of birth Age in days -------- -------- --------------------- ------------- --------------- --------------- ------------- A A-a Toy poodle 3 A-a-1 - A-a-3 5/13/2009 30 A-b Toy poodle 3 A-b-1 - A-b-3 5/15/2009 28 A-c Chihuahua 2 A-c-1 - A-c-2 4/17/2009 57 A-d Toy poodle 3 A-d-1 - A-d-3 6/23/2009 45 A-e Toy poodle 4 A-e-1 - A-e-4 7/1/2009 37 A-f Toy poodle 5 A-f-1 - A-f-5 7/9/2009 29 A-g Toy poodle 3 A-g-1-A-g-3 9/2/2009 38 A-h Toy poodle 2 A-h-1 - A-h-2 9/6/2009 35 Total 25 B B-a Maltese 2 B-a-1 - B-a-2 8/3/2009 46 B-b Beagle 4 B-b-1 - B-b-4 7/29/2009 51 B-c Miniature dachshund 5 B-c-1 - B-c-5 8/1/2009 51 B-d Papillon 3 B-d-1 - B-d-3 8/17/2009 53 B-e Chihuahua 4 B-e-1 - B-e-4 8/25/2009 55 Total 18 Informed consent from both kennels was obtained and treated in accordance with the Japanese Law Concerning the Protection of Personal Information (Law No. 57, 2003). Minimum inhibitory concentrations of ampicillin (AMP), cefazolin, ceftiofur, dihydrostreptomycin (DHS), gentamicin, kanamycin, oxytetracycline (OTC), chloramphenicol (CHL), trimethoprim-sulfamethoxazole (SXT), nalidixic acid and enrofloxacin were determined using the agar dilution method according to the Clinical and Laboratory Standards Institute (CLSI) Guidelines \[[@B8]\]. CLSI resistance breakpoints \[[@B8],[@B9]\] were used in the categorical analysis of all drugs except DHS where 32 μg/mL was used as reported elsewhere \[[@B10]\]. For quality control, E. coliATCC 25922 was used. All isolates were screened for extended-spectrum β-lactamase (ESBL) production by the combination disk test (cefotaxime and ceftazidime with or without clavulanic acid) \[[@B8]\]. ESBL-producing strains were examined for β-lactamase-encoding genes including blaTEM, blaSHV, blaCTX-M-2 group, and blaCTX-M-9 group by polymerase chain reaction and sequencing \[[@B11]\]. Pulsed-field gel electrophoresis (PFGE) was performed according to standard methods outlined by PulseNet \[[@B12]\]. DNA was digested with XbaI (Takara, Shiga, Japan) and electrophoresed on a CHEF DRII (Bio-Rad Laboratories, Hercules, CA, USA), with switch times of 2.2-54.2 s at 14°C for 20 h. PFGE profiles were analysed using BioNumerics software version 4.0 (Applied Maths, TX, USA). DNA fragments on each gel were normalised using the λ molecular weight marker on each gel to enable comparisons between different gels. Cluster analysis was performed by the unweighted pair group method using arithmetic average. DNA relatedness was calculated based on the Dice coefficient. The prevalence of antimicrobial resistance and rates of clone sharing between the two kennels were compared using Fisher\'s exact test. In this study, we investigated the prevalence of antimicrobial resistance in faecal E. coliisolates from kennel pups without a history of antimicrobial use; however, their dams may have been administered with lincomycin. If this is the case, the possibility that the pups were exposed to the agent via the milk cannot be excluded. It is generally known that lincomycin is inactive against aerobic Gram-negative bacteria including E. coli, but active against Gram-positive and/or anaerobic bacteria \[[@B13]\], possibly causing changes in the intestinal microflora composition. The effects of an altered intestinal flora on the population of resistant and susceptible E. coliare unknown, but may need to be taken into account in the present results. We found that 75.6% (n= 65) of E. coliisolates originating from 35 pups in two kennels, were resistant to one or more of the antimicrobials tested. Resistance to DHS was most frequent (66.3%), followed by AMP (60.5%), SXT (41.9%), OTC (26.7%), and CHL (26.7%) (Table [2](#T2){ref-type="table"}). Multidrug resistance defined as resistance against two or more classes of antimicrobials was observed in 60.5% (n= 52) of isolates, originating from 29 pups in two kennels. Comparison between kennels A and B revealed that the prevalence of resistance against seven of the tested antimicrobials differed significantly (P\< 0.05). Twelve and five resistance patterns were observed in kennels A and B, respectively, with the patterns differing between the two kennels, except for the AMP-DHS-SXT and DHS resistance phenotypes (Table [3](#T3){ref-type="table"}). These findings indicate that the prevalence of antimicrobial resistant E. coliin pups varies between kennels. ###### The minimum inhibitory concentration (MIC) range and resistance rates among Escherichia coliisolates from pups originating from two kennels (A and B). Substance^a^ MIC range (μg/mL) MIC~50~ MIC~90~ Resistance breakpoints (μg/mL)^b^ No. of resistant isolates (%)/No. of pups that harboured resistant isolate(s) (%) ---------------------- ------------------------ ----------------- ----------------- ------------------------------------------- ----------------------------------------------------------------------------------- ------------------------- --------------------- AMP 2 - \>512 \>512 \>512 ≥32 52 (60.5)/30 (69.8) 29 (58.0)/17 (68.0) 23 (63.9)/13 (72.2) CFZ 8 - 128 4 8 ≥32 5 (5.8)/3 (7.0) 0 (0)/0 (0) 5 (13.9)\/3 (16.7) CEF ≤0.125 - 32 0.5 1 ≥8 5 (5.8)/3 (7.0) 0 (0)/0 (0) 5 (13.9)\/3 (16.7) DHS 2 - \>512 512 \>512 ≥32 57 (66.3)/32 (74.4) 35 (70.0)/19 (76.0) 22 (61.1)/13 (72.2) GEN 0.5 - 256 1 128 ≥16 16 (18.6)/12 (27.9) 16 (32.0)\/12 (48.0)\ 0 (0)/0 (0) KAN 2 - \>512 4 16 ≥64 3 (3.5)/2 (4.7) 3 (6.0)/2 (8.0) 0 (0)/0 (0) OTC 1 - 512 2 512 ≥16 23 (26.7)/17 (39.5) 18 (36.0)\/14 (56.0)\ 5 (13.9)/3 (16.7) CHL 4 - \>512 8 512 ≥32 23 (26.7)/17 (39.5) 18 (36.0)\/14 (56.0)\ 5 (13.9)/3 (16.7) NAL 2 - \>512 4 16 ≥32 7 (8.1)/5 (11.6) 2 (4.0)/2 (8.0) 5 (13.9)/3 (16.7) ENR ≤0.03 - 256 0.06 1 ≥4 5 (5.8)/3 (7.0) 0 (0)/0 (0) 5 (13.9)\/3 (16.7) SXT ≤0.25/4.75 - \>64/1216 1/19 \>64/1216 ≥16/304 36 (41.9)/21 (48.8) 27 (54.0)\/16 (64.0)\* 9 (25.0)/5 (27.8) ^a^AMP, ampicillin; CFZ, cefazolin; CEF, ceftiofur; DHS, dihydrostreptomycin; GEN, gentamicin; KAN, kanamycin; OTC, oxytetracycline; CHL, chloramphenicol; NAL, nalidixic acid; ENR, enrofloxacin; SXT, trimethoprim-sulfamethoxazole. ^b^The breakpoints for AMP, CFZ, CEF, GEN, KAN, OTC, CHL, ENR and SXT, and that for NAL were based on CLSI document M31-A3 \[[@B9]\] and M100-S20 \[[@B8]\], respectively, whereas the breakpoint of DHS was based on an epidemiological cut-off value according to another report \[[@B10]\] ###### The distribution of resistance phenotypes among Escherichiacoliisolates from pups in two kennels (A and B). Resistance patterns^a^ No. of resistant isolates (%)/No. of pups that harboured resistant isolate (s) (%) -------------------------------- ------------------------------------------------------------------------------------ -------------------- -------------------- AMP-CFZ-CFT-OTC-CHL-NAL-ENR 5 (5.8)/3 (7.0) 0 (0)/0 (0) 5 (13.9)/3 (16.7) AMP-DHS-GEN-KAN-OTC-CHL-SXT 3 (3.5)/2 (4.7) 3 (6.0)/2 (8.0) 0 (0)/0 (0) AMP-DHS-GEN-OTC-CHL-NAL-SXT 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) AMP-DHS-GEN-OTC-CHL-SXT 11 (12.8)/9 (20.9) 11 (22.0)/9 (36.0) 0 (0)/0 (0) AMP-DHS-GEN-CHL-SXT 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) AMP-DHS-OTC-SXT 2 (2.3)/1 (2.3) 2 (4.0)/1 (4.0) 0 (0)/0 (0) AMP-DHS-CHL-SXT 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) AMP-DHS-SXT 16 (18.6)/13 (30.2) 8 (16.0)/8 (32.0) 8 (22.2)/5 (27.8) AMP-DHS-CHL 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) AMP-DHS 10 (11.6)/6 (14.0) 0 (0)/0 (0) 10 (27.8)/6 (33.3) DHS-SXT 1 (1.2)/1 (2.3) 0 (0)/0 (0) 1 (2.8)/1 (5.6) AMP 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) DHS 10 (11.6)/7 (16.3) 7 (14.0)/5 (20.0) 3 (8.3)/2 (11.1) OTC 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) NAL 1 (1.2)/1 (2.3) 1 (2.0)/1 (4.0) 0 (0)/0 (0) Susceptible 21 (24.4)/13 (30.2) 12 (24.0)/7 (28.0) 9 (25.0)/6 (33.3) ^a^AMP, ampicillin; CFZ, cefazolin; CEF, ceftiofur; DHS, dihydrostreptomycin; GEN, gentamicin; KAN, kanamycin; OTC, oxytetracycline; CHL, chloramphenicol; NAL, nalidixic acid; ENR, enrofloxacin; SXT, trimethoprim-sulfamethoxazole. Additionally, the XbaI-digested PFGE revealed 13 and 10 distinct major profiles (≥ 90% Dice similarity) in 50 and 36 isolates from 25 and 18 pups from kennels A and B (Figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"}), respectively. These PFGE profiles correlated highly with resistance phenotypes, except for profiles A-1, A-5 and A-11. Of all 43 pups, 17 pups harboured two isolates differentiated by PFGE and/or resistance phenotypic profiles, indicating that diverse E. colipopulations can colonise intestinal flora during infancy. Sixteen of 25 pups in kennel A (i.e. two, three, three, five, and three pups within litters A-a, A-b, A-d, A-f, and A-g, respectively) and 16 of 18 pups in kennel B (i.e. two, four, four, two, and four pups within from litters B-a to B-e, respectively) shared at least one E. coliclone, defined as an isolate with identical PFGE and resistance phenotypic profiles, with one or more pups of the same litters. There was no significant difference in the clone sharing rates within the same litters between the two kennels. In the two kennels, all pups of the same litters, and their respective dams, were raised together in one cage, implying that E. colimay be transmitted horizontally viafaeces. Another possibility may be vertical transfer from mothers viatheir milk and vaginal flora \[[@B14]\]. These data suggest that pups from the same litter are likely to be exposed to common sources of E. coliresulting in clonal spread of organisms, including antimicrobial resistant isolates. ![*Dendrogram of pulsed-field gel electrophoresis (PFGE) profiles from 50 Escherichia coliisolates from 25 pups originating from kennel A. A: Isolate origin. A-a, A-b, A-c, A-d, A-e, A-f, A-g, and A-h litters consisted of three (A-a-1 to A-a-3), three (A-b-1 to A-b-3), two (A-c-1 to A-c-2), three (A-d-1 to A-d-3), four (A-e-1 to A-e-4), five (A-f-1 to A-f-5), three (A-g-1 to A-g-3), and two pups (A-h-1 to A-h-2), respectively. Two isolates were obtained per pup. B: Resistance pattern. AMP, ampicillin; DHS, dihydrostreptomycin; GEN, gentamicin; KAN, kanamycin; OTC, oxytetracycline; CHL, chloramphenicol; SXT, trimethoprim-sulfamethoxazole; NAL, nalidixic acid. C: PFGE profile.](1751-0147-53-11-1){#F1} ![Dendrogram of pulsed-field gel electrophoresis (PFGE) profiles from 36 Escherichia coliisolates from 18 pups originating from kennel B. A: Isolate origin. B-a, B-b, B-c, B-d, and B-e litters consisted of two (B-a-1 to B-a-2), four (B-b-1 to B-b-4), five (B-c-1 to B-c-5), three (B-d-1 to B-d-3), and four pups (B-e-1 to B-e-4), respectively. Two isolates were obtained per pup. B: Resistance pattern. AMP, ampicillin; CFZ, cefazolin; CEF, ceftiofur; DHS, dihydrostreptomycin; GEN, gentamicin; OTC, oxytetracycline; CHL, chloramphenicol; SXT, trimethoprim-sulfamethoxazole; NAL, nalidixic acid; ENR, enrofloxacin. C: PFGE profile. Five isolates with the B-2 PFGE profile harboured the SHV-12 ESBL-encoding gene.](1751-0147-53-11-2){#F2} The following 24 pups in the two kennels shared at least one E. coliclone (i.e. the clones harbouring PFGE profiles A-1, 2, 5, 11, 12, and B-9) among the different litters; three, two, two, three, three, four, and one pups of from litters A-a to A-g in kennel A, and four and two pups of litters B-b and B-c, respectively. These litters were temporally separated (8-83 days) and originated from different mothers without direct contact, suggesting that E. coliclones may have originated from a persistent external source. One possibility, as suggested by other studies, is that the pups acquired E. colifrom their human contacts \[[@B15],[@B16]\]. Unlike clone sharing rates among the same litters, the rates among different litters were significantly different between kennels A and B \[18/25 (72.0%) vs.6/18 (33.3%) pups, respectively, P\< 0.05\]. This finding suggests that clone sharing rates among different litters can vary between kennels. Further study is needed to clarify the potential transmission route(s) between kennel pups. Overall, our data indicates that clonal spread of E. coliplays an important role in acquisition of resistant isolates by kennel pups. The prevalence of ESBL-producing isolates in companion animals and their potential impact on human health is a major issue \[[@B17]\]. In the present study, the SHV-12 ESBL-encoding gene was detected in five isolates (5.8%), exhibiting identical PFGE and resistance phenotypic profiles, from three pups within a litter (Figure [2](#F2){ref-type="fig"}). The reason for the occurrence of these resistant isolates was not apparent. To the best of our knowledge, this is the first time that SHV-12 β-lactamase has been detected in E. coliof canine origin in Japan, although it has been previously reported in other countries \[[@B18],[@B19]\]. The present findings suggest that attention needs to be paid to dogs as a potential reservoir of ESBL-producing E. coliisolates in Japan. In conclusion, our data show that pups in kennels can harbour multidrug-resistant E. coliisolates, including ESBL-producing isolates. The present results also indicate that resistant and susceptible E. coli*isolates can clonally spread not only within the same litter but also among different litters thus affecting the prevalence of resistant organisms in a kennel. Further studies are needed to fully understand the epidemiological spread of antimicrobial resistant bacteria among pups in kennels. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= KH and EM carried out antimicrobial susceptibility testing, PCR, sequencing, and PFGE. KH analysed the data. KH, YK and TT were involved in the study design and preparation of the manuscript. KH drafted the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ The authors wish to acknowledge Dr. Shiki Kawabe for providing faecal swabs from pups. This work was financially supported by a grant from the Ministry of Education, Culture, Sports, Science and Technology, Japan (Grant-in-Aid for Research Activity Start-up, No. 21880043).	{ "pile_set_name": "PubMed Central" }
Cow\'s milk allergy (CMA) accounts for most diagnosed food allergies, mainly in the first year of life.[@B1] CMA can cause severe anaphylaxis.[@B2] CMA management is based on strict avoidance and the prescription of rescue medication. Cow\'s milk (CM) contains more than 40 proteins, including casein (Bos d 8), -lactalbumin (Bos d 4), and -lactoglobulin (Bos d 5), identified as major milk allergens.[@B1] Serum specific IgE (sIgE) to casein characterized patients with persistent CMA,[@B3] CM tolerance,[@B4] and reactivity to baked milk.[@B5] Oral immunotherapy (OIT) for CMA has been used with interesting outcomes as it may induce immunologic tolerance.[@B1][@B2] However, a safety perspective represents a relevant concern, because reactions to OIT are frequent[@B1][@B2] It has been demonstrated that sIgE to raw CM \>50 kUA/L predicted not-tolerated OIT.[@B6] Molecular-based allergy diagnostics have now become recently available in clinical practice. This method allows for defining and characterizing the sensitization profile that identifies potentially dangerous proteins and suggests a more precise prognosis. In this regard, Cingolani and colleagues reported that component resolved diagnosis had a good ability to define 2 phenotypes of CMA children: \"high-anaphylaxis risk\" and \"milder-risk\".[@B7] In particular, these phenotypes can be differentiated through measuring the level of sIgE to Bos d 8. We now report our experience concerning a retrospective cohort. This observational and retrospective experience evaluated the usefulness of ImmunoCAP and ISAC (please describe company, city, country) to detect anaphylaxis after CM ingestion. We evaluated children with CMA diagnosis who consecutively visited the Istituto Gaslini (a third tier children\'s hospital) in the last year. CMA diagnosis was performed on suggestive history consistent with CM sensitization (i.e. symptom occurrence after milk ingestion), documented milk sensitization, and a positive food challenge test. Anaphylaxis was defined according to validated criteria: briefly, suggestive history (sudden occurrence of symptoms, involving at least two systems, such as gastrointestinal and/or cutaneous and/or respiratory and/or cardiovascular after milk ingestion) consistent with documented CM sensitization.[@B1] Serum sIgE to: raw CM allergen, Bos d 4, 5, and 8 were measured by the quantitative ImmunoCAP method and the semi-quantitative microarray-ISAC method (Thermo Fisher Italy). We Evaluated 53 patients (20 with anaphylaxis and 33 with CMA). Children with anaphylaxis presented: skin symptoms (89%), respiratory complaints (77%), gastrointestinal features (57%), and cardiovascular symptoms (11%). Children with CMA presented: skin symptoms (71%), gastrointestinal features (65%, mainly vomiting), and respiratory complaints (58%). The [Table](#T1){ref-type="table"} shows the clinical and immunologic characteristics of the 2 groups. The significant prevalence for male gender in anaphylaxis was consistent with that of Cingolani\'s study. The age was significantly lower in the anaphylaxis group, it conflicted with the findings of that study (surprisingly the age was particularly old in regards to CMA). Serum sIgE levels to raw CM allergen were significantly higher in the anaphylaxis group; however, there was no significant difference between the two groups for all molecular components as measured by both the ImmunoCAP and ISAC method. Receiver operating characteristic (ROC) analysis showed that raw CM ImmunoCAP had a good specificity (86.7%), but weak sensitivity (52.9%), fair positive (69.2%), and negative (76.5%) predictive value, with Diagnostic Odds Ratio (OR~Diagn~) 7.3. The assessment of molecular components by ImmunoCAP was unsatisfactory in regards to area under the ROC curve (AUC), sensitivity, and specificity, despite the fair positive and negative values. The micro-assay ISAC method was also unreliable in our setting. We believe that the diagnostic work-up for CMA should consider a molecular-based allergy diagnostic as well as a raw allergen assessment to obtain more useful information for the management and possible identification of risk factors. Partially funded by Ricerca Corrente - Italian Ministry of Health. There are no financial or other issues that might lead to conflict of interest. ###### Clinical and imunological characteristics of patients ![](aair-8-86-i001) Anaphylaxis: yes Anaphylaxis: no P^\^ Cut-off^†^ AUC Sens % Spec % OR~Diag~ (95% CI) -------------------- ------------------ ------------------ ----------- ------------ ------ -------- -------- ------------------- Number of patients 20 33 Gender: male-n (%) 15 (75.0) 15 (45.4) 0.04 Age (months) 5.5 (4.0-6.5) 32.4 (15.3-68.2) \< 0.0001 Total IgE (kU/L) 682 (165-1,439) 197 (106-459) 0.11 Cow milk (kU~A~/L) 39.6 (5.8-50.9) 4.7 (0.9-16.5) 0.02 \> 30.0 0.70 52.9 86.7 7.3 (1.8-30.2) nBos d 4 (kU~A~/L) 5.7 (0.5-32.1) 1.5 (0.7-3.5) 0.13 \> 12.2 0.65 42.9 95.2 15.0 (1.5-145.2) nBos d 5 (kU~A~/L) 7.5 (1.3-15.3) 1.3 (0.6-3.8) 0.07 \> 2.07 0.68 71.4 66.7 5.0 (1.1-21.8) nBos d 8 (kU~A~/L) 8.2 (3.4-53.0) 6.4 (1.4-17.6) 0.61 \> 36.6 0.56 36.4 83.3 2.9 (0.5-16.4) nBos d 4 (ISU) 0.9 (0.1-3.7) 0.2 (0.2-1.0) 0.58 nBos d 5 (ISU) 1.0 (0.1-5.4) 0.2 (0.2-0.6) 0.14 nBos d 8 (ISU) 1.0 (0.1-4.1) 0.2 (0.2-0.5) 0.10 ^\^P values refer to the Mann-Whitney U test; ^†^Cut-off values were determined by means of the receiver operating characteristic (ROC) curve analysis. Sens %, sensitivity; Spec %, specificity; OR~Diag~, diagnostic odds ratio; 95% CI, 95% confidence interval; AUC, area under the ROC curve.	{ "pile_set_name": "PubMed Central" }
![](hosplond72208-0026){#sp1 .182}	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-viruses-10-00353} =============== In laboratory animal facilities, efforts are made by researchers, veterinarians, and caretakers to enrich the lives of animals in captivity beyond basic biological needs, such as housing, food, water, and resting sites. Indeed, many laboratory animal facilities are mandated to provide environmental enrichment for residents \[[@B1-viruses-10-00353]\]. Environmental enrichment provides enhancements in order to meet animals' behavioral needs, providing stimulation and reducing undesirable behaviors and stress \[[@B2-viruses-10-00353]\]. The goal of enrichment is to allow animals to express natural species-specific behaviors, such as running, jumping, climbing, play, predatory simulation (when appropriate), and positive social interactions in order to reduce stress \[[@B3-viruses-10-00353]\] and improve their overall well-being \[[@B4-viruses-10-00353]\]. Enrichment comes in various forms, including contact with conspecifics, toys, human interaction, caging type and furnishings, and auditory or olfactory stimulation \[[@B4-viruses-10-00353],[@B5-viruses-10-00353]\]. In laboratory animals, enrichment must conform to facility requirements for cleaning or disposal in order to prevent enrichment items serving as fomites for infectious diseases. Enrichment recommendations for cats in laboratory settings are similar to those made to meet the needs of cats in their homes \[[@B6-viruses-10-00353]\]. In the physical environment, requirements include adequate space for the cat to move, structures for the cat to interact with, resting places, hiding places, and stimuli in order to meet the needs of the cat for exercise, activity, rest, and coping behaviors. Within the social environment, enrichment focuses on positive interactions with conspecifics and human caregivers \[[@B5-viruses-10-00353]\]. Other suggested welfare needs of cats include scratching substrates for nail conditioning and scent deposition, and an assortment of toys for play and object manipulation. While these welfare and enrichment recommendations have been evaluated for domestic housecats \[[@B4-viruses-10-00353]\], they have not been systematically evaluated for laboratory-housed cats. There is evidence that providing suitable enrichment may have effects on neurochemistry and resulting behavior, thereby improving data quality \[[@B7-viruses-10-00353]\]. In addition to meeting recommendations for best care, environmental enrichment can attenuate many stressors that exist in a laboratory environment \[[@B8-viruses-10-00353]\]. Procedures that require restraint, those that produce some level of discomfort, and the approach and presence of an unfamiliar person may contribute to laboratory stress and can negatively impact the welfare of cats housed in laboratory animal facilities. With feline models, stress may be reflected in elevated cortisol levels, diminished appetite, elective social isolation or aggression, suboptimal grooming behavior, and increased tendency to hide \[[@B9-viruses-10-00353]\]. Animals in stressful environments and those unable to express their natural behaviors can impact "baseline" laboratory and behavioral data. These outcomes can compromise the quality of the research data \[[@B10-viruses-10-00353]\]. While there are recommendations available for meeting the needs of confined cats \[[@B5-viruses-10-00353],[@B6-viruses-10-00353],[@B11-viruses-10-00353]\], very little research exists on the actual preferences of laboratory cats for specific types of enrichment, or how to assess the enrichment preferences of individual cats. What we, as humans, perceive to be enriching may not be what the animal actually prefers, and there may be individual differences in enrichment preferences. Preference testing is a way to "ask the animal" what it prefers or finds interactive \[[@B12-viruses-10-00353]\]. Preference testing is typically performed by providing an animal with a choice between two or more options, while the strength of preferences has been assessed by quantifying the amount of "work" that an animal will do to access a resource (\[[@B13-viruses-10-00353]\], and see \[[@B14-viruses-10-00353]\] for review). Enrichment is especially important in animals maintained for long-term studies, such as studies of cats infected with feline immunodeficiency virus (FIV) \[[@B15-viruses-10-00353]\]. Cats infected with FIV serve as a useful animal model for the study of human immunodeficiency virus (HIV) \[[@B16-viruses-10-00353],[@B17-viruses-10-00353]\], particularly psychomotor deficits seen with long-term infection. Cats infected with FIV for 12 or more months have been shown to be hyperactive in open field testing, more easily distracted, and more likely to have errors in completing new operant tasks compared to control cats \[[@B18-viruses-10-00353]\]. Because cats can transmit FIV via bites, FIV-infected cats are housed individually (with visual access to conspecifics). This requirement for social isolation makes provision of adequate and preferred enrichment for individual cats even more crucial. The goals of this study were, therefore, to assess the preferences for type of enrichment in laboratory-housed cats, and to determine if FIV status impacts these preferences. We predicted that individual cats do have a preference for specific types of enrichment, and that the FIV status of the cat will not systematically impact their preferences. In presenting enrichment types, we used the traditional method of presenting two choices simultaneously for some of our enrichment types; for others, we used an intensity of interaction as a surrogate for preference. Our findings supported our predictions; we found that cats had preferences for certain enrichment options, and that there were individual preferences for enrichment types. There were no differences in preferences between FIV-infected and sham-infected cats. 2. Materials and Methods {#sec2-viruses-10-00353} ======================== All protocols described in this study were approved by the North Carolina State University Institutional Animal Care and Use Committee (Protocol 12-176-B, 28 January 2013). 2.1. Experimental Animals {#sec2dot1-viruses-10-00353} ------------------------- The subjects were purpose-bred, specific-pathogen-free (SPF), castrated male domestic short hair cats (Felis catus) between 2--2.5 years of age and obtained from a commercial breeding facility. The cats were participating in a longitudinal study of the effects of FIV infection on physiological parameters and cognitive-motor skills (described in \[[@B19-viruses-10-00353]\]). Cats had been FIV-infected or sham-infected via intra-cranial inoculation 9 to 13 months prior to the initiation of the present study. The FIV status of each cat had been subsequently confirmed. Although all cats participated in a series of reward-based cognitive-motor tests on a prescribed schedule, the enrichment preference testing described here was conducted during a time when cats were not undergoing cognitive-motor testing. On physical examination, all cats were in good general health. Animal technicians and behavior team members were masked to the FIV status of the cats. 2.2. Housing and Husbandry {#sec2dot2-viruses-10-00353} -------------------------- Cats were housed in individual pens (188 cm high, 147 cm deep, 91 cm wide), each of which included an elevated perch, a hide, a litter pan (cleaned daily) and a water bowl. There were 5--6 pens located in each of four separate, identical rooms in the North Carolina State University Laboratory Animal Resources Unit (LAR). Cats were maintained on a 12/12-h light/dark cycle and fed a measured balanced feline dry ration (Hills Feline Hairball Control Diet, Hill's Pet Nutrition, Inc., Topeka, KS, USA) twice daily and weighed monthly in order to maintain body condition scores of 4--5/9, as referenced on a standard body condition score chart (Purina Body Condition Score Index, <http://www.purina.com/cat/weight-control/bodycondition.aspx>). All cats were conditioned to handling and carrier transport \[[@B20-viruses-10-00353]\], and were observed twice daily. At the start of the study, minimal untested environmental enrichment protocols were in place. Each cat was provided in its pen with a feline-appropriate play toy, which was rotated weekly. For 10 min each day, animal care technicians allowed each cat out of its pen individually for exercise, nail conditioning, and exploration. Twice a week, each cat was brushed by familiar animal technicians for examination, body care and to decrease the incidence of hairballs. Periodically, cats were taken to other rooms in carriers for physical evaluation, physiological assessment, or individual behavioral testing. Each cat room was continuously provisioned with a feline pheromone diffuser (Feliway^®^ Diffuser Ceva Animal Health, LLC, Lenexa, KS, USA), purported to reduce stress in cats \[[@B21-viruses-10-00353]\]. 2.3. Pre-Test Period {#sec2dot3-viruses-10-00353} -------------------- During the pre-test period, two months prior to data collection, interviews were conducted with animal technicians in order to learn which brushes, play toys, and nail-conditioning items already approved for safety and infectious disease control by LAR were utilized by the cats. After a literature search and discussion with technicians, specific items were chosen for their prior interest from the cats, their ease of cleaning, and their cost effectiveness. One month prior to data collection, the investigator (CK) acclimated the cats to her presence by entering each cat room for 30 min twice weekly and spending approximately 5 min brushing, playing, and individually interacting with each cat in the room. Commercially available cat treats were used to lure each cat back to his pen after the interactions. 2.4. Testing {#sec2dot4-viruses-10-00353} ------------ After the pre-test period, testing began. To evaluate behavioral choices for environmental enrichment, the cats were provided with five enrichment opportunities consisting of (1) choice of grooming brushes; (2) play with a laser light; (3) play with a 5 cm diameter ball; (4) interaction with the familiar investigator; and (5) two types of corrugated cardboard toenail conditioning objects. Each room had its own set of enrichment items, which were shared among the cats in the room, but not transferred between rooms. The order of presentation of enrichment opportunities 2--4, testing order of rooms, and testing order of cats within a room were randomized. All cats in a room were tested before moving to the next room. Each cat was individually tested a total of six times, using a standardized protocol consisting of five enrichment opportunities. A typical test protocol is described as follows: The investigator (CK) released each cat from its pen, then kneeled on the floor approximately 1 m from the pen door. Then, the cat was offered two brush types and its choice was recorded. Next, the cat was given 1 min each to interact with a laser light, directed by the investigator, a ball, and a familiar human (the investigator) (described in detail under Enrichment Choices). If the cat tried to interact with the investigator during laser light or ball testing periods, the investigator ignored the cat. The intensity of each cat's response was measured on a 0 to 2 activity scale. Zero (0) reflected no interest, in which the cat either did not interact with the object or did so only one time; one (1) reflected moderate interest, in which the cat interacted with the object multiple times, but lost interest throughout the trial period and wandered away; two (2) reflected intense interest, in which the cat interacted with the object during the entire trial period. At the end of each session, the cat was returned to his kennel. There, in the restricted space of the pen, he was offered two toenail conditioning objects for one minute and his choice was noted. 2.5. Experimental Procedures {#sec2dot5-viruses-10-00353} ---------------------------- All cats in a group were tested sequentially between the hours of 13:00 and 16:00. The trials were conducted over a period of 7 weeks when the cats were not scheduled for physiological sampling or cognitive-motor test protocols. 2.6. Enrichment Choices {#sec2dot6-viruses-10-00353} ----------------------- ### 2.6.1. Brushes {#sec2dot6dot1-viruses-10-00353} The cat was visually presented with two brushes, equidistant from the cat and held in the investigator's hands. Brush types were a 7-row Brigitte hair brush (Brigitte Brush, Brigitte's Brushes Company, Lanoka Harbor, NJ, USA) and a glove brush (Love Gloves, Four Paws Company, Neptune, NJ, USA). The cat made a selection by nudging or rubbing against one of the brushes, soliciting its use. The investigator recorded "no preference" if the cat did not allow interaction with either brush, and recorded "both brushes" if the cat alternated between brushes. The investigator was careful to not interact with the cat in other ways besides brushing, as to eliminate the cat choosing to be brushed based on human interaction. ### 2.6.2. Laser {#sec2dot6dot2-viruses-10-00353} During the laser task, the investigator turned on a handheld laser light designed to stimulate chasing activity in cats (Fun Beam Interactive Laser Toy, Petlinks System, San Rafael, CA, USA). The beam was directed around the room on the floor and walls at random for 1 min, then the cat's reaction was scored on the 0 to 2 scale. ### 2.6.3. Ball {#sec2dot6dot3-viruses-10-00353} A ping-pong ball was presented to each cat. The investigator tossed the ball around the room and recorded the intensity of the cat's interaction on the 0 to 2 scale. ### 2.6.4. Social Interaction {#sec2dot6dot4-viruses-10-00353} During the social interaction task, all other enrichment options were removed and the cat was allowed to freely interact with the investigator who was seated on the floor. The investigator encouraged interaction by calling the cat's name, patting her legs, and petting the cat if the cat chose to approach and interact. The investigator recorded the presence and intensity of the cat's interaction on the 0 to 2 scale. ### 2.6.5. Nail Conditioning Objects {#sec2dot6dot5-viruses-10-00353} In the final task, the investigator lured the cat back into his kennel with a treat and placed two nail conditioning objects into the cat's kennel: A nail conditioning circle with a ball surrounding the conditioning material (Star Chaser Scratcher Cat Toy, Coastal Pet Products, Alliance, OH, USA) and an inclined nail conditioning cardboard box (KONG Naturals Incline Scratching Cat Toy, Kong Company, Golden, CO, USA; used without catnip) in the kennel with the cat. The investigator allowed the cat 1 min to scratch on either the circle or box. The investigator recorded the cat's behavior, based on whether the cat scratched on one, both, or none of the nail conditioning objects. 2.7. Experimental Outcomes {#sec2dot7-viruses-10-00353} -------------------------- The primary objective of these trials was to determine whether or not cats demonstrate a preference toward a specific type of enrichment. The secondary objective was to determine whether or not FIV status was associated with cat preferences. 2.8. Statistical Methods {#sec2dot8-viruses-10-00353} ------------------------ Descriptive statistics were used to describe the percentages of cats showing each level of interest or preference for each enrichment type. For enrichment choice preference, a repeated-measures ordinal logistic regression model, PROG GENMOD (SAS Version 9.4, Cary, NC, USA) was used, with FIV status as a covariate for each of the enrichment types. The model allowed for the effect of day, order, and enrichment type. Linear contrasts were used to test specific effects. 3. Results {#sec3-viruses-10-00353} ========== 3.1. Animals {#sec3dot1-viruses-10-00353} ------------ Twenty-three individually-housed cats completed this study. All 23 cats participated in six trials for each of the six enrichment options, with 138 trials (in total) per enrichment type. Mean weight of the cats (measured monthly) was 3.71 kg (range 3.4--4.56 kg). There were 17 FIV positive cats and six FIV negative cats. At the start of the study, FIV positive cats had been infected for 9--13 months. No adverse events were observed by the investigator or the animal technicians as a result of these experiments. 3.2. Preferences for Enrichment Type {#sec3dot2-viruses-10-00353} ------------------------------------ ### 3.2.1. Brushes {#sec3dot2dot1-viruses-10-00353} In 109/138 trials (79.0%), cats showed interest in one or both brushes; in 29/138 trials (21.0%), cats showed no interest in either brush, preferring to explore the room or solicit attention from the investigator. When cats chose to interact with a brush, the glove brush was preferred in 44/109 trials (40.4%) while the human hairbrush was preferred in 23/109 trials (21.1%). Equal use of both brushes was found in 42/109 trials (38.5%). ### 3.2.2. Laser, Ball, and Social Interaction {#sec3dot2dot2-viruses-10-00353} For each of these enrichment types, the number and percentage of the 138 trials that received each score are shown in [Table 1](#viruses-10-00353-t001){ref-type="table"}. For each cat, the intensity scores were averaged over the six trials for each enrichment type, and the number of cats within each range of scores is shown in [Table 2](#viruses-10-00353-t002){ref-type="table"}. - There were no statistically significant differences present based on the order that each enrichment activity was presented (p = 0.16). - The only statistically significant difference found in the analysis of preference was that the laser was preferred over the ball (p = 0.04) and over the social interaction enrichment option (p = 0.02). - There were individual preferences for enrichment objects and a range of enrichment intensity scores for each cat. ### 3.2.3. Nail Conditioning Objects {#sec3dot2dot3-viruses-10-00353} Of the two toenail conditioning objects, the majority of the 23 cats chose to scratch on the inclined box or they chose not to scratch at all during the time the toenail conditioning objects were offered. Of the 138 trials, cats chose to use one or both of the objects in 85 (61.6%). Of these 85 trials, cats chose the inclined box only in 69 (81.2%), the circle only in 3 (3.5%), and both the box and circle in 13 (15.3%). The cats in our study strongly preferred the inclined box to the circular nail conditioning object. ### 3.2.4. FIV Status {#sec3dot2dot4-viruses-10-00353} No significant differences were found on any measure between FIV positive and FIV negative cats using a repeated-measures ordinal logistic regression (p = 0.57). 4. Discussion {#sec4-viruses-10-00353} ============= The aim of this study was to determine laboratory cat preferences for enrichment items (brush, laser, ball, social interaction, and toenail conditioning objects), and to determine whether FIV status affected the cats' preferences in these enrichments. We found that, in general, cats show a preference to be brushed, play with a laser, and scratch on an inclined-box nail conditioning object compared to play with a ball, interact socially with a familiar human, or scratch on a circular nail conditioning object. Further, cats appeared to have individual preferences for specific objects or interactions. When the two different brush types (glove brush and hairbrush) were introduced to the cats for tactile stimulation, cats selected either the glove brush, or both brushes. Few cats selected the human hairbrush or the no-brush option. The overall choice by the cats for brush compared to no-brush, suggests that cats favor the opportunity for the tactile stimulation of grooming. Cats groom regularly by licking their fur with their tongues, and this behavior can lead to ingestion of hair and consequent hairball formation and vomiting \[[@B22-viruses-10-00353]\]. Therefore, brushing to remove excess hairs can reduce hairball risk and promote positive interactions between caregivers and cat \[[@B1-viruses-10-00353]\]. As the cats chose to participate in the brushing, the brushing sessions may be beneficial from both a health and welfare standpoint. Caretakers may also use the brushing period to inspect the cat for any other medical problems, including injuries not readily observable. Using a comparison of the intensity of interaction scores as a surrogate for preference, we found that among the play objects, the laser was preferred over the ball and human interaction. The laser, a fast-moving, novel form of enrichment, may elicit predatory instincts of the cats, stimulating their natural species-specific hunting behaviors. Toys that move quickly seem to be especially attractive to cats, as such objects allow them to express the dual instincts to play and hunt after the play object \[[@B6-viruses-10-00353]\]. It has been argued that the laser can be frustrating since the light can never be "caught" by the cats \[[@B6-viruses-10-00353]\]. However, we did not observe signs consistent with "frustration" behavior, such as abandonment of laser-directed activity or redirected aggression. In general, the cats showed relatively little interest in the ball, although there were individual differences in cat interest. Compared to the laser, the ball was less kinetic and possibly less engaging to the cats. The cats also had more experience with the ball compared to other objects. In the facility, similar balls were occasionally thrown by caretakers during cleaning procedures, when cats were individually let out of the cages each day for exercise and social interaction. A ball was also left in the cages of the cats for enrichment by caretakers. This could make the ball less novel to the cats, in comparison to the laser. The effect of novelty, however, is inconsistent in enrichment studies, and context appears to be important. Studies in several species have shown that animals may preferentially approach \[[@B23-viruses-10-00353]\] or avoid \[[@B24-viruses-10-00353]\] novel objects, and that this effect may vary based on individual \[[@B25-viruses-10-00353]\] and environmental factors \[[@B24-viruses-10-00353]\]. Research in cats is lacking, however a study in tigers has found benefits of enrichment with novel toys \[[@B26-viruses-10-00353]\] and in one study in dogs, novel toys were chosen over familiar toys in 76% of choice trials \[[@B23-viruses-10-00353]\]. In addition, these findings must be interpreted in light of the fact that a comparison of intensity of interaction is not a true test of preference in the classical sense, where objects are compared simultaneously. As the enrichment options were presented serially, we used the intensity of interaction scores to evaluate the cats' engagement, and a comparison between these scores to indicate preference. Future work with two-choice comparisons would be useful to further evaluate preferences. Individual cats varied in their interest and the intensity of social interaction with the investigator. All cats scored higher on laser play compared to human social interaction. Social interaction with humans has been shown to positively impact the wellbeing of cats \[[@B6-viruses-10-00353]\]. In addition, cats with human interactions have been shown to be easier to handle and more sociable \[[@B27-viruses-10-00353]\]. Cat behavior can be impacted by the familiarity of the cat with the investigator \[[@B28-viruses-10-00353]\]. During the pre-testing phase, the investigator noted that with increased familiarity, the cats selected to spend more time in close proximity. The cats preferred the inclined box toenail conditioning object to the flat, circular nail conditioning object and, on the majority of trials, cats chose to use the opportunity to scratch. The cats tended to roll all over the inclined box and stretch across it. We suspect that the inclined nature of the box, allowing them to stretch and roll, made it preferable to the cats. Toenail conditioning objects allow cats to exhibit natural species-specific behaviors related to scent marking and claw-sharpening \[[@B29-viruses-10-00353]\]. Fulfilling the biological need of nail conditioning is essential in any feline welfare plan. The observed preference for the inclined box over the flat circle may be related to the manner in which wild cats are known to scratch trees, by stretching up the surface of the tree and conditioning vertically \[[@B30-viruses-10-00353]\]. The ranges in average intensity score show that cats differ individually in the intensity of their enrichment preferences, with mean intensity scores for each cat falling within a range from 0--2. This suggests that individual differences should be considered as a variable in an enrichment plan to enhance cat welfare, with preference testing performed on an individual basis from among enrichment items. We did not see differences in enrichment preferences or intensity of interaction as a function of FIV status of the feline subjects. This study was conducted on cats with and without FIV; the range of infection time was 9--13 months. Since FIV is a chronic, progressive infection, these cats might not have been infected long enough to begin to show behavioral differences. Previous studies have shown that FIV cats do not show measurable cognitive motor decline until 12--18 or more months after infection. In fact, it has been shown that cats infected with FIV for two to three years have only small declines in cortical neuronal quantity \[[@B31-viruses-10-00353]\]. Future studies should look at FIV positive cats with a longer infection time to determine if the length of infection has any impact on enrichment preferences. With no significant differences found in behavioral responses between FIV positive and FIV negative cats, the overall enrichment preference trends identified in the present study may reasonably be extended to all individually housed laboratory cats. However, a limitation is the unequal FIV status categories (FIV+, n = 17; FIV−, n = 6), thus the generalizability of our data to a larger population of non-FIV infected research cats is unknown. This study had several limitations. Care was taken to familiarize the cats with the investigator prior to the study as cats may show avoidance or inhibition of normal behavior in the presence of strangers \[[@B32-viruses-10-00353]\]. It is not known if the findings of the present study can be generalized with regard to the familiarity of the cats with a person implementing the described enrichment strategies. In addition, cats were housed in individual cages as a requirement of the FIV study, whereas group housing is recommended for compatible cats without such restrictions \[[@B1-viruses-10-00353]\]. Cats were housed in individual cages, with 5--6 cages within a room, and were tested individually in the rooms. However, in some instances when one focal animal was being tested, other cats, still in cages, became activated by the testing protocol, which could have affected the behavior of the focal animal. In addition, since a mandate of the laboratory animal facility was to provide some enrichment objects to all cats at all times, and a lack of enrichment would represent a welfare challenge, the cats were already exposed to many enrichment objects, on a rotating basis, within their cages. These objects included balls (Wiffle balls (The Wiffle Ball Inc., Shelton, CT, USA), jingle balls, golf balls, and extra small Best balls), or other objects such as crazy circles, shake rattle and roll toys, Whirlygigs, Nylabones, and PVC elbow pipes. Animal caregivers rotated the toy available to each cat weekly. The availability of a ball in the cats' cages could have contributed to a lack of interest in the ping-pong ball provided to the cats during the trials, as the cats were already familiar with a ball, and the ball could have lost its novelty compared with other play object options; however, evidence of the effect of novelty on preference is inconclusive. The cats had also been previously exposed to the circle nail conditioning object. This familiarity could have contributed to the cats' preference for the inclined box nail conditioning object. Anecdotally, since the end of this study and the recent availability of the inclined box nail conditioning object (in addition to the circle), caretakers have noted cat preference is consistent with the findings reported here. Further, the nutritional environment can be enriched through feeding techniques that extend the time the cat spends interacting with the food, such as food puzzles and delivery toys \[[@B11-viruses-10-00353]\]. While a recent study suggested that food puzzles may not increase the overall daily activity of cats \[[@B33-viruses-10-00353]\], they can increase their interaction with the food, more closely mimic natural behavior, and decrease aggressive, fearful behaviors, anxiety, and depression in housecats \[[@B11-viruses-10-00353]\]. Further work could evaluate preferences for food puzzles in laboratory-housed cats as an enrichment form that stimulates natural feeding behaviors. Finally, as behavioral scoring is inherently subjective, a future refinement would include videotaping sessions and independent scoring of preferences from video. 5. Conclusions {#sec5-viruses-10-00353} ============== Overall, both FIV positive and FIV negative cats showed a clear preference for specific enrichment items, reflected in either simultaneous presentation or higher ratings of intensity of interaction. These methods and findings may be used by laboratory animal facilities to select enrichment objects and interactions with laboratory cats. It has been shown that FIV is an effective model for studying HIV, in hopes that any vaccines developed for FIV can be extended to HIV. Therefore, it is critical to properly address the welfare needs of these models \[[@B34-viruses-10-00353]\]. Long-term confinement, common in laboratory cats, is a welfare concern for animals, and enrichment is an integral component of their welfare. Enrichment items used in this study were commercially available at reasonable cost, could be cleaned effectively according to laboratory-facility standards, and were durable. Our finding of individual differences in enrichment preferences suggest that enrichment procedures should be adjusted based on behavioral traits of individual cats. Individually tailoring enrichment, based on preferences, would be a manageable task in laboratory animal settings. We recognize Lola Hudson for her contributions to the long-term study of FIV in laboratory cats at NC State University; we thank the North Carolina State University College of Veterinary Medicine Laboratory Animal Resources Staff and veterinarians for their excellent animal care. Conceptualization, C.J.K., A.E.T., B.L.S. and M.E.G.; Data curation, E.H.G., B.L.S. and M.E.G.; Formal analysis, E.H.G.; Funding acquisition, J.F. and R.B.M.; Investigation, C.J.K.; Methodology, C.J.K., A.E.T., B.L.S. and M.E.G.; Resources, J.F., B.D.X.L. and R.B.M.; Supervision, B.L.S.; Writing---original draft, C.J.K.; Writing---review and editing, A.E.T., E.H.G., J.F., B.D.X.L., R.B.M., B.L.S. and M.E.G. This work was supported by NIH \#NS086426, "LM11A-31 Neuroprotective Efficacy in an Animal Model of HIV" (Meeker: PI). The authors declare no conflicts of interest. The funding agency had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results. viruses-10-00353-t001_Table 1 ###### Interest scores for laser, ball, and social interaction across the 138 trials. The number of trials at each score, and the percentage of the total trials is shown for each enrichment type. A score of 0 indicates no interest, while a score of 2 indicates high interest. Enrichment Type Score 0: n (%) Score 1: n (%) Score 2: n (%) -------------------- ------------------ ------------------ ------------------ Laser 18 (13.0) 28 (20.3) 92 (66.7) Ball 48 (34.8) 36 (26.1) 54 (39.1) Social Interaction 54 (39.1) 37 (26.8) 47 (34.1) viruses-10-00353-t002_Table 2 ###### Average interest scores for laser, ball, and social interaction across the 23 cats. The number of cats with averages in each range are shown for each enrichment type. A score of 0 indicates no interest, while a score of 2 indicates high interest. Enrichment Type Average Score \< 1: n (%) Average Score 1.0--1.4: n (%) Average Score 1.5--2: n (%) -------------------- ----------------------------- --------------------------------- ------------------------------- Laser 4 (17.4) 4 (17.4) 15 (65.2) ^1^ Ball 11 (47.8) 2 (8.7) 10 (43.5) Social Interaction 11 (47.8) 5 (21.7) 7 (30.4) ^1^ 11 of these cats scored 2 s in all 6 trials for the laser.	{ "pile_set_name": "PubMed Central" }
Type 2 diabetes is preceded by a long pre-diabetic state, characterized by mild elevation of fasting and/or postprandial glucose levels. This asymptomatic phase may last for years, and about one-third of these individuals finally develop type 2 diabetes ([@B1]). The pre-diabetic state, defined by an oral glucose tolerance test (OGTT), includes impaired fasting glucose (IFG), impaired glucose tolerance (IGT), or their combination ([@B2]). Epidemiological studies have shown that IFG and IGT represent two distinct subgroups of abnormal glucose tolerance ([@B1],[@B3][@B4]--[@B5]) that differ in their age and sex distribution ([@B6],[@B7]) and associated cardiovascular risk ([@B8]). Therefore, IFG and IGT are likely to have different pathophysiologies. Impaired insulin secretion and impaired insulin action are the two main pathophysiological disturbances leading to abnormal glucose tolerance. Previous studies on the role of impaired insulin secretion and insulin resistance in the development of IFG and IGT have yielded contradictory results ([@B4][@B5][@B6][@B7][@B8][@B9][@B10][@B11][@B12][@B13][@B14][@B15][@B16][@B17][@B18][@B19][@B20][@B21][@B22]--[@B23]). Inconsistencies across the studies are explained by differences in study populations, study designs and methods to assess insulin resistance and insulin secretion, and most importantly by a small sample size. Categorization of glucose tolerance is based on arbitrary cutoff points of glucose levels, and therefore different subgroups cannot fully account for changes in β-cell function and insulin action with increasing glycemia. Only a few studies have examined insulin secretion and/or insulin sensitivity as a function of glucose concentrations ([@B13],[@B24][@B25][@B26][@B27]--[@B28]). These studies have been, however, relatively small, and most of them were conducted in non-Caucasian populations. The aim of this study was to evaluate insulin sensitivity and insulin secretion across the entire range of fasting and 2-h plasma glucose (PG) from normal glucose tolerance (NGT) to type 2 diabetes to understand better the pathophysiology of the pre-diabetic state. Furthermore, we investigated the differences in insulin sensitivity and insulin release in different glucose tolerance subgroups. To address these questions, we collected a large sample of carefully phenotyped middle-aged Finnish men. RESEARCH DESIGN AND METHODS =========================== A total of 6,414 men from the ongoing population-based cross-sectional Metabolic Syndrome in Men (METSIM) study were included in the study. Subjects, aged from 45 to 70 years, were randomly selected from the population register of the town of Kuopio in eastern Finland (population 95,000). Every participant had a 1-day outpatient visit to the Clinical Research Unit at the University of Kuopio, including an interview on the history of previous diseases and current drug treatment and an evaluation of glucose tolerance and cardiovascular risk factors. Fasting blood samples were drawn after 12 h of fasting followed by an OGTT. The study was approved by the ethics committee of the University of Kuopio and Kuopio University Hospital, and it was in accordance with the Helsinki Declaration. Clinical measurements. ---------------------- Height and weight were measured to the nearest 0.5 cm and 0.1 kg, respectively. BMI was calculated as weight (kg) divided by height (m) squared. Waist (at the midpoint between the lateral iliac crest and lowest rib) and hip circumference (at the level of the trochanter major) were measured to the nearest 0.5 cm. Body composition was determined by bioelectrical impedance (RJL Systems) in subjects in the supine position after a 12-h fast ([@B29]). OGTT. ----- A 2-h OGTT (75 g of glucose) was performed, and samples for PG and insulin were drawn at 0, 30, and 120 min. Glucose tolerance was evaluated based on OGTT as follows: NGT (fasting PG \[FPG\] \<5.6 mmol/l and 2-h PG \<7.8 mmol/l), isolated IFG (FPG 5.6--6.9 mmol/l and 2-h PG \<7.8 mmol/l), isolated IGT (FPG \<5.6 mmol/l and 2-h PG between 7.8 and 11.0 mmol/l), IFG + IGT (FPG 5.6--6.9 mmol/l and 2-h PG 7.8--11.0 mmol/l), and newly diagnosed type 2 diabetic subjects (FPG ≥7.0 mmol/l and/or 2-h PG ≥11.1 mmol/l) ([@B2]). Laboratory determinations. -------------------------- Plasma glucose was measured by enzymatic hexokinase photometric assay (Konelab Systems reagents; Thermo Fischer Scientific, Vantaa, Finland). Insulin was determined by immunoassay (ADVIA Centaur Insulin IRI no. 02230141; Siemens Medical Solutions Diagnostics, Tarrytown, NY). Calculations. ------------- The trapezoidal method was used to calculate glucose area under the curve (AUC) and insulin AUC during the OGTT. Surrogate indexes of insulin sensitivity and insulin secretion were calculated according to published formulas ([Tables 1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}) ([@B25],[@B30][@B31][@B32]--[@B33]), using glucose and insulin concentrations at 0, 30, and 120 min. ###### Spearman correlation coefficients between surrogate indexes of β-cell function and insulin release during the first phase (0--10 min) and second phase (10--60 min) and total (0--60 min) during an IVGTT β-Cell function indexes Insulin release during an IVGTT ------------------------------- --------------------------------- ------- ------- HOMA-β 0.568 0.645 0.663 Insulinogenic index 0.579 0.469 0.541 InsAUC~30~/GluAUC~30~ 0.666 0.707 0.743 ΔInsAUC~120~/ΔGluAUC~120~ 0.375 0.409 0.405 InsAUC~120~/GluAUC~120~ 0.648 0.750 0.775 First-phase Stumvoll 0.651 0.663 0.702 Second-phase Stumvoll 0.662 0.692 0.730 Fasting insulin 0.548 0.747 0.735 Insulin at 30 min of an OGTT 0.615 0.691 0.715 Insulin at 120 min of an OGTT 0.429 0.673 0.645 Insulin AUC during an OGTT 0.558 0.770 0.776 P \< 0.001 for all correlation coefficients. Indexes of β-cell function were calculated as described previously: HOMA-β([@B30]), insulinogenic index ([@B31]), ΔInsAUC~120~/ΔGluAUC~120~ ([@B25]), InsAUC~120~/GluAUC~120~ ([@B31]), first phase Stumvoll ([@B32]), and second phase Stumvoll ([@B32]). InsAUC~30~/GluAUC~30~ = (insulin at 0 min + insulin at 30 min of an OGTT)/(glucose at 0 min + glucose at 30 min of an OGTT). ###### Spearman correlation coefficients between surrogate indexes of insulin sensitivity and insulin sensitivity assessed by a euglycemic-hyperinsulinemic clamp (M value) Insulin sensitivity during clamp --------------------------------------------------- ---------------------------------- ------- Insulin sensitivity indexes 1/fasting insulin 0.625 0.716 1/HOMA-IR 0.605 0.708 QUICKI[\](#TF2-1){ref-type="table-fn"} 0.605 0.708 Matsuda ISI[\](#TF2-1){ref-type="table-fn"} 0.699 0.776 MCR Stumvoll 0.701 0.683 ISI Stumvoll[\](#TF2-1){ref-type="table-fn"} 0.705 0.693 Indexes of insulin sensitivity were calculated as described previously: HOMA-IR ([@B30]), quantitative insulin sensitivity index ([@B30]), Matsuda ISI ([@B33]), MCR Stumvoll ([@B31]), and ISI Stumvoll ([@B31]). For all correlation coefficients, P* \< 0.001. \Surrogate indexes were validated against M~LBM~/I. ISI, insulin sensitivity index; MCR, metabolic clearance rate; QUICKI, quantitative insulin sensitivity index. Statistical analysis. --------------------- Data are presented as the means ± SD, median (25th, 75th percentile) for continuous variables, or as count (percentage) for categorical variables. Variables with nonnormal skewed distribution were logarithmically transformed before analysis. Continuous variables were compared across the categories of glucose tolerance by ANOVA or after adjustment for covariates using the general linear model. Pairwise comparisons between the groups were performed by Bonferroni post hoc tests (with P* value adjustment for multiple testing for each variable). Categorical variables were examined by χ^2^ test. Spearman\'s rank correlation was used to compare the surrogate indexes with the reference measures. Analyses were conducted with SPSS version 14 (SPSS, Chicago, IL). P values \<0.05 were considered statistically significant. Validation study. ----------------- A separate sample of 287 nondiabetic Finnish offspring of type 2 diabetic patients from the region of Kuopio was used to validate the OGTT-derived indexes of insulin secretion and insulin sensitivity against parameters measured by the intravenous glucose tolerance test (IVGTT) and euglycemic-hyperinsulinemic clamp. The study protocol has been described previously ([@B10]). All subjects underwent a 2-h OGTT (75 g of glucose) and on a separate occasion a frequently sampled IVGTT, followed by a euglycemic-hyperinsulinemic clamp for 120 min, as described previously ([@B10]). The study was approved by the ethics committee of the University of Kuopio and was in accordance with the Helsinki Declaration. Spearman\'s correlations were calculated between insulin AUCs during the first 10 min and during 10--60 and 0--60 min of the IVGTT, considered as the reference indexes of first-phase, second-phase, and total insulin secretion, respectively, and surrogate indexes of insulin secretion ([Tables 1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). The ratio of total insulin AUC and total glucose AUC during 0--30 min of the OGTT (InsAUC~30~/GluAUC~30~) displayed the highest correlation with the first-phase insulin release during the IVGTT (rho = 0.67, P \< 0.001) ([Table 1](#T1){ref-type="table"}). Insulin AUC during 120 min of the OGTT (InsAUC~120~), alone or adjusted for the corresponding glucose AUC (InsAUC~120~/GluAUC~120~), showed the highest correlation with the second-phase and total insulin release (rho ranging from 0.75 to 0.78, P \< 0.001) ([Table 1](#T1){ref-type="table"}). Spearman\'s correlation analysis was also performed between the ratio of whole-body glucose uptake to lean body mass and to mean insulin concentration during the last 60 min of the euglycemic clamp (M~LBM~/I), considered as a reference index of insulin sensitivity, and surrogate indexes of insulin sensitivity. Matsuda ISI (calculated from glucose and insulin levels at 0, 30, and 120 min of the OGTT) showed the highest correlation with the clamp-derived M~LBM~/I (rho = 0.78, P \< 0.001) ([Table 2](#T2){ref-type="table"}). Based on the above correlation analyses, we used Matsuda ISI as a surrogate index of insulin sensitivity, InsAUC~30~/GluAUC~30~ as a surrogate index of the early-phase insulin release, and InsAUC~120~/GluAUC~120~ as a surrogate index of total insulin release in the METSIM study. We also calculated two disposition indexes as the products of insulin sensitivity × insulin secretion (DI~30~ = Matsuda ISI × InsAUC~30~/GluAUC~30,~ DI~120~ = Matsuda ISI × InsAUC~120~/GluAUC~120~) as a measure of the β-cell response to insulin sensitivity. In a previous study analyzing several measures of disposition indexes based on insulin secretion and insulin sensitivity from an OGTT, the product of InsAUC/GluAUC and Matsuda ISI was the only combination that followed the hyperbolic relationship that is the requirement for a disposition index ([@B34]). A recent study indicated that oral disposition index (insulinogenic index × 1/HOMA-IR, where HOMA-IR is the homeostasis model assessment for insulin resistance) is predictive of diabetes over 10 years ([@B35]). We used 1/HOMA-IR as an approximate measure of insulin sensitivity in the liver ([@B33],[@B36]). The limitation of this index is that it also measures peripheral insulin sensitivity. RESULTS ======= Baseline characteristics. ------------------------- The age of 6,414 men in this study was 57 ± 7 years, and their BMI was 27.0 ± 3.9 kg/m^2^. Altogether, 2,168 subjects (34%) had NGT, 2,859 (45%) had isolated IFG, 217 (3%) had isolated IGT, 701 (11%) had a combination of IFG and IGT, and 469 (7%) had newly diagnosed type 2 diabetes. A total of 492 subjects with previously diagnosed diabetes were excluded from statistical analyses. As shown in [Table 3](#T3){ref-type="table"}, subjects with NGT and isolated IFG were significantly younger than subjects with isolated IGT, IFG + IGT, and newly diagnosed type 2 diabetes. Subjects with isolated IGT were significantly more obese than subjects with isolated IFG (according to waist-to-hip ratio and fat mass). Subjects with IFG + IGT and newly diagnosed type 2 diabetes were significantly more obese than subjects with isolated IFG or isolated IGT. ###### Clinical and anthropometric characteristics of study participants according to glucose tolerance status n NGT Isolated IFG Isolated IGT IFG + IGT Newly diagnosed diabetes P Total ------------------------------------- ------- ------------- -------------- ----------------------------------------------------------------------------- --------------------------------------------- ----------------------------------------------------------------------------- -------------- ------------- n --- 2,168 2,859 217 701 469 --- 6,414 Age (years) 6,414 57.5 ± 6.7 56.9 ± 6.8 60.3 ± 6.9 59.2 ± 6.8[\](#TF3-1){ref-type="table-fn"} 59.6 ± 6.4[\](#TF3-1){ref-type="table-fn"}[†](#TF3-2){ref-type="table-fn"} 6 × 10^−31^ 57.7 ± 6.8 BMI (kg/m^2^) 6,410 25.8 ± 3.4 27.0 ± 3.6 26.9 ± 3.7[‡](#TF3-3){ref-type="table-fn"} 28.9 ± 4.4 29.7 ± 4.9 8 × 10^−131^ 27.0 ± 3.9 Weight (kg) 6,411 79.8 ± 11.7 83.9 ± 12.7 82.4 ± 13.4[§](#TF3-4){ref-type="table-fn"}[‡](#TF3-3){ref-type="table-fn"} 88.7 ± 14.8 91.0 ± 15.6 3 × 10^−93^ 83.5 ± 13.3 Waist circumference (cm) 6,410 94.6 ± 9.6 97.9 ± 10.4 98.7 ± 10.1[‡](#TF3-3){ref-type="table-fn"} 103.5 ± 11.5 105.7 ± 12.0 2 × 10^−140^ 98.0 ± 10.9 Hip circumference (cm) 6,409 99.4 ± 6.2 101.1 ± 6.4 100.6 ± 6.7[§](#TF3-4){ref-type="table-fn"}[‡](#TF3-3){ref-type="table-fn"} 103.4 ± 7.9 104.5 ± 8.1[\](#TF3-1){ref-type="table-fn"} 3 × 10^−73^ 101.1 ± 6.8 Waist-to-hip ratio 6,409 0.95 ± 0.06 0.97 ± 0.06 0.98 ± 0.06 1.00 ± 0.06 1.01 ± 0.07 1 × 10^−119^ 0.97 ± 0.06 Fat mass (%) 6,401 22.5 ± 6.3 23.4 ± 6.0 26.2 ± 6.8 27.2 ± 6.7[\](#TF3-1){ref-type="table-fn"} 27.7 ± 6.1[†](#TF3-2){ref-type="table-fn"} 3 × 10^−108^ 23.9 ± 6.5 BMI ≥27 kg/m^2^ (median) 6,410 669 (31) 1,276 (45) 110 (51) 452 (65) 324 (69) 1 × 10^−84^ 2,831 (44) Positive family history of diabetes 6,412 889 (41) 1,340 (47) 104 (48) 348 (50) 254 (54) 5 × 10^−8^ 2,935 (46) Data are means ± SD or n (%). P values for overall comparison between five categories of glucose tolerance are shown (ANOVA for continuous variables, χ^2^ test for categorical variables). Bonferroni post hoc tests (continuous variables): all pairwise comparisons between categories of glucose tolerance were significant at P \< 0.05, except for those marked as follows: \*P \> 0.05 vs. isolated IGT; †P \> 0.05 vs. IFG + IGT; ‡P \> 0.05 vs. isolated IFG; §P \> 0.05 vs. NGT. Insulin sensitivity according to fasting and 2-h PG concentration. ------------------------------------------------------------------ We generated different categories of FPG and 2-h PG to investigate the relationship between the peripheral insulin sensitivity (Matsuda ISI) or markers of early-phase and total glucose-stimulated insulin secretion and glycemia. Categories with FPG \<5.0 mmol/l and 2-h PG \<5.0 mmol/l were set as the reference categories. We observed a considerable decrease in age- and BMI-adjusted peripheral insulin sensitivity (−17%) within the normal range of FPG, compared with the reference category. Insulin sensitivity further decreased to −50% within the range of IFG and decreased to −67% in the diabetic range of FPG ([Fig. 1](#F1){ref-type="fig"}A). A substantial decrease in insulin sensitivity (−37%) was also observed within the normal range of 2-h PG. Insulin sensitivity further decreased to −51% within the IGT range and to −57% within the diabetic range of 2-h PG ([Fig. 1](#F1){ref-type="fig"}B). When changes in insulin sensitivity according to the levels of both FPG and 2-h PG were examined, the highest insulin sensitivity was observed in subjects with FPG \<5.0 mmol/l and 2-h PG \<5.0 mmol/l, and the lowest insulin sensitivity was seen in subjects in the diabetic range of FPG and 2-h PG (supplementary Fig. 1, available in an online appendix at <http://diabetes.diabetesjournals.ord/cgi/content/full/db08-1607/DCI>). ![Insulin sensitivity (Matsuda ISI) (A and B), early-phase insulin release (InsAUC~30~/GluAUC~30~) (C and D), and total insulin release during the OGTT (InsAUC~120~/GluAUC~120~) (E and F) across the categories of FPG and 2-h PG. Bars display the value of insulin sensitivity or insulin release relative to the reference category (FPG \<5.0 mmol/l, 2-h PG \<5.0 mmol/l). Calculations were based on geometric means, adjusted for age and BMI with the general linear model. Cutoff values for different categories of FPG were (in mg/dl): 90.1 (5.0 mmol/l), 99.1 (5.5 mmol/l), 108.1 (6.0 mmol/l), 117.1 (6.5 mmol/l), 126.1 (7.0 mmol/l), 135.1 (7.5 mmol/l), 144.1 (8.0 mmol/l), 153.2 (8.5 mmol/l), and 162.2 (9.0 mmol/l). Cutoff values for different categories of 2-h PG were (in mg/dl): 90.1 (5.0 mmol/l), 108.1 (6.0 mmol/l), 126.1 (7.0 mmol/l), 144.1 (8.0 mmol/l), 162.2 (9.0 mmol/l), 180.2 (10.0 mmol/l), 198.2 (11.0 mmol/l), 216.2 (12.0 mmol/l), and 234.2 (13.0 mmol/l).](zdb0050957170001){#F1} Insulin release according to fasting and 2-h PG concentration. -------------------------------------------------------------- Age- and BMI-adjusted early-phase insulin release (InsAUC~30~/GluAUC~30~) decreased only slightly (−4%) within the normal range of FPG. It further decreased within the range of IFG and diabetes to −25 and −70%, respectively ([Fig. 1](#F1){ref-type="fig"}C). The early-phase insulin release decreased by −6% within the normal range of 2-h PG, and further decreased to −23 and −50% within the range of IGT and diabetes, respectively ([Fig. 1](#F1){ref-type="fig"}D). Age- and BMI-adjusted total insulin release (InsAUC~120~/GluAUC~120~) decreased to −13% within the range of IFG, and to −70% within the diabetic range of FPG ([Fig. 1](#F1){ref-type="fig"}E). Total insulin release increased by 14% with higher 2-h PG up to 9.9 mmol/l, and then it decreased to −45% within the diabetic range of 2-h PG ([Fig. 1](#F1){ref-type="fig"}F). The largest decreases in both early-phase (−32 to −50%) and total (−17 to −45%) insulin release were observed within the range of FPG from 7.0 to 7.9 mmol/l ([Fig. 1](#F1){ref-type="fig"}C and E). Disposition index. ------------------ The early-phase DI~30~ and total DI~120~ decreased with higher FPG within the normal range by −21 and −18%, respectively. Within the IFG range, the reduction in DI~30~ and DI~120~ reached −63 and −57% ([Fig. 2](#F2){ref-type="fig"}A). As a function of 2-h PG, DI~30~ and DI~120~ decreased to −41 and −30% in the normal range and further decreased to −60 and −48% in the IGT range ([Fig. 2](#F2){ref-type="fig"}B). ![Disposition indexes corresponding to early-phase (DI~30~ = Matsuda ISI × InsAUC~30~/GluAUC~30~) and total insulin release (DI~120~ = Matsuda ISI × InsAUC~120~/GluAUC~120~) across the categories of FPG (A) and 2-h PG (B). Cutoff values for different categories of FPG and 2-h PG are as explained in [Fig. 1](#F1){ref-type="fig"}.](zdb0050957170002){#F2} Compensatory insulin secretion. ------------------------------- Compensatory insulin secretion was not observed, despite a significant decrease in insulin sensitivity within the normal range of FPG, but, in contrast, the early-phase insulin release started to fall. However, compensatory total insulin secretion already started at low 2-h PG levels, and insulin release increased up to 10 mmol/l and then started to decrease (supplementary Fig. 2). A decrease in DI indexes was already substantial in the normal ranges of FPG and 2-h PG. Our findings remained essentially similar after adjustment for glucose level (analyses based on FPG were adjusted for 2-h PG, and analyses based on 2-h PG were adjusted for FPG) in addition to age and BMI. Insulin sensitivity and insulin release according to glucose levels in nonobese and obese individuals. ------------------------------------------------------------------------------------------------------ No significant interaction between BMI and glucose levels in determining insulin sensitivity was found. Obese subjects (BMI ≥27 kg/m^2^) within the reference categories of FPG and 2-h PG (\<5.0 mmol/l) had reduced insulin sensitivity by −45 and −41% compared with nonobese subjects (BMI \<27 kg/m^2^). The decrease in insulin sensitivity with higher FPG and 2-h PG was similar in both nonobese and obese subgroups: −17% (nonobese subjects) and −13% (obese subjects) within the normal range of FPG and −37 and −36%, respectively, within the normal range of 2-h PG ([Fig. 3](#F3){ref-type="fig"}A and B). Obese subjects within the reference categories of FPG and 2-h PG had increased early-phase (+70 and +54%) and total insulin release (+68 and +46%) compared with nonobese subjects. However, changes in both early-phase and total insulin release with higher FPG or 2-h PG were comparable in nonobese and obese subjects ([Fig. 3](#F3){ref-type="fig"}). The interaction between glucose levels and BMI (cutoff point of 27 kg/m^2^) in determining insulin release was not significant, except for the interaction between 2-h PG and BMI in determining DI~120~ (P = 0.001). ![Insulin sensitivity (Matsuda ISI) (A and B), early-phase insulin release (InsAUC~30~/GluAUC~30~) (C and D), and total insulin release in an OGTT (InsAUC~120~/GluAUC~120~) (E and F) across the categories of FPG and 2-h PG in nonobese (BMI \<27 kg/m^2^; ▲) and obese subjects (BMI ≥27 kg/m^2^; ■). Bars display insulin sensitivity or insulin release relative to the reference category (FPG \<5.0 mmol/l, 2-h PG \<5.0 mmol/l). Calculations were based on geometric means, adjusted for age and BMI, with the general linear model (performed separately in subgroups of nonobese and obese subjects). Cutoff values for different categories of FPG and 2-h PG are as explained in [Fig. 1](#F1){ref-type="fig"}. Ref., reference.](zdb0050957170003){#F3} Insulin sensitivity and insulin release in categories of glucose tolerance. --------------------------------------------------------------------------- Age- and BMI-adjusted peripheral insulin sensitivity (Matsuda ISI) was significantly decreased by −26% in isolated IFG, by −30% in isolated IGT, by −42% in IFG + IGT, and by −46% in newly diagnosed type 2 diabetes, compared with NGT ([Fig. 4](#F4){ref-type="fig"}A). Matsuda ISI was significantly lower in individuals with isolated IGT than in individuals with isolated IFG (P = 0.0016). A significantly greater decrease in isolated IFG than in isolated IGT (−31 vs. −16%, P = 0.0028) was found when insulin sensitivity was assessed with 1/HOMA-IR. 1/HOMA-IR was reduced by −39% in the IFG + IGT group and by −45% in newly diagnosed type 2 diabetic subjects ([Fig. 4](#F4){ref-type="fig"}B). Categories of glucose tolerance status differed significantly also with respect to other indexes of insulin sensitivity ([Table 4](#T4){ref-type="table"}). Compared with NGT, the age- and BMI-adjusted early-phase insulin release (InsAUC~30~/GluAUC~30~) was significantly decreased by −8% in isolated IFG, not changed in isolated IGT, decreased by −16% in IFG + IGT, and decreased by −43% in newly diagnosed type 2 diabetes ([Fig. 4](#F4){ref-type="fig"}C). The difference between isolated IFG and isolated IGT was not statistically significant (−8 vs. 0%, P = 1.0). The total insulin release (InsAUC~120~/GluAUC~120~) was significantly reduced in isolated IFG (−6%) and in newly diagnosed type 2 diabetes (−29%), whereas no significant changes were observed in isolated IGT or in IFG + IGT compared with NGT ([Fig. 4](#F4){ref-type="fig"}D). Individuals with isolated IFG had significantly lower total insulin release than individuals with isolated IGT (−6 vs. +16%, P = 0.001). The HOMA-β index, indicating basal insulin release, was significantly reduced by −17% in isolated IFG (P = 1.2 × 10^−26^), whereas it was increased by 33% in isolated IGT (P = 1.3 × 10^−15^), compared with NGT. All categories of glucose tolerance status differed significantly with respect to all other indexes of insulin secretion examined, as shown in [Table 4](#T4){ref-type="table"}. ![Insulin sensitivity (Matsuda ISI and 1/HOMA-IR) (A and B), early-phase insulin release (InsAUC~30~/GluAUC~30~) (C), total insulin release (InsAUC~120~/GluAUC~120~) (D), disposition index for early insulin release (DI~30~ = Matsuda ISI × InsAUC~30~/GluAUC~30~) (E), and disposition index for total insulin release (DI~120~ = Matsuda ISI × InsAUC~120~/GluAUC~120~) (F) in different categories of glucose tolerance. Bars show percentage of each index relative to NGT (reference, 100%). Calculations were based on geometric means, adjusted for age and BMI (ANCOVA). All pairwise comparisons were statistically significant (P \< 0.05, Bonferroni post hoc test) except for those marked: \*P \> 0.05 vs. NGT, †P \> 0.05 vs. isolated IFG, ‡P \> 0.05 vs. isolated IGT. Ref., reference. IIFG, isolated IFG; IIGT, isolated IGT; NewDM, newly diagnosed diabetes.](zdb0050957170004){#F4} ###### Surrogate indexes of insulin sensitivity and β-cell function across categories of glucose tolerance n NGT Isolated IFG Isolated IGT IFG + IGT Newly diagnosed diabetes P -------------------------------------------- ------- ------------------- ------------------------------------------------ ------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------ ---------------------------------------------------- ---------------- n --- 2,154 2,831 213 694 467 --- β-Cell function HOMA-β (mU/l, mmol/l) 6,359 60.0 (44.2--87.3) 56.4 (40.0--84.3) 78.2 (51.0--128.7) 81.7 (51.0--115.3)[\](#TF4-1){ref-type="table-fn"} 71.5 (43.3--109.4)[†](#TF4-2){ref-type="table-fn"} 9 × 10^−60^ Insulinogenic index (pmol/mmol) 6,312 100 (62--166) 96 (58--158) 75 (44--137) 84 (51--133)[‡](#TF4-3){ref-type="table-fn"} 54 (30--84) 9 × 10^−129^ InsAUC~30~/GluAUC~30~ (pmol/mmol) 6,325 24.1 (16.8--37.0) 24.8 (16.8--37.0) 24.9 (15.7--45.9)[\](#TF4-1){ref-type="table-fn"}[†](#TF4-2){ref-type="table-fn"} 26.2 (17.7--40.3)[‡](#TF4-3){ref-type="table-fn"} 21.0 (12.4--31.8) 1 × 10^−65^ InsAUC~120~ (nmol/l · min) 6,321 26.1 (17.5--40.2) 30.1 (20.7--45.4) 40.6 (24.9--68.8) 47.3 (30.6--70.1)[‡](#TF4-3){ref-type="table-fn"} 41.2 (25.9--63.7)[†](#TF4-2){ref-type="table-fn"} 9 × 10^−47^ ΔInsAUC~120~/ΔGluAUC~120~ (pmol/mmol) 6,311 143 (88--239) 153 (94--246)[\](#TF4-1){ref-type="table-fn"} 99 (59--159) 114 (73--172)[‡](#TF4-3){ref-type="table-fn"} 75 (43--124) 2 × 10^−149^ InsAUC~120~/GluAUC~120~ (pmol/mmol) 6,320 33.2 (23.2--50.3) 34.4 (24.0--50.1) 42.1 (25.5--67.8)[\](#TF4-1){ref-type="table-fn"} 42.5 (28.4--64.2)[\](#TF4-1){ref-type="table-fn"}[‡](#TF4-3){ref-type="table-fn"} 32.3 (20.0--50.8) 2 × 10^−39^ First-phase Stumvoll (pmol/l, mmol/l) 6,325 859 (640--1,185) 818 (555--1,173) 786 (560--1,342)[\](#TF4-1){ref-type="table-fn"}[†](#TF4-2){ref-type="table-fn"} 833 (546--1,283) 612 (270--1,008) 2 × 10^−88^ Second-phase Stumvoll (pmol/l, mmol/l) 6,325 234 (187--310) 232 (174--315) 221 (172--357)[\](#TF4-1){ref-type="table-fn"}[†](#TF4-2){ref-type="table-fn"} 239 (175--348) 198 (118--292) 3 × 10^−84^ Insulin sensitivity/resistance Fasting serum insulin (pmol/l) 6,360 30.6 (22.2--45.0) 41.4 (29.4--63.6) 40.5 (27.0--69.0)[†](#TF4-2){ref-type="table-fn"} 64.2 (40.5--91.8) 73.2 (43.8--116.4) 3 × 10^−122^ HOMA-IR (mU/l, mmol/l) 6,360 1.2 (0.9--1.7) 1.8 (1.3--2.8) 1.6 (1.0--2.7) 2.9 (1.8--4.2) 3.7 (2.2--6.7) 6 × 10^−266^ QUICKI (mU/l, mg/dl) 6,360 0.37 (0.35--0.39) 0.35 (0.33--0.37) 0.36 (0.33--0.38) 0.33 (0.31--0.35) 0.31 (0.29--0.34) 8 × 10^−255^ MCR Stumvoll (pmol/l, mmol/l) 6,360 8.1 (7.6--8.5) 6.9 (6.0--7.4) 7.9 (6.8--8.3) 5.9 (4.7--6.8) 4.4 (2.2--5.7) \<1 × 10^−255^ ISI Stumvoll (pmol/l, mmol/l) 6,353 0.11 (0.10--0.12) 0.10 (0.09--0.11) 0.07 (0.05--0.09) 0.06 (0.04--0.08)[‡](#TF4-3){ref-type="table-fn"} 0.04 (0.01--0.06) \<1 × 10^−255^ Matsuda ISI (mU/l, mg/dl) 6,320 8.2 (5.6--11.9) 5.8 (3.8--8.2) 5.0 (3.1--8.0) 3.2 (2.2--5.0) 2.7 (1.7--4.4) 1 × 10^−255^ Disposition index DI~30~ 6,320 195 (157--245) 139 (112--173) 123 (100--152) 88 (73--107) 55 (44--73) \<1 × 10^−255^ DI~120~ 6,320 266 (225--315) 192 (162--229) 198 (171--231)[†](#TF4-2){ref-type="table-fn"} 145 (124--168) 95 (68--116) \<1 × 10^−255^ Data are medians (25th--75th percentile). Variables with nonnormal right-skewed distribution were logarithmically transformed before analyses. P* values are based on the general linear model adjusted for age and BMI. DI~30~ = Matsuda ISI × InsAUC~30~/GluAUC~30~; DI~120~ = Matsuda ISI × InsAUC~120~/GluAUC~120~. Bonferroni post hoc tests: all pairwise comparisons between the categories of glucose tolerance were significant at P \< 0.05, except for: \*P \> 0.05 vs. NGT; †P \> 0.05 vs. isolated IFG; ‡P \> 0.05 vs. isolated IGT. Disposition index. ------------------ The early-phase DI~30~ was lower in isolated IGT than in isolated IFG (−36 vs. −29% compared with NGT, P = 0.0003). In the IFG + IGT group, DI~30~ was decreased by −53%, and in newly diagnosed type 2 diabetic subjects by −68% ([Fig. 4](#F4){ref-type="fig"}E). In contrast, the total DI~120~ was decreased to the same extent in isolated IFG and isolated IGT (both −27%). DI~120~ was reduced significantly in the IFG + IGT group by −44% and in newly diagnosed type 2 diabetic subjects by −62% ([Fig. 4](#F4){ref-type="fig"}F). All statistical analyses were also performed using a cutoff point of 6.1 mmol/l (110 mg/dl) for FPG. All results were essentially similar (supplemental Tables 1 and 2, supplemental Fig. 3). DISCUSSION ========== Our study is the largest population-based study focusing on the pathophysiology of pre-diabetes using validated surrogate markers for insulin sensitivity and insulin release. This allowed us to perform a detailed analysis on the changes in insulin sensitivity and insulin secretion from normal glucose levels to the diabetic range of hyperglycemia. We demonstrated that insulin sensitivity was already decreased substantially (by −17 and −37%, respectively) within the normal range of FPG and 2-h PG. Insulin sensitivity further decreased through the IFG and IGT range and reached a reduction of approximately −65 and −53% in the diabetic range of FPG and 2-h PG. In contrast, early-phase insulin release declined only slightly within the normal range of FPG and 2-h PG (by approximately −5%) and reached a substantial decrease in the diabetic range of FPG (∼50--70%) and 2-h PG (∼35--50%). Compensatory insulin secretion was entirely missing when FPG increased from the normal range to the IFG range, but compensatory insulin secretion was observed within the normal range of 2-h PG. We also showed that peripheral insulin resistance is the predominant feature of isolated IGT, whereas impairment in early and total insulin release characterize isolated IFG. The novel finding of our study was the observation that peripheral insulin sensitivity was already decreased considerably at relatively low glucose levels within the normal range of FPG and 2-h PG. Other studies addressing the same question have been considerably smaller in size. In a study including 188 subjects, Ferrannini et al. ([@B24]) reported a decrease in insulin sensitivity (measured by clamp) with higher 2-h PG levels within the NGT group. The increase in insulin resistance was associated with higher BMI, suggesting that obesity is the main factor responsible for the decrease in insulin sensitivity. We found that insulin sensitivity decreased comparably in both nonobese and obese individuals by −37 and −36%, respectively, within the 2-h PG range from 5.0 to 7.9 mmol/l. Thus, our findings suggest that obesity does not affect insulin sensitivity related to hyperglycemia. In another study (n = 148), Ahrén ([@B27]) reported decreased insulin sensitivity (∼14%) measured by clamp in postmenopausal normoglycemic women within the highest quartile of FPG. The mechanisms leading to a decrease in insulin sensitivity within the normal range of FPG and 2-h PG remain unclear. Although high glucose (20 mmol/l) and/or high insulin levels decrease glucose uptake in human adipose cells in vitro ([@B37]), it is not known whether this effect could be observed at moderately increased glucose concentrations in vivo. Furthermore, our results show that despite a substantial decrease in peripheral insulin sensitivity (−37%), 2-h PG could be maintained within the normal range by a compensatory increase in total insulin release. The pattern of changes in early-phase insulin release differed from that of insulin sensitivity. The decrease was small in the normal range (approximately −5%), progressed within the IFG and IGT range, and was substantial within the diabetic range (reaching −70% and −51% for diabetic FPG and 2-h PG, respectively). The defect in early insulin response manifested at lower glucose levels than the defect in total insulin response. Previous studies ([@B13],[@B25],[@B26],[@B28]) have not assessed the effect of obesity on the relationship between glucose levels and insulin secretion. We showed that both early-phase and total insulin release were substantially higher in obese subjects than in nonobese subjects within the normal range of FPG and 2-h PG (by ∼50--70%), but there was no significant interaction between obesity and glucose levels in determining insulin release. Therefore, obesity does not seem to have a major impact on further changes in insulin release in hyperglycemia. Accordingly, Camastra et al. ([@B38]) have observed that the dynamic aspects of β-cell response to glucose were unaltered in the morbidly obese nondiabetic subjects. Previous studies have been inconsistent with respect to differences in insulin sensitivity between isolated IFG and isolated IGT. Decreased peripheral insulin sensitivity in subjects with isolated IGT compared with subjects with isolated IFG has been reported in some studies using the clamp method or IVGTT ([@B9],[@B10],[@B12],[@B14]), but also similar impairment in insulin action has been found in both isolated IFG and isolated IGT in Pima Indians ([@B13]) and obese Chinese subjects ([@B16]). In two studies, the decrease in insulin sensitivity in isolated IGT compared with isolated IFG was related to obesity ([@B11],[@B17]). In our study, peripheral insulin sensitivity was significantly more reduced in isolated IGT than in isolated IFG (−30 vs. −26%, P = 0.0016). The decrease in insulin sensitivity was quite similar in nonobese and obese subjects (−27 and −31%) with isolated IGT, indicating that the reduction in peripheral insulin sensitivity in isolated IGT was not explained by obesity. Conflicting findings have been published on 1/HOMA-IR as an index of hepatic insulin sensitivity ([@B11],[@B12],[@B14],[@B18][@B19][@B20][@B21][@B22]--[@B23]). In our study, 1/HOMA-IR was more reduced in isolated IFG than in isolated IGT (−31 vs. −16%). However, HOMA-IR also reflects peripheral insulin sensitivity, and therefore reliable results on hepatic insulin sensitivity can be obtained only by using the tracer technique ([@B13],[@B39]). We observed a small reduction in early-phase insulin release in isolated IFG (−8%) but not in isolated IGT, compared with NGT. Furthermore, isolated IFG and isolated IGT differed significantly in basal and total glucose-stimulated insulin release. Basal and total insulin release were reduced in isolated IFG by −17 and −6%, respectively, but increased in isolated IGT by +33% and +16% compared with NGT. These results suggest that individuals with isolated IFG have impairment in both basal and glucose-stimulated (early and total) insulin release, whereas individuals with isolated IGT have increased basal and total insulin release. Some previous studies using IVGTT or clamp to assess insulin secretion have reported impaired insulin release in individuals with isolated IFG ([@B9],[@B10],[@B13],[@B14],[@B16]), whereas others have found impaired insulin release in individuals with IGT ([@B11],[@B15]). Studies based on OGTT measurements have reported impaired early-phase insulin secretion (insulinogenic index) in individuals with IGT only ([@B21],[@B23]) or in both individuals with IGT or IFG ([@B12],[@B18],[@B22]). Few studies have found decreased basal insulin secretion in IFG ([@B11],[@B22]). The currently accepted paradigm is that type 2 diabetes develops when the pancreas is unable to secrete more insulin to compensate for insulin resistance. However, in the nondiabetic range of glucose levels, compensatory total insulin release was observed only for increasing 2-h PG levels ([Fig. 1](#F1){ref-type="fig"}F). In contrast, in the nondiabetic range of FPG levels, this compensatory hyperinsulinemia was entirely missing, and insulin release (early-phase and total) ([Fig. 1](#F1){ref-type="fig"}C and E) linearly decreased with increasing FPG levels and was lowest in the diabetic range, indicating a major defect in insulin secretion. This observation is consistent with recent evidence from genetic studies showing that the insulin secretion defect is very often the primary cause for type 2 diabetes ([@B40]). The changes in insulin sensitivity and insulin secretion in response to higher glucose levels in our study suggest that there are two major pathways for the development of type 2 diabetes, one leading to diabetes via the elevation of FPG and another via postprandial hyperglycemia or 2-h PG. Prospective follow-up studies including subjects with isolated IFG and isolated IGT are needed, however, to confirm this hypothesis. The strengths of this study are large size, homogeneous study population, and validation of the methods used as surrogate markers for insulin sensitivity and secretion. The limitation of the study is that only middle-aged Finnish men were included in the study, and therefore we do not know if the results are valid for women or for other ethnic and racial groups. Previous studies have not reported sex differences in insulin secretion, but in some studies women were more insulin sensitive than men ([@B41][@B42]--[@B43]). Because of the large size of our cohort, we could not use the most accurate methods to evaluate insulin sensitivity (clamp) and insulin secretion (IVGTT or hyperglycemic clamp) or hepatic insulin sensitivity (tracer techniques). However, we validated our OGTT-derived indexes of insulin secretion and insulin sensitivity against the gold standard measures in an independent sample of Finnish subjects. We did not measure 1-h glucose during the OGTT, which has been shown to be a better predictor of future risk of diabetes than FPG and 2-h PG levels in a recent study ([@B44]). In summary, we showed in a large cohort of Finnish men that the impairment of peripheral insulin sensitivity started at relatively low PG levels within the normoglycemic range. In contrast, the impairment of insulin secretion progressed substantially only in the diabetic range of fasting and 2-h glucose levels. Peripheral insulin resistance was a predominant feature of isolated IGT, whereas impaired insulin secretion characterized isolated IFG. Supplementary Material ====================== ###### Online-Only Appendix The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This study was supported by a grant from the Academy of Finland (124243), the Finnish Heart Foundation, the Finnish Funding Agency for Technology and Innovation (TEKES) (1510/31/06), and the Commission of the European Community (LSHM-CT-2004-512013 EUGENE2). No potential conflicts of interest relevant to this article were reported. [^1]: A.S. and M.J. contributed equally to the work.	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== During prepuberty four types of Sertoli cells develop under hormonal control \[[@CR1]\]: fetal Sertoli cells (Sf), which are observed in the first months after birth; the Sa and Sb types, which are observed during childhood; and the Sc type, which is observed in puberty and throughout life. In the first three months after birth, activation of the hypothalamic-pituitary-gonadal (HPG) axis leads to a transient increase in gonadotropins and testosterone \[[@CR2]--[@CR4]\], termed mini-puberty. During mini-puberty, the number of Sertoli cells increases \[[@CR5], [@CR6]\]. Within the seminiferous tubules, Sertoli cells possess receptors for testosterone and follicle stimulating hormone (FSH). Gonadotropin releasing hormone (GnRH) drives testosterone and FSH signaling. In Sertoli cells, this signaling promotes the production of factors that support germ cell function and differentiation. Sertoli cells secrete anti-Müllerian hormone (AMH), activins, inhibins, glial cell-derived neurotrophic factor (GDNF), KIT ligand (KITLG), and FAS ligand (FASLG). Prepubertal Sertoli cells express subunits of the active inhibin B \[[@CR7]\], which counteracts the effect of activin A by suppressing FSH secretion in gonadotropes \[[@CR8], [@CR9]\]. The testosterone-dependent transition of gonocytes into A dark (Ad) spermatogonia during mini-puberty is an essential process for fertility \[[@CR10], [@CR11]\]. Prepubertal human testes with defective mini-puberty lacked the transition of gonocytes into Ad spermatogonia (Ad-) and showed a diminished number of Sf and Sb Sertoli cells \[[@CR12], [@CR13]\]. Those findings were consistent with recent observations of significantly reduced plasma levels of testosterone, AMH, and inhibin B in boys with cryptorchidism \[[@CR14]\]. The early postnatal increase in inhibin B during mini-puberty \[[@CR15]\] is presumably due to the activation of the HPG axis \[[@CR16]\], and it reflects the proliferation of Sertoli cells. Of interest, the early postnatal rise in inhibin B was correlated more closely with luteinizing hormone (LH) and testosterone levels than with FSH levels \[[@CR17], [@CR18]\]. That finding raised the possibility that Sertoli cell proliferation in neonatal life might depend more on LH/testosterone than on FSH. Studies in the literature have reported equivocal findings of normal, decreased, and increased inhibin B secretion in boys with cryptorchidism \[[@CR9], [@CR19]--[@CR24]\]. Cortes and colleagues found a positive correlation between inhibin B and Ad spermatogonia by analyzing 2-μm thick sections of paraffin embedded specimens \[[@CR25]\]. However, a more precise analysis method that employed 1-μm semi-thin sections of Epon embedded specimens could not confirm the earlier report of a correlation between the number of Ad spermatogonia and plasma inhibin B levels \[[@CR26]\]. Treatment with the gonadotropin-releasing hormone agonist (GnRHa), Buserelin, induced an increase in LH and testosterone \[[@CR27]--[@CR29]\] and rescued fertility in the majority of pathological cryptorchid testes. Buserelin induced expression of genes in the HPG axis \[[@CR30]\] and genes involved in gonocyte transition to Ad spermatogonia \[[@CR31]\]. Ultrastructural changes within Sertoli cells after GnRHa treatment included the presence of abundant lipid droplets, changes in nuclear form, and increased amounts of rough endoplasmic reticulum in the cytoplasm \[[@CR27]\]. In a randomized double-blind, placebo-controlled study cryptorchid boys were treated with a low dose (20 micrograms) of GnRHa, given as a nasal spray, for a short period of 28 days. Boys treated with Buserelin had the highest number and the best maturation index of germ cells \[[@CR32]\]. Furthermore, in 2007 it has been shown that patients treated with Buserelin and with a normal Leydig cell testosterone secretory capacity will have normal testicular histology and Ad spermatogonia. Those with a suboptimal Leydig cell testosterone secretory capacity will have a low Ad spermatogonia count and consequently poor prognosis for future fertility, despite successful surgery \[[@CR29]\]. In this study, we analyzed the impact of low (hypogonadotropic) LH on differential expression of specific Sertoli cell genes between Ad- testes with impaired mini-puberty and Ad+ testes that completed mini-puberty. Furthermore, we studied the effect of GnRHa on Sertoli cells by analyzing differential Sertoli cell specific gene expression after GnRHa treatment in Ad- testes. Our findings extended our understanding of how Sertoli cells and their secreted factors, including inhibins, are involved in the gonocyte transition to Ad spermatogonia. Methods {#Sec2} ======= Study population and biopsy sample collection {#Sec3} --------------------------------------------- We selected 15 patients with isolated cryptorchidism, based on histological results, and divided them into 2 groups. Seven belonged to the Ad− (lacking Ad spermatogonia) and 8 to the Ad+ (presenting Ad spermatogonia) group. The patients had a median age of 18.5 months (range 8--59 months). Data from Ad− bilateral cryptorchid boys treated with GnRHa (Buserelin) following the first orchidopexy (surgery) (4 patients) were retrieved from an ongoing randomized study. Initial biopsies revealed no Ad spermatogonia, indicating defective mini-puberty (Ad− group). The second testis was managed by orchidopexy and biopsied 6 months after the initial surgery. Thus, results from 21 biopsies were compared. Patients were age and ethnicity matched. RNA sequencing data from our two previous studies \[[@CR30], [@CR33]\] were used to analyze manually selected genes expressed specifically in Sertoli cells. A cryptorchid testis is defined as a testis localized outside of the scrotum and incapable of being brought into a stable scrotal position. All undescended testes in this study were located in the inguinal region. Testicular biopsies were taken at the time of orchidopexy. This sample was then subdivided, with one fragment fixed in glutaraldehyde for histological processing, while the other one was immediately immersed in RNAlater (ThermoFisher Scientific, Waltham, Massachusetts, USA) and stored at --25 °C until further processing (for RNA extraction and RNA- sequencing). Histological analyses {#Sec4} --------------------- Biopsies were fixed in 3% glutaraldehyde in phosphate-buffered saline (PBS, pH 7.4) and then embedded in Epon resin. Semi-thin sections (1 μm) were cut using a Reichert Om-U3 ultramicrotome (Reichert AG, Vienna, Austria). Sections were mounted on glass slides, stained with 1% toluidine blue, and examined under a Zeiss Axioskop light microscope (Carl Zeiss Microscopy GmbH, Jena, Germany) with an integrated photo-camera. Biopsies were histologically examined by two of the authors (F.H. and D.D.), each with expertise in the interpretation of semi-thin sections of prepubertal testes. During histological analyses, at least 100 tubular cross sections per biopsy were evaluated, with regard to their number of spermatogonia per tubule (S/T) and presence of Ad spermatogonia. In the prepubertal testes, Ad spermatogonia were identified according to the criteria first published by Seguchi and Hadziselimovic \[[@CR34]\]. This type of germ cell has a typical halo in the nucleus, termed the rarefaction zone, and cytoplasm with a darker aspect in comparison to Ap or fetal spermatogonia. For each biopsy, at least 100 tubular cross sections were evaluated. On the basis of this evaluation, biopsies were categorized into 2 groups, Ad- (high infertility risk, HIR) and Ad+ (low infertility risk, LIR), dependent on their infertility outcome. The Ad- group included biopsies with S/T ≤0.2 and no Ad and the Ad+ group recorded S/T scores of \>0.6 with Ad. Cryptorchid boys in the Ad- group had 8 times lower plasma LH (0.11 IU/L) compared to the Ad+ group (0.89 IU/L, p\<0.009) indicating hypogonadotropic hypogonadism \[[@CR26]\]. RNA preparation, sequencing, data analyses, and RNA expression levels {#Sec5} --------------------------------------------------------------------- The workflow from RNA isolation, through to purification, library preparation, sequencing, data analyses, and expression level analysis, has been previously described in detail \[[@CR30], [@CR33]\]. Data and differential gene expression analyses {#Sec6} ---------------------------------------------- Determination of differentially expressed genes, statistical analyses and model design were described previously \[[@CR30], [@CR33]\]. Only genes with at least one read per million, in at least two samples, were included. P values and fold-changes were calculated for the treatment factor and differentially expressed genes were defined as those displaying a false discovery rate (FDR) of less than 0.05. Raw data files are available at the Database of Genotypes and Phenotypes (dbGaP) with the accession number phs001275.v1.p1. Results {#Sec7} ======= We analyzed the expression data of 40 manually selected Sertoli cell expressing genes \[[@CR35]\], whose expression or protein products are commonly used to identify or differentiate Sertoli cells from other testicular cells at various developmental stages. The results were described as fold change (FC) of gene expression, log 2-fold-change (LogFC) and false discovery rate (FDR) between the tested groups (Ad- vs. Ad+ and GnRHa treated vs. untreated) in Table [1](#Tab1){ref-type="table"}. A positive LogFC indicated a gene upregulation and a negative LogFC a gene downregulation in Ad- testes relative to Ad+ testes or in GnRHa treated versus untreated testes (Table [1](#Tab1){ref-type="table"}).Table 1Differential expression of Sertoli cell specific genes in Ad- versus Ad+ group (Ad-/Ad+) and in the GnRHa treated versus untreated group (GnRHa)Gene IDNameFC Ad-/Ad+logFC Ad-/Ad+FDR Ad-/Ad+FC GnRHalogFC GnRHaFDR GnRHaAMHAnti-Mullerian Hormone-n.s.n.s.-n.s.n.s.ARAndrogen Receptor-n.s.n.s.1.5-0.58210.0126BMP4Bone Morphogenetic Protein 4-n.s.n.s.-n.s.n.s.BMP6Bone Morphogenetic Protein 61.7-0.76120.0030-n.s.n.s.CALB2Calbindin 2-n.s.n.s.2.2-1.15560.0034CDKN1BCyclin-Dependent Kinase inhibitor 1B (p27, Kip1)-n.s.n.s.1.7-0.78450.0016CLDN11Claudin 11-n.s.n.s.1.8-0.83150.0011CLUClusterin-n.s.n.s.1.9-0.95480.0006CREB1cAMP Responsive Element Binding protein 11.10.20000.04891.6-0.68880.0041CTSLCathepsin L-n.s.n.s.1.7-0.74920.0020DESDesmin2.8-1.48820.0014-n.s.n.s.DHHDesert Hedgehog-n.s.n.s.-n.s.n.s.DMRT1Doublesex and Mab-3 Related Transcription factor 1-n.s.n.s.1.7-0.78380.0010FASFas cell surface death receptor-n.s.n.s.-n.s.n.s.FASLGFas Ligand (TNF superfamily, member 6)-n.d.n.d.5.02.31540.0031FSHRFollicle stimulating hormone receptor1.40.49150.0363-n.s.n.s.FGF2Fibroblast Growth Factor 2-n.s.n.s.-n.s.n.s.FGF9Fibroblast Growth Factor 92.1-1.06050.0016-n.s.n.s.GATA1GATA binding protein 1 (globin transcription factor 1)-n.d.n.d.-n.d.n.d.GATA4GATA binding protein 4-n.s.n.s.-n.s.n.s.GJA1/CX43Gap Junction protein, Alpha 1/Connexin 431.30.40180.03241.9-0.95160.0002GDNFGlial cell Derived Neurotrophic Factor-n.s.n.s.2.81.46870.0036INHAInhibin, alpha-n.s.n.s.1.8-0.81580.0023INHBAInhibin, Beta A-n.s.n.s.1.6-0.63630.0124INHBBInhibin, Beta B-n.s.n.s.1.6-0.66160.0058KATNAL1Katanin p60 subunit A-like 1-n.s.n.s.1.7-0.77820.0012KITLGKIT Ligand-n.s.n.s.2.1-1.09950.0001KRT18Keratin 18, type I-n.s.n.s.1.5-0.57300.0176NR5A1/SF1Nuclear Receptor subfamily 5, group A, member 1-n.s.n.s.-n.s.n.s.PTGDSProstaglandin D2 Synthase-n.s.n.s.-n.s.n.s.SERPINA5Serpin Family A Member 5-n.s.n.s.1.7-0.75110.0076SOX8SRY (sex determining region Y)-box 82.8-1.46410.0003-n.s.n.s.SOX9SRY (sex determining region Y)-box 91.40.49830.0300-n.s.n.s.SRYSex determining Region Y-n.s.n.s.-n.s.n.s.TFTransferrin-n.s.n.s.-n.s.n.s.TJP1/ZO1Tight Junction Protein 1/Zona Occludens 1-n.s.n.s.1.6-0.68240.0102VIMVimentin-n.s.n.s.1.5-0.59010.0352WT1Wilms tumor 1-n.s.n.s.1.5-0.56540.0180ZFPM1/FOG1Zinc Finger Protein, FOG family member 2-n.s.n.s.-n.s.n.s.ZFPM2/FOG2Zinc Finger Protein, FOG family member 21.30.36950.0240-n.s.n.s.Absolute fold change (FC), Log 2-fold change (logFC), false discovery rate (FDR), not significant (n.s.), not determined (n.d.)Absolute fold changes (FC) \<2 are highlighted in italic Nine of 40 Sertoli cell specific genes were differentially expressed in Ad- testes {#Sec8} ---------------------------------------------------------------------------------- We found four genes that showed lower expression in Ad- testes compared to Ad+ testes: BMP6 with 1.7 fold-, DES with 2.8 fold-, FGF9 with 2.1 fold-, and SOX8 with 2.8 fold decreased expression (Fig. [1](#Fig1){ref-type="fig"} and Table [1](#Tab1){ref-type="table"}). The differential expression values of the genes BMP6 \[[@CR30]\], FGF9 \[[@CR33]\], and SOX8 \[[@CR30]\] between the Ad- and Ad+ group was described by us earlier.Fig. 1Graphical abstract of up- and downregulated genes in Ad- compared to Ad+ testes and after GnRHa treatment Five genes (GJA1, CREB1, FSHR, ZFPM2/FOG2, and SOX9 (described by us earlier \[[@CR30]\])) displayed increased RNA levels in Ad- testes compared to Ad+ testes (Fig. [1](#Fig1){ref-type="fig"} and Table [1](#Tab1){ref-type="table"}). We noted that 29 Sertoli cell specific genes were not significantly differentially expressed. Moreover, the expression levels of FASLG and GATA1 were not detectably different between the two groups. GnRHa affects Sertoli cell specific genes {#Sec9} ----------------------------------------- We found that GnRHa treatment had mostly a negative effect (48%) by reducing expression profiles of Sertoli cell specific genes. GnRHa treatment downregulated 19 Sertoli cell specific genes (AR, CLU, INHA, INHBA, INHBB, KRT18, VIM, KATNAL1, GJA1, TJP1, CLDN11, SERPINA5, CDKN1B, CALB2, CTSL, CREB1, DMRT1 \[[@CR31]\], WT1, and KITLG \[[@CR31]\]) (Fig. [1](#Fig1){ref-type="fig"} and Table [1](#Tab1){ref-type="table"}). In an earlier study, we reported that GnRHa treatment had a negative effect on the gene expression of secreted Sertoli cell factor, KITLG, and the transcription factor, DMRT1, which are essential during germ cell development and differentiation \[[@CR31]\]. Interestingly, only two genes (CREB1 and GJA1) were differentially expressed between Ad- and Ad+ testes. We found that GnRHa treatment highly upregulated the gene expression of FASLG (fivefold) and GDNF (2.8 fold). The strongly upregulated RNA expression of GDNF after GnRHa treatment was described by us earlier \[[@CR31]\]. Eighteen genes did not show significant differential RNA expressions, and GATA1 expression was not detected. Discussion {#Sec10} ========== Positive and negative regulators mediate LH-dependent Sertoli cell development {#Sec11} ------------------------------------------------------------------------------ Contrary to our expectation, only nine out of 40 Sertoli cell specific genes were differentially expressed in LH-deficient Ad- testes compared to Ad+ testes (Table [1](#Tab1){ref-type="table"}). Inhibin B, a dimer of α and β subunits (INHA and INHBB), is mainly produced by Sertoli cells; it negatively regulates the release of pituitary FSH and antagonizes the stimulating action of activin A \[[@CR36]--[@CR39]\]. Of importance, no differential expression was observed in the single inhibin subunits, INHA, inhibin βA (INHBA), and INHBB, between the two studied groups. This finding supported the observation that plasma values for inhibin B were similar between Ad- and Ad+ testes \[[@CR26]\]; moreover, this finding brings into question the hypothesis that inhibin B is essential for the transformation of gonocytes into Ad spermatogonia \[[@CR25]\]. Alternatively, we propose that, during mini-puberty, LH and testosterone, but not inhibin B, drive the differentiation of gonocytes into Ad spermatogonia. Bone morphogenetic protein 6 (BMP6) was reported to inhibit apoptosis and to influence the production of tight junction protein 1 (TJP1), GDNF, and KITLG in Sertoli cells \[[@CR40]\]. Although we observed a downregulation of BMP6, the expression levels of TJP1, KITLG, and GDNF were not significantly (FDR\>0.05) altered between the two groups. Fibroblast growth factor 9 (FGF9) is a downstream effector of GATA4, ZFPM2, SRY, and SOX9 signaling pathway during male sex determination. FGF9 regulates Sertoli cell differentiation through FGFR2 signaling \[[@CR41], [@CR42]\]. Both transcription factors, SOX8 and SOX9, can induce AMH expression, and it was suggested that these transcription factors might be redundant in testis differentiation \[[@CR43], [@CR44]\]. In Ad- testes, FGF9 and SOX8 expression levels were reduced more than two-fold, but ZFPM2 and SOX9 were slightly increased (1.3 fold and 1.4 fold, respectively) (Table [1](#Tab1){ref-type="table"}, \[[@CR30]\]); this result suggested that function of Sertoli cells was impaired in Ad- testes. The cAMP response element binding protein 1 (CREB1) is a member of the family of transcription factors that are responsive to hormones, which are critical in nearly all mammalian cell types for development and differentiation. FSH \[[@CR45]\] and testosterone (non-classical) activate signaling pathways \[[@CR46]\] that lead to phosphorylation/activation of CREB1 in the nucleus of Sertoli cells. Similarly, GDNF-signaling leads to phosphorylation of CREB1 in germ cells \[[@CR47]\]. We observed 1.1 fold increased CREB1 RNA expression in Ad- testes; however, we did not evaluate the activation state of CREB1 protein. Alterations in gap junction protein alpha 1 (GJA1, also known as Connexin 43) expression were previously associated with different forms of spermatogenic impairment in men \[[@CR48]--[@CR54]\]. Furthermore, GJA1-based gap junctions have been reported to form a transverse and longitudinal intercellular communication network within seminiferous tubules, which ensures the synchronization of germ cell proliferation and differentiation \[[@CR55]\]. However, the effect of an absolute fold change of 1.3 in GJA1 expression in Ad- testes remains unexplained. GnRHa-treatment suppresses gene expression in Sertoli cells {#Sec12} ----------------------------------------------------------- GnRHa treatment caused downregulation of the cytoskeleton-related genes, KRT18, VIM, and KATNAL1, and the cell junction protein encoding genes, GJA1, TJP1, and CLDN11. This downregulation pointed to a structural remodeling process, both within Sertoli cells and in cell-cell connections between individual Sertoli cells and between Sertoli cells and germ cells. Clusterin (CLU) was originally identified as an androgen/testosterone-repressed gene in the prostate \[[@CR56]\] and reduced CLU protein levels in testis have been linked to male infertility \[[@CR57], [@CR58]\]. In a recent report, it was demonstrated that CLU conferred protection against apoptosis and appeared to induce meiosis in male germ cells via VLDLR/LPR8 receptors in rats \[[@CR59]\]. GnRHa treatment suppressed the expression of both these receptors (VLDLR logFC^GnRHa^ -0.74 and LPR8 logFC^GnRHa^ -0.76). We suggest that the observed reduced levels of androgen-sensitive CLU expression resulted from GnRHa-induced increases in testosterone, and that this activity inhibited germ cells from entering premature meiosis. Furthermore, we observed reductions in RNA expression of CDKN1B, CALB2, CTSL, CREB1, DMRT1 \[[@CR31]\], and WT1. All these genes are involved in Sertoli cell function and differentiation. This finding suggested that GnRHa induced transcriptional changes in multiple target genes. Cells that express AR in human testis include Sertoli cells, Leydig cells, peritubular myoid cells, arteriole smooth muscle cells, and vascular endothelial cells. Immunostaining methods for detecting AR protein showed rather poor expression in human prepubertal Sertoli cells, but AR expression increased after the age of four years \[[@CR60]--[@CR63]\]. Furthermore, AR expression is modulated by testosterone and FSH \[[@CR64], [@CR65]\]. Therefore, the observed reduction in AR expression could be due to a GnRHa-dependent increase in testosterone in treated testes. Analyzing ARKO mice, Yue and coworkers compared expression levels of Ddx4, Amh, c-Kit, Mmp14/Mt1-mmp, Rhox5, Zbtb16, and Pou5f1 genes. They concluded that gonocyte transformation to A-type spermatogonia was androgen-independent in mouse and human. However, they did not analyze genes specific for Ad spermatogonia, like T/BRACHYURY, TERT, PAX7, DMRTC2, and FGFR3. Furthermore, Li et al. performed immunofluorescence histochemistry to detect the proliferation marker, Ki67, the Sertoli cell marker, AMH, and the germ-cell marker, DDX4. They did not observe any differences in gonocyte migration from the center to the tubular basement membrane, or in spermatogonial stem cell transformation \[[@CR66]\]. Thus, they concluded that gonocyte transition to Ad spermatogonia and gonocyte migration towards the basement membrane were androgen-independent processes in both mouse and human. Although the expression of spermatogonial markers is conserved in species ranging from rodents to primates, species differences have been reported, including the succession of markers and their correlations with the differentiation state of spermatogonia (reviewed \[[@CR67], [@CR68]\]). This species discrepancy was true for DDX4; in human, DDX4 was not localized in gonocytes, but its onset of expression was associated with the formation of spermatogonia \[[@CR69]\]. Notably, there are different types of spermatogonia \[[@CR34]\]. It is of paramount importance to realize that all spermatogonia are DDX4-positive cells, including spermatogonial stem cells (SSCs). In contrast to the findings of Li and co-workers, our results led us to conclude that, in infants with cryptorchidism and defective mini-puberty, testosterone is essential for the gonocyte-to-Ad spermatogonia transition; moreover, we concluded that testosterone had an effect after GnRHa treatment on expression of Sertoli cell specific genes, such as AR. GnRHa treatment also had a negative effect on RNA levels of the inhibin-subunit encoding genes, INHA, INHBA, and INHBB. Biologically active activins and inhibins are dimers composed of different α (encoded by INHA) and β subunits (βA encoded by INHBA and βB encoded by INHBB); the dimers include activin A (βAβA dimer), activin B (βBβB dimer), activin AB (βAβB dimer), inhibin A (αβA dimer), and inhibin B (αβB dimer). In the prepubertal testis, both INHA and INHBB subunits are expressed in Sertoli cells \[[@CR7]\]. However, in pubertal and adult testis, INHBB is no longer expressed in Sertoli cells, but it continues to be expressed in germ cells and Leydig cells \[[@CR7], [@CR70], [@CR71]\]. Serum inhibin B levels reflect Sertoli cell function \[[@CR8], [@CR37]\]. Serum inhibin B levels were reported to be normal in boys with cryptorchidism, including patients in the high infertility risk group (HIR/Ad-) \[[@CR7], [@CR21], [@CR26]\]. During mini-puberty, inhibin B levels are elevated. Earlier studies of gonadotropin treatment in prepubertal boys with cryptorchidism showed that gonadotropin stimulation increased serum inhibin B levels in an age-dependent manner \[[@CR9], [@CR21]\]. In the present study, we did not evaluate inhibin B serum levels after GnRHa treatment, but the differential gene expression data of the single subunits (INHA, INHBA, and INHBB) pointed to gonadotropin-dependent transcriptional repression of both inhibin B and activin A. Importantly, upon treatment, although the expression of the protein encoding gene INHBA was repressed, the respective long non-coding RNA INHBA-AS1 (logFC^GnRHa^ +2.02; \[[@CR30]\]) was upregulated. Both INHBA and INHBA-AS1 were previously associated with dental caries in genome-wide association studies \[[@CR72]\], and INHBA was postulated to play a role in early tooth development \[[@CR73]\]. Our results suggested that INHBA-AS1 could be involved in the development of Sertoli cells. It was previously shown that activin A was involved, together with the germ-cell secreted protein, NODAL, in the upregulation of NANOS2 during germ cell differentiation. Both activin A and NODAL bind activin A receptor type 2 (ACVR2A/2B) and type I receptors (ACVR1B/1C), but NODAL requires concomitant binding of the transmembrane co-receptor, CFC1/CRIPTO1, for activation. After GnRHa treatment, ACVR2A was found to be downregulated (logFC^GnRHa^ -0.60), and ACVR1C was upregulated (logFC^GnRHa^ +0.85). The overall negative effect of GnRHa, except for its effect on ACVR1C, suggested that the treatment-induced Ad spermatogonial development in infants was independent of the activation of inhibin B and the activin A pathway in Sertoli cells \[[@CR26]\]. This finding was inconsistent with the hypothesis that inhibin B might be important for gonocyte transformation into Ad spermatogonia \[[@CR25]\]. FASLG, GDNF, and other genes with positive responses to GnRHa {#Sec13} ------------------------------------------------------------- FASLG expression was reported to be regulated by gonadotropin in human testis \[[@CR74]\], and it was testosterone-dependent in rhesus monkeys \[[@CR75]\]. With GnRHa treatment, we observed an increase in FASLG expression. This change could point to elevated apoptosis of Sertoli and germ cells; furthermore, it supported the notion that FASLG expression was dependent on testosterone and gonadotropin. Additionally, previous studies identified FASLG as target of matrix metallopeptidase 7 (MMP7), which proteolytically processes FASLG into a soluble form \[[@CR76]\]. Interestingly, we also observed increases in MMP7 RNA levels upon GnRHa treatment (logFC^GnRHa^ +2.52, FDR 0.0016), which suggested that soluble FASLG might be involved in GnRHa-induced germ cell development. The influence of GDNF on the transition of gonocytes to spermatogonia in Ad- testes and the GnRHa dependence of GDNF expression was recently described \[[@CR31]\]. GDNF was also reported to stimulate the proliferation of cultured, immature mouse Sertoli cells via its receptor subunit, NCAM, and the ERK1/2 signaling pathway \[[@CR77]\]. Moreover, gonadotropin-dependent FSH stimulation was shown to induce increases in GDNF levels \[[@CR78]\]; that interaction supported our findings that GDNF expression increased with GnRHa treatment. Furthermore, GDNF produced by peritubular myoid cells was reported to be testosterone-regulated and to be essential for repopulating the undifferentiated spermatogonial pool \[[@CR79]\]. Therefore, a testosterone-dependent GDNF expression in Sertoli cells was likely, but a contribution of peritubular myoid GDNF RNA signaling to our data was also likely and has to be considered. Sex hormone-binding globulin (SHBG/ABP), also known as androgen binding protein, is produced by Sertoli cells in most mammals. SHBG controls the availability of androgens in seminiferous tubules and in the epididymis (reviewed \[[@CR80]\]). Transgenic mice that overexpressed rat SHBG developed progressive impairment of spermatogenesis \[[@CR81]\]. Studies on human testis-specific SHBG revealed alternative splicing in exon 1. The Sertoli cell-specific SHBG transcripts included an exon 1 that lacked the secretion signal, and SHBG expression was repressed; the SHBG isoform with the other exon 1 was expressed in germ cells \[[@CR82], [@CR83]\]. Based on those findings, we excluded SHBG as a Sertoli cell specific gene in the present study. Nonetheless, we would like to point out that SHBG expression was found four fold increased after GnRHa treatment (logFC^GnRHa^ +2.00, FDR 0.0037). This finding supported the notion that SHBG expression was gonadotropin-dependent in the testis, which suggested that SHBG might be involved in the gonocyte transition to Ad spermatogonia. Limitations and outlook {#Sec14} ----------------------- To get an indication of differential expression of specific Sertoli cell expressing genes we selected for this study testicular RNA sequencing data of 40 Sertoli cell expressing genes whose expression profiles and protein products are commonly used as Sertoli cell specifiers. However, a contribution of RNA signals of other testicular cells in this study could not be excluded. Furthermore, changes at the mRNA level do not necessarily mean changes at the protein level. Nonetheless, differential RNA levels as determined by RNA profiling are a good indication that a given gene is implicated in a certain process. Therefore, it is convincing that, despite the above mentioned limitations, the observed differential expressions reflect the pathological state in Ad- testes and is the consequence of GnRHa treatment. In future, protein and gene expression of single cells should be tackled. Conclusion {#Sec15} ========== This study showed that in Ad- testes with lower LH levels the expression of Sertoli cell specific genes, involved in cell morphology through complex gene networks, was affected positively and negatively. Based on these results, it is indicated that reduced LH and concomitantly reduced testosterone levels caused impaired Sertoli cell function and increased apoptosis. GnRHa treatment mostly resulted in suppression of Sertoli cell specific genes. Based on our observations of differential gene expression of intermediate filaments and cell-cell junction proteins, we suggested that GnRHa triggered a restructuring process in Sertoli cells. Furthermore, we proposed that GnRHa induced GDNF and FASLG expression, which supported Sertoli cell proliferation and the gonocyte transition to Ad spermatogonia. Finally, our results indicated that, during mini-puberty, LH and testosterone, but not inhibin B, were critical factors in driving the gonocyte transition to spermatogonia. ACVR : Activin A receptor AR : Androgen receptor AMH : Anti-Mullerian hormone BMP6 : Bone morphogenetic protein 6 CALB2 : Calbindin 2 CDKN1B : Cyclin-dependent kinase inhibitor 1B CLDN11 : Claudin 11 CLU : Clusterin CREB1 : cAMP response element binding protein 1 CTSL1 : Cathepsin L dbGaP : Database of Genotypes and Phenotypes DES : Desmin DMRT1 : Doublesex and mab-3 related transcription factor 1 FDR : False Discovery Rate FASLG : FAS ligand FGF9 : Fibroblast growth factor 9 FGFR2 : Fibroblast growth factor receptor 2 FSH : Follicle stimulating hormone FSHR : Follicle stimulating hormone receptor GATA4 : GATA Binding Protein 4 GDNF : Glial cell derived neurotrophic factor GJA1 : Gap junction protein alpha 1 GnRHa : Gonadotropin-releasing hormone agonist HIR : High infertility risk HPG : Hypothalamic-pituitary-gonadal INHA : Inhibin alpha INHBA : Inhibin beta A INHBA-AS1 : Inhibin beta A antisense RNA 1 INHBB : Inhibin beta B KATNAL1 : Katanin p60 subunit A-like 1 KITLG : KIT Ligand KRT18 : Keratin 18 LIR : Low infertility risk LH : Luteinizing hormone LogFC : Absolute fold change MMP7 : Matrix metallopeptidase 7 RHOX5 : Reproductive homeobox 5 SERPINA5 : Serpin peptidase inhibitor family A member 5 SOX8 : SRY-Box 8 SOX9 : SRY-Box 9 SSC : Spermatogonial stem cell TJP1 : Tight junction protein 1 VIM : Vimentin WT1 : Wilms tumor 1 ZFPM2 : Zinc finger protein, FOG family member 2 This study was supported, in part, by the European Social Fund under the Global Grant measure. Work in the Stadler group was supported by funding from the MetastasiX project of SystemsX.ch. We thank Manuel Kohler, Department of Biosystems Science and Engineering (D-BSSE) ETH Zurich, for technical assistance with RNA sequencing. Funding {#FPar1} ======= The funding sponsors played no role in the design of the study; in the collection, analyses, or interpretation of data; in writing the manuscript; or in the decision to publish the results. KG-S analyzed and interpreted the data and contributed to writing the paper. FH conceived and designed the research, performed experiments, analyzed and interpreted the data, and contributed to writing the paper. GV conceived and designed the research, performed experiments, and analyzed the data. PD produced RNA-Sequencing data. VB performed experiments. DD performed experiments and analyzed the data. MBS analyzed and interpreted the data and contributed analysis tools. All authors read and approved the final manuscript. Ethics approval and consent to participate {#FPar2} ========================================== Investigations were carried out in accordance with the Declaration of Helsinki of 1975, revised in 2008. All aspects of this study were approved by the Institutional Review Board and the Independent Ethics Committee of Vilnius University. Approval was also provided for research involving the use of material (data records or biopsy specimens) that had been collected for non-research purposes (Vilnius Regional Biomedical Research Ethics Committee, No. 158200-580-PPI-17, 11 June 2013). Competing interests {#FPar3} =================== The authors declare that they have no competing interests. Publisher's Note {#FPar4} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.	{ "pile_set_name": "PubMed Central" }
慢性髓性白血病（CML）是一种起源于造血干细胞的血液系统恶性疾病，Ph染色体\[t（9；22）（q34；q11）\]是其特征性细胞遗传学标志，存在于90%\~95%CML患者中[@b1]。额外染色体异常（ACA）是CML病程演进中出现的额外或多重细胞遗传学改变[@b2]。ACA在初诊CML慢性期患者中的发生率为5%\~10%，在急变期患者中则上升至60%\~80%，往往预示疾病进展[@b1],[@b3]。以伊马替尼为代表的酪氨酸激酶抑制剂（TKI）的广泛应用，使绝大多数CML慢性期患者的生存和生活质量得到改善。TKI治疗过程中，ACA的出现与CML耐药及病程演进密切相关[@b4]。我们通过回顾性分析我院387例CML患者在TKI治疗过程中的细胞遗传学及ABL激酶区点突变资料，旨在分析细胞遗传学变化与CML病程演变的关系。病例与方法 {#s1} ========== 1．病例资料：2007年1月至2014年12月于我院血液科就诊的387例应用TKI治疗的CML患者纳入研究。间接荧光原位杂交（FISH）技术检测BCR-ABL融合基因均为阳性。其中男254例（65.6%），女133例（34.4%），中位年龄36（15\~80）岁。中位随访时间42.5（2.0\~120.5）个月。CML的诊断、分期及疗效评价参照《血液病诊断及疗效标准》[@b5]。 2．染色体核型分析：染色体制备采用骨髓细胞直接法或24 h短期培养法，常规染色体G显带技术分析核型，所有标本分析细胞数至少为20个。核型异常参照《人类细胞遗传学国际命名体制（2013）》的有关规定进行描述。ACA中，常见异常如+8、+Ph\[+der（22）t（9; 22）（q34; q11）\]、等臂17\[ⅰ（17）（q10）\]、+19及ider（22）（q10）t（9; 22）（q34;q11）称为"主要路径"异常[@b3]。其他少见的染色体异常称为"次要路径"异常。 3．FISH法检测BCR-ABL基因：融合基因探针（双色双融合探针）购自美国Abbott公司，BCR探针显绿色荧光信号，ABL探针显红色信号。实验步骤参照产品说明书。在荧光显微镜下，同一视野下采用两种滤片分别观察杂交信号。每例患者标本用计算机成像系统分析间期细胞200个。 4．ABL激酶区点突变检测：采集外周血或骨髓标本，以EDTA抗凝，用淋巴细胞分离液分离单个核细胞，采用TRIzol试剂提取总RNA。逆转录合成体系及PCR试剂购自日本TaKaRa公司，具体步骤参见产品说明书。巢式PCR扩增ABL激酶区第237\~486位氨基酸片段，终产物864 bp，对PCR产物进行10 g/L琼脂糖凝胶电泳分析。阳性结果送至上海英骏生物技术有限公司进行分析，测序结果应用chroma软件分析。 5．疗效分析：总生存（OS）时间指自接受治疗开始至患者死亡或随访终止；无事件生存（EFS）时间指自接受治疗开始，至进展为加速期或急变期，发生血液学或细胞遗传学复发，任何原因导致TKI停药、死亡或随访终止等事件发生日期为止；无病生存（DFS）时间指从获得完全缓解到复发或死亡或末次随访之日。 6．统计学处理：采用SPSS 18.0软件进行统计学分析，两组之间均数比较应用独立样本t检验；率的比较采用χ^2^检验或Fisher确切概率法。生存分析采用Kaplan-Meier法，两组之间比较采用Log-rank检验。P\<0.05为差异有统计学意义。结果 {#s2} ==== 一、初诊CML患者合并ACA情况及其对生存的影响 1．初诊CML患者染色体核型特征：387例初诊CML患者，慢性期352例（90.9%），加速期8例（2.1%），急变期27例（7.0%）。Ph^+^CML 364例（94.1%），Ph^−^CML 23例（5.9%）。Ph^+^CML中，标准易位t（9; 22）（q34；q11）320例（87.9%）；变异易位5例（1.4%），均为复杂变异易位；Ph^+^CML合并ACA 39例（10.7%）。Ph^−^CML患者中，17例（73.9%）为正常核型，Ph^−^合并其他染色体异常者6例（26.1%）。39例初诊Ph^+^CML合并ACA患者中，−Y者2例（5.1%），−Y以外的染色体异常37例，其中，"主要路径"异常22例（56.4%），包括：der（22）t（9; 22）（q34；q11）10例，+8 7例，i（17q）3例，+Ph 2例；"次要路径"异常15例（38.5%），包括：平衡易位6例，复杂变异3例，−7、+17各1例，其他染色体结构异常4例。 2．ACA伴ABL激酶区点突变情况：39例初诊合并ACA及5例变异易位患者中，慢性期24例，进展期20例；1例慢性期、6例进展期患者放弃治疗。可随访的患者37例，7例进展期患者在初诊时行ABL激酶区点突变检测，3例点突变阳性，包括E459K、C475Y、M351T各1例。 3．初诊CML合并ACA对生存的影响：接受随访的标准易位患者320例，其中慢性期305例，进展期15例；合并ACA患者37例，其中慢性期23例，进展期14例。慢性期患者与进展期患者相比，ACA发生率差异有统计学意义（7.3%对57.1%，χ^2^=73.971，P\<0.001）。慢性期患者诊断时基本特征见[表1](#t01){ref-type="table"}，标准易位组与合并ACA组相比性别、年龄、WBC、PLT、EUTOS评分差异均无统计学意义（P值均\>0.05）。两组完全血液学反应（CHR）率分别为97.7%和95.7%（Fisher确切概率法，P=0.445），治疗12个月时完全细胞遗传学反应（CCyR）率分别为77.4%和65.2%（χ^2^=1.760，P=0.185），差异均无统计学意义。累计总的CCyR率分别为88.5%和73.9%（χ^2^=4.175，P=0.041），累计总的主要分子学反应（MMR）率分别为84.6%和65.2%（χ^2^=5.751，P=0.016），差异有统计学意义。进一步比较两组长期生存，累计3年OS率分别为93.9%和87.1%（P=0.209）；累计3年EFS率分别为71.1%和53.6%（P=0.037）；累计3年DFS率分别为81.0%和62.5%（P=0.003）（[图1](#figure1){ref-type="fig"},[图2](#figure2){ref-type="fig"},[图3](#figure3){ref-type="fig"}）。 ![慢性髓性白血病慢性期患者标准易位和合并额外染色体异常（ACA）组总生存比较](cjh-38-02-112-g001){#figure1} ![慢性髓性白血病慢性期患者标准易位和合并额外染色体异常（ACA）组无事件生存比较](cjh-38-02-112-g002){#figure2} ![慢性髓性白血病慢性期患者标准易位和合并额外染色体异常（ACA）组无病生存比较](cjh-38-02-112-g003){#figure3} ###### Ph^+^慢性髓性白血病慢性期患者诊断时基本临床特征病例特征标准易位组（305例）合并ACA组（包含变异易位）（24例）检验值 P值 ------------------------------ ------------------------ ----------------------------------- ---------- ---------- 性别\[例（%）\] 0.308^a^ 0.579 男 199（65.2） 17（70.8）女 106（34.8） 7（29.2）年龄\[岁，M（范围）\] 36（16\~69） 35（11\~70） 0.283^b^ 0.782 WBC\[×10^9^/L，M（范围）\] 212.94（5.49\~808.00） 193.97（8.40\~513.66） 0.338^b^ 0.734 PLT\[×10^9^/L，M（范围）\] 611（91\~3 065） 416（24\~1 226） 0.945^b^ 0.264 EUTOS评分\[例（%）\] 0.263^c^ 低 278（91.1） 20（83.3）高 27（8.9） 4（16.7）注：^a^为χ^2^值，^b^为t值，^c^采用Fisher确切概率法进展期患者（标准易位15例，合并ACA 14例）中，8例加速期患者选择转换TKI治疗，4例在治疗过程中进展为急变期，中位时间5（2\~11）个月。21例急变期患者，1例放弃治疗，2例患者因严重肺部感染而死亡，9例选择异基因造血干细胞移植（allo-HSCT），5例（55.6%）在移植后12个月处于CCyR状态；另外9例未接受移植患者，在治疗后2年内因疾病进展或严重并发症而死亡。二、TKI治疗CML过程中ACA的发生情况及对生存的影响 1．TKI治疗CML过程中ACA发生类型和频率：320例初诊标准易位患者，17例失访；随访的303例患者中，232例（76.6%，慢性期217例，进展期15例）无ACA发生，71例（23.4%）出现ACA。出现ACA患者中CML慢性期占18.3%（13例），类型包括：−Y（2例）、−17（3例）、−6（2例）、−5（1例）、变异易位（3例）、+8（2例）；CML进展期占81.7%（58例），类型包括：复杂变异（11例）、−Y（8例）、−17（6例）、−5（8例）、+Ph（6例）、+8（5例）、−7（5例）、−6（4例）、der（22）t（9; 22）（q34；q11）（3例）、−21（2例）。仍处于慢性期患者与出现疾病进展患者相比，ACA发生率差异有统计学意义（5.7%对79.5%，χ^2^=168.213，P\<0.001）。ACA患者与无ACA患者相比，前者更易出现疾病进展（81.7%对6.5%，χ^2^=168.21，P\<0.001）。 2．ACA伴ABL激酶区点突变情况：71例TKI治疗过程中出现ACA的患者中，5例进展期患者放弃治疗；3例合并脑出血、多器官功能衰竭死亡；随访的63例患者（慢性期13例，进展期50例），29例（46.0%）ABL激酶区点突变检测阳性，其中慢性期5例（38.5%），进展期24例（48.0%）。慢性期患者与进展期患者相比，突变发生率差异无统计学意义（χ^2^=0.378，P=0.539）。232例无ACA患者，34例（14.7%）合并ABL激酶区点突变，与ACA组相比，突变发生率差异有统计学意义（χ^2^=29.04，P\<0.001）。 3．TKI治疗CML过程中出现的ACA对生存的影响：可随访的63例治疗过程中出现ACA患者中，13例慢性期患者应用TKI治疗，治疗12个月后9例（69.2%）获得CCyR，与217例TKI治疗过程中无ACA的慢性期患者相比，两组CCyR率差异无统计学意义（69.2%对78.8%，Fisher确切概率法，P=0.487）。进一步比较两组长期生存，累计3年OS率分别为84.6%和98.9%（P\<0.001），累计3年EFS率分别为41.7%和92.9%（P\<0.001），累计3年DFS率分别为83.3%和98.5%（P\<0.001），差异均有统计学意义（[图4](#figure4){ref-type="fig"},[图5](#figure5){ref-type="fig"},[图6](#figure6){ref-type="fig"}）。 ![慢性髓性白血病慢性期患者酪氨酸激酶抑制剂治疗过程中出现与未出现额外染色体异常（ACA）组总生存（OS）比较](cjh-38-02-112-g004){#figure4} ![慢性髓性白血病慢性期患者酪氨酸激酶抑制剂治疗过程中出现与未出现额外染色体异常（ACA）组无事件生存比较](cjh-38-02-112-g005){#figure5} ![慢性髓性白血病慢性期患者酪氨酸激酶抑制剂治疗过程中出现与未出现额外染色体异常（ACA）组无病生存比较](cjh-38-02-112-g006){#figure6} 50例进展期且治疗过程中出现ACA患者，26例接受allo-HSCT，24例采用TKI联合化疗或转换TKI治疗，15例（30.0%）在治疗12个月时获得CCyR，与15例进展期无ACA患者相比，两组CCyR率差异无统计学意义（30.0%对46.7%，χ^2^=1.431，P=0.232）。进一步比较两组长期生存，进展期出现ACA患者累计OS率更低（39.3%对73.9%，P=0.086）（[图7](#figure7){ref-type="fig"}）。 ![慢性髓性白血病接受酪氨酸激酶抑制剂治疗患者进展期出现与未出现额外染色体异常（ACA）组总生存比较](cjh-38-02-112-g007){#figure7} 讨论 {#s3} ==== 以伊马替尼为代表的TKI推广及allo-HSCT方案的优化应用，将CML变为一种可长期管理的慢性疾病，但仍有部分患者难以从目前的治疗中获益。越来越多的研究证实，ACA的出现与CML病程的进展密切相关，如何尽早优化此类患者的治疗方案，成为目前国内外学者研究的焦点。CML病程中，初诊CML合并ACA者常见"主要路径"异常改变。本研究中，387例初诊CML患者，Ph^+^CML占94.1%，合并ACA者占10.7%，其中"主要路径"异常占56.4%。与文献[@b6]报道相一致。另外，23.4%的标准易位患者在治疗过程中出现ACA，CML慢性期患者主要为染色体数目异常，进展期患者除染色体数目改变外，更易伴"主要路径"异常及复杂核型异常。与文献[@b7]报道相比，本组患者治疗过程中ACA的发生率较高，且复杂变异、−17、−5、−6、−7等"次要路径"异常显示较高频率，分析其原因可能与部分患者在接受TKI治疗过程中存在治疗依从性差相关。因此，治疗过程中对患者加强教育，提高依从性，对CML患者诊疗及预后具有重要意义。 CML慢性期进展到加速或急变期往往对TKI反应欠佳[@b8]--[@b10]，对于这类患者，尽管选择转换二代TKI或TKI联合化疗能够使部分患者再次获得治疗反应，但大部分患者很快再次出现疾病进展[@b9],[@b11]--[@b13]，allo-HSCT仍然是目前唯一治愈CML的方案[@b14]--[@b15]。2013年欧洲白血病指南指出，初诊CML合并ACA，尤其是"主要路径"异常往往提示不良预后，但目前的研究不足以提供针对性的治疗计划[@b3]。近来研究报道，唯有"主要路径"提示不良预后，其他ACA改变，不需要更换治疗方案[@b6],[@b16]。本研究中，初诊CML慢性期患者，合并ACA者与标准易位者相比，累计3年OS率差异无统计学意义（P=0.209）；但前者EFS及DFS率显著降低（P值分别为0.037和0.003），进展期合并ACA者长期生存率下降。与文献[@b4],[@b17]报道相一致。对于治疗过程中出现的ACA，Lee等[@b17]认为，除-7异常外其他染色体改变不需要更改治疗策略；2013年欧洲白血病指南指出,治疗过程中出现"主要路径"异常是证实疾病进展的标志。我们的研究中，治疗过程中出现ACA者，疾病进展率高（χ^2^=168.21，P\<0.001），CML慢性期治疗过程中出现ACA者与无ACA者相比，前者OS、EFS、DFS率显著降低（P值均\<0.001））；进展期治疗过程中出现ACA者总生存率显著下降，预后较无ACA者差（P=0.086），与文献[@b18]报道一致。由此可知，初诊或是治疗过程中出现ACA者，更易发生不良事件或疾病进展，随访中需增加病情监测频率，必要时更改治疗策略。 ABL激酶区点突变是CML病程中导致患者复发耐药、治疗失败的重要危险因素，直接影响CML生存和预后[@b19]--[@b20]。本研究进一步分析ABL激酶区点突变与ACA的关系，结果显示初诊合并ACA者行点突变检测阳性率较高，我们推测，部分患者在初诊时可能已存在点突变，成为影响TKI疗效的隐匿因素，对初诊合并ACA患者行ABL激酶区点突变检测可能能为临床治疗提供指导。但本研究初诊ACA患者检测ABL激酶区点突变病例数少，尚需更多的临床数据予以证实。我们的研究结果显示，治疗过程中出现ACA者46.0%ABL激酶区点突变检测阳性，CML慢性期与进展期相比，点突变发生率差异无统计学意义（χ^2^=0.378，P=0.539）。与文献[@b21]报道相比，本组患者突变率高，可能与研究对象为ACA患者相关。进一步比较ACA与无ACA者点突变发生率，前者更易合并点突变（χ^2^=29.04，P=0.000），提示细胞遗传学不稳定者可能更易出现ABL激酶区点突变改变。因此，对于这类患者，积极行点突变检测对治疗方案的选择及预后判断同样具有极其重要的作用。综上所述，CML在病程演进中往往合并ACA，此类患者更易出现不良事件或疾病进展，对患者加强教育，提高依从性，治疗过程中严格按照诊疗监测指南[@b22]行细胞遗传学监测，以及TKI反应欠佳或治疗失败时及时行染色体核型分析及ABL激酶区点突变检测对疾病治疗及预后具有极其重要的作用。	{ "pile_set_name": "PubMed Central" }
Introduction {#section1-s-0036-1583289} ============ Continuing medical education has gained importance over the last decades. Nowadays it is considered a valuable tool in the knowledge acquisition process for residents, fellows, and young surgeons. The major purpose is to produce changes in knowledge to improve patient care.^[@bibr1-s-0036-1583289]^ Different strategies are available to teach specific topics for surgeons such as courses, cadaver laboratories, and online presentations. However, the learning process is complex, and sometimes the knowledge transfer strategy is not as successful as expected. In an attempt to improve teaching tools, de Boer et al developed a set of instruments to provide insights into the effectiveness of surgical education in the field of orthopedic trauma, termed the Learning Assessment Toolkit. It was developed to supplement the judgment of surgical educators before and after a teaching event with real evidence of need, motivation, and outcomes of educational programs.^[@bibr1-s-0036-1583289],[@bibr2-s-0036-1583289]^ The AOSpine educational program was developed to expand knowledge and to promote integration among spine surgeons around the world. It is based on curriculum development tools, by means of congresses, courses, seminars, webinars, cadaveric courses, and fellowship programs.^[@bibr3-s-0036-1583289]^ The number of spine surgery teaching activities has gradually increased over the last decades. However, there is a paucity of data about the effectiveness of the learning process and the level of competence acquired by participants in such activities. The present study is aimed at evaluating objectively the knowledge transfer provided by theoretical and practical activities during AOSpine courses for spine surgeons. Methods {#section2-s-0036-1583289} ======= This pilot test was conducted during two AOSpine principles courses in Brazil, supported by AOSpine Latin America during 2014. The events targeted spine fellows and young spine surgeons. The participants completed pre- and postcourse questionnaires, consisting of nine questions about their professional experience, preferences regarding adolescent idiopathic scoliosis (AIS) classification, and two AIS clinical cases. The students were asked to classify the curves by means of the Lenke classification. The questionnaires were developed based on the Learning Assessment Toolkit.^[@bibr1-s-0036-1583289]^ Learning Assessment Toolkit {#section3-s-0036-1583289} --------------------------- The Learning Assessment Toolkit provides information to educators about the strengths and weaknesses of their program, evaluates if educational goals were met, demonstrates objectively if knowledge was transferred, provides objective evidence of the success or failure of an educational strategy, and finally offers evidence about which future changes can be made.^[@bibr1-s-0036-1583289],[@bibr2-s-0036-1583289]^ The key competencies guide the teaching and learning process during a course. A key competency was defined as a piece of knowledge and/or skill that educators expected the course participants to have or be able to do after the course. Lenke Classification {#section4-s-0036-1583289} -------------------- Developed in 2001, the Lenke classification system provides a comprehensive and reliable means to categorize all AIS curves.^[@bibr4-s-0036-1583289]^ This classification system requires the analysis of both upright coronal and sagittal radiographs along with the supine side bending views. The triad classification system consists of a curve pattern (1 to 6), a lumbar spine modifier (A, B, C), and a sagittal thoracic modifier (−, N, +). The ultimate goal of this classification system is to organize similar curve patterns to provide comparisons of various treatment methods and to give optimal treatment for each surgical patient with AIS.^[@bibr4-s-0036-1583289]^ Precourse Assessment {#section5-s-0036-1583289} -------------------- The precourse questionnaire was given online 1 week before the event. It consisted of nine questions about the following topics: baseline characteristics of the participants: (1) type of specialty (orthopedic or neurosurgeon) and (2) experience in spine surgery (years); experience in deformity spine surgery: (3) percentage of outpatient deformity cases and (4) percentage of surgeries performed on such cases; (5) scoliosis classification system routinely used and (6) difficulties when applying Lenke classification; and classification of (7) curve pattern, (8) lumbar spine modifier, and (9) sagittal thoracic modifier of two AIS clinical cases according to the Lenke classification system. Learning Strategies Used during the Course {#section6-s-0036-1583289} ------------------------------------------ Two learning strategies were used during the course. Initially, oral lectures focused on AIS radiographic evaluation and the Lenke classification system, and then practical exercises were presented. Participants were divided into small groups (five or six per group) guided by one or two faculty members ([Fig. 1](#fig1-s-0036-1583289){ref-type="fig"}). Each participant had his or her own printed version of the two cases ([Fig. 2](#fig2-s-0036-1583289){ref-type="fig"}), a goniometer, and a pencil. The activity consisted of measuring the coronal and sagittal Cobb angles and lumbar and sagittal modifiers and finally describing Lenke\'s classifications. The exercise was performed within 1 hour. After that, an oral review was performed to explain the rationale of the proper case classification. The expected key competencies with this strategy were to improve the knowledge of the Lenke classification system and its proper application. ![During the course, practical exercises using printed X-rays of scoliotic spines were performed by participants and overseen by one or two faculty members.](10.1055_s-0036-1583289-fig1){#fig1-s-0036-1583289} ![(A) X-rays of case one. (B) X-rays of case two.](10.1055_s-0036-1583289-fig2){#fig2-s-0036-1583289} Postcourse Assessment {#section7-s-0036-1583289} --------------------- One week after the course, participants were contacted by email to reclassify the same deformity cases. Statistical Analysis {#section8-s-0036-1583289} -------------------- The descriptive analysis was performed for baseline characteristics, experience in spine surgery deformity, routinely used scoliosis classification systems, and difficulties when applying the Lenke classification. The differences in correct answers between the pre- and postcourse were analyzed to reveal the number of participants whose accuracy in classification improved after the course. The Fisher exact test or Pearson chi-square test was used to generate p values. A p value of less than 0.05 was considered statistically significant. Results {#section9-s-0036-1583289} ======= A total of 62 participants were included in the study, of whom 51 (82%) were orthopedic surgeons and 11 (18%) were neurosurgeons. Their experience in spine surgery as reported by the participants is presented in [Fig. 3](#fig3-s-0036-1583289){ref-type="fig"}. The experience in deformity spine surgery measured through the number of deformity cases seen in outpatient clinic and surgeries performed annually by the course participants varied ([Table 1](#table1-s-0036-1583289){ref-type="table"} and [Table 2](#table2-s-0036-1583289){ref-type="table"}, respectively). ![Years of experience in spine surgery.](10.1055_s-0036-1583289-fig3){#fig3-s-0036-1583289} ###### Percentage of deformity cases seen in outpatient clinic annually ![](10.1055_s-0036-1583289-table1) 1--20%, n (%) 21--50%, n (%) 51--100%, n (%) No deformity cases treated, n (%) ----------------- ------------------ ------------------- ------------------------------------- 47 (76) 11 (18) 2 (3) 2 (3) ###### Percentage of deformity surgeries performed annually ![](10.1055_s-0036-1583289-table2) 1--20% n (%) 21--50%, n (%) 51--100%, n (%) No deformity cases operated on, n (%) ---------------- ------------------ ------------------- ----------------------------------------- 47 (76) 7 (11) 3(5) 5 (8) Eighty-nine percent of the participants were familiar with the Lenke classification system and used it on a routine basis. The King classification was used by 11%. Difficulties using the Lenke classification were reported by 55% of the participants. [Table 3](#table3-s-0036-1583289){ref-type="table"} shows the parameters with which participants had difficulties within this classification. ###### Which parameter of the Lenke classification do you find difficult? ![](10.1055_s-0036-1583289-table3) n \% ---------------------------- ----- ------ Curve pattern 17 27.4 Lumbar spine modifier 1 1.6 Sagittal thoracic modifier 0 0.0 More than one parameter 16 25.8 No difficulty 17 27.4 No answer 11 17.7 The precourse assessment evidenced gaps in knowledge regarding the Lenke classification system. In the first case, the curve pattern was correctly identified by 50 participants (80.6%), lumbar spine modifier by 47 participants (75.8%), and sagittal thoracic modifier by 51 participants (82.3%; [Table 4](#table4-s-0036-1583289){ref-type="table"}). In the second case, the curve pattern was correctly identified by 29 participants (46.8%), lumbar spine modifier by 54 participants (87.1%), and sagittal thoracic modifier by 49 participants (79%; [Table 5](#table5-s-0036-1583289){ref-type="table"}). In the postcourse assessment, a decrease in the number of participants with wrong answers was observed in both cases regarding all three Lenke parameters. In the first case, statistically significant differences were observed in both curve pattern (83.3%, p = 0.005) and lumbar spine modifier (46.6%, p = 0.049). No statistically significant improvement was seen in sagittal thoracic modifier (33.3%, p = 0.309) ([Table 6](#table6-s-0036-1583289){ref-type="table"}). In the second case, there was a statistical improvement in the curve pattern (27.4%, p = 0.018). No statistically significant improvement was seen regarding lumbar spine modifier (9.8%, p = 0.121) and sagittal thoracic modifier (12.9%, p = 0.081; [Table 7](#table7-s-0036-1583289){ref-type="table"}). ###### Pre- and postcourse answers of participants in case one ![](10.1055_s-0036-1583289-table4) Parameter Precourse, n (%) Postcourse, n (%) ---------------------------- -------------------- --------------------- Curve pattern 1^a^ 50 (81) 60 (96) 2 4 (6) 1 (2) 3 6 (10) 1 (2) 4 2 (3) 0 5 0 0 6 0 0 Lumbar spine modifier A^a^ 47 (76) 55 (89) B 9 (14) 4 (6) C 6 (10) 3 (5) Sagittal thoracic modifier − 8 (13) 5 (8) N^a^ 51 (82) 54 (87) + 3 (5) 3 (5) ^a^Correct answer. ###### Pre- and postcourse answers of participants in case two ![](10.1055_s-0036-1583289-table5) Parameter Precourse, n (%) Postcourse, n (%) ---------------------------- -------------------- --------------------- Curve pattern 1 2 (3) 2 (3) 2 2 (3) 1 (2) 3 13 (21) 9 (14) 4 5 (8) 1 (2) 5 11 (18) 7 (11) 6^a^ 29 (47) 42 (68) Lumbar spine modifier^b^ A 5 (8) 1 (2) B 3 (5) 2 (3) C^a^ 54 (87) 58 (95) Sagittal thoracic modifier − 10 (16) 3 (5) N^a^ 49 (79) 56 (90) + 3 (5) 3 (5) ^a^Correct answer. ^b^One missed piece of data in the postcourse assessment. ###### Comparison between pre- and postcourse of participants with wrong answer in case one ![](10.1055_s-0036-1583289-table6) Parameter Precourse (n) Postcourse (n) Improvement, n (%) p Value ---------------------------- ----------------- ------------------ ---------------------- ----------- Curve pattern 12 2 10 (83.3%) 0.005 Lumbar spine modifier 15 7 8 (46.6%) 0.049 Sagittal thoracic modifier 12 8 4 (33.3%) 0.309 ###### Comparison between pre- and postcourse of participants with wrong answer in case two ![](10.1055_s-0036-1583289-table7) Parameter Precourse (n) Postcourse (n) Improvement, n (%) p Value ---------------------------- ----------------- ------------------ ---------------------- ----------- Curve pattern 33 20 13 (39.3%) 0.018 Lumbar spine modifier 8 3 5 (62.5%) 0.121 Sagittal thoracic modifier 13 6 7 (53.8%) 0.081 Discussion {#section10-s-0036-1583289} ========== Interest is increasing in the improvement of medical education quality through measurable effects of the teaching strategies.^[@bibr5-s-0036-1583289]^ The present study was based on the Learning Assessment Toolkit proposed by de Boer et al.^[@bibr1-s-0036-1583289]^ This method provides accurate information regarding acquired competencies that could help in clinical judgment and the doctor\'s decision-making process.^[@bibr1-s-0036-1583289],[@bibr6-s-0036-1583289]^ To the authors\' knowledge, this study is the first to describe objective data about knowledge acquisition among young spine surgeons. As a result of the learning strategies used by educators during the course, the participants improved their knowledge of the Lenke classification system. This knowledge acquisition was demonstrated through the decreased number of participants with wrong answers in two cases after the course. More satisfactory results in the curve pattern of case one were observed than in case two, possibly because the correct answer for case two is Lenke 6. This curve pattern is frequently misclassified as curve 3 because both are double curves. The difference between the two is that in Lenke 3, the main curve is thoracic, and in Lenke 6, it is thoracolumbar/lumbar. Although orthopedic surgeons are classically more familiar with AIS treatment, every year more neurosurgeons are involved with the disease. Hence, it is necessary to consolidate principles and concepts to create a universal language among spine surgeons that could improve the literature reports. The proper classification of the curve is essential, especially when fusion is considered. The Lenke classification system is commonly used for that purpose.^[@bibr4-s-0036-1583289]^ However, despite its broad application, interobserver reliability is considered poor to fair when the degree of professional training is taken into account.^[@bibr4-s-0036-1583289],[@bibr7-s-0036-1583289]^ Surprisingly, 55% of the participants of the present study reported difficulties when applying it, although 89% use it on a routine basis. The authors applied two learning strategies during two different AOSpine principles courses: oral presentations and a practical exercise using printed X-rays of patients with AIS. The improvement of the students\' ability to properly classify AIS curves ranged from 33 to 83%. As previously reported, if only traditional lectures are chosen as an isolated learning strategy, minor changes are expected in the students\' behavior as a health professional.^[@bibr8-s-0036-1583289]^ In such cases, the amount of information retained is considered low (up to 20%). Educators could expect higher knowledge absorption when students are actively involved and are stimulated to find their own solutions to problems.^[@bibr8-s-0036-1583289]^ The relevance of the active participation of the students during AOSPINE diploma curriculum was previously discussed. In the present study, the students were encouraged to think by themselves and, after that, to convince the rest of the participants of the rationale for their thoughts. That fact may have had the most significant effect on the learning process. The present study has some limitations, such as the small number of participants, which did not allow performing a subgroup analysis. Furthermore, the participants were asked to answer the same questions before and after the course. That fact could have generated a test--retest bias, giving a false impression about knowledge acquisition occurring as a result of the course. Conclusions {#section11-s-0036-1583289} =========== This pilot test showed objectively that the learning strategies used during AOSpine courses improved the participants\' knowledge. This knowledge acquisition was demonstrated through the decreased number of participants with wrong answers of two clinical cases after the course. Teaching strategies must be continually improved to ensure an optimal level of knowledge transfer. Disclosures: Alberto Ofenhejm Gotfryd: none Jose Alfredo Corredor: none William Jacobsen Teixeira: none Delio Eulálio Martins: none Jeronimo Milano: none Alexandre Sadao Iutaka: none The authors would like to thank AOSpine Latin America for their contribution to the copyediting of this manuscript.	{ "pile_set_name": "PubMed Central" }
Background ========== Patellofemoral pain is a common disorder whose aetiology is multifactorial and is often attributed to foot function. Foot orthoses are commonly prescribed for this condition; however the mechanisms by which they work are poorly understood. Previous studies using single segment foot models have hypothesised that it may be control of the midfoot which holds the key to understanding orthotic control. Over the last decade it has become possible to divide the foot into multiple segments, however little work exists investigating the use of orthoses on different segments of the foot in this patient group. The aim of this study was to investigate the differences in the kinematics and kinetics of the lower limb during walking and step descent between patellofemoral patients and normal subjects and the effect of ¾ and full length foot orthoses versus no intervention. Method ====== Kinematic and kinetic data were recorded from 15 healthy subjects and 15 patients diagnosed with patellofemoral pain using 10 Oqus cameras and 4 AMTI force platforms. Subjects were asked to walk at a self-selected pace and complete a 20cm step down. The foot was modelled using a three segment 6 degrees of freedom model by fixing the marker set directly to the shoes and the lower limb was modelled using the calibrated anatomical systems technique. Results ======= Significant differences were seen between the healthy subjects and the patellofemoral pain patients during both tasks at the midfoot and rearfoot movement in the sagittal and coronal planes (p=0.003 to 0.016); at the knee joint significant differences were seen in the sagittal, coronal and transverse plane movement (p=0.001 to 0.01); and in the moments about the ankle and knee joints in the sagittal and coronal planes (p=0.012 to 0.035). The orthoses produced statistically significant differences in the movement in the forefoot, midfoot and rearfoot across all three planes for both tasks (p=0.001 to 0.032). The orthoses showed no change in the knee kinematics, although a significant reduction in the knee coronal plane moments during step descent was seen in both the ¾ and full length foot orthoses (p=0.019, p=0.028). Conclusions =========== Despite placing markers on the shoes this study was able to detect significant differences within the foot segments and identified potentially clinically important differences between patellofemoral pain patients and normal subjects and was able to determine clinical important changes due to treatment.	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== Avian infectious bronchitis (IB) is caused by a virus in the Coronaviridae family, genera Gammacoronavirus. It is a highly contagious disease with a short incubation period \[[@CR1]\]. The Avian coronavirus was previously classified, and is most commonly referred to, as avian infectious bronchitis virus (IBV). The IBV is responsible for respiratory disease, which manifests in clinical symptoms such as sneezing and tracheal-bronchial rales that can lead to the development of more severe symptoms \[[@CR2], [@CR3]\]. Infected birds exhibit reduced performance, consequently leading to a reduction in weight gain and deterioration in egg quality and quantity. Secondary bacterial infections will also contribute to economic losses. Carcass condemnation due to the development of airsacculitis \[[@CR4], [@CR5]\] negatively impacts commercial sales of bird meat and eggs. Brazil was once the world's largest exporter of poultry and currently the world's third largest producer of bird meat \[[@CR6]\]. The consequences of IBV are a significant threat to Brazil's poultry industry. The IBV genome consists of a non-segmented positive-sense single-stranded RNA that is approximately 27.6 kb in length. It encodes non-structural (accessory proteins) and four structural proteins: the nucleocapsid protein (N), the spike protein (S), the envelope protein (E), and the matrix protein (M). The nucleocapsid protein, or N protein, consists of 409 amino acids. It has a molecular mass of approximately 50 kDa and directly binds with the viral genome to form the virion nucleocapsid \[[@CR6], [@CR7]\]. Its structure is highly conserved, with different strains of IBV sharing a high degree of identity (94--99%) \[[@CR8]\]. The N protein is also known for its immunogenicity, inducing specific antibody and cytotoxic T-cells mediated responses \[[@CR9], [@CR10]\]. There is significant interest in the use of the IBV N protein as an important target for diagnosis since it possesses the antigenic characteristics required for the development of serological assays that can be applied to detect or quantify antibodies against the IBV \[[@CR11]\]. The laboratory diagnosis of IB is dependent on direct and indirect techniques. The direct techniques are employed for viral isolation and genomic or phenotypic identification of the virus, while the indirect methods are used to detect specific antibodies \[[@CR12]\]. In addition to being applied for serodiagnosis, serological techniques can also be employed to evaluate the immune responses stimulated by vaccines. Commercial ELISA kits are typically used to indirectly diagnose IBV. These kits, however, are expensive when large number of samples require screening and they are not acessible for applications with the scale of the Brazilian poultry industry \[[@CR13]--[@CR15]\]. ELISA techniques currently available are designed to detect polyclonal antibodies that target the whole virion. The use of nucleoprotein as the antigen for diagnosis and evaluation of vaccine immune responses is an interesting target to explore since this protein plays a important role in IBV virus replication and the induction of a specific immune response in infected birds \[[@CR16], [@CR17]\]. The use of recombinant antigens in the design of a specific diagnostic technique facilitates the development of highly sensitive and specific assays that display a high antigen concentration and, thereby, reduce or eliminate background reactions. The use of recombinant antigens also represents a viable method of reducing immunoassay development costs. Easy production of antigens in expression systems leads to simple and efficient antigen development which can reduce the production costs associated with diagnosis \[[@CR18]\]. The aim of the current study was to evaluate the combined use of an ELISA and Western blot (WB) to detect antibodies against the nucleocapsid protein of IBV. Methods {#Sec2} ======= Virus strain and viral RNA extraction {#Sec3} ------------------------------------- A previously characterized Brazilian viral sample of IBV Strain Massachusetts 41 (M41- CNPSA -- EMBRAPA -- Concórdia, SC, Brazil) was propagated after 9 days of incubation in the chorioallantoic cavity of specific pathogen free (SPF) embryonated chicken eggs. The allantoic fluid was then collected and stored at − 70 °C. Viral RNA extraction was carried out with TRIzol® LS reagent (Invitrogen™, EUA), according to the manufacturer's instructions. N protein coding sequence amplification and cloning {#Sec4} --------------------------------------------------- Extracted RNA from IBV strain M41 was used for cDNA synthesis with random oligonucleotides. Reverse transcription (RT) was carried out using SuperScript® One-Step RT-PCR System (Invitrogen, USA). The resulting cDNA samples were used to for PCR amplification of the whole orf of the N protein gene. Primers based on the IBV M41 N protein gene sequence available at GenBank (accession number M28566) were designed to align between 102 and 120 and 1312--1331 bp of the gene and include cleavage sites for restriction enzymes. There was a restriction site for XhoI in the forward primer (5′ -- CCGCTCGAGATGGCAAGCGGTAAGGCAA -- 3′) and a restriction site for KpnI in the reverse primer (5′ -- GGGGTACCTCAAAGTTCATTCTCTCCTA -- 3′). The PCR reaction was performed with approximately 25 ng of the extracted cDNA, 3.5 mM MgCl, 0.2 mM dNTPs, 2 units of Taq DNA polymerase, 1X reaction buffer, 1 pmol of each primer, and 5 M N,N,N-trimethylglycine (betaine) under the following conditions: 1 cycle of 95 °C for 7 min, 1 cycle of 70 °C for 1 min, then 45 cycles of 94 °C for 1 min, 50 °C for 1 min, and 72 °C for 4 min, and a final extension of 72 °C for 10 min. The PCR amplification product was confirmed on a 1% agarose gel and purified using GFX PCR DNA and Gel Band Purification kit (GE Healthcare, Chicago, USA), according to the manufacturer's instructions. Recombinant N protein (rN) expression {#Sec5} ------------------------------------- The PCR product was cloned into pAE vectors by a T4 DNA ligase (Invitrogen) binding reaction after cleavage with restriction enzymes KpnI and XhoI. The constructed recombinant pAE/n expression vector was used to transform E. coli BL21(DE3) Star competent cells (Invitrogen). The resulting recombinant clones were cultivated in 10 mL of LB broth medium with 100 μg/mL of ampicillin (37 °C, 16 h, 250 rpm). The whole culture volume was transferred to flasks containing 200 mL of LB and incubated at 37 °C with agitation (200 rpm) until the optical density (O.D.) at 600 nm reached 0.8. The expression of the recombinant N protein (rN) was induced by adding isopropyl-β-d-thiogalactopyranoside (IPTG, 0.5 mM final concentration) to the culture and incubating for 3 h at 37 °C with agitation (200 rpm). The cells were harvested at 10,000 x g for 10 min at 4 °C and the culture pellet containing the rN protein was subjected to a solubilization procedure in ÄKTA wash buffer (0.234% NaH~2~PO~4~, 2.92% of NaCl, 0.068% Imidazole, pH 8.0, supplemented with 20 μg/mL lysozyme). The resuspended cells were submitted to seven sonication cycles of 20 s at 60 Hz and centrifuged again. The rN was purified by affinity chromatography using HisTrap HP 1 mL columns pre-packed with pre-charged Ni Sepharose in the ÄKTAprime Automated Liquid Chromatography system (GE Healthcare). The protein concentration was determined with Qubit™ Protein Assay Kits (ThermoFisher Scientific, USA) according to the manufacture's instructions. Characterization of rN by SDS-PAGE and Western blot {#Sec6} --------------------------------------------------- The SDS-PAGE was carried out on 12% polyacrylamide gel. The gels were either stained with Coomassie Blue R-250 (Bio-Rad, California, USA) or electroblotted onto Hybond-ECL 0.45 μm nitrocellulose membranes (GE Healthcare) using the Bio-Rad Mini Trans-Blot Cell (Bio-Rad) for Western blot. Briefly, the membrane was blocked with 5% non-fat milk in phosphate buffer saline containing 0.05% Tween-20 (PBS-T) (137 mM NaCl, 2.7 mM KCl, 100 mM Na~2~HPO~4~, 2 mM KH~2~PO~4~, pH 7.4), incubated with mouse monoclonal antibody (MAb) anti-6xHis (Sigma, USA) diluted to 1:10000 in PBS-T, and incubated with polyclonal antibody Anti-Mouse IgG conjugated to HRP (Sigma). All incubation steps were performed at 37 °C for 1 h under slight agitation followed by three washes with PBS-T. The immunoblot was developed using 3,3′-diaminobenzidine (Sigma). Avian serum sample {#Sec7} ------------------ A total of 389 chicken serum samples (n = 389) were used to develop and evaluate the ELISA-rN. The serum samples were kindly provided by MercoLab Laboratories (Garibaldi, Brazil), a laboratory accredited by the MAPA (Ministry of Agriculture, Livestock and Food Supply) that uses the commercial ELISA kit IBV Ab Test (IDEXX) for characterization. Serum samples from chickens with a positive diagnosis for Newcastle disease were used for the evaluation of the ELISA-rN specificity. The samples used in this study originated from birds with a known history of vaccination, including vaccination against the Newcastle disease virus, and were categorized into three groups: commercial egg-layers, broiler breeders, and meat-producing chickens. This information was further used to improve the evaluation of the ELISA-rN test. Development of the indirect ELISA-rN {#Sec8} ------------------------------------ To standardize the indirect ELISA development, serum dilutions (1:100, 1:200 and 1:400) and different concentrations of purified rN antigen (50, 100 and 200 ng/well) were tested. Positive and negative sera for IBV were used as controls. The 96-well microtiter plates (Nunc MaxiSorp®, Thermo Fisher, USA) were coated with 100 ng/well of rN diluted in 0.05 M carbonate-bicarbonate buffer (pH 9.6) and incubated overnight at 4 °C. The plates were incubated at 37 °C for 1 h with a blocking solution (5% non fat dry milk in PBS-T). Serum samples diluted 1:200 in PBS-T were added in duplicate and the plates were again incubated for 1 h at 37 °C. After this period, the secondary antibody horseradish peroxidase (HRP) - rabbit anti-chicken IgY peroxidase conjugate (Sigma) diluted 1:10,000 in PBS-T was added and plates were once more incubated at 37 °C for 1.5 h. After each incubation, the plates were washed three times with PBS-T. In the final step, plates were washed five times and 100 μl/well of peroxidase substrate o-phenylenediamine dihydrochloride (Sigma Aldrich) was added. The reaction was interrupted with 2 N H~2~SO~4~ and the results were read as O.D. using a spectrophotometer at 492 nm. To determine the cutoff value, negative sera were used, totalizing 20 sera, considering the mean of sera added of two standard deviation. The ROC analysis was used to simulate the influence of different cutoff values on the sensitivity and specificity of the test. The results were compared with those obtained with the commercial kit. Serum that scored as negative in the commercial kit but as positive in the ELISA-rN were submitted to Western blot analyses using the rN protein as antigen. Repeatability and specificity of ELISA-rN {#Sec9} ----------------------------------------- In order to evaluate the repeatability of the ELISA-rN, three separate batches of recombinant N protein were produced and purified following the methodology described above with distinct purification times, to demonstrate the reproducibility of the recombinant protein expression procedure. Serum samples were selected for testing against each batch of antigen and the averages, and standard deviations, of the O.D. at 492 nm were calculated. Of the samples used for this test, one was strongly reactive positive, four were moderately reactive positive, and four were negative serum samples. To evaluate the specificity of the test, negative serum samples to IBV and positive serum samples to Newcastle disease were tested in the ELISA-rN. Development of the Western blot-rN {#Sec10} ---------------------------------- The western blot was performed almost as described above. The rN was deposited in all wells of the SDS-PAGE gel and, after transfer to the nitrocellulose membrane, it was cut to obtain 14 strips. Each strip was incubated with a 1:200 serum dilution after blocking with 5% non-fat milk. The antibody horseradish peroxidase (HRP) - rabbit anti-chicken IgY peroxidase conjugate (Sigma) (1:10000) was added to all membranes and the reaction was revealed using 3,3′-diaminobenzidine (Sigma). All incubation steps were performed at 37 °C for 1 h under slight agitation and were followed by three washes with PBS-T. Statistical analysis {#Sec11} -------------------- The results of the ELISA-rN compared to the commercial kit results, the Receiver-operating characteristics (ROC) curves, the area under the curve (AUC) and the confidence intervals were obtained through the software GraphPad Prism version 6.0 for Windows (GraphPad Software). Results {#Sec12} ======= Cloning and expression of recombinant N protein {#Sec13} ----------------------------------------------- The expected cDNA fragment of approximately 1240 bp was successfully amplified by PCR using primers designed to obtain the coding region of the N protein. Successful recombinant pAE/n vector construction was confirmed through cleavage with the same restriction enzymes used for the plasmid construction (date not shown). The recombinant N protein (rN) was expressed as a soluble protein, dispensing refolding steps, and, after purification, a yield of 10 mg/L of rN was obtained. The SDS-PAGE results demonstrated the presence of two distinct bands that had a molecular mass of approximately 45 kDa and 50 KDa (Fig. [1a](#Fig1){ref-type="fig"}). The same bands were observed in the Western blot and confirmed the presence of bands that corresponded to the protein of interest (Fig. [1b](#Fig1){ref-type="fig"}).Fig. 1Expression of the recombinant N protein. a The rN protein expressed in E. coli Unstained MW Marker (ThermoFisher Scientific); Lane 2: Recombinant E. coli strain BL21 star (DE3) before protein expression induction; Lane 3: Recombinant E. coli strain BL21 star (DE3) after protein expression induction; Lane 4: Purified rN protein. b Western blot characterization of the rN protein performed with MAb Anti6xHis. Lane 1: Spectra Multicolor Broad Range Protein Ladder (ThermoFisher Scientific); Lane 2: Recombinant N Protein; Lane 3: Negative control E. coli strain BL21 star (DE3) Other studies have also described the cloning of the 1200 bp corresponding to the N protein gene from IBV \[[@CR18], [@CR19]\]. They reported that the recombinant protein presented as two distinct molecular masses, one 50 kDa and other 45 kDa. The 45 kDa mass is the truncated form of the N protein \[[@CR18], [@CR19]\]. This corroborates with the results observed through SDS-PAGE and Western blot in the current study (Fig. [1](#Fig1){ref-type="fig"}). Indirect ELISA-rN and ROC analysis {#Sec14} ---------------------------------- The results of the ELISA-rN were compared with the results obtained with the IBV Ab Test (IDEXX). Receiver-operating characteristics (ROC) with 95% confidence intervals were used to analyse these results (Fig. [2](#Fig2){ref-type="fig"}). Out of 244 sera that scored positive in the IBV Ab Test, 220 also scored positive in the ELISA-rN test. The sensitivity of the ELISA-rN test was 90.16%. Out of the 145 sera that scored negative in the IDEXX Test, 131 also scored negative in the ELISA-rN, indicating a specificity of 90.34%. The area under the curve (AUC) was 0.9588 (P \< 0.001 and cutoff 0.5415).Fig. 2Interactive dot diagram based on ELISA-rN outcomes in relation to ELISA kit IDEXX IBV Ab Test. The results of the commercial test ELISA kit IDEXX IBV Ab Test (IDEXX) were compared with that of the ELISA-rN, confirming 90.16% sensitivity and 90.34% specificity Test repeatability and specificity {#Sec15} ---------------------------------- Three different protein batches were used as coating antigens for testing the ELISA-rN and the results from all three were similar (P \> 0.05) and, thus, indicate antigen stability (Fig. [3](#Fig3){ref-type="fig"}). All Newcastle disease positive sera tested negative in the developed ELISA-rN, providing evidence of test specificity (data not shown).Fig. 3Evaluation of different batches of rN protein production used as antigen in the indirect ELISA development and tested with positive and negative sera for IBV. Sample 1: strongly reactive positive, samples 2--5: moderately reactive positive and samples 6--9: negative serum samples Western blot for analysis of discrepant results {#Sec16} ----------------------------------------------- All sera that tested negative in the commercial test and positive in the ELISA-rN were submitted for Western blot analyses using the rN protein as an antigen. All serum samples reacted with the two bands of rN, confirming the results obtained in the developed ELISA (Fig. [4](#Fig4){ref-type="fig"}). Sera that tested negative in the ELISA-rN and positive in the commercial test also reacted with the rN in WB (data not shown).Fig. 4Western blot analyses of the avian serum samples that scored negative for the IDEXX IBV Ab Test and positive in the ELISA-rN. Lane 1: Spectra Multicolor Broad Range Protein Ladder (ThermoFisher Scientific); Lanes 2--15: Serum samples with discrepant ELISA results Discussion {#Sec17} ========== Previous studies that focused on the development of more efficient diagnostic techniques were not limited to avian infectious bronchitis detection, but also aimed to control and monitor chickens vaccination \[[@CR12], [@CR20]\]. The nucleoprotein from IBV is widely considered the choice protein for the development of immunoassays for antibody detection since this protein plays an important role in inducing an antibody response in IBV infected or vaccinated animals \[[@CR10], [@CR13]\]. Other studies have demonstrated that the nucleocapsid protein is highly conserved among IBV isolates (91--96.5% similarity), immunogenic, and abundantly expressed during infection \[[@CR8], [@CR10], [@CR21]\]. These features make this protein an interesting candidate for use in diagnostic techniques, such as ELISA and WB \[[@CR4]\]. The ELISA is a powerful tool because it provides a safe and an easy way to evaluate the IBV vaccine efficiency, and perform serological diagnosis as well as epidemiological surveillance \[[@CR18]\]. In contrast to the ELISA presented in this study, the ELISA developed by Lugovskaya et al. \[[@CR15]\] used two fragments of the recombinant N protein expressed in E. coli as an antigen, achieved a specificity of 87.36%, and a sensitivity of 93.81%. These results are comparable with the commercial test that is currently used for routine IBV diagnosis that has a specificity of 88.97% and a sensitivity of 92.86% \[[@CR15]\]. The ELISA-rN developed in this study presented similar specificity and sensitivity results (90.16 and 90.34%, respectively), and thus highlights that a unique IBV nucleoprotein fragment can be employed to efficiently detect specific antibodies in poultry serum. Our study uses a Western blot technique to complement the results obtained through ELISA and further support the IBV diagnosis. The Western blot technique may prove useful for epidemiological studies, for monitoring specific pathogen free farms, and in vaccine potency tests. The association of ELISA and Western blot can be seen as a tool to be considered to increase sensitivity for serodiagnosis purposes \[[@CR22]\]. Since positive sera react with two bands in the Western blot, the technique is even more specific. The recombinant N protein produced in this study was expressed in the soluble fraction. Our rN was easily recoverable from the culture without using denaturing agents. In contrast to previous studies where the N protein was expressed in its insoluble form \[[@CR10], [@CR14]\], our procedure avoids the necessity to refold the recombinant protein. The production of rN was repeatable since the same ELISA-rN test results, and about the same yield, were obtained when different production and purification times were used. Besides the use of the N protein for the diagnosis of IBV, an ELISA that was developed using the S protein was also described and compared to a commercial test \[[@CR23]\], and achieved a specificity and sensitivity of 89.83 and 92.38%, respectively. However, the N protein offers specific advantages when used for the purpose of diagnosis during viral infection since it is produced in larger quantities than the S protein (the N proteins is produced at a ratio of 6:1 relative to the S protein \[[@CR24]\] and plays an important role in the virus's replications and assembly process \[[@CR25]\]. Additionally, the S protein has hypervariable regions that cause mutations in its sequence and therefore offers low efficiency as a protein for diagnosis \[[@CR1]\]. The results from the ELISA test developed using the rN protein indicate that the test could be effectively applied for IBV diagnosis \[[@CR17]\]. The yield of rN per litre of LB broth culture would coat approximately 1000 96-well microtiter plates allowing for the diagnostic analysis of approximately 45,000 serum samples in duplicate. Also noteworthy is the production cost reduction from using rN in a soluble form, since this eliminates costs associated with the refolding step while it also preserves important conformational epitopes. The currently available commercial test on the other hand, employs the whole IBV as an antigen which requires viral propagation and the implementation of robust laboratory biosafety standards \[[@CR23]\]. It is worth mentioning that, by comparing the developed ELISA-rN and WB with the IDEXX IBV Ab Test, the analysis conducted in the current study indicated that it was possible that some serum identified as negative for anti-IBV antibodies on the commercial test be positive when tested using the ELISA-rN and by WB. This could be of concern as false negative results are undesirable, especially if a lot of birds are misdiagnosed. Thus, the ELISA-rN could be applied together with the WB developed in this study for the routine detection of IBV in diagnostic laboratories. False negative results that contribute to the spread of the disease can be avoided, or at least decreased, when the two tests are applied together. Conclusion {#Sec18} ========== The indirect ELISA developed here with rN as an antigen allowed for the detection of anti-IBV antibodies in chicken serum at high specificity and sensitivity. The association between ELISA and Western blot techniques developed with a subunit of IBV (rN) were able to detect antibodies that were not detected with the commercial ELISA test suggesting greater sensitivity in the developed ELISA-rN. In addition, the ELISA-rN with the advantages of easy preparation and improved safety could be a promising alternative to the whole live virus ELISA. We thank Mercolab laboratory (Garibaldi, RS, Brazil) for giving the sera that were analyzed in this study. Funding {#FPar1} ======= Not applicable. Availability of data and materials {#FPar2} ================================== All data generated or analyzed during this study are included in this article. All authors read and approved the final manuscript. Ethics approval and consent to participate {#FPar3} ========================================== This article does not contain any studies with human participants or animals performed by the author. Consent for publication {#FPar4} ======================= The authors authorize the publication. Competing interests {#FPar5} =================== No known competing interest. Publisher's Note {#FPar6} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#sec1-1} ============ Renal cell carcinoma (RCC) is referred to as the 'internists' tumor' due to its varied presentation and often unpredictable course. Every urologist has known patients with small renal masses but distant metastases, and at the other end of the spectrum, those who survive years with extensive disease. Some may even have seen spontaneous regression of metastatic RCC. Prognostication of these cases has remained difficult, and a bewildering number of markers and prognostic systems have been used to better characterize prognosis in RCC.\[[@ref1]\] The need for accurate prognostication became even more acute with the advent of targeted therapy for advanced RCC, and here too, new systems were devised.\[[@ref2]\] Of these, the International Metastatic RCC Database Consortium prognostic model has proven accurate and has recently been externally validated.\[[@ref3][@ref4]\] There is now a growing body of the literature, suggesting that thrombocytosis could prove a simple, accurate and inexpensive prognostic marker in many malignancies, including RCC.\[[@ref5]\] We, therefore, decided to review our experience to determine the prognostic significance of the platelet count in all stages of RCC. MATERIALS AND METHODS {#sec1-2} ===================== Ours is tertiary medical center in southern India with a large uro-oncology practice. We undertook a retrospective review of our electronic medical record database. All patients with biopsy or histopathologically proven RCC from January 2004 to June 2011 were eligible for inclusion. Patients undergoing either partial or radical nephrectomy were included. Patients with no record of a preoperative platelet count were excluded. Those with bilateral or multiple tumors \[including von Hippel Lindau (vHL) disease\] were also excluded from the study. TNM staging was reviewed according to the American Joint Committee for Cancer 2009 revision of the International TNM classification. Tumor and nodal status were based on histopathology of the surgical specimen with radiological input. Metastases status was assessed using imaging studies, namely contrast-enhanced computed tomography (CECT) of the abdomen and chest X-ray, with CT chest in selected cases. Baseline data collected on all patients included demographic variables, laboratory parameters, TNM stage, pathological features, and available follow-up and survival data. The Fuhrmann system was used to assess tumor grade. Disease progression was defined as the new appearance of local recurrence or distant metastasis on a follow-up visit. Thrombocytosis was defined as a platelet count above 400,000/cubic milliliter (cuml). Statistics {#sec2-1} ---------- Initially, the correlation between baseline patient and tumor characteristics with the platelet count was studied. Patients were then classified into two arms: Those with and without thrombocytosis. These groups were then compared with respect to gross and microscopic tumor characteristics, TNM stage, and disease progression (as defined above). Standard statistical tests of significance were applied. Furthermore, multivariate analysis by logistic regression was performed using thrombocytosis, TNM stage, and tumor size as variables. A P \< 0.05 was considered statistically significant. Statistical analysis was performed using SPSS version 16. RESULTS {#sec1-3} ======= [Figure 1](#F1){ref-type="fig"} describes the derivation of the study group after applying the exclusion criteria. A total of 649 patients with RCC were identified during the study period. There was no record of a platelet count in 308. Of the remaining 341 patients, 11 were diagnosed cases of vHL disease, and a further 8 had bilateral tumors, leaving 322 in the study group. ![Flowchart depicting derivation of the study group](IJU-31-42-g001){#F1} Baseline demographic and tumor variables of the study population are represented in [Table 1](#T1){ref-type="table"}. The association of clinical and staging variables with the platelet count is represented in [Table 2](#T2){ref-type="table"}. We discovered that the platelet count was significantly higher in females and those with Fuhrmann grade 3-4 tumors. The platelet count also showed a significant progressive increase with TNM stage. ###### Baseline demographic and tumor factors (n=322) ![](IJU-31-42-g002) ###### Association of clinical and staging variables with the platelet count ![](IJU-31-42-g003) Patients were subsequently classified into those with thrombocytosis (n = 35) and compared with those with a normal platelet count (n = 287) \[[Table 3](#T3){ref-type="table"}\]. Both groups were similar with respect to age and gender. Patients with thrombocytosis had significantly larger tumors. Their tumors were also significantly more likely to have sarcomatoid and rhabdoid features, lymphovascular invasion, and capsular invasion at diagnosis. Patients with thrombocytosis had a significantly higher tumor stage; a significantly lower hematocrit and higher total leucocyte count than those with a normal platelet count. The rate of progression was higher in those with thrombocytosis with a significantly quicker median time to progression in this group. ###### Results ![](IJU-31-42-g004) Multivariate analysis of time to progression revealed TNM stage as the only significant predictor \[[Table 4](#T4){ref-type="table"}\]. The possible reason for this is the relatively short follow-up in our study. ###### Multivariate analysis ![](IJU-31-42-g005) DISCUSSION {#sec1-4} ========== Thrombocytosis has been implicated as an adverse prognostic factor in a number of malignancies including RCC.\[[@ref6]\] In these situations, the exact cause of the secondary thrombocytosis is unclear, but it appears that circulating cytokines and growth factors play an important role.\[[@ref6]\] Of these, interleukin (IL)-6 appears to be the most potent stimulator of megakaryocyte progenitors, and a raised level has been demonstrated in the majority of patients with malignancy-associated thrombocytosis.\[[@ref7]\] Other cytokines that have been implicated include macrophage-colony stimulating factor, thrombopoietin, and IL-11.\[[@ref7]\] Platelets have also been implicated in tumor growth as they can secrete large concentrations of vascular endothelial growth factor (VEGF) that is crucial for tumor angiogenesis, while also stimulating megakaryocyte maturation at the bone marrow level.\[[@ref7][@ref8]\] Activated platelets also secrete thrombopoietin, which reinforces platelet formation in the bone marrow.\[[@ref8]\] Platelets and their secretory product thrombospondin have also been implicated in tumor metastasis. The purported mechanisms include allowing adhesion of tumor cells to the vascular endothelium, penetration through the endothelial barrier, and preventing malignant cells from being cleared from the circulation.\[[@ref6]\] Over the past decade, a few studies have studied the prognostic significance of thrombocytosis in RCC. In the largest study so far Bensalah et al. reported that a platelet count \> 450,000/cuml positively correlated with worsening T stage, Fuhrman grade, tumor size, lymph node status, and distant metastasis in 804 patients with RCC.\[[@ref8]\] Patients with thrombocytosis also had a significantly worse 5-year survival on both univariate and multivariate analyses.\[[@ref8]\] This impact on prognosis was seen for both localized and metastatic disease.\[[@ref8]\] Inoue et al. also showed a positive correlation between thrombocytosis and tumor size and stage.\[[@ref7]\] In their study, thrombocytosis was associated with a worse prognosis, but when adjusted for stage or tumor size, this was limited to pT1-2 tumors.\[[@ref7]\] In the study by Cho et al., thrombocytosis significantly correlated with tumor size and metastasis.\[[@ref9]\] It was a predictor of recurrence-free survival on univariate but not multivariate analysis.\[[@ref9]\] Patel et al. reported a retrospective study on 237 patients who underwent radical nephrectomy for clinically localized disease.\[[@ref5]\] They concluded that an increase in platelet count of \>20% following radical nephrectomy could reliably predict recurrence and cancer-free survival.\[[@ref5]\] The need for prognostication in metastatic RCC has gained immense significance with the birth of targeted therapy, with accurate prognostication allowing individualization of therapy. In an early study, Symbas et al. studied 259 patients with metastatic RCC who received a variety of adjuvant therapies.\[[@ref6]\] They discovered that those whose platelet count remained persistently normal had a 64% longer survival than those with thrombocytosis, even on multivariate analysis.\[[@ref6]\] Perhaps the most important system for prognostication, the International Metastatic RCC Database Consortium prognostic model includes platelet count and neutrophil count above the upper limit of normal, as independent adverse factors.\[[@ref3]\] The other components of this system are similar to those of the Memorial Sloan Kettering Cancer Centre (MSKCC) model and include hemoglobin less than the lower limit of normal, corrected calcium greater than the upper limit of normal, Karnofsky performance status \<80%, and time from diagnosis to treatment of \<1 year.\[[@ref3]\] This model has recently been externally validated and has shown a concordance index of 0.68-0.73, which was better than other similar systems.\[[@ref4]\] Other prognostic systems for metastatic RCC, including a revised model from the MSKCC (for patients treated with sunitinib), and one from the Cleveland Clinic also include the platelet count as one of the parameters.\[[@ref10]\] While there is no doubt that a combination of factors will be most reliable in prognostication, the platelet count alone is a simple investigation that can give the clinician an indicator of the gravity of disease, and we believe that further prospective studies could help better establish its exact role. The success of targeted therapy has also opened up the possibilities of neoadjuvant and adjuvant therapy for locally advanced RCC. Research in this area is underway and three randomized trials are ongoing to determine the efficacy of these drugs in the adjuvant setting.\[[@ref11]\] The fact that thrombocytosis correlates well with adverse prognosis, even in patients with clinically nonmetastatic disease, may possibly be used to select patients who would benefit from further treatment. Our study is limited by its retrospective nature and the relatively short follow-up of patients (median of 7 months). Only 18% (n = 58) of our patients had more than 3 years follow-up, and only 6.8% (n = 22) had more than 5 years follow-up. This could be the reason that platelet count was not deemed significant on multivariate analysis of time to progression. However, platelet count correlated well with other adverse clinicopathological factors and was associated with a significantly shorter time to progression. Our study remains the first such study from the Indian subcontinent and has the second largest sample size from similar papers. CONCLUSION {#sec1-5} ========== Rising platelet count correlated significantly with advancing stage and grade of disease in RCC. Patients with thrombocytosis were significantly more likely to have advanced tumors at presentation, poorer histological features, and more rapid disease progression. Multivariate analysis shows TNM stage to be the only reliable predictor of time to progression and further prospective studies are needed to establish the utility of the platelet count in this setting, and in potentially predicting the need for adjuvant therapy. Source of Support: Nil Conflict of Interest: None declared.	{ "pile_set_name": "PubMed Central" }
CONTEXT AND CAVEATSPrior knowledgeIndividuals with MUTYH-associated polyposis, a recessively inherited disorder, have a lifetime risk for colorectal cancer that approaches 100%. It is not known whether specific histological and molecular genetic features of such cancers influence tumor behavior and patient survival.Study designCharacteristics and survival of European patients with MUTYH-associated polyposis colorectal cancer and matched population-based control patients with colorectal cancer were compared.ContributionThe survival was statistically significantly better for patients with MUTYH-associated polyposis colorectal cancer than for control patients, even after adjustment for differences in age, stage, sex, subsite, country, and year of diagnosis.ImplicationsProspective studies are needed to further investigate this survival difference between MUTYH-associated colorectal cancer patients and colorectal cancer patients from the general population and to investigate whether disease-specific interventions, such as timing and type of surgery and chemotherapy, are warranted.LimitationsSeveral types of bias are possible, including selection, lead-time, and length-time biases. Treatment was not reported for many patients in this study.From the Editors Worldwide, colorectal cancer accounted for about one million newly diagnosed cancers in 2002, representing approximately 10% of all new cancers ([@bib1]). Estimated 5-year survival for colorectal cancer is approximately 54% in Western Europe ([@bib1]). Tumors in patients with inherited cancer syndromes may arise through distinct molecular genetic pathways and show histological features that are different from those in most sporadic tumors. These differences might, at least in part, influence tumor behavior and patient survival. For instance, mismatch repair--deficient tumors (associated with Lynch syndrome or sporadic microsatellite instability) have been reported to have a decreased likelihood of metastasizing, and patients with such tumors have better survival than patients with sporadic colorectal cancer ([@bib2]--[@bib8]), although some reports have not confirmed this finding ([@bib9]--[@bib12]). In 2002, the first autosomal recessive inherited form of colorectal cancer, MUTYH-associated polyposis (Mendelian Inheritance in Man \#608456), was described ([@bib13]). MUTYH-associated polyposis is believed to be responsible for 0.3%--1% of all colorectal cancers ([@bib14],[@bib15]). The MUTYH protein is a base excision repair glycosylase that is involved in the repair of DNA damage resulting from the oxidation of guanine nucleotides. The oxidation product of guanine, 8-oxo-7,8-dihydro-2′-deoxyguanosine can mispair with adenine, leading to a transversion in which a G:C base pair is replaced with a T:A base pair. The MUTYH protein prevents these transversions by scanning the newly synthesized DNA strand for any mispaired adenines, with guanines or 8-oxo-7,8-dihydro-2′-deoxyguanosines, and excising them. The risk of colorectal cancer in individuals with biallelic MUTYH mutations is high. The penetrance of colorectal cancer in patients with MUTYH-associated polyposis at age 60 years was estimated to be 100% in one study ([@bib16]) and 43% in another ([@bib14]). We hypothesized that survival of patients with MUTYH-associated polyposis and colorectal cancer might differ from that of colorectal cancer patients from the general population because of the distinct mutational mechanism underlying MUTYH-associated polyposis. The purpose of this study was to compare survival between patients with MUTYH-associated polyposis colorectal cancer and matched control patients with colorectal cancer from the general population. Subjects and Methods ==================== Study Population ---------------- This multicenter study was collaboration between three research groups from the Institute of Human Genetics (University of Bonn, Bonn, Germany), the Institute of Medical Genetics (School of Medicine, Cardiff University, Cardiff, United Kingdom), and the Department of Clinical Genetics (Leiden University Medical Center, Leiden, the Netherlands). The study population contained 147 patients with MUTYH-associated polyposis colorectal cancer and 272 matched patients with colorectal cancer from the general population. Informed consent was obtained according to protocols approved by the appropriate national and/or local ethic review boards (the Multi-Centre Research Ethics Committee for Wales, ref. 06/MRE09/19; University of Bonn Ethics Review Board No. 063/04; and Leiden University Medical Center Ethics Review Board No. P01.019). The Patients with MUTYH-associated polyposis were all biallelic MUTYH mutation carriers and included 113 index patients and 34 of their affected siblings. Siblings were selected and tested for MUTYH mutations in case they had developed colorectal cancer and/or polyps. Genotyping was performed as described previously ([@bib17]--[@bib19]) \[see the Leiden Open Variation Database database for all reported MUTYH mutations ([@bib20])\]. The time of diagnoses ranged from June 15, 1967, through August 13, 2001, for Dutch patients; from October 15, 1977, through March 10, 2006, for German patients; and from February 12, 1970, through February 14, 2006 for patients from the United Kingdom. Colon cancer was defined by use of the code C18 and rectal cancer was defined by use of the codes C19--C20, according to the International Classification of Diseases for Oncology, Edition 3 ([@bib21]). Tumor localization was categorized by the following anatomical subsites: proximal colon (consisting of the cecum, appendix, ascending colon, hepatic flexure, transverse colon, and splenic flexure; C18.0--C18.5), distal colon (consisting of descending colon and sigmoid; C18.6--C18.7), colon not otherwise specified (C18.8--C18.9), and rectum (consisting of rectosigmoid and rectum; C19.9--C20.9). Tumor stage was classified according to pathological TNM stage ([@bib22]). When the pathological stage was unknown, clinical stage was used. For most patients in this study, treatment information was not known and could, therefore, not be included as a determinant influencing survival. Year of diagnosis was used as a proxy of treatment because treatment changed during the study period. Survival time was defined as the time from the date of diagnosis until death or the end of the study (July 1, 2006). Patients who were still alive at the end of the study were censored on July 1, 2006. The control patients from the general population were patients who were diagnosed with colorectal cancer and whose data were derived from the Saarland Cancer Registry in Germany, the Eindhoven Cancer Registry in the Netherlands, or the Northern and Yorkshire Cancer Registry and Information Service in the United Kingdom. The Saarland Cancer Registry is the only population-based cancer registry in Germany, and it has provided internationally accepted high-quality data throughout the past 35 years ([@bib23]). Saarland is a state located in southwestern Germany with a population of approximately 1.1 million or approximately 1.3% of the total German population. The population structure and the health-care system in Saarland are very similar to Germany as a whole. The Eindhoven Cancer Registry is the oldest population-based cancer registry in the Netherlands that collects data from an area of 2.4 million inhabitants in southern Netherlands ([@bib24]). The Northern and Yorkshire Cancer Registry and Information Service is one of the 11 UK registries and collects data from a population of 6.6 million in the center of the United Kingdom. We aimed to select two control patients with colorectal cancer for each patient with MUTYH-associated polyposis colorectal cancer who were matched for country, stage at diagnosis, age at diagnosis, year of diagnosis, and cancer subsite. The age of diagnosis in the matched German and Dutch control patients was between 7 years younger and 7 years older than that in the case patient. The age of diagnosis in the matched UK control patients was between 4 years younger and 4 years older than that in the case patient. Cancer subsite was defined as either colon or rectum for German and Dutch control patients or as one of the first three characters of International Classification of Diseases for Oncology coding---C18, C19, and C20---for UK control patients. Patients with MUTYH-associated polyposis colorectal cancer from the United Kingdom were matched by the year of diagnosis for the period from January 1, 1996, through December 31, 2004. For UK patients with MUTYH-associated polyposis colorectal cancer who were diagnosed before 1996 (n = 19), we used control patients who were diagnosed in 1996 because the Northern and Yorkshire Cancer Registry and Information Service did not have data before 1996. Also, no control data were available for patients with MUTYH-associated polyposis colorectal cancer who were diagnosed after 2004, and so these patients were matched with control patients from 2004. We selected only control patients without second tumors because otherwise control patients might be included with a possible inheritable form of colorectal cancer that might influence the outcome of the survival analysis. Statistical Analysis -------------------- Differences in patient and tumor characteristics between patients with MUTYH-associated polyposis colorectal cancer and control patients were analyzed by use of the χ^2^ test. Survival analysis was performed with Kaplan--Meier curves and Cox regression. The Cox model accounted for the clustering effect of sibling pairs. Hazard ratios (HRs) and 95% confidence intervals (CIs) were produced with robust standard errors by comparing patients with MUTYH-associated polyposis with control patients. Regression analysis was adjusted for the matching variables (ie, age, period of diagnosis, site of colon tumor, center, and stage). Moreover, all analyses were adjusted for sex. Stratified analyses were performed by adjusting for the same set of variables (ie, age as continuous variable, period of diagnosis \[1967--1979, 1980--1989, 1990--1999, or 2000--2006\], site of colon tumor \[colon or rectum\], center \[Germany, United Kingdom, or the Netherlands\], stage \[I, II, III, or IV\], and sex \[male or female\]). We used STATA software, version 10.0 (Stata Corp LP, College Station, TX). The proportional hazard assumption of MUTYH-associated polyposis was evaluated by applying Kaplan--Meier curves. The effect of MUTYH-associated polyposis over time satisfied the assumption of proportionality because the graphs of the log\[−log(survival)\] vs log(survival time) resulted in graphs with parallel lines. All statistical tests were two-sided. Results ======= Crude survival for patients with MUTYH-associated polyposis was statistically significantly better than for control patients with colorectal cancer from the general population (log-rank test 5-year survival, P = .002) ([Figure 1](#fig1){ref-type="fig"}). Five-year survival was 78% (95% CI = 70% to 84%) for patients with MUTYH-associated polyposis colorectal cancer compared with 63% (95% CI = 56% to 69%) for control patients with colorectal cancer. ![Crude survival of patients with MUTYH-associated polyposis colorectal cancer and control patients with colorectal cancer in the United Kingdom, Germany, and the Netherlands (including a total of 419 participants, 147 patients with MUTYH-associated polyposis colorectal cancer and 272 control patients). Survival estimates and the corresponding 95% confidence intervals (gray dotted lines) for MUTYH-associated polyposis patients with colorectal cancer (black continuous line) and control patients with colorectal cancer (black dotted line). After adjustment for differences in age, stage, sex, subsite, country, and year of diagnosis, survival remained better for MUTYH-associated polyposis colorectal cancer patients than for control patients (hazard ratio of death = 0.48, 95% CI = 0.32 to 0.72, P \< .001).](jncidjq370f01_ht){#fig1} Perfect matching of all patients with MUTYH-associated polyposis and control patients was not feasible. There were some differences between patients with MUTYH-associated polyposis and control patients, including the number of positive lymph nodes (N stage), for which 69 (25%) of the 272 control patients had mismatches or missing information; whether metastasis occurred (M stage), in which 81 (30%) had mismatches or missing information; tumor subsite for which 33 (12%) had mismatches; exact year of diagnosis, for which 76 (28%) had mismatches; and sex, for which 128 (47%) had mismatches. Among the 272 control patients, there was a larger proportion of unknown N or M stage (39% or 106 patients) than among the 147 patients with MUTYH-associated polyposis (10% or 14 patients). In addition, patients with MUTYH-associated polyposis had statistically significantly more tumors located in the proximal colon (52% or 76 patients) than control patients (39% or 107 patients) (P = .015) and diagnosis before 1989 (19% or 28 patients vs 16% or 44 patients, respectively) (P = .046) ([Table 1](#tbl1){ref-type="table"}). ###### Characteristics of the total study population (n = 419), including 147 patients with MUTYH-associated polyposis colorectal cancer (MAP CRC) and 272 control patients with CRC Characteristic Patients with MAP CRC Control CRC patients Total population ------------------------------------ ----------------------- ----------------------------------------------------------------------------- ------------------- Male, No. (%) 82 (56) 124[\](#tblfn1){ref-type="table-fn"} (46)[†](#tblfn2){ref-type="table-fn"} 206 (49) Index patient, No. (%) 113 (77) 272 385 Siblings, No. (%) 34 (23) 0 34 Method of detection CRC Symptomatic, No. (%) 109 (74) 272 381 Surveillance, No. (%) 34 (23) 0 34 Unknown, No. (%) 4 (3) 0 4 Median age (range), y 54.0 (32.1--81.1) 52.1 (28.5--79.1) 53.1 (28.5--81.1) Location, No. (%) Proximal colon 76 (52) 107 (39)[‡](#tblfn3){ref-type="table-fn"} 183 (44) Distal colon 21 (14) 73 (27) 94 (22) Rectum 38 (26) 65 (24) 103 (25) Colon, not otherwise specified 12 (8) 18 (7) 30 (7) Unknown 0 (0) 9 (3) 9 (2) T stage, No. (%) 0 or in situ 6 (4) 6 (2) 12 (3) 1 15 (10) 33 (12) 48 (12) 2 22 (15) 38 (14) 60 (14) 3 50 (34) 118 (43) 168 (40) 4 8 (5) 18 (7) 26 (6) Unknown 46 (31) 59(22) 105 (25) N stage, No. (%) 0 85 (58) 136 (50) 221 (53) 1 33 (23) 54 (20) 87 (21) 2 15 (10) 29 (10) 44 (11) Unknown 14 (10) 53 (20) 67(16) M stage, No. (%) 0 125 (85) 161 (59)[§](#tblfn4){ref-type="table-fn"} 286 (68) 1 12 (8) 25 (9) 37 (9) Unknown 10 (7) 86 (32) 96 (23) Period of diagnosis, No. (%) 1967--1979 12 (8) 20 (7) 32 (8) 1980--1989 16 (11) 24 (9) 40 (10) 1990--1999 69 (47) 136 (50) 205 (49) 2000--2006 50 (34) 92 (34) 142 (34) Country, No. (%) Germany 55 (37) 106 (39) 161 (38) United Kingdom 42 (29) 66 (24) 108 (26) the Netherlands 50 (34) 100 (37) 150 (36) Unknown for six patients. P* = .046. χ^2^ test was used. All statistical tests were two-sided. P = .015. P \< .001. After adjustment for age, country, period of diagnosis, stage, subsite, and sex, risk of death was statistically significantly lower among patients with MUTYH-associated polyposis colorectal cancer than among control patients (HR = 0.48, 95% CI = 0.32 to 0.72, P \< .001) ([Table 2](#tbl2){ref-type="table"}). ###### Cox regression analysis for patients with MUTYH-associated polyposis (MAP) colorectal cancer (CRC) compared with control CRC patients[\](#tblfn5){ref-type="table-fn"} Analysis Hazard ratio (95% confidence interval) Hazard ratio (95% confidence interval) ------------------------------ ---------------------------------------- ---------------------------------------- Overall (MAP vs control CRC) 0.48 (0.32 to 0.72) 0.48 (0.32 to 0.73) Stratified Stage I or II 0.45 (0.23 to 0.91) 0.47 (0.21 to 1.04) III or IV 0.64 (0.34 to 1.20) 0.64 (0.33 to 1.22) Site of diagnosis Colon 0.42 (0.26 to 0.67) 0.40 (0.24 to 0.66) Rectum 0.48 (0.22 to 0.1.02) 0.55 (0.24 to 1.25) Country United Kingdom 0.66 (0.22 to 1.97) 0.76 (0.25 to 2.32) Germany 0.28 (0.10 to 0.74) 0.28 (0.10 to 0.82) the Netherlands 0.49 (0.31 to 0.79) 0.45 (0.27 to 0.74) Calendar period 1967--1989 0.49 (0.20 to 1.17) 0.40 (0.15 to 1.03) 1990--2006 0.51 (0.30 to 0.85) 0.52 (0.31 to 0.88) Data are from the Cox model with robust standard errors. The model was adjusted for the matching variables, age, country, period of diagnosis, stage, and site of colorectal cancer, and also for sex. Symptomatic patients are MUTYH-associated patients who underwent colon screening because of symptoms (eg, anemia, nausea, diarrhea, or blood in the stool) and had colorectal cancer at presentation. MUTYH-associated patients in whom colorectal cancer was detected during surveillance because of a positive family history or previously identified polyps were excluded from this analysis. When the analysis was stratified by stage, the survival benefit was higher among patients with stage I and II disease (HR = 0.45, 95% CI =0.23 to 0.91) than for stage III and IV disease (HR = 0.64, 95% CI = 0.34 to 1.20) ([Table 2](#tbl2){ref-type="table"}). The survival benefit was similar among patients with MUTYH-associated polyposis colorectal cancer whose tumor was in the colon (HR = 0.42, 95% CI = 0.26 to 0.67) and among those whose tumor was in the rectum (HR = 0.48 for rectum, 95% CI = 0.22 to 1.02). Increased survival was observed among patients with MUTYH-associated polyposis from all three countries (compared with control patients), with that for the German group being the highest. When the analysis was stratified by the period of diagnosis, similar survival benefits were observed for the period 1967--1989 (HR = 0.49, 95% CI = 0.20 to 1.17) and 1990--2006 (HR = 0.51, 95% CI = 0.30 to 0.85). In this study, colorectal cancer was detected during surveillance in 25 of the 113 index patients with MUTYH-associated polyposis and in nine of the 34 siblings with MUTYH-associated polyposis. Colon surveillance was initiated in these 25 index patients and nine siblings because of previously identified polyps that caused symptoms, including constipation, diarrhea, or blood in the stool (n = 16), or because of a family history of colorectal cancer, most often in a parent (n = 18). In four patients (three index patients and one sibling), the mode of detection of colorectal cancer was not known. When we excluded patients with MUTYH-associated polyposis colorectal cancer detected during surveillance from the analysis, we still observed statistically significant better survival among MUTYH-associated polyposis patients than among control patients with colorectal cancer ([Table 2](#tbl2){ref-type="table"}). Discussion ========== In a European cohort, survival of MUTYH-associated polyposis patients with colorectal cancer was statistically significantly better than that of control patients with colorectal cancer. This advantage in survival remained statistically significant after adjustments for age, stage, colon site, period of diagnosis, country, and sex. In a stratification analysis for early-stage (ie, stages I and II) vs late stage (ie, stages III and IV) cancers, the survival benefit for patients with MUTYH-associated polyposis colorectal cancer compared with control patients with colorectal cancer was slightly higher among patients with early-stage colorectal cancer (HR = 0.45, 95% CI = 0.23 to 0.91) than among those with later stage colorectal cancer (HR = 0.64, 95% CI = 0.34 to 1.20). To our knowledge, this is the first study to examine survival of MUTYH-associated polyposis patients with colorectal cancer compared with that of matched control patients with colorectal cancer from the general population. Patients with MUTYH-associated polyposis colorectal cancer were recruited from the largest MUTYH-associated polyposis cohort so far assembled. Given the retrospective character of the study, there are many possible biases and limitations that might lead to an overestimation or an underestimation of survival benefits (eg, selection, lead-time, and length-time biases). Treatment was not reported for many patients in this study. Selection Bias -------------- It can be expected that patients from families with several affected members who survived their cancer may be more likely to come to the attention of clinical geneticists than those from families in which all affected members died from their disease. Therefore, cohorts of patients who are recruited through genotyping studies could be biased toward those with better prognosis. This form of bias may have been operating in previous studies of Lynch syndrome--specific survival ([@bib9],[@bib25]). However, a number of observations are counter to this argument. First, Hampel et al. ([@bib26]) reported that index patients from Lynch syndrome families are younger at diagnosis of colorectal cancer than other mutation-positive patients in their family. Therefore, patients who come to medical attention through genetic testing do not necessarily have a milder phenotype. Second, although patients who die young or shortly after their diagnosis of colorectal cancer might not come to the immediate attention of clinical geneticists, the nonaffected members of their family may be referred for genetic counseling. MUTYH* genotyping can be done on DNA isolated from archived formaldehyde-fixed tumor tissue ([@bib27]) of a deceased patient or in the DNA isolated from blood of parents and/or siblings who are still alive. Third, patients with MUTYH-associated polyposis who have a relatively mild phenotype (eg, nonaggressive colorectal cancer at a later age) are likely be underrepresented in our cohort of patients with MUTYH-associated polyposis because the likelihood that they could have inherited a predisposition toward colorectal cancer may be lower. Another selection bias might be that patients with MUTYH-associated polyposis who have no polyps or only a few polyps (eg, 0--10 polyps) are likely to be underrepresented in our cohort, particularly when there is no family history of colorectal cancer in a sibling. MUTYH mutation screening in population-based colorectal cancer patients has shown that one-third of biallelic mutation carriers with colorectal cancer have no or only few polyps (eg, 0--10 polyps) ([@bib15],[@bib28]). Such patients are less frequently referred for molecular genetic analysis than patients with more florid forms of polyposis (eg, more than 10 polyps, numerous or multiple polyps). It is not known whether prognosis of patients with no or only few polyps (eg, 0--10 polyps) differs from that of patients with MUTYH-associated polyposis with colorectal cancer and polyposis. Finally, for UK patients with MUTYH-associated polyposis colorectal cancer who were diagnosed before 1996 (n = 19), we used control patients who were diagnosed in 1996 because the Cancer Registry did not have data before 1996. This procedure could have lead to better survival in control patients because treatment of cancer is expected to have improved between 1970 and 1996, the period in which the 19 UK case patients were diagnosed with colorectal cancer. However, after adjustment for date period in the multivariable Cox regression analysis, results remained unchanged. Lead-Time and Length-Time Bias ------------------------------ Heightened awareness among and surveillance of high-risk patients lead to diagnoses at an earlier stage of disease (lead-time bias) and might account for an apparent survival advantage. Length-time bias is also a consideration if screening tests lead to detection of asymptomatic indolent tumors. Patients with MUTYH-associated polyposis who are enrolled in surveillance programs could gain a survival benefit by early detection. As expected, there are differences in stage between patients whose disease was detected by surveillance and those whose disease was detected symptomatically; stage I disease was diagnosed in 14 (41%) of the 34 patients during surveillance and in 21 (19%) of the 109 patients diagnosed symptomatically. It should be noted that, after adjustment for other factors including stage, exclusion of patients with MUTYH-associated polyposis diagnosed during surveillance, survival benefit (ie, hazard ratio) did not change ([Table 2](#tbl2){ref-type="table"}, second column). Other Possible Biases --------------------- Another explanation of the better survival of patients with MUTYH-associated polyposis compared with that of control patients might be that patients with MUTYH-associated polyposis receive more extensive surgery because they usually have more polyps. However, the overall survival of patients with MUTYH-associated polyposis might actually be worse because they are prone to develop multiple cancers. Indeed, 46 (31%) of the 147 patients with MUTYH-associated polyposis in this study actually had two or more colorectal cancers at the time of diagnosis or developed a second colorectal cancer later on in life. Immune Response Differences and Survival Advantage -------------------------------------------------- An active immune response (represented by a high number of tumor-infiltrating lymphocytes) is strongly associated with better survival rates in control patients with colorectal cancer ([@bib29]--[@bib31]). It has been proposed that the immune system of patients with high microsatellite instability and mismatch repair--deficient tumors might be more active than that of colorectal cancer patients in the general population, which would lead to better survival ([@bib32],[@bib33]). Because of a defect in the DNA repair, more mutant proteins are expected in the mismatch repair--deficient tumors than in sporadic colon tumors. As a result, more peptide fragments of mutant proteins might be presented at the cell surface of the mismatch repair--deficient cancer cells, which activate the immune system. Furthermore, the enhanced mutation rate in these tumors may also induce a mutation burden that is not compatible with tumor cell survival. We have previously shown ([@bib34]) that MUTYH-associated polyposis colorectal cancers share similar characteristics with mismatch repair--deficient cancers, including a preferential proximal location, a high rate of mucinous morphology, and an increased level of tumor-infiltrating lymphocytes. The disruption of MUTYH protein function in MUTYH-associated polyposis carcinoma cells might lead to more oxidative DNA damage and generation of mutant peptides that could be presented to cytotoxic T cells through the expression of HLA class I receptors. It has, indeed, been shown that loss of expression of HLA class I receptors has been frequently identified in MUTYH-associated polyposis colorectal cancers and in mismatch repair--deficient colorectal tumors ([@bib35],[@bib36],[@bib37]), indicating that these tumors may be subject to strong selective pressure that favors outgrowth of cancer cells that acquire an immune-evasive phenotype ([@bib37]). In conclusion, in this study, patients with MUTYH-associated polyposis colorectal cancer had statistically significantly better survival than matched control patients. The reasons for this difference remain unknown, but a compromised base excision repair system could render MUTYH-associated polyposis colorectal cancers more immunogenic than sporadic colorectal cancers, which are characterized predominantly by chromosomal instability. This survival difference may have implications for clinical decision making in relation to the timing and type of interventions required, such as surgery and chemotherapy. Future prospective studies are needed to confirm this survival difference between MUTYH-associated colorectal cancer patients and colorectal cancer patients from the general population. Funding ======= This study was supported by the Cancer Research Wales, the Wales Office of Research and Development through the Wales Gene Park; the German Cancer Aid (grant 106244); and the Dutch Digestive Diseases Foundation (grant MWO 0355). Neither of these funds had any role in the design and conduct of the study; collection, analysis, or interpretation of the data; the decision to submit the manuscript for publication, and the writing of the manuscript. The authors had full responsibility for all activities above. We thank the patients and families who agreed to participate in the study and all the doctors who assisted with recruitment. The authors also thank the Mallorca Group of European experts on hereditary gastrointestinal tumors for their helpful discussions. The authors have no conflicts of interest to declare.	{ "pile_set_name": "PubMed Central" }
Diabetes was originally identified by the presence of glucose in the urine. Almost 2,500 years ago it was noticed that ants were attracted to the urine of some individuals. In the 18th and 19th centuries the sweet taste of urine was used for diagnosis before chemical methods became available to detect sugars in the urine. Tests to measure glucose in the blood were developed over 100 years ago, and hyperglycemia subsequently became the sole criterion recommended for the diagnosis of diabetes. Initial diagnostic criteria relied on the response to an oral glucose challenge, while later measurement of blood glucose in an individual who was fasting also became acceptable. The most widely accepted glucose-based criteria for diagnosis are fasting plasma glucose (FPG) ≥126 mg/dL or a 2-h plasma glucose ≥200 mg/dL during an oral glucose tolerance test (OGTT) on more than one occasion ([@B1],[@B2]). In a patient with classic symptoms of diabetes, a single random plasma glucose ≥200 mg/dL is considered diagnostic ([@B1]). Before 2010 virtually all diabetes societies recommended blood glucose analysis as the exclusive method to diagnose diabetes. Notwithstanding these guidelines, over the last few years many physicians have been using hemoglobin A1C to screen for and diagnose diabetes ([@B3]). Although considered the "gold standard" for diagnosis, measurement of glucose in the blood is subject to several limitations, many of which are not widely appreciated. Measurement of A1C for diagnosis is appealing but has some inherent limitations. These issues have become the focus of considerable attention with the recent publication of the Report of the International Expert Committee that recommended the use of A1C for diagnosis of diabetes ([@B4]), a position that has been endorsed (at the time of writing) by the American Diabetes Association (ADA) ([@B1]), the Endocrine Society, and in a more limited fashion by American Association of Clinical Endocrinologists/American College of Endocrinology ([@B5]). This review will provide an overview of the factors that influence glucose and A1C testing. FACTORS CONTRIBUTING TO VARIATION IN RESULTS {#s1} ============================================ Before addressing glucose and A1C, it is important to consider the factors that impact the results of any blood test. While laboratory medicine journals have devoted some discussion to the sources of variability in results of blood tests, this topic has received little attention in the clinical literature. Factors that contribute to variation can conveniently be divided into three categories, namely biological, preanalytical, and analytical. Biological variation comprises both differences within a single person (termed intraindividual) and between two or more people (termed interindividual). Preanalytical issues pertain to the specimen before it is measured. Analytical differences result from the measurement procedure itself. The influence of these factors on both glucose and A1C results will be addressed in more detail below. GLUCOSE MEASUREMENT {#s2} =================== FPG {#s3} --- Measurement of glucose in plasma of fasting subjects is widely accepted as a diagnostic criterion for diabetes ([@B1],[@B2]). Advantages include inexpensive assays on automated instruments that are available in most laboratories worldwide ([Table 1](#T1){ref-type="table"}). Nevertheless, FPG is subject to some limitations. One report that analyzed repeated measurements from 685 fasting participants without diagnosed diabetes from the Third National Health and Nutrition Examination Survey (NHANES III) revealed that only 70.4% of people with FPG ≥126 mg/dL on the first test had FPG ≥126 mg/dL when analysis was repeated ∼2 weeks later ([@B6]). Numerous factors may contribute to this lack of reproducibility. These are elaborated below. ###### FPG for the diagnosis of diabetes ------------------------------------------------------------------------------------------ Advantages • Glucose assay easily automated • Widely available • Inexpensive • Single sample Disadvantages • Patient must fast ≥8 h • Large biological variability • Diurnal variation • Sample not stable • Numerous factors alter glucose concentrations, e.g., stress, acute illness • No harmonization of glucose testing • Concentration varies with source of the sample (venous, capillary, or arterial blood) • Concentration in whole blood is different from that in plasma • Guidelines recommend plasma, but many laboratories measure serum glucose • FPG less tightly linked to diabetes complications (than A1C) • Reflects glucose homeostasis at a single point in time ------------------------------------------------------------------------------------------ ### Biological variation {#s4} Fasting glucose concentrations vary considerably both in a single person from day to day and also between different subjects. Intraindividual variation in a healthy person is reported to be 5.7--8.3%, whereas interindividual variation of up to 12.5% has been observed ([@B6],[@B7]). Based on a CV (coefficient of variation) of 5.7%, FPG can range from 112--140 mg/dL in an individual with an FPG of 126 mg/dL. (It is important to realize that these values encompass the 95% confidence interval, and 5% of values will be outside this range.) ### Preanalytical variation {#s5} Numerous factors that occur before a sample is measured can influence results of blood tests. Examples include medications, venous stasis, posture, and sample handling. The concentration of glucose in the blood can be altered by food ingestion, prolonged fasting, or exercise ([@B8]). It is also important that measurements are performed in subjects in the absence of intercurrent illness, which frequently produces transient hyperglycemia ([@B9]). Similarly, acute stress (e.g., not being able to find parking or having to wait) can alter blood glucose concentrations. Samples for fasting glucose analysis should be drawn after an overnight fast (no caloric ingestion for at least 8 h), during which time the subject may consume water ad lib ([@B10]). The requirement that the subject be fasting is a considerable practical problem as patients are usually not fasting when they visit the doctor, and it is often inconvenient to return for phlebotomy. For example, at an HMO affiliated with an academic medical center, 69% (5,752 of 8,286) of eligible participants were screened for diabetes ([@B11]). However, FPG was performed on only 3% (152) of these individuals. Ninety-five percent (5,452) of participants were screened by random plasma glucose measurements, a technique not consistent with ADA recommendations. In addition, blood drawn in the morning as FPG has a diurnal variation. Analysis of 12,882 participants aged 20 years or older in NHANES III who had no previously diagnosed diabetes revealed that mean FPG in the morning was considerably higher than in the afternoon ([@B12]). Prevalence of diabetes (FPG ≥126 mg/dL) in afternoon-examined patients was half that of participants examined in the morning. Other patient-related factors that can influence the results include food ingestion when supposed to be fasting and hypocaloric diet for a week or more prior to testing. Glucose concentrations decrease in the test tube by 5--7% per hour due to glycolysis ([@B13]). Therefore, a sample with a true blood glucose value of 126 mg/dL would have a glucose concentration of ∼110 mg/dL after 2 h at room temperature. Samples with increased concentrations of erythrocytes, white blood cells, or platelets have even greater rates of glycolysis. A common misconception is that sodium fluoride, an inhibitor of glycolysis, prevents glucose consumption. While fluoride does attenuate in vitro glycolysis, it has no effect on the rate of decline in glucose concentrations in the first 1 to 2 h after blood is collected, and glycolysis continues for up to 4 h in samples containing fluoride ([@B14]). The delay in the glucose stabilizing effect of fluoride is most likely the result of glucose metabolism proximal to the fluoride target enolase ([@B15]). After 4 h, fluoride maintains a stable glucose concentration for 72 h at room temperature ([@B14]). A recent publication showed that acidification of the blood sample inhibits glycolysis in the first 2 h after phlebotomy ([@B16]), but the collection tubes used in that study are not commercially available. Placing tubes in ice water immediately after collection may be the best method to stabilize glucose initially ([@B2],[@B16]), but this is not a practical solution in most clinical situations. Separating cells from plasma within minutes is also effective, but impractical. The nature of the specimen analyzed can have a large influence on the glucose concentration. Glucose can be measured in whole blood, serum, or plasma, but plasma is recommended by both the ADA and World Health Organization (WHO) for diagnosis ([@B1],[@B2]). However, many laboratories measure glucose in serum, and these values may differ from those in plasma. There is a lack of consensus in the published literature, with glucose concentrations in plasma reported to be lower than ([@B17]), higher than ([@B16],[@B18],[@B19]), or the same as ([@B20]) those in serum. Importantly, glucose concentrations in whole blood are 11% lower than those in plasma because erythrocytes have a lower water content than plasma ([@B13]). The magnitude of the difference in glucose between whole blood and plasma changes with hematocrit. Most devices (usually handheld meters) that measure glucose in capillary blood use whole blood. While the majority of these report a plasma equivalent glucose value ([@B21]), this result is not accurate in patients with anemia ([@B22]) (unless the meter measures hematocrit). The source of the blood is another variable. Although not a substantial problem in the fasting state, capillary glucose concentrations can be 20--25% higher (mean of 30 mg/dL) than venous glucose during an OGTT ([@B23]). This finding has practical implications for the OGTT, particularly because the WHO deems capillary blood samples acceptable for the diagnosis of diabetes ([@B2]). ### Analytical variation {#s6} Glucose is measured in central laboratories almost exclusively using enzymatic methods, predominantly with glucose oxidase or hexokinase ([@B24]). The following terms are important for understanding measurement: accuracy indicates how close a single measurement is to the "true value" and precision (or repeatability) refers to the closeness of agreement of repeated measurements under the same conditions. Precision is usually expressed as CV; methods with low CV have high precision. Numerous improvements in glucose measurement have produced low within-laboratory imprecision (CV \<2.5%). Thus, the analytical variability is considerably less than the biological variability, which is up to 8.3%. Nevertheless, accuracy of measurement remains a problem. There is no program to standardize results among different instruments and different laboratories. Bias (deviation of the result from the true value) and variation among different lots of calibrators can reduce the accuracy of glucose results. (A calibrator is a material of known concentration that is used to adjust a measurement procedure.) A comparison of serum glucose measurements (target value 98.5 mg/dL) was performed among ∼6,000 laboratories using 32 different instruments ([@B25]). Analysis revealed statistically significant differences in bias among clinical laboratory instruments, with biases ranging from −6 to +7 mg/dL (−6 to +7%) at a glucose concentration of 100 mg/dL. These considerable differences among laboratories can result in the potential misclassification of \>12% of patients ([@B4]). Similarly, inspection of a College of American Pathologists (CAP) survey comprising \>5,000 laboratories revealed that one-third of the time the results among instruments for an individual measurement could range between 141 and 162 mg/dL ([@B26]). This variation of 6.9% above or below the mean reveals that one-third of the time the glucose results on a single patient sample measured in two different laboratories could differ by 14%. OGTT {#s7} ---- The OGTT evaluates the efficiency of the body to metabolize glucose and for many years has been used as the "gold standard" for diagnosis of diabetes. An increase in postprandial glucose concentration usually occurs before fasting glucose increases. Therefore, postprandial glucose is a sensitive indicator of the risk for developing diabetes and an early marker of impaired glucose homeostasis ([Table 2](#T2){ref-type="table"}). Published evidence suggests that an increased 2-h plasma glucose during an OGTT is a better predictor of both all-cause mortality and cardiovascular mortality or morbidity than the FPG ([@B27],[@B28]). The OGTT is accepted as a diagnostic modality by the ADA, WHO/International Diabetes Federation (IDF) ([@B1],[@B2]), and other organizations. However, extensive patient preparation is necessary to perform an OGTT. Important conditions include, among others, ingestion of at least 150 g of dietary carbohydrate per day for 3 days prior to the test, a 10- to 16-h fast, and commencement of the test between 7:00 [a.m.]{.smallcaps} and 9:00 [a.m.]{.smallcaps} ([@B24]). In addition, numerous conditions other than diabetes can influence the OGTT ([@B24]). Consistent with this, published evidence reveals a high degree of intraindividual variability in the OGTT, with a CV of 16.7%, which is considerably greater than the variability for FPG ([@B6]). These factors result in poor reproducibility of the OGTT, which has been documented in multiple studies ([@B29],[@B30]). The lack of reproducibility, inconvenience, and cost of the OGTT led the ADA to recommend that FPG should be the preferred glucose-based diagnostic test ([@B1]). Note that glucose measurement in the OGTT is also subject to all the limitations described for FPG ([Table 1](#T1){ref-type="table"}). ###### OGTT for the diagnosis of diabetes ------------------------------------------------------------------------------------------------------------------ Advantages • Sensitive indicator of risk of developing diabetes • Early marker of impaired glucose homeostasis Disadvantages • Lacks reproducibility • Extensive patient preparation • Time-consuming and inconvenient for patients • Unpalatable • Expensive • Influenced by numerous medications • Subject to the same limitations as FPG, namely, sample not stable, needs to be performed in the morning, etc. ------------------------------------------------------------------------------------------------------------------ A1C MEASUREMENT {#s8} =============== A1C is formed by the nonenzymatic attachment of glucose to the N-terminal valine of the β-chain of hemoglobin ([@B24]). The life span of erythrocytes is ∼120 days, and consequently A1C reflects long-term glycemic exposure, representing the average glucose concentration over the preceding 8--12 weeks ([@B31],[@B32]). Both observational studies ([@B33]) and controlled clinical trials ([@B34],[@B35]) demonstrate strong correlation between A1C and retinopathy, as well as other microvascular complications of diabetes. More importantly, the A1C value predicts the risk of microvascular complications and lowering A1C concentrations (by tight glycemic control) significantly reduces the rate of progression of microvascular complications ([@B34],[@B35]). ### Biological variation {#s9} Intraindividual variation of A1C in nondiabetic people is minimal ([@B36]) ([Table 3](#T3){ref-type="table"}), with CV \<1% ([@B37]). Variability between individuals is greater. Data derived from several investigators imply that A1C values may not be constant among all individuals despite the presence of similar blood glucose or fructosamine concentrations ([@B38]). Some investigators have termed this a "glycation gap" and proposed that there are differences in the rate of glycation of hemoglobin ("low and high glycators") ([@B39]). Studies of twins with type 1 diabetes support a genetic contribution to A1C values ([@B40]), and heritability of the glycation gap was observed in healthy female twins ([@B41]). However, the glycation gap is essentially a measure of A1C adjusted for fructosamine. Importantly, measurement of fructosamine, which is glycated albumin and protein, suffers from several limitations ([@B42]). In addition, some authors have questioned the statistical analysis (which is not standard) used in determining the glycation gap and noted the statistical tautology that the outcome is correlated with the residual from a regression ([@B43]). Importantly, the postulate of a glycation gap remains unsubstantiated by data because glycation rates cannot be measured accurately in vivo. In addition, the hemoglobin glycation index (difference between observed A1C and that predicted from blood glucose) is not an independent predictor of the risk of microvascular complications ([@B43]), and the possible clinical significance of the glycation gap is unclear. ###### A1C for the diagnosis of diabetes --------------------------------------------------------------------------------------------------- Advantages • Subject need not be fasting • Samples may be obtained any time of the day • Very little biological variability • Sample stable • Not altered by acute factors, e.g., stress, exercise • Reflects long-term blood glucose concentration • Assay standardized across instruments • Accuracy of the test is monitored • Single sample, namely whole blood • Concentration predicts the development of microvascular complications of diabetes • Used to guide treatment Disadvantages • May be altered by factors other than glucose, e.g., change in erythrocyte life span, ethnicity • Some conditions interfere with measurement, e.g., selected hemoglobinopathies • May not be available in some laboratories/areas of the world • Cost --------------------------------------------------------------------------------------------------- Accumulating evidence supports the hypothesis that race influences A1C. Initial studies in patients with diabetes reported statistically significant differences in A1C concentrations among races ([@B44]). While adjusted for factors that may influence glycemia, it remains possible that these differences may be due to variations in glycemic control. More compelling support was provided in NHANES III where Mexican Americans and blacks had higher average A1C values than whites ([@B45],[@B46]). Similar findings were observed in adults with impaired glucose tolerance in the Diabetes Prevention Program ([@B47]) and validated in a cross-sectional analysis of two studies ([@B48]). Collectively these data suggest that there are differences in A1C concentrations among racial groups. However, it is not clear that these changes have clinical significance. A1C was measured in the Atherosclerosis Risk in Communities (ARIC) study in 11,092 adults who did not have a history of diabetes or cardiovascular disease ([@B49]). Consistent with prior publications, blacks had mean A1C values 0.4% higher than whites. Nevertheless, race did not modify the association between the A1C value and adverse cardiovascular outcomes and death ([@B49]). Because follow-up revealed that blacks with biochemically defined incident diabetes were significantly less likely than whites to report having received a diagnosis of diabetes by a physician, the authors speculate that delays in diagnosis may explain the higher A1C values in blacks. The molecular mechanism underlying the racial and ethnic differences remains to be established. Possibilities include differences in rates of glucose uptake into erythrocytes, rates of intraerythrocytic glucose metabolism, rates of glucose attachment to or release from hemoglobin or erythrocyte life span ([@B50],[@B51]). Regardless of the mechanism, the variations in A1C concentrations are relatively small (≤0.4%), and no consensus has been reached on whether different cutoffs should be used for different races. ### Preanalytical variation {#s10} Most factors that alter FPG do not significantly affect A1C concentrations. Acute illness, short-term lifestyle changes (e.g., exercise), recent food ingestion, and sample handling do not significantly alter A1C values ([Table 3](#T3){ref-type="table"}). Importantly, whole blood samples are stable for 1 week at 4°C and for at least 1 year at −70°C or colder ([@B13],[@B52]). The interpretation of A1C depends on the erythrocytes having a normal life span. Patients with hemolytic disease or other conditions with shortened erythrocyte survival have a substantial reduction in A1C ([@B53]). Similarly, individuals with acute blood loss have spuriously low A1C values because of an increased fraction of young erythrocytes. False increases in A1C have been reported with some methods in patients with hypertriglyceridemia, hyperbilirubinemia, uremia, chronic alcoholism, or chronic ingestion of salicylates ([@B13]). Because most interferences are method specific, in many cases they can be overcome by selecting an appropriate method that is not subject to the interference. Individuals with iron deficiency anemia have increased A1C and fructosamine concentrations ([@B54]), both of which are reduced by therapy with iron ([@B54],[@B55]). A mechanism for the higher A1C was recently identified by the demonstration that malondialdehyde, which is increased in subjects with iron deficiency anemia ([@B54]), augments glycation of hemoglobin ([@B56]). However, the magnitude of the increase in A1C is probably small. Examination of 10,535 adults without self-reported diabetes in NHANES III revealed that while 13.7% of women had iron deficiency, only 4.74% and 0.48% had A1C ≥5.5% or ≥6.5%, respectively ([@B57]). Iron deficiency in women was associated with a small (odds ratio 1.39) yet significant greater odds of A1C ≥5.5% but not with greater odds of A1C ≥6.5%. Iron deficiency was rare in men (\<0.5%) ([@B57]). Nevertheless, it would seem prudent to correct the iron deficiency before measuring A1C in individuals with severe iron deficiency anemia. ### Analytical variation {#s11} There are ∼100 different methods used to measure A1C. The most widely used commercial methods use either antibodies (immunoassays) or cation-exchange chromatography (most commonly high-performance liquid chromatography) to separate the glycated (A1C) from the nonglycated hemoglobin ([@B24]). The National Glycohemoglobin Standardization Program (NGSP) has been instrumental in standardizing A1C testing among laboratories ([@B58],[@B59]), particularly (but not exclusively) in the U.S. The NGSP has markedly improved the performance of A1C testing ([@B58]). At the time of writing, the vast majority (93%) of clinical laboratories that participate in CAP surveys use methods with between-laboratory CVs \<5% ([www.ngsp.org](www.ngsp.org)). Within laboratory CVs for some methods are as low as \<0.5%. In addition, the International Federation for Clinical Chemistry (IFCC) developed a reference method using mass spectrometry (or capillary electrophoresis) for A1C measurement, which should result in international harmonization as it facilitates traceability to a metrologically sound accuracy base. It is important to emphasize that the IFCC method is technically complex, time consuming, and expensive and is not designed for routine analysis of patient samples. Hemoglobin variants affect some A1C measurements. The most common variants are HbS, HbE, HbC, and HbD. A1C measurement is not appropriate in subjects homozygous for HbS or HbC, with HbSC or with any other variant that alters erythrocyte survival. However, A1C can be measured accurately in individuals heterozygous for HbS, HbE, HbC, or HbD and in those with increased HbF, provided an appropriate assay is used ([@B53],[@B60]). Only ∼4% of the 3,378 clinical laboratories that participated in the 2010 GH2 College of American Pathologists survey (which measures A1C) use methods in which HbAS or HbAC has clinically significant interference. In addition, if the sample is analyzed by high-performance liquid chromatography method, careful inspection of the chromatogram usually reveals the aberrant peaks produced by the variant hemoglobin. The presence of a hemoglobin variant should be considered if A1C is \>15% or if a large change in A1C coincides with a change in laboratory A1C method ([@B53]). In these situations, hemoglobin electrophoresis should be performed. It is important to emphasize that, like any other test, A1C results that are inconsistent with the clinical presentation should be investigated. PERSPECTIVE {#s12} =========== Notwithstanding the use of glucose (FPG and/or the OGTT) as the "gold standard" for the diagnosis of diabetes for many years, glucose testing suffers from several deficiencies. The requirement that the subject be fasting at the time the blood is drawn is a considerable inconvenience. While our ability to measure glucose has improved, inherent biological variability can produce very large differences within and among individuals. In conjunction with lack of sample stability, which is difficult to overcome in clinical practice, these factors results in lack of reproducibility of glucose testing. A1C, which reflects chronic blood glucose values, is routinely used in monitoring glycemic control and guiding therapy. The significant reduction in microvascular complications with lower A1C and the absence of sample lability, combined with several other advantages ([Table 3](#T3){ref-type="table"}), have led to the recommendation by some organizations that A1C be used for screening and diagnosis of diabetes ([@B1]). Accumulating evidence suggests that racial differences in A1C values may be present, and the possible clinical significance of this needs to be determined. Importantly, A1C cannot be measured in certain conditions. Despite these caveats, A1C can be measured accurately in the vast majority of people. A comprehension of the factors that influence A1C values and the conditions where it should not be used will produce accurate and clinically meaningful results. The convenience of sampling at any time without regard to food ingestion makes it likely that measurement of A1C will result in the detection of many of the millions of people with diabetes who are currently undiagnosed. See accompanying editorial, p. 524. No potential conflicts of interest relevant to this article were reported. D.B.S. researched data and wrote the manuscript. The author thanks M. Sue Kirkman of the American Diabetes Association for insightful comments and Rob Krikorian of Brigham and Women\'s Hospital for expert assistance with preparing the manuscript.	{ "pile_set_name": "PubMed Central" }
Literature in visual perception has identified that there are cross-cultural differences in visual perception \[[@B1]\]. Research comparing members of interdepended and collectivist East Asian cultures with independent and individualist European American cultures into picture perception showed that East Asians are more likely to attend the perceptual field as a whole and to focus on context and Westerns to focus on the salient foreground objects \[[@B1]\]. Research on cross-cultural differences has focused on investigating cross-cultural differences related to bottom-up information. Furthermore, research that experimentally manipulated the cultural norms of individualism and collectivism groups managed to attenuate cultural-specific preferences for social factors beneficial in human motivation \[[@B2]\]. Investigating the underlying mechanisms involved in these differences is very important as it can affect everyday tasks, advertisement and many other aspects of our everyday life. Here we present the first steps of this work, investigating the underlying processes in cross-cultural differences using computational modelling studies. The computational model is based on the spiking Search over Space and Time (sSoTS) model \[[@B3]\], that has been used to simulate Visual Attention task. sSoTS has incorporated mechanisms that allows us to investigate both bottom-up and top-down processes. We show that sSoTS can successfully simulate cross-cultural differences in Visual attention involving bottom-up tasks. Moreover, we expand the studies by making predictions from the computational modelling studies for cross-cultural differences and top-down tasks. Acknowledgements ================ The authors would like to acknowledge the use of the Advanced Research Computing (ARC) in carrying out this work.	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== Over the past 3 decades the north-western Pakistani province of Khyber Pakhtunkhwa, with Peshawar as its capital, has been affected by political instability, economic uncertainty, regional conflict, and continuous daily violence \[[@CR1]\]. In such settings, common mental disorders are highly prevalent \[[@CR2]\]. In Pakistan alone, the estimated prevalence of common mental disorders ranges between 10 and 16 % \[[@CR3]\] and major depression is one of the major causes of disability \[[@CR4]\]. These mental health conditions impair basic functioning required for survival and livelihoods. Accordingly, there is a great need for effective psychosocial interventions that can reduce symptoms of common mental disorders and improve daily functioning. A major obstacle to addressing the mental health challenges in many humanitarian settings is that they occur in low and middle income countries (LMICs). LMICs typically lack sufficient numbers of specialized mental health care professionals, have poor systems to deliver evidence-based interventions, and cannot afford costs associated with intensive treatment programs \[[@CR5]\]. Although psychological interventions, such as cognitive behavioural therapy (CBT), are effective in the treatment of common mental disorders that are evident in the wake of adversity \[[@CR6], [@CR7]\], these typically require expert mental health professionals providing treatments that are usually lengthy and costly to the health service. Overall, there is little evidence for treating mental health conditions in humanitarian settings \[[@CR2]\], and the trials that have been conducted have focused on treatments for posttraumatic stress disorder (PTSD) \[[@CR2]\]. Limitations of this approach are that PTSD treatments are usually relatively intensive and lengthy, and individuals affected by adversity and trauma typically have a range of other problems than PTSD, including depression and anxiety \[[@CR2]\]. For interventions to be scalable in LMIC settings, they should be of short duration and sufficiently simple that they can be carried out by lay people in the community \[[@CR8]\], and they should address a broad range of mental and psychosocial health problems relevant to communities affected by adversity. To meet these demands, WHO has developed a low-intensity 5-sessions program termed Problem Management Plus (PM+) which may be delivered by trained lay people \[[@CR9]\]. This program is aimed at reducing symptoms of depression, anxiety, PTSD, stress in the wake of adversity and trauma. It comprises evidence-based techniques: of (a) problem solving, (b) stress management, (c) behavioural activation, and (d) accessing social support. Although developed following considerable international consultation, the WHO has adopted the position that implementation of the protocol should not occur until it has been proven to be effective via controlled trials in LMICs. To this end, this paper provides an overview of the trial protocol for the initial study of PM+ in the Peshawar District. The PM+ manual has been translated into Urdu and adapted to the local culture in the Peshawar District. Qualitative methods including free listing and key informant interviews \[[@CR10]\] identified priority problems and health concepts from local perspectives which were used to inform intervention adaptation. Subsequent to this process, this study aims to (a) evaluate the effectiveness of the locally adapted version of PM+ in Peshawar in reducing symptoms of common mental disorders; and (b) evaluate the cost-effectiveness of PM+ in this setting. Methods {#Sec2} ======= Design {#Sec3} ------ The study is a randomised controlled trial (RCT) that compares PM+ to treatment as usual (TAU) in primary care attendees in three primary care centres in Peshawar District, Pakistan. The primary outcomes are psychological distress in terms of states of anxiety and depression at 20 weeks after inclusion. Secondary outcomes are functional disability, PTSD symptoms, and cost-effectiveness. Participants {#Sec4} ------------ Participants will include primary care attendees who fulfil the following inclusion criteria: (a) score above 2 on a screening questionnaire for common mental disorders (General Health Questionnaire-21; GHQ-12) \[[@CR11], [@CR12]\] and (b) score above 16 on a screening questionnaire for functional impairments (WHO Disability Assessment Schedule 2.0; WHODAS) \[[@CR13]\]. Exclusion criteria are: (a) suicide risk as defined in the mhGAP Intervention Guide \[[@CR14]\]; (b) individuals with a severe mental disorder (e.g., psychotic disorders, substance dependence) or severe cognitive impairment (e.g., severe intellectual disability or dementia), based on definitions in the mhGAP Intervention Guide \[[@CR14]\]. Procedure {#Sec5} --------- The study will be carried out in three peri-urban primary healthcare centres (PHCs) in Peshawar District in Pakistan, overseen by staff of Lady Reading Hospital. All research assessments and PM+ sessions are conducted at the PHCs. The study flow diagram is presented in Fig. [1](#Fig1){ref-type="fig"}.Fig. 1Flow diagram Potential participants are informed about the study. Informed consent entails a two-step procedure: 1. Informed consent for screening and 2. Informed consent for taking part in the PM+ trial. The latter is only required for participants meeting inclusion criteria. For each step respondents who decide to participate will be asked to complete a written consent form. For participants who are illiterate, witnessed oral consent and a thumb print in lieu of a signature will be asked, in line with recommendations from WHO \[[@CR15]\]. The witness will be any staff member of at the PHC (except the medical officer or health care professional involved directly in the care of the patient in order to avoid potential explicit or implicit coercion); or any adult person not related to the participant who the participant is comfortable having present during consent. Following informed consent for screening, a research assistant will record demographic characteristics and participants will be administered the GHQ-12, the WHODAS, and the suicide screening questions. If they are not at imminent risk of suicide, do not meet other exclusion criteria, and score above both the cut-offs of the GHQ-12 and the WHODAS, they will be provided informed consent information about the RCT. At least 24 h later, a research assistant will conduct the pre-trial assessment, which will comprise the Hospital Anxiety and Depression Scale (HADS) \[[@CR16], [@CR17]\], the Patient Health Questionnaire (PHQ-9) \[[@CR18]\], the Harvard Trauma Questionnaire (HTQ) \[[@CR19]\] events list, Life Events List for Pakistan (LELFP) \[[@CR20]\], the PTSD Checklist for DSM-5 (PCL-5) \[[@CR21]\], the Psychological Outcome Profiles instrument (PSYCHLOPS) \[[@CR22]\] and the Service Receipt Inventory (SRI) \[[@CR23]\]. If participants are not selected because they score below the cut-offs for the GHQ-12 or the WHODAS, they will be provided feedback on their test outcomes and reasons why they are not eligible for the study will be explained to them. The post-intervention assessment (WHODAS, HADS, PHQ-9, PCL-5, HTQ events list, LELFP, PSYCHLOPS) will be scheduled 7 weeks after the pre-intervention assessment (i.e., 1 week after the 5^th^ PM+ session), and the follow-up assessment (WHODAS, HADS, PHQ-9, PCL-5, HTQ events list, LELFP, PSYCHLOPS, SRI) will be scheduled at 13 weeks after the post-intervention assessment (i.e., 20 weeks after inclusion, in line with the timing of the follow-up assessment for the PM+ participants). The participants who receive PM+ will also be administered the PSYCHLOPS by the lay-counsellor at the beginning of each PM+ session. All assessments will be delivered in interview format as many participants are expected to be illiterate. All instruments will be administered by independent assessors blind to the allocation status of the participants. All assessors have received a 2-day training covering administering the instruments, common mental disorders, general interview techniques, and ethical research conduct. Ongoing monitoring of assessors' competency will be conducted through regular supervision by the trial manager. Sample size {#Sec6} ----------- A total number of 346 participants will be included in this study. Since we are not aware of similar intervention studies that have been carried out in this population, and we expect the population to be heterogeneous with respect to the types of common mental disorders, we aimed for a relatively conservative estimate of a 50 % reduction in HADS overall score in the PM+ group as compared to a 30 % reduction in the control group at 3 months follow-up (this estimate was extrapolated from a trial involving lay workers in India) \[[@CR24]\]. This corresponds with an odds ratio for the PM+ program of 2.3. Power calculations suggest a minimum sample size of 133 participants per group (power = 0.95, alpha = 0.05, two-sided). Taking into account an expected 30 % attrition at 3 months follow-up, we aim to include a total number of 346 participants (173 in the PM+ group and 173 in the TAU control group). Randomisation {#Sec7} ------------- Randomisation will occur following pre-assessment. This will be conducted by an independent research assistant located off-site (Human Development Research Foundation, Islamabad, Pakistan) and not involved in the delivery of PM+, clinical supervision, assessments or other aspects of the day-to-day running of the study. Randomisation will be performed using computerized software on a 1:1 basis. Participants randomised into the PM+ condition will be allocated to a PM+ lay-counsellor by the same independent researcher who performed the randomisation. This lay-counsellor will plan five consecutive meetings with the participant, with the first session being scheduled no longer than 1 week after the pre-intervention assessment. PM+ {#Sec8} --- Developed by WHO, PM+ is a new, brief, psychological intervention program based on established problem-solving and behavioural therapy techniques \[[@CR9]\]. The program consists of 5 weekly sessions lasting 90 min. Session one orients participants to the program with motivational interviewing techniques to improve engagement, provides psychoeducation about common reactions to adversity, and teaches participants a basic stress management strategy. The latter strategy is practiced at the conclusion of every subsequent session to enhance learning. Session two addresses a participant-selected problem through the provision of problem solving techniques and introduces commencement of behavioural activation procedures. Sessions three and four continue to support participants' application of problem solving, behavioural activation, and relaxation exercises, and introduces strategies to strengthen social support networks. In session five, education about retaining treatment gains is given, all learned strategies are reviewed, and the program is finished. For a more detailed description of the development of the manual, see Dawson et al. \[[@CR9]\]. PM+ providers are male and female lay-counsellors who will have received 8 days of training in basic counselling skills and delivery of PM+ (inclusive of 6-day initial training and a 2-day refresher training before the start of the trial). Weekly supervision will be provided by three supervisors who have received an additional 1-day intensive training in supervisory techniques. In addition, supervisors will receive monthly supervision with the PM+ master trainer, a clinical psychologist (KD), to ensure treatment adherence and provide support. Treatment fidelity will be ensured by an independent assessor who has been trained in PM+. Independent assessors will also directly observe a sample of 10 % of lay-counsellor's sessions and will systematically assess which elements of the program have been carried out by the PM+ provider using a checklist. Treatment-as-usual (TAU) {#Sec9} ------------------------ TAU in primary healthcare centres in Peshawar for common mental disorders usually consists of no treatment, or a range of alternate treatment regimes, such as vitamin injections; evidence-based mental health care is currently not available in PHCs. For this study, participants in the TAU condition will be referred to their primary care physicians for treatment. These primary care physicians will receive the standard training in treatment of common mental disorders that is routinely taught by the Lady Reading Hospital. If, during this treatment or during the study's assessments participants in TAU arm show severe psychiatric disorders (e.g., psychosis) or problems (e.g., imminent suicidality) that require immediate specialist treatment and follow-up, they will be referred to a psychiatrist within the Lady Reading Hospital. Throughout the study, we will carefully keep track of the types and amount of support participants receive through the SRI instrument (see measures section below). Screening measures {#Sec10} ------------------ Risk of suicide and presence of severe psychiatric symptoms as defined by the International Statistical Classification of Diseases and Related Health Problems-10th revision (ICD-10) \[[@CR25]\] will be assessed using the assessment tools consistent with the mhGAP Intervention Guide \[[@CR14]\]. The WHODAS \[[@CR13]\] is a generic assessment instrument assessing health and disability. Simple to administer, it is applicable across all health states, including mental disorders, and across cultures. The WHODAS assesses difficulties people have due to their illness across six domains of functioning (cognition, mobility, self-care, getting along, life activities, and participation). Difficulties are scored over the last 30 days on a five-point Likert scale as none, mild, moderate, severe, or extreme. The 12-item interviewer administered version will be used in this study. The GHQ-12 \[[@CR11], [@CR12]\] assesses level of general psychological distress during screening. It consists of 12 questions that are scored on a four-point Likert scale ranging from 0 to 3. When used as a screening tool, the GHQ-12 is usually scored bi-modally (i.e.,-0-0-1-1)., and the total score ranges between 0 and 12, with higher scores representing higher levels of distress. In previous studies in Pakistan, cut-offs of 1 or higher and 2 or higher have been reported and used for determining clinical caseness of common mental health disorders \[[@CR12], [@CR26], [@CR27]\]. Primary outcomes {#Sec11} ---------------- In addition to the WHODAS, psychological distress in terms of states of anxiety and depression is measured using the HADS \[[@CR16], [@CR17]\]. The HADS is a well-established 14-item scale consisting of two sub-scales: HADS-A (anxiety, seven items, range 0--21) and HADS-D (depression, seven items, range 0--21). Higher scores indicate more anxiety and/or depression. The Urdu version of the HADS showed satisfactory reliability and validity \[[@CR17]\]. Secondary outcomes {#Sec12} ------------------ The PHQ-9 is nine-item instrument measuring presence and severity of depression during the past 2 weeks \[[@CR18]\]. The PHQ-9 questions are derived from the 16-item version. Participants rate their responses on a four-point Likert scale ranging from "not at all" to "nearly every day". The the PHQ-9 total severity score ranges from 0 to 27. The PHQ has been validated in Urdu \[[@CR28], [@CR29]\]. DSM-5 posttraumatic stress disorder (PTSD) symptoms during the past week will be measured using the PCL-5 \[[@CR21]\], which is a 20-item checklist. Items are rated on a 0--4 scale and add up to a total severity score of 80. The previous version of the PCL based on DSM IV PTSD symptoms, the PCL-C, has been used previously in Pakistan \[[@CR30]\] and has been found to have acceptable psychometric properties (Mushtaq, unpublished data, 2013). The PCL-5 will be adapted to ask for symptoms in the last week (rather than month) to enhance sensitivity to change. Finally, PSYCHLOPS \[[@CR22]\] assesses progress on problems for which the person seeks help. It consists of four questions that encompass three domains: problems (2 questions), functioning (1 question) and wellbeing (1 question). Participants are asked to give free text responses to the problem and function domains. Responses are scored on an ordinal six-point scale producing a maximum score of 18 (six points per domain). The PSYCHLOP version administered at posttreatment and follow-up also includes an overall valuation question (determining self-rated outcome ranging from "much better" to "much worse"). PSYCHLOPS has been validated in primary care populations across several countries \[[@CR31], [@CR32]\]. Other measures {#Sec13} -------------- To assess the experience of potentially traumatic events, part one of the HTQ \[[@CR19]\] will be administered. This includes 17 items describing a range of traumatic events, such as: "lack of food and water", "forced separation from family members", and "being close to death". Each event is rated as either present (1) or absent (0). The HTQ has been validated and applied in many countries, including Pakistan \[[@CR33]\]. Life events other than potentially traumatic events (e.g., loss of job, housing problems, financial difficulties, problems with the law, marital problems, bereavement, etc.) are assessed using a life events measure previously developed for the Pakistani population (LELFP \[[@CR20]\]). Life events are rated as either present (1) or absent (0). Economic evaluation {#Sec14} ------------------- The Service Receipt Inventory (SRI) assesses service utilization during the time preceding the assessment and related characteristics of people with mental disorders, as the basis for calculating the costs of care for mental health cost-effectiveness research \[[@CR23]\]. It has been previously used in Pakistan and India \[[@CR23], [@CR34]\]. Adverse events reporting {#Sec15} ------------------------ All adverse reactions and serious adverse events (SAEs) that are reported spontaneously by the participant or observed by the investigators or other staff members during the trial will be recorded by the research team and will be reported to the local independent advisory board. The chair of the advisory board will review SAEs within 48 h and the advisory board will review all AEs twice a month, where necessary determining any appropriate action in respect of ongoing trial conduct. The consent process includes informing participants whom they can contact if they become distressed or are displeased with any aspect of the study. Depending on the nature of the adverse event, follow up may require additional tests or medical procedures as indicated, and/or referral to the general physician or a medical specialist. All adverse events will be followed until specialist care (including referrals, additional tests or medical procedures) is in place for the client, or until a stable situation has been reached. The principal investigator will inform the participants and the local independent advisory board if anything occurs, when it appears to the project group that the disadvantages of participation may be significantly greater than was foreseen. Analysis plan {#Sec16} ------------- To determine comparability between the conditions at baseline, multiple planned comparisons will be conducted for continuous variables and chi-squared tests for categorical ones. Hierarchical linear modelling (HLM) analysis will be carried out to assess differential change over time in mean HADS anxiety, depression and total score, WHODAS, and PCL scores between groups. For each outcome, the effects of time of measurement, group, and the group-by-time interaction will be analysed. We will add the following covariates at baseline to examine subgroup effects: gender, education, and severity of symptoms. HLM presumes intent-to-treat analyses as HLM allows the number of observations to vary between participants and effectively handles missing data. Time (linear and quadratic), treatment condition, and their interaction will be included in the models. Fixed effects parameters will be tested at 95 % CI. The Level 1 model will represent within-patient change over time, and the Level 2 model will predict variation in within-patient change over time and encompass between-patient variables. We will also analyse aggregated health care costs, computed from costs of treatment (primary care, outpatient hospital visits, impatient admission, diagnostic tests and investigations, drug prescriptions), and the aggregated patient and family costs (number of days with reduced working hours, informal caregiving time by relatives or friends), and travel costs and time spent travelling to or waiting for consultations. Cost data will also be analysed using HLM, analysing effects of time of measurement, group, and the group-by-time interaction on aggregated health care costs and patient and family costs. Descriptive analyses will be carried out in SPSS and HLM analyses in Stata version 11.2. Across all analyses, two-tailed tests will be reported with Cronbach alpha = .05. Ethics {#Sec17} ------ The project has been approved locally by the Institutional Review and Ethics Board of the Postgraduate Medical Institute, Lady Reading Hospital, Peshawar, Pakistan and by the WHO Ethical Review Committee (Protocol ID: RPC627, January 20, 2015). Discussion {#Sec18} ========== This RCT examines the effectiveness and cost-effectiveness of PM+ in Peshawar, Pakistan. PM+ is a brief transdiagnostic psychological intervention delivered by paraprofessionals for people in low-income communities affected by adversity. It aims to fill an urgent need by providing an evidence-based low-intensity mental health intervention that is amenable to LMICs. A second RCT on the effectiveness of individually delivered PM+ is planned to be carried out in women affected by adversities in Nairobi, Kenya. A third RCT on the effectiveness of group delivered PM+ is planned to be carried out Swat, Pakistan. If proven effective, PM+ may not only be used in similar areas in Pakistan, but rolled out to other affected areas for further adaptation and testing in diverse humanitarian settings. The PM+ manual, if proven effective, will be published by WHO and will be made available (with accompanying training materials) for free on WHO's website. Atif Rahman and Mark van Ommeren share last authorship Competing interests The authors declare that they have no competing interests. Authors' contributions All authors were involved in the design of the study; MS, SF, RB, KD, SUH, AC, AR, & MvO were involved in drafting the study protocol for ethics review, and all authors were involved in drafting and commenting on the paper. The authors alone are responsible for the views expressed in this article and they do not necessarily represent the views, decisions or policies of the institutions with which they are affiliated. All authors read and approved the final manuscript. The study is funded by USAID, Washington, USA, and ELRHA, Research for Health in Humanitarian Crises (R2HC), London, UK. The funding bodies had no role in the design of the this protocol.	{ "pile_set_name": "PubMed Central" }
It is now common knowledge that we are in the midst of a worldwide epidemic of type 2 diabetes (T2D). This disease occurs disproportionately in African Americans, Hispanic Americans, and other minorities. T2D is a major cause of cardiovascular disease morbidity and mortality, and these unfavorable outcomes are greatly increased in the presence of T2D ([@B1]--[@B4]). Impaired glucose tolerance (IGT) is itself a risk factor for cardiovascular disease ([@B5]). Obesity and IGT are each major risk factors for T2D, with yearly incidence of T2D in individuals with IGT ranging from 3.9 to 8.7% in various American populations, with an average of 5.7% ([@B6]), and averaging 6.5% in a meta-analysis of global studies ([@B7]). Both insulin resistance and diminished β-cell function (insulin secretion) are crucial pathogenetic factors in the progression from IGT to T2D, with the proportional impact of these defects varying in different reports and populations ([@B8]). The rate of progression can be reduced by lifestyle modification leading to weight loss ([@B9]) and the use of metformin ([@B9]), the thiazolidinedione drugs pioglitazone (PIO) ([@B10]) and rosiglitazone ([@B11]), α-glucosidase inhibitors ([@B12]), sulfonylureas ([@B13]), and insulin ([@B14]). Selection of therapeutic agents could be enhanced by matching their dominant actions on insulin resistance and/or β-cell dysfunction to the relative dominance of these abnormalities in specific individuals or populations with IGT. For the above reasons, the study by DeFronzo et al. ([@B15]) in this issue is of particular interest. In the Actos Now for the prevention of diabetes (ACT NOW) study ([@B10]), which included 602 high-risk IGT patients, PIO reduced the progression to T2D over 2.4 years from an annual incidence of 7.6% in the placebo (PLAC) group by 72% (95% CI 51--84%); 48% of the PIO group reverted to normal glucose tolerance (NGT) compared with 28% in the PLAC group (P \< 0.001). In the current report, 441 (73%) of these subjects had oral glucose tolerance tests at baseline and at study completion, with sampling of plasma glucose, insulin, and C-peptide. The protocol permitted calculation of insulin sensitivity by the Matsuda Index ([@B16]) as well as insulin secretion rate using plasma insulin and deconvolution of plasma C-peptide ([@B17]). Insulin sensitivity increased 92% in the PIO group compared with 17% in the PLAC group (P = 0.002), whereas insulin secretion rate increased 33% in the PLAC group compared with 11% in the PIO group (P \< 0.001). The Disposition Index (insulin secretion rate/insulin resistance) increased 66% with PIO compared with 12% with PLAC (P \< 0.001). The focus of DeFronzo et al., however, was on the relative changes from baseline of these derived parameters of glucose metabolism. Subjects were grouped into three categories according to their follow-up oral glucose tolerance tests: progression to T2D (N = 15 PIO and 45 PLAC), no change in IGT status, and reversion to NGT. In subjects who reverted to NGT, insulin sensitivity was increased 96% with PIO compared with PLAC, whereas insulin sensitivity changed little in those who progressed to T2D in either group. Insulin secretion in response to glucose declined with PIO and PLAC in those who progressed to T2D. Of importance, the Disposition Index fell slightly and similarly in both these groups. In an analytical tour de force, a plot of incidence of T2D in the combined PIO and PLAC groups was negatively associated with octiles of the decrease in the natural logarithm of the Disposition Index with an extraordinary correlation coefficient of 0.99. The use of the somewhat unusual octiles reflected a curvilinear association between the variables. While this exceedingly strong association was driven by an outlier value at the lowest octile of the Disposition Index that corresponded to the highest incidence of diabetes, the remaining 7 points yielded a visually convincing but much more shallow curvilinear association. IGT subjects who experienced a \>80% improvement in the Disposition Index had an approximate 2% incidence of diabetes whereas the incidence was 14% in those whose Disposition Index declined 60--80%. Better final glucose tolerance status (NGT \> IGT \> T2D) was categorically associated with increase in insulin sensitivity (odds ratio \[OR\] = 0.61), insulin secretion (OR = 0.61), and Disposition Index (OR = 0.26). In multivariate analyses of factors associated with end of study outcome, NGT and IGT data were combined and compared with T2D. In the PIO group, the combined favorable outcome was associated with a 21% increase in insulin secretion rate, and a 70% increase in insulin secretion in response to glucose. There was a 109% increase in insulin sensitivity comparing only the group who reverted from IGT to NGT to the group that did not. In assessing all of these impressive results, several points should be considered. The authors strongly assert that the overall conclusions indicate that improvement in β-cell function, i.e., the Disposition Index, is the dominant factor associated with final glucose tolerance status, and that the beneficial effect of PIO on reduced progression to T2D or increased reversion to NGT is largely attributable to the thiazolidinedione effect on this factor (OR = 0.26). This conclusion is consistent with most previous reports showing that insulin resistance is a factor in the earlier stages of IGT ([@B8],[@B18]) whereas deteriorating insulin secretion in response to glucose is a later determinant of worsening glucose tolerance and severity of T2D ([@B19]). Hence, the decrease in conversion rate of IGT to T2D reflects less deterioration of β-cell function in the PIO group. However, the Disposition Index incorporates both of these functional parameters of glucose metabolism, and increases in both insulin sensitivity and insulin secretion were individually associated with the final impact of PIO therapy. This should not be surprising. Although the temporal sequence of insulin resistance and diminished β-cell responsiveness cited above is widely accepted and appears logical, this does not mean that they are independent events. Several animal models support a homeostatic relationship between insulin sensitivity and insulin secretion ([Fig. 1](#F1){ref-type="fig"}). Decreased sensitivity to insulin elicits production, especially by the liver, of betatrophin, a molecule that stimulates proliferation of β-cells, expansion of β-cell mass, and improvement in glucose tolerance in experimental animals ([@B20]). Moreover, insulin resistance---specifically at the level of the β-cell---impairs first phase insulin secretion in response to glucose ([@B21]). ![Established and speculative beneficial effects of PIO.](3663fig1){#F1} Thus, there are two ways to interpret PIO's effects on the Disposition Index and its individual components ([Fig. 1](#F1){ref-type="fig"}). The most straightforward is that PIO is a thiazolidinedione and has positive actions on both insulin sensitivity in its target tissues as well as on β-cell function ([@B22]). Plasma glucose homeostasis is thereby maintained at an acceptable level because PIO only rarely causes hypoglycemia. A second speculative construct is to posit a brain regulating center with overall control of total body insulin action ([Fig. 1](#F1){ref-type="fig"}). When there is need for more total body insulin action on all its target tissues, in addition to its effects on glucose metabolism, this center would simultaneously increase insulin sensitivity and insulin secretion. Of course, there would be need for a negative feedback plasma/intracellular glucose brake on the output of such neurons to prevent hypoglycemia. PIO has been shown to affect glucose metabolism in brain cells ([@B23]). The possibility that the ultimate seat of T2D could lie in the brain has also been suggested before ([@B24]), given the importance of insulin in regulating energy expenditure and storage through central as well as peripheral effects. PIO could presumably act by stimulating the output of such putative neurons. These two alternative mechanisms of PIO action, peripheral and central, need not be mutually exclusive. These speculative ruminations aside, the main importance of the observations by DeFronzo et al. may be to encourage further development of T2D treatments that have ameliorative effects on glucose dysmetabolism, but less severe and less frequent adverse effects on the risks of congestive heart failure ([@B25]), fractures ([@B26]), and possibly bladder cancer ([@B27]). Their observations on the natural history of progression or reversion of IGT could also be applied to modification or development of other therapeutic agents. From bedside to bench and back to bedside should be a productive therapeutic path to follow. See accompanying original article, p. 3920. No potential conflicts of interest relevant to this article were reported.	{ "pile_set_name": "PubMed Central" }
![](edinbmedj74775-0081){#sp1 .271} ![](edinbmedj74775-0082){#sp2 .272} ![](edinbmedj74775-0083){#sp3 .273} ![](edinbmedj74775-0084){#sp4 .274}	{ "pile_set_name": "PubMed Central" }
![](amjdentsci81223-0045){#sp1 .285} ![](amjdentsci81223-0046){#sp2 .286}	{ "pile_set_name": "PubMed Central" }
Introduction {#Sec1} ============ Anterior cruciate ligament reconstruction (ACLR) surgery is a common procedure to improve functional stability after an anterior cruciate ligament (ACL) injury. The successful rate of primary ACLR is approximated to be 75% to 97% among patients^[@CR1],[@CR2]^. This means that almost one-fourth of primary ACLR surgeries bound to fail despite improved surgical methods^[@CR3]^. The causes associated with failed ACLR surgery are surgical factors and repetitive injury to the knee^[@CR4],[@CR5]^. Among these patients, approximately 2.9 to 8.9% underwent the following revision ACLR surgery^[@CR6],[@CR7]^. Compared to primary ACLR surgery, revision surgery is more technically demanding. In revision surgery, accessibility to original tunnels is a key factor of success since the removal of hardware is often required and bone grafting is likely to be necessary^[@CR8]^. This contributes to an estimated 35% failure rate of revision ACLR surgery, which translated to approximately 54% of patients returning to pre-injury levels^[@CR9]^. Therefore, with these technical difficulties, the success rate of revision ACLR surgery is commonly less compared to primary ACLR surgery^[@CR10]^. In addition to regain the joint stability, the goal of ACLR surgery is also to recover the function of the knee joint and muscle strength^[@CR10]^. To maximize the outcomes of surgery and enhance the functional recovery, an appropriate rehabilitation program is incorporated into the routine postoperative care. The impaired knee functions associated with ACL injury include instability in dynamic movement and quadriceps weakness. As being critical to dynamic joint stability, the quadriceps muscle weakness eventually leads to decreased knee function, and poor exercise performance, and may contribute early onset of osteoarthritis^[@CR11]--[@CR13]^. Thus, the strength of muscle around the knee joint, especially the quadriceps muscle, was the main target of training and keep indicator for monitoring the functional recovery after ACLR surgery^[@CR13]^. Thus, to evaluate the success of primary ACLR surgery, joint stability, muscle strength and knee function were common outcomes reported in previous studies^[@CR1],[@CR10],[@CR11]^. However, there were limited studies which have assessed muscle strength recovery following revision ACLR surgery. Thus, we performed this study to assess the stability and functional recovery after primary and revision ACLR surgery under the scheduled rehabilitation program. This study aimed to (1) compare recovery of knee extensor muscle (quadriceps) and flexor muscle (hamstring) strength in primary versus revision ACLR surgery, and (2) compare clinical functional and stability outcomes in primary versus revision ACLR surgery. Results {#Sec2} ======= Demographics {#Sec3} ------------ There were 80 male participants included in this study. There were no statistically significant differences in age, height and body weight between primary and revision groups (p = 0.538, p = 0.105, p = 0.969 respectively) (Table [1](#Tab1){ref-type="table"}).Table 1Demographic characteristics of study subjects.PrimaryRevisionp valueGroup size (n)4040Age (years)31 (23--35)29.5 (24--38)0.538Height (cm)173.6 (171.1--176.0)171.2 (170.0--174.6)0.105Weight (kg)76.3 (72.1--81.7)76.3 (71.7--81.5)0.969Results are shown as median with interquartile range. Ligament stability {#Sec4} ------------------ The AP laxity between groups was statistically different (p = 0.0022) (Fig. [1](#Fig1){ref-type="fig"}). The median laxity was 2.0 mm (IQR, 1.0 mm--2.5 mm) in primary group and 3.0 mm (IQR, 2.0 mm--3.25 mm) in revision group. AP laxity between involved and healthy limb was also recorded. For the primary group, the difference was \<3 mm in 30 cases, 3--5 mm in 10 cases, and \>5 mm in none of the cases. For the revision group. The difference was \<3 mm in 7 cases, 3--5 mm in 28 cases, and \>5 mm in 5 cases. The degree of side-to-side laxity difference was significantly different between two groups (X^2^ test, p \< 0.0001) (Table [2](#Tab2){ref-type="table"}).Figure 1Box-and-whisker plot for anteroposterior laxity in the primary and the revision ACL reconstruction groups. The revision ACLR group had greater instability comparing with primary group. AP anteroposterior.Table 2Anteroposterior ligament laxity.LaxityPrimary, n (%)Revision, n (%)\<3 mm30 (75.0%)7 (17.5%)3--5 mm10 (25.0%)28 (70.0%)\>5 mm0 (0%)5 (12.5%)Results are shown as number with percentage.Chi-Squared test: p \< 0.0001. Isokinetic knee strength {#Sec5} ------------------------ Deficits in knee extensor strength were not significantly different between primary and revision ACL reconstruction groups, showing 18.0% (IQR, 13.0--28.0%) and 22.0% (IQR, 14.0--33.0%) respectively, at 60°/sec. The knee extensor also showed 17.0% (IQR, 7.0--24.5%) and 14.5% (IQR, 10.0--21.5%) deficit in primary and revision group, respectively at 180°/sec. Moreover, there were no differences in knee flexor strength between primary and revision ACLR groups, showing 9.5% (IQR, 1.5--14.5%) and 12.5 (IQR, 6.0--21.5%), respectively, at 60°/sec and 4.5% (IQR, −2.0--16.5%) and 11.5% (IQR, −0.5--15.0%), respectively, at 180°/sec (Table [3](#Tab3){ref-type="table"}).Table 3Isokinetic knee strength deficits.Isokinetic strength deficitsPrimary (%)Revision (%)p valueExtensor 60°/sec18.0 (13.0--28.0)22.0 (14.0--33.0)0.308Extensor 180°/sec17.0 (7.0--24.5)14.5 (10.0--21.5)0.931Flexor 60°/sec9.5 (1.5--14.5)12.5 (6.0--21.5)0.091Flexor 180°/sec4.5 (−2.0--16.5)11.5 (−0.5--15.0)0.343Results are shown as median with interquartile range. Knee functional score {#Sec6} --------------------- For primary reconstruction, the median IKDC subject score was 81.05 (IQR, 71.7--89.7) and the median Lysholm score was 95.0 (IQR, 89.0--99.0) at the last follow-up. For the revision reconstruction group, the median IKDC subject score was 80.4 (IQR, 67.85--83.85) and the median Lysholm score was 92.5 (IQR, 88.0--95.0) at the last follow-up. There were no differences in functional scores including IKDC score and Lysholm score for primary reconstruction versus revision surgery (p = 0.154, p = 0.324, respectively) (Table [4](#Tab4){ref-type="table"}).Table 4Functional scores.PrimaryRevisionp valueIKDC score81.05 (71.7--89.7)80.4 (67.85--83.85)0.153Lysholm score95.0 (89.0--99.0)92.5 (88.0--95.0)0.324IKDC: International Knee Documentation Committee.Results are shown as median with interquartile range. Discussion {#Sec7} ========== The principal findings of this study are patients with revision ACL reconstructed knees had a higher AP translation and a higher percentage of instability than those with primary ACLR surgery. However, after planned rehabilitation, patients in revision groups can have non-inferior results in the isokinetic knee strength and knee functional outcomes compared to those with primary ACLR surgery. Revision ACLR surgeries are known to show inferior results compared with those for primary ACLR surgeries with respect to postoperative instability, return to sports, and patients' satisfaction^[@CR2],[@CR10],[@CR14]--[@CR16]^. However, it is unclear that whether the poor clinical outcomes are from knee joint instability or other factors such as muscle strength. Gifstad et al. found an overall reduction in muscle strength for the injured knee compared with the uninjured knee in the patients with revised ACLR, which correspond to the decreased performance in various knee functional scores^[@CR10]^. Meanwhile, several studies showed loss of knee flexion strength only^[@CR17]--[@CR20]^ or both flexion and extension muscle power in patients with the revision ACLR surgery but not in those with the primary ACLR surgery^[@CR10]^. In the current study, though with higher degrees of joint instability, patients in the revision ACLR group still had similar performance in the knee flexion and extension muscle power, and had similar clinical outcomes when compared with those in the primary ACLR group. The ideal graft for ACLR surgery remains controversial. Nevertheless, the degree of joint instability is reported to be associated with graft laxity, whereas allografts can theoretically trigger immune responses and show relatively slower postoperative reformation^[@CR21],[@CR22]^. In a study comparing autografts and allografts in ACLR surgery, a significant difference of ≥3 mm was found in ligament laxity measurements in 14.9% of autografts and 31.1% of allografts^[@CR23]^. In contrast, there was no difference in laxity between the revision group and primary group when only autologous hamstring tendon was used as the graft material^[@CR24]^. Accordingly, in this study, we used double-looped semitendinosus and gracilis autografts for primary ACLR compared to tibialis anterior tendon allografts for revision ACLR. Considering significant laxity as side-to-side difference more than 5 mm in AP laxity^[@CR25],[@CR26]^, none in primary group and five patients of the revision group (12.5%) showed laxity of ≥5 mm in this study. As such, we found a statistically significant difference in the joint laxity between two groups. To sum up, this difference between the two groups may possibly attribute to the different graft materials used. Muscle strength is a factor greatly influencing knee function. Therefore, it is important to accurately determine the level of muscular strength during postoperative recovery. The quadriceps muscle plays a key role in maintaining dynamic stability of the knee joint, whereas hamstring muscles act as an agonist to the ACL preventing anterior dislocation of the knee^[@CR27]^. Atrophy of these two muscles causes more functional instability and impairment. Although isokinetic equipment is widely used for accurate and objective assessment of muscle strength in the knee joint^[@CR1],[@CR10],[@CR11]^, evidence on revision ACLR surgery using isokinetic equipment to assess muscle strength is lacking. In this study, we used isokinetic equipment to assess the degree of muscle strength recovery in the knee joint between the two groups at the same time point of 1 year postoperatively. The two groups had similar outcomes in isokinetic muscle strengths, which corresponded to similar results in functional scores. Uribe et al.^[@CR28]^ reported that the outcomes of revision ACLR, in that only about half patients returned to their pre-injury activity level which was not as good as primary ACLR, regardless of the type of graft materials. However, we observed similar results in extensor and flexor muscle defect rates between primary and revision ACLR groups. Moreover, there were also no differences in functional knee scores (IKDC and Lysholm score) between the two groups. That is, after planned training and rehabilitation, patients with revision ACR still can have good results in knee strength and functions. Our study had four limitations. First, there is the small number of revision ACLR cases, and all patients are males. The findings of our study cannot apply to the female patients. Further studies on females are needed. Second, the fixation methods of tendon graft were not identical with and between two groups, which cause inherent bias of this study. Third, the pre-injury status of each patient was hardly recorded, which might make the interpretation of findings difficult, especially muscle strength. However, we presented the data of isokinetic muscle strength deficit by comparing the healthy limb, which could minimize this potential bias. Fourth, we cannot address the impact of potential covariates including commitment meniscus injury and critical physical findings status due to incomplete recording and limited sample size. Further study with a large sample will validate the findings of comparison after adjustment of potential covariates. In conclusion, patients with revision ACLR surgery had significantly higher AP laxity than those with primary ACLR surgery but can have similar results in knee strength and functions after rehabilitation. More intensive rehabilitation is suggested for patients with revision ACLR surgery. Methods {#Sec8} ======= Subjects and demographics {#Sec9} ------------------------- This is a cross-sectional study with 80 male patients with arthroscopic ACL reconstruction surgery (40 primaries and 40 revisions) recruited from Samsung Medical Center between April 2013 and May 2016. Hamstring (semitendinosus-gracilis) tendon autografts were used in the primary ACL reconstruction group, while tibialis anterior tendon allografts were used in revision ACL reconstruction group. The grafts were fixed by a cortical suspensory device (ENDOBUTTON CL, Smith-Nephew, London, UK) for the femoral side and bioabsorbable interference screws with a post-tie for the tibial side. The exclusion criteria were that patient who had more than one revision ACL reconstruction surgery on the same knee, combined/multiple ligamentous injuries or previous surgery on the same lower limb. All participants provided informed consent prior to testing, and the study was approved by the Sungkyunkwan University School of Medicine Clinical Research Ethics Board (SMC 2016-09-074-001). The committee that approved the research, confirm that all research was performed in accordance with relevant guidelines/regulations. Informed consent was obtained from all participants and/or their legal guardians. Rehabilitation {#Sec10} -------------- The same postoperative rehabilitation program was used for primary and revision groups. At the first week right after the operation, the general light intensity of isometric exercises and range of motion (ROM) exercises were performed. The angle of knee flexion increased 15° each week in ROM exercise program. At the fourth week, the knee joint range of motion achieved at 90°, and reached 130° at the sixth week. The patients subjected to partial weight bearing for 4--6 weeks postoperatively and progressively switched to full weight-bearing. By the 6--12 weeks, the patient would achieve combined strength, endurance and balance exercise without pain. At six months, light running was allowed. By nine month after surgery, the patient could receive sports-related training if there were no problems such as effusion, pain, or knee instability. Assessment {#Sec11} ========== Anteroposterior (AP) knee translation was assessed via the KT-2000 arthrometer (MEDmetric, San Diego, CA, USA) with maximal manual tension and a knee flexion angle of 20° and 30 lb (134 N) anterior force at 12 months after surgery. The measurements were done three times, and a mean value was calculated. These examinations were done by the same personnel who have several years of arthrometry experience to minimize the errors of measurements. Side-to-side AP laxity difference between injured and noninjured knee was also recorded. At 12 months after index surgery, Isokinetic knee strength with peak extensor and flexor torque (Nm/kg) was evaluated via a CSMI dynamometer (CSMI Medical Solutions, Stoughton, MA, USA) whereby angular velocities of 60°/sec and 180°/sec were measured to determine isokinetic knee strength. After warming up on a cycle ergometer for 5 min, subjects were allowed to practice the test protocol for familiarization. Subsequently, patients allowed to perform the strength test 3 times with a 1 minute washout period between each test session. While doing the sessions, patients were encouraged to perform with the maximum effort by the examiner. Isokinetic concentric measurements of quadriceps and hamstring strength demonstrated excellent reliability^[@CR29]^. Both injured and non-injured side were measured for three times, and a mean difference was calculated as$$\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$strength\,deficit=(noninjured\,limb-injured\,limb/noninjured\,limb)\ast 100 \% $$\end{document}$$ Knee joint functional scores were evaluated at 12 months after index surgery using International Knee Documentation Committee (IKDC)^[@CR30]^ and Lysholm knee scales^[@CR31]^. The IKDC knee scale quantifies symptoms, sports activities, and function^[@CR30]^. The Lysholm knee scale quantifies pain, instability, locking, swelling, stair-climbing, squatting abilities, and need for support^[@CR31]^. Statistical analysis {#Sec12} -------------------- A priori calculated power \>0.80 at an alpha level equal to 0.05 was used to determine that a sample size of 35 for each group was necessary to determine in KT-2000 arthrometry value^[@CR32]^. All data statistical analysis was performed using SPSS version 13.0 (SPSS Inc., Chicago, IL, USA). Chi-Squared test and Mann-Whitney U test were used throughout to compare variables between two groups, with p \< 0.05 considered statistically significant. All data were reported with median and interquartile range (IQR). Data availability {#Sec13} ----------------- The datasets generated during and analysed during the current study are not publicly available due to policy of the Sungkyunkwan University School of Medicine Clinical Research Ethics Board but are available from the corresponding author on reasonable request. Liang-Tseng Kuo and Won Hah Park contributed equally to this work. Publisher\'s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. D.K.K. and W.H.P. designed the study and directed the project. D.K.K. and G.P. performed the test and recorded the data. D.K.K. and L.T.K. analyzed the data and prepared the tables and figures. D.K.K., G.P. and K.B.H.M.K. wrote the paper. D.K.K., G.P. and L.T.K. contributed to interpretation of the results. All authors provided critical feedback and helped shape the research, analysis and manuscript. Competing Interests {#FPar1} =================== The authors declare no competing interests.	{ "pile_set_name": "PubMed Central" }
Introduction {#sec1} ============ The choroidal blood supply to the outer retina is essential for the health of photoreceptors (Bill, [@ref3]; Yancey & Linsenmeier, [@ref98], [@ref99]). Choroidal vessels are innervated by parasympathetic, sympathetic, and sensory nerve fibers, which regulate choroidal vessel dilation and thereby choroidal blood flow (ChBF) (Bill, [@ref3]; Stone et al., [@ref89]; Cuthbertson et al., [@ref16], [@ref17], [@ref18]; Reiner et al., [@ref80]). We have previously described the central and peripheral components of the neural pathway in birds by which retinal information can reflexively increase ChBF (Gamlin et al., [@ref30]; Fitzgerald et al., [@ref25], [@ref26], [@ref27]; Reiner et al., [@ref76]; Cuthbertson et al., [@ref16]). This neural circuit arises from retinal ganglion cells ([Fig. 1](#fig1){ref-type="fig"}) that project to the contralateral visual suprachiasmatic nucleus (vSCN), which itself then projects to the medial part of the nucleus of Edinger--Westphal (EWM), mainly on the same side as the eye of origin of the circuit (Cantwell & Cassone, [@ref8],[@ref9]). The EWM in turn projects to the ipsilateral ciliary ganglion (CG), where it terminates as boutonal endings on choroidal neurons of the CG, which innervate choroidal blood vessels (Gamlin et al., [@ref30]; Reiner et al., [@ref75]; Reiner et al., [@ref76]; Cuthbertson et al., [@ref16]). The CG terminals in the choroid produce vasodilation using muscarinic-endothelial nitric oxide mechanisms (Zagvazdin et al., [@ref103], [@ref104]). This circuit is activated by retinal illumination (Fitzgerald et al., [@ref26]), and it may thus serve to match ChBF to retinal activity-dependent need.Fig. 1.Schematized horizontal views of midbrain and eye in pigeon showing the central (A) and peripheral (B) circuitry for the visual pathways to the nucleus of Edinger--Westphal (EW) that drive ChBF increases and pupil constriction. (A) The pathway shown with red lines depicts the crossed projection from the retina to the vSCN that, in turn, has a mainly contralateral projection to EWM, which controls ChBF via its ipsilateral projection to choroidal neurons of the CG, as depicted in (B). The pathway depicted with blue lines in (A) shows a crossed projection from the retina to AP, which then projects to the contralateral EWLcl, which controls the pupillary light reflex (PLR) via an ipsilateral projection to pupilloconstrictive neurons of the CG, as depicted in (B). (B) The peripheral circuitry controlling ChBF and PLR, with EW, the CG, and the eye all drawn in horizontal view. The subdivisions of EW project ipsilaterally via the oculomotor nerve to the CG, where the projection from EWM terminates on choroidal neurons that project to choroidal blood vessels. Projections from both the rostromedial part of lateral EW and from EWLcl terminate on ciliary neurons that project to the ciliary body and the iris, and control accommodation and the PLR, respectively. The subdivisions of EW are color-coded in (A and B), and the projections of each to the eye via the CG in (B) are as well. Other abbreviations: lateral subdivision of the nucleus of Edinger--Westphal (EWL); lateral reticular formation (LRF); optic tectum (TeO). EWM lesions or choroidal nerve transections reduce basal ChBF in birds (Shih et al., [@ref83]; Fitzgerald et al., [@ref26]), and prevent light-mediated reflexive increases in ChBF (Fitzgerald et al., [@ref26]), thereby likely causing chronic choroidal insufficiency. Consistent with this premise, we have found that destruction of EWM in young adult pigeons maintained in normal diurnal light (DL) greatly increases Müller cell expression of glial fibrillary acidic protein (GFAP) throughout the entire depth and topographic extent of the ipsilateral retina, up to a year post-lesion (Fitzgerald et al., [@ref24]; Kimble et al., [@ref105]). Moreover, EWM lesions in young adult pigeons reduce visual acuity at about one year after the lesions (Hodos et al., [@ref46]). Thus, our findings support the view that impairment of parasympathetic control of ChBF may harm the retina. Age-related decline in parasympathetic choroidal innervation has been observed in humans as well as in pigeons, in association with declines in basal ChBF (Grunwald et al., [@ref38]; Fitzgerald et al., [@ref27], [@ref28]; Ito et al., [@ref49]; Jablonski et al., [@ref50]). The precise consequences of impaired parasympathetic control of ChBF for the health of any particular retinal cell type have not, however, been established. Because of their critical dependence on ChBF, we conducted blinded quantitative analysis of photoreceptor outer segment abundance in pigeon retina following electrolytic destruction of EWM. Since it seemed possible that any protective benefit of ChBF regulation by EWM for photoreceptor health might be especially manifested under more stressful lighting conditions, we studied pigeons housed in both normal 400 lux 12 h light to 12 h dark cycle, as well as pigeons housed in constant 400 lux light. The results show that short wavelength (blue/violet) cones and principal cones are particularly vulnerable to ChBF deficiency caused by disrupted parasympathetic choroidal regulation, and constant light (CL) accentuates the vulnerability. Materials and methods {#sec2} ===================== Subjects {#sec2-1} -------- Fifty-three male and female White Carneaux pigeons (400--600 g), obtained from the Palmetto Pigeon Farm, the University of Maryland, or Duke University, were used. All animal studies were performed in accordance with a protocol approved by the Institutional Animal Care and Use Committee of the University of Tennessee Health Science Center (UTHSC) and complied with the National Institutes of Health and Society for Neuroscience guidelines, and the ARVO statement on the Use of Animals in Ophthalmic and Vision Research. Prior to the study, birds were maintained on a 12 h 400 lux light to 12 h dark photoperiod (12L--12D) in a fly cage, and had food and water access ad libitum. Forty-three of the animals received an electrolytic lesion that targeted either EWM or area pretectalis (AP). Ten pigeons received either a sham lesion or no surgery, and served as control birds. The EWM lesions were made to disrupt parasympathetic control of ChBF by the EWM circuit shown in [Fig. 1](#fig1){ref-type="fig"}. The AP lesions served as a control for the inadvertent but generally unavoidable effects of EWM lesions on the lateral pupil control part of EW (i.e., EWL), specifically the caudolateral part of EWL (EWLcl) that controls pupil constriction (Reiner et al., [@ref75]; Gamlin et al., [@ref31]). In principle, damage to the EWLcl that causes a fixed, dilated pupil could by itself result in light-mediated damage to the retina (Li et al., [@ref56]; de Raad et al., [@ref19]). To assess the effects of a fixed dilated pupil on the retina, the AP, which is the retinorecipient pretectal source of visual input to the EWLcl ([Fig. 1](#fig1){ref-type="fig"}) and the homologue of the mammalian olivary pretectal nucleus, was unilaterally destroyed in some pigeons (Reiner et al., [@ref75]; Gamlin et al., [@ref31]). Since AP receives luminance-related retinal input from the contralateral eye and in turn projects contralaterally to the pre-pupillomotor neurons of the EWLcl, AP mediates the pupillary light reflex for the contralateral eye and provides tonic drive to the caudolateral part of the contralateral EWL (Reiner et al., [@ref75]; Gamlin et al., [@ref31]). The destruction of AP, consequently, dilates the pupil and eliminates the pupillary light reflex in the contralateral eye, as does a lesion of the EWLcl ipsilateral to that eye. Lesions of the nucleus of EW or AP {#sec2-2} ---------------------------------- For electrolytic lesions, animals were deeply anesthetized with ketamine (66 mg/kg) and xylazine (6.6 mg/kg), and secured in a stereotaxic device. Pigeons ranged from 0.5--3 years of age at the time of lesion. The EW and AP were targeted using coordinates from the stereotaxic atlas of the pigeon brain of Karten & Hodos ([@ref52]). Body temperature was maintained using a Harvard heating blanket. Electrode placement in EW or AP was confirmed by monitoring the effects of electrical activation (100 Hz, 0.5 ms pulse duration, 40--100 µA pulse amplitude) via an insulated stainless steel electrode (AM Systems, Carlsborg, WA) on the ipsilateral pupil in the case of EW activation, and contralateral pupil in the case of AP activation. Upon electrode placement that yielded low threshold brisk pupil constriction, 1 mA constant anodal current was passed through the electrode for 30 seconds to destroy EW or AP. In a few cases, bilateral lesions were made. Animals were monitored for 24 h post surgery, and then placed in individual cages until fully recovered from surgery and anesthetic (postoperative analgesic was given). The birds were housed subsequently in either cyclic 12 h 400 lux to 12 h dark (12L/12D) for up to 16.5 months or in constant 400 lux for up to three weeks, using fluorescent lighting. Birds surviving for more than 4 months in DL were, in some cases, moved to fly cages with fluorescent lighting after individual cage housing. The abbreviation DL is used for the 12L--12D condition, and CL for the 24L--0D condition. Histology and immunohistochemistry---brain {#sec2-3} ------------------------------------------ The birds were deeply anesthetized and transcardially perfused with fixative consisting of 4% paraformaldehyde--0.1 [m]{.smallcaps} lysine--0.01 [m]{.smallcaps} sodium periodate in 0.1 [m]{.smallcaps} sodium phosphate buffer (pH 7.4). After perfusion, the brains were removed, cryoprotected in 20% sucrose--10% glycerol in 0.1 [m]{.smallcaps} sodium phosphate buffer, and sectioned at 40 µm in six series. One series was mounted on slides as brains were sectioned frozen on a sliding microtome, and stained with 0.2% cresyl violet to determine lesion accuracy. If needed to determine if the lesion damaged all or only part of EW, additional series of sections were processed for immunocytochemistry using an antibody against choline acetyltransferase (ChAT) to detect EW neurons, which are cholinergic (Reiner et al., [@ref76]). To determine if EWM or its input from the vSCN was destroyed, immunolabeling for substance P (SP), which is enriched in terminals of the vSCN input to EWM and thus delineates EWM and its input from the vSCN (Gamlin et al., [@ref30]), was also carried out on one or more series from animals in which the lesion was in the region of the tract from the vSCN to EWM. Histology---retina {#sec2-4} ------------------ After transcardial perfusion, eyes were also removed from the head, the corneas cut away, and the eyecups transferred to an electron microscopy grade fixative consisting of 2% gluteraldehyde--2% paraformaldehyde--0.05% acrolein in a 0.1 [m]{.smallcaps} sodium cacodylate buffer. The eyecups were stored in this solution until processed for plastic embedding, at which time they were divided into four quadrants (superior, temporal, nasal, and inferior), with the red field occupying the superior quadrant (Fitzgerald et al. [@ref27]). These quadrants were further cut into 2--3 separate wedges each for the central and peripheral retina, and rinsed in 0.1 [m]{.smallcaps} sodium cacodylate 3--5 times, immersed in 1% osmium tetroxide solution (in 0.1 [m]{.smallcaps} sodium cacodylate), dehydrated in an ascending series of alcohols, infiltrated in an increasing percentage of epoxy resin, and embedded in plastic molds. One-half to one-micron sections were obtained from blocks containing superior central retina using an Ultracut E (Reichert, Vienna, Austria), and stained with toludine blue/azure II solution. All analyses were conducted on the superior central retina, as in Fitzgerald et al. ([@ref27]), since it is the high acuity area in pigeon retina (Hodos et al., [@ref45]). Classification into lesion groups {#sec2-5} --------------------------------- Brain histological outcome and pupil light reflex assessment were used to categorize animals into lesion outcome groups ([Fig. 2](#fig2){ref-type="fig"}). Birds with lesions targeting EW or AP were categorized as a lesion miss if EW or AP was undamaged, if the lesion was without effect on the pupil light reflex, and if the lesion did not affect suprachiasmatic nucleus input to EWM (Gamlin et al., [@ref30]). Based on our analysis, birds were classified into three groups: (1) control birds that either had no surgery (nine birds), had a sham lesion (one bird), or had a lesion that missed the intended EW or AP target and had no effect on the pupil light reflex or input to EWM (two birds); (2) AP or EWLcl lesion birds that sustained an AP lesion (13 birds), or EW damage that only impaired the pupillary light reflex (i.e., destruction of EWLcl with less than 50% involvement of EWM) (six birds); and (3) EWM lesion birds that sustained greater than 90% EWM destruction irrespective of encroachment on EWLcl (22 birds). For the birds housed in 12L--12D lighting, there were 12 birds in the control category, 14 birds in the AP/EWLcl (eight AP and six EWLcl) category, and 15 in the EWM category. In the case of the AP lesions (all to the left AP), the contralateral eye (i.e., right) was used in analysis and is termed the experimental affected eye, while in the case of EW lesions, the ipsilateral eye was used for analysis and is termed the experimental affected eye. In the case of control eyes from birds housed in 12L--12D lighting, both eyes were used in the analysis for four birds. In these cases, data for the two eyes were averaged together per bird and used in analysis of group effects. In the text, table and graphs, we will refer to this group as control-DL. We did not use eyes contralateral to EWM lesions as controls because our prior study showed that EWM destruction slightly affects the contralateral eye (Kimble et al., [@ref105]). Three EWM birds among those housed in DL had bilateral EWM lesions. In these cases as well, data for the two eyes were averaged together and used in analysis of group effects. In the text, table and graphs, we will refer to the birds with EWM destruction housed in DL as the EWM-DL group. In the case of those birds with AP or EWLcl lesions housed in DL, we use the term AP-DL for simplicity, and because the EWLcl lesions are AP-like.Fig. 2.Images showing examples of lesions used to classify pigeons into groups. Image (A) shows a complete right EW lesion (EW-Lx) in a bird housed in 12L/12D DL, and for comparison image (B) shows normal EW on both sides of the brain from a bird that received a left AP lesion and was subsequently housed in 12L/12D DL. Images (C and D) show the left lesioned AP (AP-Lx) and right unlesioned AP, respectively, from the same bird as shown in (B). The lesions in (A and C) resulted in a fixed dilated pupil in the right eye, and a loss of the pupil light reflex. The images in E and F show sections immunostained for ChAT and SP, respectively, from a pigeon that received a left AP lesion. The images show that SP+ fibers, which arise from the vSCN, terminate in medial EW. Images (G and H) show immunostained sections for ChAT and SP, respectively, from a pigeon that received a bilateral EW lesion and was subsequently housed in 12L/12D DL. In this case, the immunostaining revealed a complete lesion on the left but substantial sparing of the right EWM and its input from the vSCN. Due to the complete destruction of the right EWL (with concomitant loss of the pupil light reflex) and the substantial sparing of the right EWM and its input from vSCN, the right eye for this bird was classified as AP-like. The magnification is the same in all images. For the CL condition, there were five birds in the AP category and seven in the EWM category. In the case of the AP lesions (all to left AP), the contralateral eye (i.e., right) was again the experimentally affected eye used in analysis, while in the case of EWM lesions, the affected ipsilateral eye was again used in the analysis. These groups will be referred to as AP-CL and EWM-CL, respectively. As we had no CL birds without either AP or EWM lesions, we used the eyes ipsilateral to the AP lesion (i.e., the left eye) as the control eyes. We expected this to serve effectively as a control eye since pupil control in pigeons is entirely crossed, and there was no pupil impairment in eyes ipsilateral to AP lesions. Moreover, we found no statistically significant differences between the left (i.e., unaffected) eyes of AP birds housed in 12L--12D lighting and control birds housed in 12L--12D lighting for any of our endpoints. We refer below to the left control eyes of the AP-CL birds as the control-CL group, for symmetry to the control-DL group. A summary of these groups is presented in [Table 1](#tab1){ref-type="table"}.Table 1.Summary of the groups, showing the number of birds per group, and the ages and survival times for the animals whose eyes are used in these groups.GroupNumber of pigeonsMean survival postlesion in monthsShortest survival postlesion in monthsLongest survival postlesion in monthsMean age at sacrifice in monthsYoungest age at sacrifice in monthsOldest age at sacrifice in monthsControl-DL126.590.5012.5026.39.560AP-DL148.230.5016.5026.87.554EWM-DL155.600.2516.2516.78.530Control-CL50.450.250.7510.39.510.8AP-CL50.450.250.7510.38.2511.8EWM-CL70.460.250.7510.59.510.8 Photoreceptor analysis {#sec2-6} ---------------------- The pigeon retina contains six different cone photoreceptor types as defined by their lipid droplet--photopigment combination and one type of rod photoreceptor ([Fig. 3](#fig3){ref-type="fig"}) (Mariani & Leurre-Dupree, [@ref106]; Bowmaker et al., [@ref5]). Cone oil droplets serve as narrow long-pass filters that together with the cone photopigment determine the spectral responses of the cone type (Bowmaker et al., [@ref5]; Vorobyev, [@ref96]; Hart & Vorobyev, [@ref43]). In our sections, although oil droplet color was not discernible, the size and darkness of the oil droplet, and its relative location in the inner or outer row of oil droplets served to distinguish photoreceptor outer segment types. Of the six cone types, two always occur as a gap junction-coupled pair termed the double cone, whose individual members are the principal cone and accessory cone (Smith et al., [@ref85]; Bowmaker et al., [@ref5]). We recognized principal cones by the location of their large, pale oil droplet in the outer row of oil droplets. The accessory cone possesses a very narrow outer segment with a small circular entity below the outer segment that some have called a clear oil droplet (Meyer & Cooper, [@ref64]; Meyer & May, [@ref65]; Mariani & Leurre-Dupree, [@ref106]; López-López et al., [@ref61]), and some have not (Morris & Shorey, [@ref66]; Bowmaker et al., [@ref5]; Hart et al., [@ref42]; Kram et al., [@ref54]). Because of its narrow outer segment, an accessory cone is not always evident in the same plane of section as its associated principal cone. Due to their thin outer segment and indistinct droplets, we did not count accessory cones. Among the remaining four single cone types, red cones have a large dark red oil droplet in the outer row of oil droplets, and green cones have a large, dark green oil droplet in the inner row of oil droplets. Thus, the outer segments of these two cone types can be readily distinguished by the size and laminar position of their oil droplets. The blue and violet cones, however, possess moderately sized, pale oil droplets in the inner oil droplet row (Vorobyev, [@ref96]). Thus, although we could identify blue cone and violet cone outer segments, we could not distinguish between them, and thus treat them here as one class. Finally, rod outer segments are easily recognized by their wide and untapered shape and lack of an oil droplet.Fig. 3.Schematic (A) illustrating the six different cone photoreceptor types as defined by their lipid droplet--photopigment combination, and the one type of rod photoreceptor present in the pigeon retina, as identified and characterized in prior studies noted in the text. Image (B) shows retinal photoreceptor outer segments in normal pigeon retina. Red cones (red R), principal cones (white P), green cones (green G), violet/blue cones (blue B), and rods (black R) are identifiable in B by the traits shown in (A). Accessory cones (black A) are also evident in (B). Schematic (A) is adapted and modified for pigeon from [Fig. 1](#fig1){ref-type="fig"} of Morris and Shorey ([@ref66]). Based on these characteristics, we quantified the outer segment abundance of principal cones, blue--violet cones, green cones, red cones, and rods. Images of the outer retina were captured at high magnification (400×), with at least 1 mm of retina photographed for each eye and each animal. Blinded analysis was then conducted on coded images of the abundance of the outer segments of the different photoreceptor types. To avoid planar counting artifacts, we only counted photoreceptors with evident outer segments just above the oil droplet, and only counted rods traversing the oil droplet rows. Thus, our measure of photoreceptor abundance is not an oil droplet or outer segment count per se and may provide a lower estimate of photoreceptor abundance per length of retina than cell body, oil droplet, or outer segment counts of photoreceptor abundance do (Meyer & May, [@ref65]; Bowmaker et al., [@ref5]; Kram et al., [@ref54]). Our goal was, however, to detect relative changes in specific photoreceptor types between groups as a function of lesion or lighting condition, rather than provide photoreceptor type counts per unit of retina per se. Photoreceptor abundance is presented per 100 µm length of retina sampled. Statistical analysis {#sec2-7} -------------------- The effects of post-lesion survival time on photoreceptor abundance for each type were assessed by regression analysis for DL-housed birds, and CL-housed birds. For birds sustaining no lesion, the time spent in housing since arrival at UTHSC was regarded as the survival time. The regression analysis in general showed an effect of post-lesion survival time on abundance per photoreceptor type in only a limited number of cases. Any significant effects revealed by regression analysis are described below. In light of the limited effects of post-lesion survival time on photoreceptor abundance, as the main approach for determining the effects of the lesion manipulations or lighting conditions, we analyzed effects across groups using mixed-model 2-way Analysis of Variance (ANOVA), performed using Statistical Analysis Software (SAS), with three lesion levels (control, AP and EWM) and two lighting levels (diurnal versus constant), with individual comparisons between groups assessed by Fischer PLSD. Our approach for assessing effects of lesion and/or lighting condition was to compare each experimental condition to the control-DL condition to determine if it was significantly different. Because of the large number of comparisons, to limit false detection of differences (type 1 error), we set the significance level at 0.0125. Results are presented as mean ± SEM (Standard Error of the Mean). Results {#sec3} ======= Principal cone outer segments {#sec3-1} ----------------------------- No significant differences were found for principal cone outer segment abundance between the two control groups (control-DL and control-CL), nor between either of the AP groups and the control-DL group ([Figs. 4](#fig4){ref-type="fig"} and [5](#fig5){ref-type="fig"}). Thus, CL per se was not found to adversely affect principal cone outer segment abundance, since neither control-CL nor AP-CL principal cone outer segments showed significant loss. Although no significant change was seen in principal cone outer segment abundance in the EWM-DL group compared to the control-DL group, a significant 42.2% reduction was seen in principal cone outer segment abundance in the EWM-CL group compared to the control-DL group (P = 0.0096). To further evaluate the lesion effects on photoreceptor outer segment abundance, we performed regression analysis to determine if principal cone outer segments were undergoing survival-related decline that was perhaps not yet evident at the group level in some cases. We, however, found no significant correlation of principal cone outer segment loss with post-lesion survival for any of the DL or CL groups, indicating that loss of principal cones was only evidenced in the EWM-CL group and it was not markedly progressive over the 3 weeks of CL.Fig. 4.Images showing photoreceptor outer and inner segments in an eye unaffected by any lesion (A), in an eye affected by a contralateral AP lesion from a bird housed in normal DL for 0.5 months (B), in an eye affected by EWM destruction from a bird housed in normal DL for 1.5 months (C), in an eye affected by EWM destruction from a bird housed in normal DL for 2.7 months (D), in an eye affected by EWM destruction from a bird housed in CL for 1 week post-lesion (E), and in an eye affected by EWM destruction from a bird housed in CL for 3 weeks post-lesion (F). Red cones (red R), principal cones (white P), green cones (green G), violet/blue cones (blue B), and rods (black R) are identifiable by the traits shown in [Fig. 2A](#fig2){ref-type="fig"}. Note that all photoreceptor types are identifiable in (A and B), but fewer violet/blue cones are evident in the eyes affected by EWM destruction from birds housed in normal DL (C and D). Violet/blue cones are also absent, and principal cones sparse or absent in eyes affected by EWM destruction from birds housed in CL (E and F). Note also that outer and inner segments in eyes affected by EWM destruction in the eye from the bird housed in CL for 3 weeks (F) showed additional abnormalities not seen in the one week eye, including shrinkage, vacuolation, and darkening of inner segments. By comparison to the retina shown in [Fig. 2B](#fig2){ref-type="fig"} and in images (A and B), the inner segments in (C and D) are largely normal in appearance, reflecting the more limited effects of EWM destruction on photoreceptors in birds housed in normal DL. All images are at the same magnification.Fig. 5.Graph illustrating the effects of the various manipulations performed on the abundance of outer segments of the different photoreceptor types in the superior central pigeon retina, per 100 µm length of retina. Photoreceptor outer segments were identified and counted according to criteria and approaches described in the text. The greatest losses observed were for blue/violet cone outer segments following EWM destruction and housing in either DL or CL. Principal cone outer segment loss following EWM destruction and housing in CL was also seen. Error bars represent SEMs. Asterisks indicate significant differences from the control-DL group at the P \< 0.0125 level. Blue/violet cone outer segments {#sec3-2} ------------------------------- Blue/violet cones showed a significant outer segment loss following EWM destruction in both lighting conditions ([Figs. 4](#fig4){ref-type="fig"} and [5](#fig5){ref-type="fig"}). For example, blue/violet cone outer segment abundance in eyes affected by EWM destruction in birds housed in DL was 48.5% of that in the eyes of control-DL birds (P = 0.0095). A similar effect was seen for blue/violet cone outer segment abundance in EWM lesion-affected eyes in the CL-housed birds, whose blue/violet cone abundance was 34.0% of that in the control-DL eyes (P = 0.0070). The blue/violet cone outer segment abundance was, however, not significantly less in AP experimental eyes than that in control-DL eyes for either lighting group ([Fig. 5](#fig5){ref-type="fig"}). The blue/violet cone outer segment loss was significantly correlated with survival time in EWM experimental eyes in birds housed in CL (r = −0.867), but not in any of the other groups, indicating it was progressive in the EWM-CL group. Green cone outer segments {#sec3-3} ------------------------- Green cones did not show any reduction in outer segment abundance in AP or EWM lesion groups under either the DL or CL conditions, compared to the control-DL eyes ([Figs. 4](#fig4){ref-type="fig"} and [5](#fig5){ref-type="fig"}). We also found no significant correlations of green cone outer segment abundance with post-lesion survival in any of the DL groups, or for the control-CL and AP-CL groups. Although green cone outer segment loss per se was not significant in eyes affected by EWM destruction in CL birds compared to control-DL eyes at a group level, green cone outer segment abundance in the EWM-CL group tended to be less than in the other CL groups. Moreover, green cone outer segment abundance in EWM-CL trended toward a significant inverse correlation with post-lesion survival (r = −0.817), which however did not achieve the 0.8433 required for significance. These results suggest that some decline in green cones may have been in progress in EWM-CL eyes. Red cone outer segments {#sec3-4} ----------------------- Red cones did not show any reduction in outer segment abundance in AP or EWM lesion groups under either the DL or CL conditions, compared to the control-DL eyes ([Figs. 4](#fig4){ref-type="fig"} and [5](#fig5){ref-type="fig"}). We also found no significant correlations with post-lesion survival in any of the groups for red cone outer segment abundance. Rod outer segments {#sec3-5} ------------------ No significant differences were seen in the comparisons between AP-DL, EWM-DL, and control-CL eyes to control-DL eyes for rod outer segment abundance ([Figs. 4](#fig4){ref-type="fig"} and [5](#fig5){ref-type="fig"}). Although rod outer segment abundance in EWM-CL eyes was about 70% of that in the control-DL eyes, the difference was not significant (P = 0.0816), although it trended in that direction. The rod outer segment abundance in the AP-CL group also appeared to trend toward reduction ([Fig. 5](#fig5){ref-type="fig"}), but the difference was not significant compared to the control-DL eyes (P = 0.2731). We found no significant correlations with post-lesion survival in any of the DL or CL groups for rod outer segment abundance, indicating no prominent trend toward progressive loss of rod outer segments over the survival times used. Although rod outer segments did not yet show significant loss after three weeks of CL, rod photoreceptor pathology in the form of darkened and degenerating inner segments was seen ([Fig. 4](#fig4){ref-type="fig"}). Discussion {#sec4} ========== Our findings show that EWM lesions that disrupt parasympathetic control of ChBF adversely affect short wavelength-sensitive blue/violet cones and principal cones, and this effect is exacerbated by CL. These photoreceptor types have been shown in prior studies in mammals and birds to be preferentially vulnerable as well to light, hypoxia, and aging (Greenstein et al., [@ref37]; Hodos et al., [@ref45]; Fite et al., [@ref23]; Machida, [@ref62]; Curcio, [@ref15]; Algvere et al., [@ref1]; Organisciak & Vaughn, [@ref68]; Hovis et al., [@ref47]). Our results and their implications are discussed in more detail below. Prior studies in birds {#sec4-1} ---------------------- The circuit from the retina to the EWM via the vSCN ([Fig. 1](#fig1){ref-type="fig"}) mediates increases in ChBF in response to retinal illumination (Fitzgerald et al., [@ref24],[@ref25], [@ref26]; Shih et al., [@ref82]). Lesions of EWM or sectioning of the choroidal nerve from CG to the choroid reduces basal ChBF and prevent light-mediated reflexive increases in ChBF (Shih et al., [@ref83]; Fitzgerald et al., [@ref26]). Our prior studies in pigeons show that the ChBF insufficiency caused by EWM destruction in young adult pigeons causes progressive increases in Müller cell GFAP, so that by one year after lesion GFAP immunolabeling of Müller cells extends throughout the entire depth and extent of the ipsilateral retina (Fitzgerald et al., [@ref24]; Kimble et al., [@ref105]). GFAP upregulation in retinal Müller cells is a well-known correlate of retinal injury, stress, or disease (Bignami & Dahl, [@ref2]; Eisenfeld et al., [@ref20]; Burns & Robles, [@ref6]; Osborne et al., [@ref69]; Humphrey et al., [@ref48]; Sarthy & Egal, [@ref81]; Tanihara et al., [@ref91]; Chen & Weber, [@ref10]), and its occurrence after an EWM lesion thus implies retinal injury. Consistent with this interpretation, increases in GFAP expression have also been found following transient choroidal and/or retinal ischemia (Gay et al., [@ref33]; Hayreh & Weingeist, [@ref44]; Osborne et al., [@ref69]; Tanihara et al., [@ref91]; Kim et al., [@ref53]). We have further shown that the retinal abnormalities caused by an EWM lesion in young adult pigeons housed in DL result in reduced visual acuity by about one year post-lesion (Hodos et al., [@ref46]). Our present studies suggest that blue/violet and principal cone loss may have contributed to these acuity deficits. Given that loss of principal cones and accessory cones occurs together with age in pigeons (Hodos et al., [@ref45]), it seems likely that EWM destruction may have affected accessory cones as well. In our prior study reporting retinal GFAP increases after EWM destruction (Kimble et al., [@ref105]), we saw no significant GFAP increase over the two months after lesions that disrupted the PLR without affecting ChBF control (i.e., by destruction of either AP or EWLcl). In the present study as well, PLR disruption alone had no significant effect on photoreceptors in either normal DL or in CL. The prominent outer segment loss seen in EW lesion birds thus appeared to require disruption of ChBF regulation by EWM. Our observation that retinal pathology was insignificant after lesions that disrupt pupil alone is consistent with our acuity data, showing that acuity is unimpaired in pigeons one year after pupil light reflex elimination by means of AP destruction (Hodos et al., [@ref46]). It remains possible that longer post-lesion survivals would reveal an impact on photoreceptor health after pupil light reflex disruption alone, perhaps progressing to prominent outer segment loss. Studies in humans with diseases that impair pupil constriction are consistent with this possibility (Laor & Korczyn, [@ref55]; Gräf & Jungherr, [@ref35]). Light damage {#sec4-2} ------------ The damaging effects of transient exposure to extremely bright light, or sustained exposure to moderately bright or normal light, on photoreceptors have been demonstrated in a variety of species (Marshall et al., [@ref107]; Fite et al., [@ref23]; Machida, [@ref62]; Pérez & Perentes, [@ref72]; Thomson et al., [@ref94]; Algvere et al., [@ref1]; Organisciak & Vaughn, [@ref68]; Thomas et al., [@ref93]). Rods are particularly vulnerable to light damage, especially in nocturnal species (Cicerone, [@ref11]; Tanito et al., [@ref92]; Organisciak & Vaughn, [@ref68]; Okano et al., [@ref67]). By contrast, rod vulnerability to CL appears to be lower in pigeons than in nocturnal rodents, as true for other diurnal species in general, since no significant rod loss was seen for any of the bird groups housed in CL for up to three weeks. A trend in that direction was, however, observed for the EWM-CL eyes. We also observed low cone vulnerability in control eyes or AP-lesion affected eyes from birds housed in CL. By contrast, Marshall et al. ([@ref107]) reported cone outer segment loss in pigeons that were exposed to 1000 lux light for at least 24 h. It is possible then that more sustained and/or brighter lighting than we used (400 lux for up to 3 weeks) is needed to cause cone outer segment loss in normal or AP lesion pigeon eyes. Our prior study (Kimble et al., [@ref105]) and the current one indicate that the adverse effects of EWM circuit destruction on the retina are exacerbated by light, suggesting that ChBF regulation by the EWM circuit depicted in [Fig. 1](#fig1){ref-type="fig"} acts to mitigate light damage to photoreceptors. Photoreceptors show high metabolic activity in response to flickering light (Bill & Sperber, [@ref4]), and ChBF in pigeons has been shown to increase in response to flickering light (Shih et al., [@ref84]). It may be then that the unrelenting photoreceptor metabolic demand from a constantly changing retinal image in an illuminated cage, especially one that is constantly illuminated, may require sustained high ChBF. Moreover, inner retinal demand may exacerbate outer retinal vascular insufficiency, since the choroid is the sole vascular supply for the entire depth of the avian retina (Bill & Sperber, [@ref4]). As a result, the outer retina in pigeons following EWM destruction would experience ongoing choroidal insufficiency, causing hypoxic-ischemic injury (Gaudric et al., [@ref34]; Yu & Cringle, [@ref101]). Hypoxic insults to outer retina and/or diminished ChBF are known to heighten Müller cell GFAP expression (Penn et al., [@ref71]; Canady et al., [@ref7]), and cause RPE and photoreceptor dysfunction and loss (Linsenmeier et al., [@ref59]; Linsenmeier & Steinberg, [@ref60]; Yancey & Linsenmeier, [@ref98]; Ciulla et al., [@ref12]; Johnson et al., [@ref51]; Yu & Cringle, [@ref102]). Differential photoreceptor vulnerability {#sec4-3} ---------------------------------------- In the present study, we found that short wavelength cones and principal cones were most vulnerable to a disturbance in parasympathetic regulation of ChBF by the EWM circuit. Prior studies have also found these to be among the most vulnerable of the photoreceptor cell types. For example, short wavelength cones in diurnal species, including avian species, have been reported to be vulnerable to: (1) constant or bright light (Sperling et al., [@ref87]; Sykes et al., [@ref90]; Sperling, [@ref88]; Machida, [@ref62]); (2) hypoxia (Smith et al., [@ref86]; Algvere et al., [@ref1]; Connolly et al., [@ref13]; Hovis et al., [@ref47]); and (3) aging (Eisner et al., [@ref21]; Haegerstrom-Portnoy, [@ref41]; Gao & Hollyfield, [@ref32]; Curcio et al., [@ref14]; Gray et al., [@ref36]; Algvere et al., [@ref1]; Okano et al., [@ref67]). Blue cone vulnerability and loss in humans have also been observed in diabetic retinopathy, glaucoma, and retinitis pigmentosa (Greenstein et al., [@ref37]). The particular vulnerability of short wavelength cones may involve the greater photic energy of short wavelength light or the greater metabolic requirements of blue cones (Penn & Anderson, [@ref70]; Young, [@ref100]; Algvere et al., [@ref1]; Connolly et al., [@ref13]). As EWM destruction would render the affected eye hypoxic-ischemic, it is not surprising that this cone type should be vulnerable following disabling of parasympathetic control of ChBF. Accessory and principal cones in pigeons are preferentially vulnerable to age-related loss (Hodos et al., [@ref45]; Fitzgerald et al., [@ref27]). Hodos et al. ([@ref45]) suggested that age-related accessory and principal cone loss might reflect the cumulative effects over a lifetime of vulnerability to light damage, because they lack the protection provided by a colored oil droplet and are highly light sensitive (Mariani & Leure-Dupree, [@ref106]; Bowmaker et al., [@ref5]). Moreover, their coupling to one another via gap junctions (Smith et al., [@ref85]) may cause them to exacerbate the light damage to each other. In the present study, we also observed a suggestion of rod vulnerability in EWM-lesioned birds housed in CL, in that we observed a trend toward rod outer segment loss and we also observed darkened and degenerating rod inner segments. Rods are known to be particularly vulnerable to light damage, especially in nocturnal species (Cicerone, [@ref11]; Tanito et al., [@ref92]; Organisciak & Vaughn, [@ref68]; Okano et al., [@ref67]). Their particular vulnerability to light damage has been attributed to their light sensitivity, which is especially acute in nocturnal species. Light-induced generation of reactive oxygen species has been thought to be a major contributor to rod phototoxicity, with subsequent peroxidation of outer segment lipids (Organisciak & Vaughn, [@ref68]). It may be that diminished parasympathetic control of ChBF heightens sensitivity of rods to light damage by impairing energy-dependent anti-oxidant defenses (Organisciak & Vaughn, [@ref68]). Disease implications {#sec4-4} -------------------- A link between deficient parasympathetic ChBF regulation and outer retinal pathology and dysfunction has been shown in our prior studies of aging pigeons (Fitzgerald et al., [@ref27]). In humans, parasympathetic innervation of the choroid by the pterygopalatine ganglion is diminished in normal aging, as is basal ChBF (Grunwald et al., [@ref38]; Ito et al., [@ref49]; Jablonski et al., [@ref50]). Moreover, age appears to impair the adaptive stabilization of ChBF by the parasympathetic nervous system in response to systemic blood pressure fluctuations (Reiner et al., [@ref79]), which appears to be a major function of the pterygopalatine parasympathetic input to the choroid (Cuthbertson et al., [@ref18]; Li et al., [@ref57]; Reiner et al., [@ref77], [@ref78]). The present results raise the possibility that age-related impairments in adaptive vasodilatory ChBF control may contribute to the age-related waste accumulation in and along Bruch's membrane and photoreceptor loss observed in humans (Potts, [@ref74]; Tso, [@ref95]). Yet more profound declines in ChBF, as well as in its baroregulation, occur in dry and wet age-related macular degeneration (AMD), with the ChBF declines increasing in severity with AMD severity, and predicting the development of neovascularization (Friedman et al., [@ref29]; Ciulla et al., [@ref12]; Grunwald et al., [@ref39]; Pournaras et al., [@ref73]; Metelitsina et al., [@ref63]; Feigl, [@ref22]). In individuals with pro-AMD genetic predispositions in the alternate complement cascade or in lipid metabolism, such waste accumulation may trigger an inflammatory response that damages RPE and photoreceptors (Winkler et al., [@ref97]; Hageman et al., [@ref40]; Feigl, [@ref22]). Thus, deficient vasodilatory ChBF regulation may play a role in age-related retinal decline and be a risk factor for AMD (Hageman et al., [@ref40]). This research was supported by UTHSC Department of Ophthalmology unrestricted grant from Research to Prevent Blindness (MECF), NIH/NEI RO1EY05298 (AR), and The Methodist Hospitals Endowed Professorship in Neuroscience (AR). Thanks to Marion Joni, Betty Cook, Sherry Cuthbertson, Aminah Henderson, Jennifer White, and Leslie Kidd for technical assistance. Contacts: <malinda@cbu.edu>; <tingwong88@gmail.com>; <cnazor@uthsc.edu>; <ndelmar@uthsc.edu>; <areiner@uthsc.edu>. The authors have no commercial relationships.	{ "pile_set_name": "PubMed Central" }
What is known? {#S0001} ============== The majority (99%) of the children who die during the first 4 weeks of life do so in the poorer parts of the world, especially in sub-Saharan Africa and South Asia.In 2018, the Neonatal Mortality Rate (NMR) for Kenya was 19.6 deaths per 1000 live births.In 2018, the NMR for South Africa was 10.7 deaths per 1000 live births. What is new? {#S0002} ============ The most neonatal deaths resulted from preterm birth complications followed by intrapartum-related events for the two countries.Within the sub-Saharan region, disparities exist as neonates born in South Africa are more likely to survive during the neonatal period compared to Kenya. Introduction {#S0003} ============ A newborn infant, or neonate, is a child under 28 days of age.[@CIT0001] During these first 28 days of life, the child is at the highest risk of dying.[@CIT0001] Globally 2.5 million children died in the first month of life in 2018, an estimate that may be significantly higher considering the likelihood of under-reporting, especially in low-income and middle-income countries.[@CIT0002],[@CIT0003] There are approximately 7000 newborn deaths every day.[@CIT0002] On current trends, neonatal mortality declined more slowly than mortality among children aged 1--59 months.[@CIT0002] As a result, the share of neonatal deaths among all under-five deaths increased from 40% in 1990 to 47% in 2018.[@CIT0002] Reducing neonatal mortality is increasingly important not only because the proportions of under-five deaths that occur during the neonatal period are increasing as under-five mortality declines but also because the health interventions needed to address the major causes of neonatal deaths generally differ from those needed to address other under-five deaths.[@CIT0004] Every year, 135 million newborns enter the world, each arriving naked and apparently equal.[@CIT0005] Yet, their chances of surviving and thriving vary dramatically depending on which world these babies are born into---ranging from high-income countries with universal neonatal intensive care to the world of home births without midwives, medical supplies, or health system support.[@CIT0005] Skilled health care during pregnancy, childbirth and in the postnatal (immediately following birth) period prevents complications for mother and newborn, and allows for early detection and management of problems.[@CIT0001] In addition, WHO and UNICEF now recommend home visits by a skilled health worker during a baby's first week of life to improve newborn survival.[@CIT0001] Among the Sustainable Development Goal (SDG) regions, sub-Saharan Africa had the highest neonatal mortality rate in 2018 at 28 deaths per 1000 live births, followed by Central and Southern Asia with 25 deaths per 1000 live births.[@CIT0006] A child born in sub-Saharan Africa is 10 times more likely to die in the first month than a child born in a high-income country.[@CIT0006] With modest trends in reducing neonatal mortality in sub-Saharan Africa and high levels of neonatal mortality, many countries in the region are at risk of missing the SDG target on neonatal mortality.[@CIT0006] Estimates by UNICEF show that 42 of 48 countries in the region are projected to miss the SDG neonatal mortality target by 2030.[@CIT0006] About half are projected to meet the SDG target sometime after 2050 if progress is not accelerated.[@CIT0006] In the African Region, 1.12 million newborn deaths occur annually.[@CIT0001] The main causes include prematurity and low-birth-weight, infections, lack of oxygen at birth, and birth trauma.[@CIT0001] These causes account for nearly 80% of the deaths in this age group.[@CIT0001],[@CIT0007] Neonates, especially in sub-Saharan Africa (SSA), are challenged with a diversity of lethal clinical conditions, especially infectious diseases, which demand urgent interventions.[@CIT0008] When these clinical conditions, such as birth asphyxia, septicemia, jaundice, meningitis and tetanus are not appropriately diagnosed and attended to in time, they worsen the statistics of neonatal mortality in SSA.[@CIT0008] The purpose of the current review was to explore and describe the trends of neonatal mortality within the two sub-Saharan countries that is Kenya and South Africa. Methods {#S0004} ======= Inclusion and Exclusion Criteria {#S0004-S2001} -------------------------------- All descriptive or cross-sectional or observational, retrospective, national, regional studies reporting on the present research problem were included. An article was included if it met the following criteria: 1) Conducted in Kenya, South Africa, or within the sub-Saharan Region and published between Jan 1, 1975, and November 30, 2019, 2) had study participants as deceased neonates and 3) published in English. Conference abstracts, letters to editors, review, and commentary articles were excluded. Data Sources and Search Strategies {#S0004-S2002} ---------------------------------- A systematic search of articles was from the following databases; Scopus, Web of Science, Science Direct, Cochrane Library, PUBMED, OVID and Google scholar ([Figure 1](#F0001){ref-type="fig"}). The search comprised both Medical Subject Headings (MeSH) and free text words (title and abstract word searches). The following were the keywords used in the search: "Neonatal Survival", "Sub-Sahara", "Kenya" and "South Africa". Additionally, researchers searched grey literature resources such as a database/website of dissertations and theses and WHO websites. The reference list of included studies was manually searched for possible additional eligible articles. The searches were conducted from February 2019 to January 2020. In particular, the researchers used the following search terms in PUBMED database; "Neonatal Mortality" "causes of neonatal deaths" AND "Kenya" AND "South Africa" "Sub-Sahara"Figure 1PRISMA flow diagram.Note: Adapted from Liberati A, Altman DG, Tetzlaff J, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009;151(4):65--94.[@CIT0009] Selection of Studies for Inclusion in the Review {#S0004-S2003} ------------------------------------------------ Titles and abstracts of studies retrieved from each database search were stored in Mendeley Research Manager. Duplicates were removed before the screening process was initiated. Two review authors (BBM and RM) independently reviewed the titles and abstracts of all studies and disagreements were solved by discussion. The same steps were taken for full-text screening of the results. Data Extraction and Management {#S0004-S2004} ------------------------------ Using a standardized data extraction form, two review authors (BBM and RM) independently extracted data from eligible studies including the first author's last name, year of publication, study location, participant characteristics, study design utilized, major findings of the study. Quality Assessment {#S0004-S2005} ------------------ Two review authors (BBM & RM) independently assessed the quality of all included studies using the Critical Appraisal skills programme. BBM assessed the quality of the included studies while RM checked the assessed studies. Disagreements were resolved by discussion between the four review authors. No study was excluded at this point after the quality assessment. Synthesis of Results {#S0004-S2006} -------------------- The qualitative analysis was approved by the technical task team who got expert experience in biomedical systematic review. This was a secondary analysis and all identifiers of the individual participants were removed or presented using unique codes. This study adopted PRISMA ([Figure 1](#F0001){ref-type="fig"}) reporting as demonstrated previously by Liberati et al.[@CIT0009] Data was presented and organized under two thematic domains: 1) trends of national neonatal mortality rates in South Africa and Kenya and 2) causes of neonatal mortality. Results and Discussion {#S0005} ====================== The search generated 2209 articles of which 92 duplicate articles were removed. After the screening process of their titles and abstracts, only 31 articles were identified and were thereafter included in the full-text review. The identified articles were further assessed using the adopted inclusion criteria. Among 31 articles only 27 met the inclusion criteria ([Figure 1](#F0001){ref-type="fig"}). According to [Table 1](#T0001){ref-type="table"}, the specific study settings of the articles utilised were as follows: 10 studies covered Sub-Sahara, 9 were conducted in Kenya and 8 were carried out in South Africa. The study participants were deceased neonates. [Table 1](#T0001){ref-type="table"} gives a further description of the articles utilised. The present study aimed to systematically explore and describe the trends of neonatal mortality within the two sub-Saharan countries. The findings were presented and discussed under two thematic domains: 1) trends of national neonatal mortality rates in South Africa and Kenya and 2) causes of neonatal mortality Table 1A Description of the Articles UtilisedArticleYearDesignSettingResultsMasaba and Mmusi-phetoe, 2019[@CIT0010]2019Cross-sectionalKenyaA multifaceted approach is needed to reduce neonatal mortalities in KenyaTank, Omar, and Musoke, 2019[@CIT0011]2019Prospective auditKenyaMortality was high among preterm neonatesShikuku, Milimo, Ayebare, Gisore, and Nalwadda, 2018[@CIT0012]2018ObservationalKenyaEffective resuscitation for birth asphyxiaGitobu, Gichangi, and Mwanda, 2018[@CIT0013]2018SurveyKenyaFree maternal services increased hospital deliveriesAyodo, Samba, Amimo, and Asweto, 2018[@CIT0014]2018Survey designKenyaPreterm birth is one of the main causes of neonatal deathWagura, Wasunna, Laving, Wamalwa, and Ng, 2018.[@CIT0015]2018Cross-sectionalKenyaAt-risk mothers should receive intensified antenatal care to mitigate preterm birthOkube and Sambu, 2017[@CIT0016]2017Cross-sectionalKenyaPrevalence rate of preterm birth was 20.2%Ministry-of-Health-\[MOH\], 2015[@CIT0017]2015National surveysKenya1.5 million babies are born in Kenya annuallyYego et al, 2013[@CIT0018]2013RetrospectiveKenyaPre-term birth and asphyxia are the leading causes of neonatal deathsVelaphi et al, 2019[@CIT0019]2019Prospective studySouth AfricaUsing a combination of blood culture and a PCR-based test the common pathogens isolated in neonates with sepsis were Ureaplasma spp. and GBSRhoda, Velaphi, Gebhardt, Kauchali, and Barron, 2018.[@CIT0020]2018South Africa Demographic Health SurveySouth AfricaSouth Africa has a double burden of disease driving neonatal mortality; term babies are dying owing to intrapartum-related events, and preterm deaths are linked to related complications.Brits et al, 2015.[@CIT0021]2015Case controlSouth AfricaCo-morbidity as a risk to preterm birthRhoda and Velaphi, 2015[@CIT0022]2015National surveySouth AfricaPreterm infants are at high risk of developing hypothermiaBruckmann and Velaphi, 2015.[@CIT0023]2015Medical record analysisSouth AfricaThe incidence of asphyxia ranged from 8.7 to 15.2/1000 live birthsBamford, 2015.[@CIT0024]2015National DHS analysisSouth AfricaSouth Africa continues to face a high burden of disease and high maternal and child mortality rates.Mabaso, Ndaba, and Mkhize-kwitshana, 2014.[@CIT0025]2014Analysis of National Department of HealthSouth AfricaRe-engineering of the Primary Health Care (PHC) services to reduce neonatal mortalitySave the Children Fund \[SCF\], 2013.[@CIT0026]2013National analysisSouth Africa84,000 preterm births occur in South Africa annuallyWHO, 2019[@CIT0001]2019Regional estimatesSub-Sahara15 million babies are born too early every yearChawanpaiboon et al, 2019.[@CIT0027]2019Multi-regional analysisSub-Sahara12·0 million (81.1%) of these preterm births occurred in Asia and sub-Saharan Africa.Amare, Mela, and Dessie, 2019.[@CIT0028]2019ReviewDeveloping countriesNeonatal sepsis was found to be high which accounted for a third of the neonates.Ranjeva, Warf, and Schiff, 2018.[@CIT0029]2018Global health agencies data analysisAfricaNeonatal sepsis presents a \$10--\$469 billion financial burden that could be alleviated through successful treatment and preventionMongella, Songane, and Akande, 2010[@CIT0030]2010Regional estimatesAfricaEach year at least 1.16 million newborns die in Sub-Saharan AfricaUNICEF, 2019.[@CIT0006]2019Global surveyUnited Nation regionsIn 2018, the NMR for Kenya was 19.6 deaths per 1000 live birthsLawn et al, 2013.[@CIT0005]2013Systematic estimates12 countriesInvesting in care of every woman and every newborn will reduce both deaths and disabilitiesUNICEF, 2013.[@CIT0031]2013Regional estimatesGlobal analysisGlobally, an estimated 10 million babies have breathing difficulties at birth.Gillam-Krakauer and Gowen, 2018.[@CIT0032]2018Regional estimates--15--20% of babies diagnosed with birth asphyxia die during the neonatal periodLawn, Mongi, and Cousens, 2010[@CIT0033]2010Regional estimatesAfricaBirth asphyxia also accounts for 280,000 deaths a year in sub-Saharan Africa Kenya and South Africa National Neonatal Mortality Rate {#S0005-S2001} ------------------------------------------------------- In 2018, the NMR for Kenya was 19.6 deaths per 1000 live births ([Figure 2](#F0002){ref-type="fig"}). The NMR has fallen gradually from 35.4 deaths per 1000 live births in 1975, but the country is yet to achieve the SDG3 target 2 of reducing neonatal mortality to 12 deaths per 1000 live births.[@CIT0006] South Africa compared to Kenya has been having lower neonatal mortality rates for the past three decades. Its NMR has fallen gradually from 27.9 deaths per 1000 live births in 1975 to 10.7 deaths per 1000 live births in 2018. Of note also is that South Africa has achieved the set goal of SDG 3.2.[@CIT0006]Figure 2Kenya and South Africa National Neonatal Mortality rate, (per 1000 live births). Causes of Neonatal Deaths in 2018: South Africa and Kenya {#S0005-S2002} --------------------------------------------------------- [Table 2](#T0002){ref-type="table"} shows that most neonatal deaths resulted from preterm birth complications followed by intrapartum-related events for South Africa, while in Kenya intrapartum-related complications were the leading cause. Neonatal sepsis, congenital abnormalities and pneumonia ranked 3rd, 4th, and 5th in both countries as diarrhoea was insignificant as a cause of neonatal death. South Africa has been able to manage intrapartum-related complications better than Kenya. However, South Africa continues to struggle with preterm births. Because South Africa's economy is more developed than Kenya's, their increased death rate from more preterm birth was unexpected. More developed income countries tend to have better universal neonatal intensive care to care for the preterm births than their counterparts.[@CIT0005] Table 2Percentage Distribution of Causes of Neonatal Deaths in 2018: South Africa and KenyaCausesSouth AfricaKenyaPreterm birth complications36%28%Intrapartum related events20%29%Sepsis/tetanus14%16%Congenital abnormalities10%13%Diarrhoea0%0%Pneumonia5%7%Other conditions15%8% ### Preterm Birth Complications {#S0005-S2002-S3001} Findings from this review reveal that preterm birth complications were the top cause of neonatal mortalities in South Africa while in Kenya it ranked second.[@CIT0006],[@CIT0017],[@CIT0021] This finding implies that preterm births dying in South Africa were higher than in Kenya. Preterm birth (PTB) refers to babies born alive before 37 weeks of pregnancy are completed.[@CIT0014] Estimated 15 million babies are born too early every year.[@CIT0001] That is more than 1 in 10 babies.[@CIT0001] Approximately 1 million children die each year due to complications of preterm birth.[@CIT0001] Comparing with developed countries, preterm babies born from developing countries face higher morbidity and mortality rates.[@CIT0001] In low-income settings, half of the babies born as a preterm die due to a lack of feasible, cost-effective care, like provision of basic warmth, supporting breathing difficulties and taking preventive and control measures of infection.[@CIT0001] In high-income countries, almost all of these babies survive.[@CIT0016] The risk of neonatal death due to complications of preterm birth is at least 12 times higher for an African baby than for a European baby.[@CIT0015] Out of the 1.5 million babies born in Kenya annually, around 134,000 come too soon, according to the Ministry of Health's Division of Family Health.[@CIT0017] Moreover, a hospital-based study by Yego et al[@CIT0018] found that pre-term birth and asphyxia are the leading causes of early neonatal deaths in a teaching and referral hospital in Kenya.[@CIT0018] In South Africa, more than 1.2 million babies are born annually, of whom 14% are born prematurely.[@CIT0021] Sadly, about 10% of these babies do not survive despite most births occurring at health-care facilities.[@CIT0026] Common complications of preterm birth are high rates of respiratory distress syndrome, bronchopulmonary dysplasia, necrotising enterocolitis, sepsis, periventricular leucomalacia, seizures, intraventricular haemorrhage, cerebral palsy, infections, feeding difficulties, hypoxic ischaemic encephalopathy, and visual and hearing problems.[@CIT0027] Preterm infants are at high risk of developing hypothermia because they do not have enough energy stores and brown fat to produce heat.[@CIT0022] Hypothermia is an independent risk factor for mortality -- the more severe the hypothermia, the greater the mortality.[@CIT0022] Prematurity is a major cause of neonatal mortality and morbidity as well as a significant contributor to long-term adverse health outcomes.[@CIT0015],[@CIT0027] ### Intrapartum-Related Events {#S0005-S2002-S3002} The study revealed that intrapartum-related complications accounted for the highest neonatal mortalities in Kenya while it ranked second in South Africa ([Table 2](#T0002){ref-type="table"}). This implies South Africa is better prepared to handle newborn complications immediately after birth compared to Kenya. Birth asphyxia, defined as the failure to initiate and sustain breathing at birth by WHO, causes about a quarter of all the neonatal deaths.[@CIT0012] Effective resuscitation at birth can prevent a large proportion -- approximately 30% -- of these deaths.[@CIT0012] Clinically, the need for bag-and-mask ventilation (BMV) at birth and/or an Apgar score \<7 have been used to define intrapartum asphyxia.[@CIT0023] A report titled "Birth Asphyxia Complications" estimated 10 million babies have breathing difficulties at birth, mainly resulting in brain injury.[@CIT0031] This is caused by obstructed labour or acute haemorrhage during birth.[@CIT0031] Globally, more than 700,000 newborns died of birth complications and among survivors, 233,000 had a moderate or severe disability and another 181,000 had learning problems.[@CIT0031] According to Gillam-Krakauer and Gowen, the incidence of birth asphyxia is 2 per 1000 live births in developed countries, but the rate is up to 10 times higher in developing countries where there may be limited access to maternal and neonatal care.[@CIT0032] Gillam-Krakauer and Gowen further noted that of those babies affected 15--20% dies during the neonatal period and up to 25% of the survivors are left with permanent neurologic deficits.[@CIT0032] Birth asphyxia also accounts for 280,000 deaths a year in sub-Saharan Africa.[@CIT0033] Babies born in sub-Saharan Africa have a very high risk of birth asphyxia and intrapartum stillbirth.[@CIT0033] The best intervention is prevention through improved Antenatal Care (ANC) and, particularly, skilled attendance and emergency obstetric care.[@CIT0033] ### Sepsis \\| Tetanus {#S0005-S2002-S3003} Systemic infections of the newborn such as septicaemia refer to generalised bacterial infection documented by a positive blood culture in the first 28 days of life.[@CIT0019] It encompasses various systemic infections of the newborn such as septicaemia, meningitis, pneumonia, arthritis, osteomyelitis and urinary tract infections.[@CIT0019] Sepsis accounts for around one-third of deaths in neonates worldwide.[@CIT0011] Neonatal sepsis usually has nonspecific presentation; hence, delay in treatment, and especially initiation of effective antibiotic therapy results in serious consequences ranging from neuro-developmental deficits to death.[@CIT0011] Sepsis was attributed to cause approximately 400,000 neonatal deaths in 2015 globally, half of which occurred in sub-Saharan Africa where 34.6% to 66.0% of neonatal deaths reportedly occur within the first 24 hours of life.[@CIT0019] Bacteria are the leading cause of neonatal sepsis.[@CIT0028] Early empirical antimicrobial treatment is associated with better outcomes in neonatal sepsis, but antimicrobials must be discontinued timeously to prevent the emergence of further antimicrobial resistance.[@CIT0034] Despite its disproportionate burden on childhood mortality, neonatal sepsis receives vastly less investment internationally as a public health priority compared with other major conditions.[@CIT0029] Studies by Save the Children Fund corroborate this by noting that neonatal tetanus is one of the most cost-effective conditions to prevent -- yet it still kills tens of thousands of African babies each year.[@CIT0030] Conclusion {#S0006} ========== This systematic review noted that, although South Africa has achieved the WHO target of 12 neonatal deaths per every 1000 live births by 2030, it still has to work towards the challenges managing the preterm births. Additionally, review results noted that in 2018, the neonatal mortality rate for Kenya was 19.6 deaths per 1000 live births. The neonatal mortality rate has fallen gradually from 35.4 deaths per 1000 live births in 1975. On the other hand, South Africa has had its neonatal mortality rate fall from 27.9 deaths per 1000 live births in 1975 to 10.7 deaths per 1000 live births in 2018. Most neonatal deaths resulted from, preterm birth complications followed by intrapartum-related events for the two countries. Within the sub-Saharan region, disparities exist as neonates born in South Africa are more likely to survive during the neonatal period compared to Kenya. The authors would like to thank the University of South Africa for their support to the study. Author Contributions {#S0007} ==================== All authors contributed towards data analysis, drafting and critically revising the paper, gave final approval of the version to be published, and agreed to be accountable for all aspects of the work. Disclosure {#S0008} ========== The authors have declared that no conflicts of interest exist.	{ "pile_set_name": "PubMed Central" }
Key points {#Sec1} ========== Non-invasive imaging offers a solution for parents declining conventional autopsy.Postmortem ultrasound can delineate main body organs and identify congenital anomalies.Postmortem ultrasound can be useful where access to MRI is unavailable.Postmortem ultrasound may aid image-guided organ biopsies in the future. Background {#Sec2} ========== The chronic decline in parental consent rates for perinatal autopsies has contributed to the rise of non-invasive methods for death investigation \[[@CR1]\]. Traditionally, whole body radiography (i.e. skeletal surveys) has been the mainstay of postmortem imaging; however, their usage is limited outside detecting inheritable bone diseases \[[@CR2]\]. Large cohort studies have investigated the use of postmortem CT (PMCT) \[[@CR3], [@CR4]\] and MRI \[[@CR5]--[@CR7]\], with better concordance seen in perinatal postmortem MRI (PMMR) of up to 90% when compared to autopsy \[[@CR6]\]. Unfortunately, both PMCT and PMMR can be costly and difficult to access, thereby limiting more widespread usage. Ultrasound is widely used in both the antenatal and perinatal imaging for live foetuses and children. It is familiar and understandable to parents and referring clinicians and has also been shown to identify abnormalities in major body organs in the perinatal postmortem setting. Early studies have demonstrated reasonable sensitivity and specificity for whole body diagnoses of approximately 75% and 83.3%, respectively \[[@CR8]\], although some abnormalities may be challenging to detect, namely cardiac anomalies with sensitivity rates ranging from 18.2--50% \[[@CR8]--[@CR10]\]. Given the lower cost of ultrasound imaging, and therefore greater affordability and ease of access, it may provide a pragmatic solution to parents who decline conventional invasive autopsy but lack access to other cross-sectional imaging techniques. Whilst many healthcare professionals in theory possess the technical skills to perform postmortem ultrasound (e.g. sonographers, radiologists, foetal medicine doctors, obstetricians), many are unsure where to begin, what to look for and how to report such studies. This article therefore aims to provide such information to those wishing to start performing perinatal postmortem ultrasound examinations. We provide advice on imaging techniques, views to acquire, technical considerations and some commonly encountered 'normal postmortem appearances'. The methods described for conducting a thorough ultrasound examination are based upon our own experience of postmortem sonography in a paediatric tertiary referral centre and radiology research unit, with experience of postmortem imaging in a wide range of settings. Perinatal deaths {#Sec3} ================ Perinatal death rates and reasons for their occurrences vary considerably across Europe, ranging from 4.6 per 1000 births in Germany to 12.4 per 1000 births in Latvia, with approximately 36,000 annual perinatal deaths across Europe \[[@CR11]--[@CR13]\]. The reasons for these losses can be broadly classified into those relating to maternal health issues (e.g. thrombophilia), placental and cord abnormalities, obstetric complications, acquired (infection) or congenital foetal anomalies \[[@CR14]\]. Of these classifications, the commonest indication for postmortem imaging is in assessment of the latter subgroup---namely developmental foetal anomalies and perinatal complications (e.g. intra-cranial haemorrhage). The identification of these pathologies can help account for foetal demise and further define or confirm antenatal imaging findings, particularly following termination of pregnancy. It is also important to bear in mind when counselling parents and clinicians prior to imaging that despite full and thorough investigations, there are still a significant number of perinatal deaths where a cause for foetal demise is not discovered and the cause remains 'undetermined' \[[@CR15], [@CR16]\]. Consent and referral indication {#Sec4} ------------------------------- Parental consent is vital in all aspects of a perinatal autopsy and, at our institution, permission to perform all postmortem imaging studies are included in our standard autopsy consent forms \[[@CR17]\]. Whilst postmortem ultrasound is non-invasive and does not alter the child's body in any way, we require parental consent for use of any images to be retained and used for research, teaching and educational content. At present, there are no clear indications or referral criteria for perinatal postmortem ultrasound, as there is limited data to support in which cases it may be of most use. Clearly structural abnormalities such as those identified on antenatal imaging would be most readily identifiable, but confirming the presence or absence of infection may be challenging \[[@CR7], [@CR8], [@CR13]\]. Timing after death {#Sec5} ------------------ Immediately after receipt of the body to our mortuary, the child is kept in a body bag and in cold storage, at a temperature of 4 °C until imaging can be performed. From personal experience, optimal timing for ultrasonography should be as close as possible to the time of death or delivery; however, there is no evidence of a decline in diagnostic accuracy rates with increasing postmortem interval timing (i.e. time between delivery and imaging). Ultrasound performed for intra-uterine deaths are usually of a lower diagnostic quality in cases where a prolonged intra-uterine retention period has elapsed (due to maceration related changes \[[@CR18]\]). These factors are usually unavoidable, given logistical delays in referrals from other hospitals or maternity units. Nevertheless, we still obtain adequate imaging with postmortem intervals of up to 14 days. Patient preparation {#Sec6} ------------------- The body can be scanned immediately following removal from cold storage and does not strictly speaking need to be left to warm at room temperature. Nevertheless, one should be aware that cold storage has been reported to alter tissue properties at CT and MRI imaging \[[@CR19], [@CR20]\], and Okuda et al. \[[@CR21]\] have postulated that a reduction in the reflection coefficient between fat and soft tissues at colder temperatures could result in diminished ultrasound image contrast. In reality, the changes in image quality between a cold and warm foetus are probably not large enough to present diagnostic challenges, although this has not been comprehensively measured. Tissue fixation in formalin prior to imaging is also unnecessary in our experience, but may allow for reduced body tissue laxity and is used in some centres for foetuses \< 15 weeks gestation in order to preserve intra-cranial anatomy which is prone to maceration \[[@CR9]\]. Scanning environment, equipment and operator {#Sec7} -------------------------------------------- We use standard paediatric postmortem sonographic equipment, but have a dedicated machine for postmortem imaging which is kept in the mortuary. In the absence of a separate machine, reserving certain ultrasound probes for postmortem use may be prudent to minimise cross-contamination and transmission of infection, or else the use of antibacterial wipes and transducer covers will suffice. It is also important to be mindful of adequate privacy in the scanning cubicle, and in some hospitals, this may necessitate postmortem imaging being performed outside of routine working hours. In general, a high-resolution probe is most effective at visualising the internal body structures for perinatal cases. We recommend tailoring the probe choice to the case, but we have obtained diagnostic image quality using both 2--8 MHz and 7--16 MHz high-frequency linear probes, although other authors have used high-frequency curvilinear probes (such as those used for transvaginal studies due to their smaller footprint), or a lower frequency curvilinear probe for larger subjects \[[@CR22]\]. 3D/4D volume acquisitions have been assessed \[[@CR22]\], but not widely adopted, and the additional benefit of postmortem 3D ultrasound imaging over 2D imaging has not been thoroughly investigated. Conventional B-mode imaging is therefore suitable and recommended for the entire study. Due to lack of patient movement and cardiovascular output, colour Doppler and M-mode function is not required. The ultrasound imaging pre-set we use is the same as for imaging superficial musculoskeletal lesions (termed 'MSK Enhanced' setting on our system), given the small size of our foetuses and superficial nature of many of the organs. The image depth, time gain control, overall gain and focal zones (both position and number) are amended on a case-by-case basis to acquire the best possible diagnostic quality images. In general, harmonic imaging has not been found to be very useful given the reduced depth and small size of our subjects \[[@CR23]\]. Our postmortem studies are performed by a paediatric radiologist although similarly trained sonographers and foetal medicine clinicians may also be suitable. Imaging tip {#Sec8} =========== Due to the early gestational ages and therefore smaller size of many perinatal cases (compared with live neonatal imaging), image optimisation and visualisation can be an issue, particularly, if there is a lack of complete contact between the footprint of the ultrasound transducer and the foetal body. In order to overcome such a problem, it is possible to use the 'water bath' technique (similar to the method used for imaging small inflamed and tender joints in live children \[[@CR24]\] Figure [1](#Fig1){ref-type="fig"}). This is performed by placing the foetus in a container of cold, still water at a depth just sufficient to submerge the body and head. A gel pad can be used and placed beneath the foetus to allow for stabilisation of the body underwater and prevention of body movement by ripples during scanning. The images obtained may be of slightly reduced spatial resolution; but there is usually much better visualisation of the internal organs owing to improved sonographic wave transmission through the water to greater skin surface area and, since there is no direct pressure of the probe on the body, there is reduced distortion and compression of internal structures (Fig. [2](#Fig2){ref-type="fig"}).Fig. 1Photograph of an empty 2 L plastic rectangular container (a) with a pink silicon gel pad for postmortem ultrasound water bath. Diagram demonstrating the set-up of the water bath (b). The container is filled with water, left to rest for 15 min and then the foetus placed in the bath with neck supported by gel pad. The ultrasound transducer is partly submerged in the waterFig. 2Two transverse postmortem ultrasound images of the chest in the same 23-week gestational-aged foetus, obtained 4 days after death. Postmortem ultrasound performed outside the water bath (a) and with foetus in the water bath (b). Notice how the thoracic cage was collapsed both due to body laxity and transducer compression in image (a), but how there is improved visualisation of the heart (asterisk) and lungs (white arrows) using the water bath (b) as the chest expands An alternative approach to the water bath technique is covering a small foetus in a layer of ultrasound transmission gel (at approximately 0.5 cm depth, as described by Votino et al. \[[@CR22]\]). We would not recommend this technique as large quantities of ultrasound gel causes more difficulty in cleaning the body after scanning, and residual gel left on the body can contribute to fungal growth on the skin. For this reason, antiseptic cleaning after scanning is essential, as is ensuring that all gel is removed and the body is dry. Imaging and reporting algorithm {#Sec9} =============================== Given the lack of aeration of foetal lungs in the majority of perinatal deaths, a whole body ultrasound including imaging of the mediastinum and lungs is possible and visualisation of the majority of the internal body organs can be comprehensively studied. A complete postmortem ultrasound assessment usually takes between 15 and 30 min depending on patient size and complexity. As with all types of imaging assessment, a consistent and methodical approach should be adopted. We recommend performing and reporting the whole body ultrasound on a 'systems' basis, as this is similar in structure to the autopsy reports provided by pathologists. By this method, the body is divided into the following systems: 'neurological', 'cardiac/thoracic', 'abdominal' and 'musculoskeletal'. We report each system separately with abnormalities pertaining to each area documented. A table listing these systems with recommended ultrasound views, orientation and diagnoses to consider are given in the Additional file [1](#MOESM1){ref-type="media"}: Tables S1--S3). We also further discuss these below. Brain {#Sec10} ----- Coronal and sagittal views can be performed through the anterior fontanelle (Fig. [3](#Fig3){ref-type="fig"}), as in live neonatal cranial imaging. Additional views via the sphenoidal fontanelle (to obtain 'trans-temporal' views) and via the mastoid fontanelle (for 'trans-mastoid' views) provide imaging of the posterior fossa and brain stem including the upper cervical spine (Fig. [4](#Fig4){ref-type="fig"}).Fig. 3Paired coronal postmortem ultrasound images (obtained in the water bath) and T2-weighted postmortem MRI at 1.5 T in the same 20-week gestational-aged foetus. The images were obtained 4 days after death, demonstrating normal anatomy. Images through the frontal horns (a, b) and the posterior horns (c, d) of the lateral ventricles demonstrate how detailed anatomy of the brain, including sulcation and appearances of the ventricles can be adequately reviewed by ultrasound, and not necessarily require further cross-sectional imagingFig. 4Paired postmortem ultrasound and T2-weighted MRI imaging at 1.5 T are shown in this 20-week gestation foetus, 4 days after death (same case as Fig. [3](#Fig3){ref-type="fig"}). The inclusion of additional sonographic views via the left sphenoid fontanelle (a) provides clear images of the midbrain and occipital lobes (white arrows), as the comparison with similar view on MRI (b) demonstrates. The left trans-mastoid view of the brain at ultrasound (c) shows clearly the cerebellum (asterisk) and upper cervical spinal canal (dotted white arrow), comparable in detail as the matched T2-weighted postmortem MRI (d) It is important to note that due to the laxity of soft tissues, head moulding and overlapping of cranial sutures in postmortem cases, the quality of images may not always be diagnostic. Intra-cranial gas from postmortem changes and extraction-related deformities can obscure views of the brain (Fig. [5](#Fig5){ref-type="fig"}). Typically, the coronal views are the most diagnostically useful (rather than the sagittal views) given distortion of the midline.Fig. 5Images of delivery and maceration related changes that degrade postmortem ultrasound image quality in a 22-week gestational-aged foetus. Sagittal T2-weighted postmortem MRI at 1.5 T (a) through the midline shows soft tissue oedema of the scalp, overlapping cranial sutures and underlying brain maceration. On the corresponding left trans-temporal view of the brain at postmortem ultrasound (b) and at axial postmortem MRI (c) at the level of the midbrain, it is difficult to delineate normal brain parenchymal architecture. In such cases, the imaging is described as 'non-diagnostic' and useful information cannot be gleaned by either imaging modality When diagnostic quality images are obtained, an assessment can be made on the gestation-appropriate gyration and sulcation of the brain, assessment of midline structures such as corpus callosum (with all components typically visible on antenatal ultrasound imaging by 18 weeks gestation \[[@CR25]\]), intra-cranial masses, haemorrhage and ventricular size. With regards to ventricular dilatation, it is worth noting that despite an antenatal history of ventriculomegaly, shifts in cerebrospinal fluid dynamics occurring after death can result in an apparent 'resolution' of this appearance and should not be interpreted as a discrepancy or false positive result by prenatal imaging \[[@CR26]\]. Spine {#Sec11} ----- The views of the spine obtained at postmortem ultrasound are similar to those in live neonatal imaging---namely sagittal views of the conus medullaris, sacrum and transverse images of the cervical, thoracic and lumbar spine (Fig. [6](#Fig6){ref-type="fig"}). The pulsatile movements of the filar roots will be absent.Fig. 6Sagittal postmortem ultrasound images of a normal cervical spine (a), thoracolumbar spine (b) and lumbosacral spine (c) in a 21-week gestational-aged foetus, obtained 12 days after death. Corresponding transverse views of the cervical (d), thoracic (e) and lumbar (f) spinal cord are also shown from a routine normal postmortem ultrasound examination Assessment should be directed at excluding neural tube defects \[[@CR27]\] or caudal regression syndrome, identifying vertebral segmental anomalies and spinal dysraphism. Where these are suspected, further characterisation with skeletal radiographs (i.e. babygrams, skeletal survey) can help delineate the pathology better, if not already performed \[[@CR2]\]. A useful tip to remember to avoid misdiagnosis of a tethered cord is that gestational age can have a significant impact upon the normal expected level of the conus medullaris. One postmortem series reported that only 50% of foetuses at 26 weeks gestational age have a conus at the L3 vertebral level or above. With earlier gestations, the conus can be normally expected to lie at the L4/5 level or lower \[[@CR28]\]. Thorax {#Sec12} ------ All foetuses who have not breathed (miscarriage, stillbirths or following termination of pregnancy) will still have unaerated, fluid-filled lungs, and some neonates may have made respiratory effort at birth or undergone failed resuscitation and thus present at postmortem with partially aerated lungs. The air within the lungs can obscure adequate sonographic imaging of the lobar anatomy and also clear views of the heart. Assessment of the thorax is usually limited to the assessment and symmetry of lobes of each lung (Fig. [7](#Fig7){ref-type="fig"}) and presence of pulmonary lesions (e.g. cystic structures which may include bronchopulmonary foregut malformations) or congenital diaphragmatic hernias. Small pericardial and pleural effusions are common and attributable to expected postmortem change \[[@CR29], [@CR30]\] (Fig. [8](#Fig8){ref-type="fig"}). It is sometimes possible to visualise the fluid-filled trachea, bronchi and also a collapsed oesophagus.Fig. 7Paired sagittal postmortem ultrasound (a, b) and T2-weighted postmortem MRI images at 1.5 T (c, d) of the right lung in the same 18-week gestational-aged foetus, 12 days after death. These images clearly demonstrate the three different pulmonary lobes (RUL---right upper lobe, RML---right middle lobe, RLL---right lower lobe). The solid white arrows on the labelled images show normal expected physiological pleural fluid seen as part of postmortem change and should not be described as pathologicalFig. 8Transverse postmortem ultrasound image through the chest in a 19-week gestational-aged foetus, obtained 8 days after death showing normal postmortem fluid accumulation within the pleural spaces (solid white arrows) and pericardium (dotted arrows) bilaterally. These are commonly seen and described as 'postmortem changes' and should not alarm the ultrasound operator Termination of pregnancy involving foeticide (i.e. intra-cardiac injection of a toxic agent, such as potassium chloride) can cause haemorrhage (i.e. fluid with internal debris) and gas within the pericardium and pleural spaces, and this should be recognised as iatrogenic and correlated with the antenatal history \[[@CR31]\] (Fig. [9](#Fig9){ref-type="fig"}).Fig. 9Transverse imaging through the lower thoracic cavity in a 20-week gestational-aged foetus, 2 days after death following termination of pregnancy for suspected renal agenesis. Termination was performed by in utero foetal intra-cardiac injection of potassium chloride. Postmortem ultrasound (a) and postmortem T2-weighted MRI at 1.5 T (b) both demonstrate haemorrhage within the right pericardial (yellow asterisk) and pleural space (yellow dotted arrow). It is also possible to visualise the 'normal' consolidated right lower lobe of the lung (white solid arrow) and a small pericardial effusion (white dotted arrow). These are typical foeticide-related changes and iatrogenic in nature Cardiac and vascular imaging {#Sec13} ---------------------------- Conventional cardiac views can be difficult to obtain at postmortem ultrasound. This may be in part due to body distortion and the small size of the heart at the limits of ultrasound resolution at earlier gestation. Lack of cardiovascular output and therefore colour Doppler flow also hamper structural assessment of the heart, which may account for the lower diagnostic accuracy rates for this body area, as alluded to in the introduction. Nevertheless, by obtaining transverse and sagittal views through the heart, it is still possible to glean useful information regarding ventricular wall thickness, the orientation of the outflow tracts (Fig. [10](#Fig10){ref-type="fig"}) and presence of septal defects.Fig. 10Labelled transverse postmortem ultrasound images of the chest in a 20-week gestational-aged foetus, obtained 4 days after death demonstrating normal anatomy. The labelled images are taken at level of the aortic arch (a), level of the main outflow tracts (b) and at the biventricular level of the heart (c). They serve to demonstrate the detail that can be gained with ultrasound imaging in a well-preserved foetus Where there is a large intra-thoracic mass (e.g. congenital diaphragmatic hernia), it is also possible to make a comment regarding the displacement of the mediastinum. Imaging of the aortic arch, superior vena cava and pulmonary veins are sometimes possible (Fig. [11](#Fig11){ref-type="fig"}), but usually collapsed.Fig. 11Paired sagittal midline imaging of the chest, in a 20-week gestational-aged foetus obtained approximately 1 week after death. The postmortem ultrasound imaging (a) and T2-weighted MRI imaging at 1.5 T (b) both clearly demonstrate the usual appearances of haemostasis and collapse of the descending aorta (solid white arrows) and also prominent patent ductus arteriosus (dotted arrow) Abdomen {#Sec14} ------- All intra-abdominal organs are methodologically assessed in both the transverse and sagittal planes, as in live cases. The hepatic, renal and intra-abdominal vessels are commonly collapsed and therefore their assessment is difficult (Figs. [12](#Fig12){ref-type="fig"} and [13](#Fig13){ref-type="fig"}). The common bile duct is also nearly impossible to identify; however, the presence or absence of the gallbladder is usually possible.Fig. 12Matching transverse upper abdominal views in labelled diagram format (a), postmortem ultrasound (b) and T2-weighted MRI at 1.5 T imaging (c) of the same 36-week gestational-aged foetus 3 days after death demonstrating haemostasis in the hepatic veins. The corresponding sagittal postmortem ultrasound (d) and matched T2-weighted MRI (e) views of the thrombosed main portal vein (white arrows) are also shown. This imaging example shows how even with decent sonographic windows in a well-preserved foetus, the common bile duct is quite difficult to delineateFig. 13Sagittal postmortem ultrasound images of the upper abdomen in a 19-week gestational-aged foetus, obtained 8 days after death. Images through the abdominal aorta (solid arrows) (a) and ductus venosus (arrow heads) are both collapsed. The imaging through the inferior vena cava (dotted arrows) (b) show similar appearances. Another finding of note is the echogenic, prominent appearing right adrenal gland (asterisk). This is the usual expected appearance of the adrenal gland on perinatal postmortem ultrasound examinations Findings that may be pathological in live neonates, but which are frequently physiological in the postmortem setting, include a small amount of simple ascites, periportal echogenicity in the liver (Fig. [14](#Fig14){ref-type="fig"}) and some loss of corticomedullary differentiation of the kidneys (especially with prolonged intra-uterine retention) (Fig. [15](#Fig15){ref-type="fig"}).Fig. 14Transverse postmortem ultrasound images of a normal liver in two different foetuses, both acquired 12 days after death. The liver in image (a) demonstrates marked periportal echogenicity and was obtained in a 32-week gestational-aged stillborn foetus. In image (b), there are fewer periportal echoes in this 21-week gestational-aged foetus following termination of pregnancy for oligohydramnios. The causes for these differences is unknown and could relate to different phases of decomposition or gestational age, as both livers were histologically normal at autopsyFig. 15Normal sagittal appearances of the kidneys at postmortem ultrasound in two different foetuses (left column (a, c)---left kidney, right column (b, d)---right kidney). The solid white arrows in each of the images demonstrate the upper and lower poles of the kidneys. Images (a, b) were obtained in a 20-week gestational-aged foetus, 4 days after death. They show the normal expected corticomedullary differentiation. Images (c, d) were obtained from a stillborn 25-week gestational-aged foetus, 10 days after death. These kidneys lack corticomedullary differentiation and are harder to identify when seen against the background of other solid abdominal viscera and bowel. All kidneys were histologically unremarkable both on antenatal ultrasound imaging and autopsy. The lack of corticomedullary differentiation is likely to relate to autolysis or maceration-related changes, rather than pathological causes Septated fluid in the abdomen and large-volume ascites causing abdominal distension is however abnormal and, where present, should raise concern for hydrops or underlying sepsis. In addition, kidneys that appear enlarged, markedly echogenic or have internal cystic structures and pelvicalyceal dilatation are suggestive of underlying renal anomalies. Where kidneys are not identified in the retroperitoneum, a detailed examination of the pelvis should be performed to exclude an ectopic kidney. One common difficulty in perinatal postmortem ultrasound imaging is in the identification and location of the spleen, particularly in earlier gestation foetuses. The reasons for this are twofold---the first being that the liver can be relatively quite large and the spleen relatively small in size \[[@CR32]\] during foetal life and, secondly, the echotexture of the hepatic parenchyma can be very similar to that of the spleen (Fig. [16](#Fig16){ref-type="fig"}). Nevertheless, an enlarged spleen can often be identified, which could signify underlying infection or an inborn error of metabolism (e.g. perinatal Gaucher's \[[@CR33]\] and other alloimmune diseases \[[@CR34]\]). The pancreas is also occasionally a difficult organ to identify given that it can autolyse early on in the maceration process and appear absent. When present in a non-macerated foetus, the appearances are very similar to ante-mortem ultrasound images (Fig. [17](#Fig17){ref-type="fig"}).Fig. 16Normal appearances of the liver and spleen in a 19-week gestational-aged foetus following miscarriage, imaged 6 days after death. The transverse postmortem ultrasound imaging (a) through the upper abdomen demonstrates a very similar echogenicity of the liver with the spleen (white arrow). The differences between the two organs is much better seen at T2-weighted postmortem MRI (b), where both organs demonstrate different signal intensitiesFig. 17Normal appearances of the pancreas (white arrows) on transverse postmortem ultrasound (a) in a 20-week gestational-aged foetus, 4 days after death with corresponding postmortem T2-weighted MRI at 1.5 T (b) obtained 12 days after death Finally, after all solid viscera are identified and assessed, the bowel and pelvis are examined. Assuming that the anterior abdominal wall is intact and there is no diaphragmatic hernia, commenting upon normal bowel rotation is possible. In this instance, the lack of intraluminal gas is helpful as it allows one to trace the path of the duodenum and assess where it crosses the midline within the retroperitoneum (Fig. [18](#Fig18){ref-type="fig"}). This is easier than identifying the orientation of the superior mesenteric artery and vein on postmortem ultrasound which are commonly thrombosed and small in size.Fig. 18Normal transverse postmortem ultrasound appearances of the bowel in a 19-week gestational-aged foetus, obtained 8 days after death. The first three images (a--c) demonstrate normal rotation of the bowel. In image (a), the gastric pylorus becomes the first part of the duodenum (solid white arrows) and is seen to cross the midline to the right of the abdomen. In image (b), the second part of the duodenum is shown (white arrows), and finally, in image (c), this becomes the third part of the duodenum as it crosses back to the left side of the abdomen (white arrows). In image (d), normal small and large bowel loops in the lower abdomen are seen and meconium filled, without any significant bowel wall thickening or interloop separation The pelvic structures are best visualised on a sagittal view of the lower abdomen (Fig. [19](#Fig19){ref-type="fig"}). It is possible to appreciate the rectum, urinary bladder and uterus (if female). Ovaries are not usually visualised in early gestational aged foetuses, and the testes (in males) can frequently be beyond the limits of image resolution for adequate assessment.Fig. 19These images demonstrate normal appearances of the pelvic structures in a female 18-week gestational-aged foetus. Sagittal postmortem ultrasound of the pelvis (a) and corresponding postmortem T2-weighted MRI at 1.5 T (b) have been obtained 12 days after death. The tubular appearances of a normal pre-pubertal uterus, rectum and urinary bladder are all well seen and labelled in these images. Using a high-frequency linear probe, the pelvic anatomy can be better demonstrated on ultrasound rather than MRI Musculoskeletal {#Sec15} --------------- Postmortem ultrasound assessment for the musculoskeletal system is best reserved for assessment of soft tissue masses, particularly of the head and neck such as venolymphatic (and other vascular) malformations or teratomas \[[@CR35]\]. Where further skeletal imaging is needed, radiography or CT for ossified structures can be more helpful. In the perinatal setting, a skeletal survey alone is sufficient to diagnose inheritable bone disorders (i.e. skeletal dysplasias). Whilst several studies have demonstrated the relative futility in performing radiography routinely for all perinatal deaths \[[@CR2], [@CR36], [@CR37]\], it still remains common practice in most centres and therefore can be easily referenced if there is any doubt regarding bony appearances during sonographic imaging. Conclusions {#Sec16} =========== Perinatal postmortem ultrasound is an easily accessible and simple imaging tool that can allow for visualisation of internal organs and aid the perinatal autopsy. This article has highlighted the key aspects for approaching a postmortem ultrasound examination including patient preparation, imaging protocol and organ assessment. It serves as a guideline for radiologists, sonographers, foetal medicine clinicians and potentially other allied health professionals who have ultrasound skills and are keen to provide a non-invasive imaging autopsy service. In early cohort studies, postmortem ultrasound has shown high concordance rates with autopsy although certain anomalies, such as thoracic and cardiac pathologies may still pose a challenge and remain a major limitation. The future of postmortem ultrasound may include usage as a first line 'screening' technique to determine which patients could benefit from further postmortem cross-sectional imaging (e.g. PMMR), or for assistance during image-guided organ biopsies in a 'minimally invasive autopsy', where tissue samples for genetic or histopathological analysis is required. Both of these topics represent areas of further research on this subject. Additional file =============== {#Sec17} Additional file 1:Table S1. Imaging protocol for perinatal postmortem ultrasound study---neurological system. Table S2. Imaging protocol for perinatal postmortem ultrasound study---cardiothoracic system. Table S3. Imaging protocol for perinatal postmortem ultrasound study---abdominal system. (DOCX 24 kb) CT : Computed tomography MRI : Magnetic resonance imaging PACS : Picture archiving and communications system PMCT : Postmortem computed tomography PMMR : Postmortem magnetic resonance imaging PMUS : Postmortem ultrasound Funding {#FPar1} ======= SCS is supported by a RCUK/ UKRI Innovation Fellowship and Medical Research Council (MRC) Clinical Research Training Fellowship (Grant Ref: MR/R00218/1). This award is jointly funded by the Royal College of Radiologists (RCR). OJA is funded by a National Institute for Health Research (NIHR) Career Development Fellowship (NIHR-CDF-2017-10-037), and NJS funded by an NIHR Senior Investigator award. The authors receive funding from the Great Ormond Street Children's Charity and the Great Ormond Street Hospital NIHR Biomedical Research Centre. This article presents independent research funded by the MRC, RCR, NIHR and the views expressed are those of the author(s) and not necessarily those of the NHS, MRC, RCR, the NIHR or the Department of Health. Availability of data and materials {#FPar2} ================================== Data sharing is not applicable to this article as no datasets were generated or analysed during the current study. SCS performed the ultrasound examinations and obtained the images included in this educational review. OJA and SCS report on the postmortem MRI and US examinations. OJA and NJS conceived of the manuscript concept and idea. SCS, NJS, and OJA all contributed to the writing and editing of the final manuscript. All authors read and approved the final manuscript. Ethics approval and consent to participate {#FPar3} ========================================== The imaging experience and cases reported upon are part of an ethically approved larger study investigating minimally invasive autopsy techniques and newer methods of post mortem imaging. This has ethical approval from the National Health Service (NHS) Health Research Authority, Research Ethics Committee (REC) ID: CE13/LO/1494 and CE2015/81. Consent for publication {#FPar4} ======================= All parents of children who underwent postmortem imaging and less invasive autopsy techniques have signed consent forms agreeing for imaging to be used for educational and research purposes. Competing interests {#FPar5} =================== The authors declare that they have no competing interests. Publisher's Note {#FPar6} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.	{ "pile_set_name": "PubMed Central" }
###### Key messages - Interstitial Lung Diseases (ILDs) are characterised by symptoms such as chronic, dyspnea and cough. These lead to significant psychological distress, which is scarcely manageable. - This study represents the first attempt to conduct a Mindfulness-Based Stress Reduction (MBSR) program in a group of patients with ILD. - MBSR appears to be a safe and feasible approach in patients with ILD at different stages of severity, and it might produce significant improvement of mood and perception of stress. Introduction {#s1} ============ Interstitial lung diseases (ILDs) are a large and heterogeneous group of conditions characterised by the diffuse involvement of the pulmonary interstitial space, with different patterns of inflammation and fibrosis. Many ILDs are rare entities if considered separately, although together they account for one-third of the overall respiratory morbidity. Chronic ILDs may result in the onset of pulmonary fibrosis, consisting in the aberrant deposition of connective tissue (scar) in the interstitium, possibly leading to impairment of gas exchange and ultimately death from respiratory failure. Idiopathic pulmonary fibrosis (IPF), one of the most common ILDs, is a specific form of chronic, progressive interstitial pneumonia of unknown cause, associated with a poor prognosis with an average duration from diagnosis to time of death of 2--3 years.[@R1] [@R2] Chronic, progressive shortness of breath is nearly universal in patients with fibrotic ILDs, and may be accompanied by other invalidating symptoms such as fatigue and cough. These affect the quality of life of these individuals by limiting daily life activities, with consequences on the psychological and emotional levels. The anxiety related to fear of such symptoms and consequent loss of self-confidence can generate further disability, thus triggering a dangerous vicious circle. Basal dyspnoea is negatively correlated with quality of life and survival in IPF,[@R3] [@R4] and longitudinal changes in dyspnoea also predict prognosis.[@R5] However, despite the clinical significance of psychological and emotional disturbance in ILD, there is paucity of data about its appropriate management; disease-centred pharmacological therapies have shown little benefit, and complementary management strategies are frequently needed, including pulmonary rehabilitation, supplemental oxygen and opioids.[@R6] Available interventions that could have an impact on psychological aspects of the disease may include individual psychological counselling, participation inpatient support groups or attendance at mindfulness programs: however, to the best of our knowledge, the effect of these interventions has never been objectively measured. The term 'mindfulness' refers to a conscious modality characterised by a state of mental presence and attention to the present moment.[@R7] Mindfulness-based techniques teach the capacity of discriminating thoughts, emotions and actions with positive effects from those with harmful effects, increasing the awareness that thoughts are not reality but only one of its representations. Mindfulness-Based Stress Reduction (MBSR) is a standardised mindfulness training program created in 1979 by J Kabat-Zinn to integrate the ancient practices of meditation with contemporary clinical and psychological practice.[@R8] Initially, it was developed in an experimental way to treat groups of patients suffering from chronic pain,[@R9] but after its standardisation it has been largely adopted for treating other disorders in the medical and psychological fields. In cancer, mindfulness groups were shown to develop acceptance to the progression of diseases, reduction of fear and avoidance drives.[@R10] Critical reviews of the evidence of the effect of mindfulness intervention in this field are encouraging,[@R11] [@R12] with clinically relevant implications to alleviate psychological and physical suffering of persons living with cancer. MBSR has also demonstrated efficacy in reducing negative moods and the perceived stress related to dyspnoea in asthmatic patients in a pilot study.[@R13] MBSR programs show efficacy in improving the health-related quality of life,[@R14] confirming the reliability of these studies and suggesting long-term follow-up and the use of active control groups. Fibrotic lung diseases, often characterised by the impairment of quality of life due to progressive symptoms and worsening emotional status,[@R15] [@R16] may represent another potential application field for MBSR, but has not currently been investigated. In this study, we sought to assess the safety and the feasibility of an MBSR program in a group of patients affected by different ILDs. We investigated the efficacy of MBSR on mood, quality of life and also pulmonary function, as we hypothesised that mindfulness practice might indirectly trigger a positive, measurable effect on pulmonary performance and exercise tolerance. Methods {#s2} ======= Study population and design {#s2a} --------------------------- Nineteen adult patients were recruited between November and December 2012 at the Centre for Rare Lung Diseases of the University Hospital of Modena, Italy. Inclusion criteria included a confirmed diagnosis of ILD according to international guidelines. Exclusion criteria were the inability to express a valid consent, the inability to perform pulmonary function tests, the use of long-term oxygen therapy for more than 18 h/day and the regular practice of meditation or Yoga, as the study was meant to evaluate the efficacy of mindfulness-based practice including meditation and Hatha Yoga techniques introduced for the first time. Patients of all levels of disease were included, as the purpose of the study was to evaluate for the first time the feasibility of an MBSR program in a heterogeneous sample of patients with ILD. We conducted a prospective, observational study with a duration of 1 year, comprising three phases: a preintervention observational period (2 months), an intervention period (2 months) during which patients were trained in the MBSR program, and a postintervention observational period (8 months; [figure 1](#BMJRESP2014000065F1){ref-type="fig"}). ![Study design. SOB, Shortness Of Breath questionnaire; CASA-Q, Cough And Sputum Assessment Questionnaire; POMS, Profile Of Mood State test; PSS, Perceived Stress Scale; PFT, pulmonary function test; 6MWT, 6 min walk test; MBSR, Mindfulness Based Stress Reduction program.](bmjresp2014000065f01){#BMJRESP2014000065F1} Ethics and recruitment {#s2b} ---------------------- The conduct of the study was approved by the Ethics Committee of Modena, Italy. Twenty-five patients followed up at the Centre for Rare Lung Diseases of the University Hospital of Modena were invited to take part and provided with a participant information sheet. An orientation meeting was also held in order to introduce candidates to the basic concepts of mindfulness and give information about the MBSR program. Written informed consent was collected from all patients who decided to participate. Overall, six patients declined participation, mainly due to the practical difficulties they would have encountered in attending the MBSR program (many patients lived far away from the centre). MSBR program {#s2c} ------------ The program was conducted by an MBSR instructor with a personal history of 20 years of supervised meditation practice and who was specifically trained for 2 years in mindfulness methods at the University of Florence (Italy). It was divided into eight weekly meetings of 150 min each, to which all the patients were invited to join as a single group. Meetings were held in a comfortable quiet room located inside the hospital building. Three different techniques were taught in sequence: body scan focusing on sensations from different parts of the body, with verbal reminders from the instructor to focus on abdominal breathing; sitting meditation on a chair with a straight back, bringing not judgemental awareness of natural breath and on mental events; light yoga based on instruction in simple body movements and postures with controlled breathing. Instructions were given to patients to continue performing exercises daily at home for 45 min between the weekly meetings and after the training programme was completed; patients were also provided with CDs as a guide for home practice. An all-day meeting of 7 h between the sixth and the seventh session completed the course. Overall, the instructor total contact time was 28 h in 8 weeks. At the end of the program, patients were asked to continue practising alone at home every day. To support patients' home practice, three quarterly group meetings of 150 min each were added during the postintervention observational period. Outcome measures {#s2d} ---------------- The primary outcomes of the study were safety, measured as the frequency of adverse events related to the practice of mindfulness techniques, and feasibility, measured as the rate of dropouts and adherence to the MBSR program. Other outcomes included changes in mood, perceived stress and respiratory symptoms (measured by patient-reported outcomes-questionnaires), change in lung function (assessed by the rate of relative change from baseline in forced vital capacity (FVC, % predicted), and rate of change in Single Breath-Diffusion Lung Capacity of Carbon Oxide (SB-DLCO) and change in exercise tolerance measured as changes in the 6 min walking test (6MWT). The Profile Of Mood State (POMS) test,[@R17] measuring the improvement of mood, consists of 40 items clustered into 6 scales (anger, confusion, depression, fatigue, tension, Vigour). The Perceived Stress Scale (PSS)[@R18] measures the reduction of stress and consists of 10 items corresponding to 10 different domains (anger, perception of control, irritability, self-trust, expectations, efficacy, control of irritation, self-control, uncontrollable anger, excessive difficulties). The University of California San Diego Shortness Of Breath Questionnaire (UCSD-SOB)[@R19] and Cough And Sputum Assessment Questionnaire (CASA-Q),[@R20] respectively assess self-reported shortness of breath and impact of cough while performing a variety of activities of daily living. Questionnaires were administered at: enrolment (T0), at the beginning and at the end of the intervention (T1 and T2), and then at each quarterly meeting until the end of the follow-up (T3, T4 and T5). Statistical analysis {#s2e} -------------------- For POMS and PSS tests, average scores and changes in scores were evaluated at the beginning and the end of the study and also between the different intervals of the follow-up. For SOB and CASA-Q, total scores for each patient were calculated and then changes in the scores were evaluated for statistical significance between the beginning and the end of the study. Lung function and exercise tolerance were also measured for each patient approximately every 3 months. Spirometry and SB-DLCO were performed according to the available international guidelines,[@R21] as well as the 6MWT (distance covered).[@R24]The Wilcoxon test was used to calculate the statistical significance of the changes in the average scores of the POMS and PSS questionnaires between the different times of the follow-up. A non-parametric test for repeated measures was used to assess changes in lung function tests (FVC and SB-DLCO), 6MWD and questionnaire scores for SOB and CASA-Q between T0, T2 and T4. To handle missing answers in items of the SOB and CASA-Q, we chose to calculate the total score relatively to the maximum theoretical score based on the number of answers provided. Results {#s3} ======= Baseline characteristics {#s3a} ------------------------ The main baseline characteristics are shown in [table 1](#BMJRESP2014000065TB1){ref-type="table"}. Eleven men and eight women (mean age±SD: 65±8 years) formed a single MBSR training group. The majority of participants (12 of 19) had a diagnosis of IPF. The remaining seven patients represented a broad spectrum of ILD. ###### Demographic and clinical features of the population ----------------------------------------------------------------------- Patients enrolled (N=19) -------------------------------------------- -------------------------- Age, years (mean±SD) 65±8 M/F, n 11/8 IPF, n 12 Other ILDs, n 7 NSIP 2 RA-ILD 1 Chronic-HP 1 CTD-ILD 1 Asbestosis 1 DIPNECH 1 Lung function tests at baseline * Per cent of predicted FVC\ 82±26 (mean±SD) Per cent of predicted SB-DLCO (mean±SD) 51±19 6MWD at baseline, metres\ 499±87 (mean±SD) Dropouts, n (%) 2 (10.5%) Average number of sessions attended (of 9) 8 ----------------------------------------------------------------------- 6MWD, 6-min walking test distance; CTD-ILD, connective tissue disease associated interstitial lung disease; DIPNECH, diffuse idiopathic pulmonary neuroendocrine cell hyperplasia; FVC, forced vital capacity; HP, hypersensitivity pneumonitis; ILDs, interstitial lung diseases; IPF, idiopathic pulmonary fibrosis; NSIP, non-specific interstitial pneumonia; RA-ILD, rheumatoid arthritis associated interstitial lung disease; SB-DLCO, single breath diffusion lung capacity for CO. Safety and feasibility {#s3b} ---------------------- Two patients (10.5%) dropped out early from the study for non-respiratory causes (low back pain and chronic fatigue, respectively) before they had started the intervention, and were not included in the analysis. The remaining 17 patients completed the 8-week intervention period. One patient suffered from acute deterioration of clinical conditions during the intervention period and was admitted into the ward of the clinic of respiratory diseases, but still managed to attend the meetings during the period of hospitalisation. There were no adverse events related to the activities performed, either during the intervention period or during the follow-up, when patients continued the daily home practice. The average number of sessions attended per patient was eight of nine. Adherence to homework was not recorded. POMS and PSS {#s3c} ------------ The results for the average scores of POMS and PSS tests are reported in [figure 2](#BMJRESP2014000065F2){ref-type="fig"}. The POMS test showed significant changes in the mean scores indicating an improvement of mood throughout the study period. A total score was calculated by the addition of all the variables except for vigour, whose scores were subtracted. A statistically significant (z=2.197, p=0.028) reduction can be found between the start and the end of the study. This reduction becomes more evident after the intervention (from T2 on), while a slight trend towards a worsening of mood was seen during the intervention period. Between T0 and T5, that is, between enrolment and the end of the study, the scores for three scales showed a statistically significant reduction: fatigue (z=2.362, p=0.018), depression (z=2.1977 p=0.028) and anger (z=1.960, p=0.05). The scale vigour showed a significant increase for the same period (z=2278, p=0.023). The scale confusion showed a significant reduction for the intervals T1--T2 (z=2.701, p=0.007) and T3--T5 (z=1.979, p=0.048). Tension showed a steady reduction between T0 and T5, never reaching statistical significance though. ![Change in average scores of Profile Of Mood State test (A) and Perceived Stress Scale (B) single items between T0 and T5.](bmjresp2014000065f02){#BMJRESP2014000065F2} Perceived stress (PSS score) showed improvement in all items from T0 and T5. Nevertheless, only four items showed a statistically significant improvement: control of irritation increasing between T3 and T5 (z=2000, p=0.048), Uncontrollable Anger parallel decreasing between T1 and T2 (z=0.680, p=0.046), Anger decreasing between T1 and T4 (z=2.212, p=0.034) and Irritability decreasing between T1 and T5, respectively (z=2.111, p=0.046). All changes suggest a positive effect in terms of reduction of stress. There was also a significant decrease in Self-control before the start of the intervention, between T0 and T1 (z=2.008, p=0.045), while it showed a trend towards improvement thereafter. Respiratory symptoms questionnaires and pulmonary function tests {#s3d} ---------------------------------------------------------------- Pulmonary function tests were performed at T0, T2 and T4, and were analysed together with respiratory symptoms scores at the same time points. For each variable, only patients for whom all three time points were available have been included. Changes in the % relative scores in the Shortness of Breath Questionnaire and the Cough and Sputum Assessment Questionnaire are shown for each patient in [figure 3](#BMJRESP2014000065F3){ref-type="fig"}. No statistically significant differences in questionnaire scores were observed between different time points. ![Change in % relative scores of Shortness of Breath (SOB) questionnaire and Cough And Sputum Assessment Questionnaire (CASA-Q) between T0 and T4 (N=12).](bmjresp2014000065f03){#BMJRESP2014000065F3} Similarly, changes in the serial measurements of % predicted FVC and % predicted SB-DLCO, or in the distance walked at the 6MWT were not observed, other than those due to deteriorations associated with the progression of the underlying fibrotic lung disease ([figure 4](#BMJRESP2014000065F4){ref-type="fig"}). ![Change in % predicted forced vital capacity FVC (N=13), % predicted SB-DLCO (N=11) and 6 min walking distance (N=14) between T0 and T4. FVC, forced vital capacity; DLCO, diffusion lung capacity of CO; 6MWD, 6 min walking distance.](bmjresp2014000065f04){#BMJRESP2014000065F4} Discussion {#s4} ========== The findings of this 1-year prospective observational pilot study support the feasibility and safety of an MBSR program in patients affected by ILDs. Feasibility needed to be established, since one of the main practices of MBSR involves paying attention to breathing, which potentially could increase stress in these patients. The dropout rate was low (10.5%), with two patients withdrawing from the study during the preintervention observational period, as they felt unable to attend the MBSR program regularly. All 17 patients who entered the program managed to complete the study, attending an average of eight meetings of the nine scheduled. This high completion rate is particularly pertinent as many patients had to travel (in some cases, they lived hundreds of kilometres away) and the intervention was held in winter, between January and March, and patients had to face bad weather. Although adherence to homework was not recorded (see limitations of the study below), during the 2 months program and in the follow-up period a regular daily home practice of 30--45 min was informally reported by 2/3 of the patients; 1/3 of patients admitted being less regular with home exercises. We believe that the reasons for the good adherence to such a new approach are several. First, the setting chosen for the MBSR program was a conference room located in the Clinic for Respiratory Diseases, a familiar place for the patients. As such, the proximity to a well-known secure environment and to the familiar and trusted medical staff (although no doctors were admitted to take part in the MBSR sessions) might have worked as a facilitating factor. Second, the paucity of treatment options for many ILDs may have encouraged these greatly motivated patients to search for new approaches of care to improve their performances and quality of life. Third, feelings of loneliness and abandonment are often present in patients with rare diseases, and sharing such an experience with people affected by the same health problems, as with dedicated support groups, may be helpful. In fact, MBSR could represent a complementary strategy for developing new potential synergies and forming peer groups, as well as for helping patients acquire self-awareness and feel more empowered. Furthermore, the supportive role and empathy with the mindfulness instructor may act as a non-specific factor[@R13] improving program adherence and outcomes. The results obtained from the questionnaires on mood and perceived stress showed a potential efficacy of the MBSR program in improving the emotional status in these patients. The improvements in the POMS and PSS questionnaires throughout the 1-year follow-up suggests that the application of mindfulness techniques may play a role in helping patients to reduce mood disturbance and in adjusting better to their incurable disease. The mechanism of action of mindfulness is obviously different from 're-framing' in which thoughts are challenged and restructured.[@R25] The shift of perspective, named 're-perceiving', is a more reliable mechanism. However, there might be another way how mindfulness can work, as we observed in patients an increased capacity to stay exposed to simple and preordained mental life, disrupting the strenuous process of holding on themselves and their world. This attitude of 'bearing the void' has a curious, committed approach; it is effectively sustained by deep yoga practice, that allow to listen physical sensations in a new interesting way, not confirming usual mindsets. It is noticeable that the improvement of mood had a constant trend over time ([figure 2](#BMJRESP2014000065F2){ref-type="fig"}) right after the end of the intervention period. In this period, we recorded from many patients' statements about the attractiveness of the method and their regular practice---developed in a personal blend of meditation, yoga, body scan and informal exercises. Therefore, it is possible that the follow-up periodical meetings worked as a booster for the patients' will in pursuing their home practice. It is not possible, though, to exclude some positive seasonal influence on the mood acting as a confounding factor, as the postintervention observational period covered the whole spring and summer, potentially bearing a more positive attitude. The positive trend in the emotive sphere was not reflected by changes in respiratory symptoms of these patients, as measured by the Shortness of Breath and the CASA-Q questionnaires. Lung function and exercise tolerance were also unaffected. This discrepancy was not unexpected, however. The respiratory questionnaires focused on the ability of patients to perform daily life activities that become limited as the disease progresses, specifically dyspnoea and cough. It would be surprising if an intervention aimed at modifying the perception of a symptom had such an effect that could even alter the symptom itself and hence improve physical performances. Indeed, it has to be pointed out that there was no significant change in respiratory symptoms over the study period, which should be considered as a positive outcome in a 12-month follow-up study including patients with progressive respiratory diseases. This pilot study has several limitations. First of all, the sample size of the study population is small. For the POMS and the PSS tests, we performed the analysis across different time points. Although the overall change in scores between the beginning and the end of the observation period is consistent, the adoption of multiple testing carries a potential risk of generating false positives. This issue might be addressed by larger studies. ILDs are rare conditions, though, and the program itself was designed for small groups. As such, future research requires international multicentre populations. The advantage of exploring the effect of the MBSR program in people belonging to different cultural and social settings would add significant strength to the results. Second, assessment of mood or stress is very challenging and sensible to the influence of many confounding factors: age, comorbidities (physical or psychological) and external factors (such as seasonality) can easily affect the perception of mood states and symptoms. Hence, the adoption of a properly matched control arm and randomisation would be needed to determine the real efficacy of the MBSR program in the emotional field. Another limitation is represented by the substantial heterogeneity of the population recruited into the study in terms of disease severity and diagnosis, although the majority of patients (63%) were affected by IPF. Again, this limitation is strictly related to the single-centre design of the study, and larger randomised and possibly multicentre studies would allow a proper stratification of patients in order to seek how this approach can impact different stages of severity in specific conditions. Finally, in this pilot study, adherence to homework was not rigorously recorded in diaries, usually adopted for recording mindfulness practice at home. This was done in order to not apply pressure on patients with breathing difficulties. Moreover, the sample size was too small for making adequate comparisons between more and less compliant patients. Indeed, measures to monitor the adherence of participants to daily home exercises should be adopted in confirmatory studies in order to state with certainty whether pursuing mindfulness practice after the end of the program results in a consistent, long-lasting effect. Conclusions {#s5} =========== This pilot, single-centre, 1-year observational study demonstrates that MBSR is a feasible and safe approach in patients affected by ILDs. Our results suggest a positive effect on patients' moods, lending support for larger controlled studies with a stratified population to assess the efficacy of MBSR programs and evaluate their complementarity with existing dedicated support groups. RF and MPR wish to sincerely thank their meditation and yoga teacher M. Franco Bertossa of the Associazione Asia (Bologna, Italy) for his teaching and for providing inspiration and support for this work. Contributors:* GS and RF have revised this version of the manuscript. Conception and design, GS, SC, RF, MPR and LR. MBSR intervention program, RF, MPR. Patients support, MO, MG. Acquisition of clinical data, GS and SC. Data analysis and interpretation, GS, SC, RF, MPR, SP, GM and KP. First drafting of manuscript GS, SC, RF. Revision of manuscript, all authors. Project supervision, LR and FL. Competing interests: None. Ethics approval: This study was granted approval from Research Ethics Committee of Modena, Italy. Provenance and peer review: Not commissioned; externally peer reviewed. Data sharing statement: No additional data are available.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#sec1-1} ============ Malaria remains a significant health burden in sub-Saharan Africa. However, there has been a major decline in malaria morbidity (by 42%) and mortality (by 66%) between 2000 and 2015. To further address malaria-related challenges, many African countries have adopted WHO\'s Global Technical Strategy for Malaria 2016-2030.\[[@ref1][@ref2]\] Nigeria is among the 15 African countries responsible for 80% of global malaria cases and 78% of deaths.\[[@ref1]\] Nonetheless, the country has recently experienced a significant reduction in malaria parasite rate in children aged \<5 years, from 42% in 2008 to 27% in 2015, coupled with a reduction in the overall malaria deaths from 300,000 to 122,800 in 2016.\[[@ref3][@ref4]\] Notably, the Nigerian government has also adopted the global malaria elimination strategy through its National Malaria Strategic Plan 2014--2020.\[[@ref5]\] For eliminating malaria, its silent reservoirs should be identified in high-risk populations using highly sensitive and accurate diagnostic tools, in addition to sustaining its clinical management and integrated vector control efforts in transmission-prone areas.\[[@ref6][@ref7]\] High-risk populations often comprise adults with asymptomatic malaria (ASM) who do not seek treatment, remain undiagnosed and serve as hotspots of malaria transmission all year round, with varying intensity.\[[@ref6]\] Therefore, ASM plays a critical role in delaying the global elimination of malaria.\[[@ref8][@ref9]\] ASM is also of clinical importance because of its health impact in humans, including pregnant women.\[[@ref8]\] Sub-Saharan Africa is parallelly also facing a diabetes crisis, wherein \>95% of the cases are of type 2 diabetes mellitus (T2DM).\[[@ref10]\] An estimated 15.5 million Africans aged 20--79 years were reported to be suffering from diabetes in 2017.\[[@ref11]\] The global excess mortality attributed to diabetes in adults has been estimated to be 3.8 million deaths.\[[@ref10]\] A significant proportion of such mortality is attributed to poor glycemic control, in addition to other factors for which mechanisms of occurrence are not completely understood.\[[@ref12]\] Recently, a study from Central Africa reported a significant relation between insulin resistance and malaria infection.\[[@ref13]\] However, despite the high burden of malaria and T2DM in sub-Saharan Africa, limited data exist regarding their co-occurrence and its associated clinical effects.\[[@ref13][@ref14][@ref15][@ref16]\] In Nigeria, T2DM currently affects an estimated 4 million adults,\[[@ref17]\] making the country one of the most burdened countries in sub-Saharan Africa. However, there is currently paucity of information on the co-occurrence of ASM and T2DM among Nigerian patients. Further, the impact of such a co-occurrence on diabetes management outcomes such as glycemic control remains unknown. Understanding patterns of interactions between malaria and T2DM in Nigeria may guide development of strategies to reduce the increasing T2DM trend and improve its case management. Accordingly, the primary objective of the present study was to determine the extent of ASM parasite reservoirs among a cohort of adult T2DM patients in Lagos, Nigeria. As secondary objectives, the effects of ASM on glycemic control and anemia and the performance of the following diagnostic tools were also evaluated: histidine-rich protein 2 rapid diagnostic test (HRP2-RDT), light microscopy and polymerase chain reaction (PCR). MATERIALS AND METHODS {#sec1-2} ===================== Ethical approval {#sec2-1} ---------------- The Lagos State Hospital\'s Ethics Committee approved this study (Ref. no.: LSHMB 027) on February 4, 2015. All study participants provided their written informed consent prior to inclusion in this study. Study site {#sec2-2} ---------- This study was conducted in Alimosho, a local government area of Lagos, Nigeria. The area is a malaria-endemic area with a catchment population of 1,277,714 in 2006. In the study area, similar to other parts of Lagos, malaria transmission is perennial during the rainy (April--October) and dry (November--March) seasons, with Plasmodium falciparum being the major causative agent and Anopheles gambiae the major vector.\[[@ref18]\] T2DM is also a health problem among adults living in Alimosho, and cases of poor glycemic control among patients attending public health facilities have been reported.\[[@ref19]\] Based on the availability of health information system to confirm diagnosis, track the duration of T2DM and provide evidence of treatment with antidiabetic therapies, six private health facilities providing T2DM care with laboratory support were purposively selected \[[Figure 1](#F1){ref-type="fig"}\]. They included Talent Specialist Hospital, Egbeda; Santa Maria hospital, Igando Egan; Jamtec Clinicals, Ikotun; Crest Specialist Hospital, Isuti; Blessing Clinic, Akesan and Pakal Hospital, Obadore. The locations of these health facilities were mapped using geographical positioning system \[[Figure 1](#F1){ref-type="fig"}\]. ![Location map showing Alimosho local government area and the study sites. ![](SJMMS-8-32-g001.jpg) Talent specialist hospital; ![](SJMMS-8-32-g002.jpg) Santa Maria hospital;![](SJMMS-8-32-g003.jpg) J Jamtec Clinicals; ![](SJMMS-8-32-g004.jpg) Crest specialist hospital; ![](SJMMS-8-32-g005.jpg) Blessing Clinic; ![](SJMMS-8-32-g006.jpg) Pakal hospital](SJMMS-8-32-g007){#F1} Study population {#sec2-3} ---------------- The study population comprised afebrile patients of both genders, aged ≥40 years, who had a confirmed diagnosis of T2DM and had been receiving its medications for at least 3 months before the study. Patients with T2DM were defined as those on oral hypoglycemic agents and with a homeostasis model assessment for insulin resistance (HOMA-IR) value of ≥2.5 in at least two measurements.\[[@ref20]\] This was above the mean HOMA-IR of 1.78 found in 10 randomly selected age-matched nondiabetic controls in the study settings. Other inclusion criteria were a ≥95% treatment adherence in the previous 2 months (i.e., not missing more than six doses during this period for twice-daily or thrice-daily regimens) and no history of malaria. These patients were enrolled by convenience on their clinic days during the study period. Patients who were obese (body mass index of ≥30 kg/m^2^), hypertensive (mean blood pressure of \>140/90 mmHg for two measurements) and had other diabetes-related complications were not enrolled in this study. Sample size calculation {#sec2-4} ----------------------- The minimum number of T2DM patients required for this study was determined using Epi Info 7.0 (CDC, Atlanta, GA, USA) with significance (α) set at 5% and power (1 -- α) at 80%, as recently described by Sharma.\[[@ref21]\] Assuming that an estimated 10% of the T2DM patients on oral hypoglycemic agents for at least 3 months before the study have malaria parasites in their blood with a confidence limit of 95%, the minimum sample size was calculated using the formula: N = z^2^pq/d^2^, where N = sample size, z = 1.96, p = 0.22, q = 0.78 and d = 0.05 Considering an attrition rate of 10%, the minimum sample size was calculated to be 153 patients. In this study, a total of 208 eligible patients who met all the inclusion criteria for this study were enrolled. Study design {#sec2-5} ------------ This was a cross-sectional, prospective study of ASM infection during March--August 2015, among T2DM patients receiving care at selected private health facilities in Alimosho, Lagos, Nigeria. After providing their written informed consent, the patients were asked to fast, not consume alcohol, smoke tobacco or engage in strenuous work activity/exercise after 10 pm the day before examination. On the examination day, between 8 and 10 am, a pretested semi-structured questionnaire was administered to the patients for obtaining the sociodemographic information, including the number of children aged \<5 years in their household and frequency of fruits and vegetable servings per week. The patients were also clinically examined, underwent anthropometric measurements and provided venous blood samples for laboratory measurement of packed cell volume (PCV), fasting blood glucose (FBG) estimation and malaria diagnosis. Two different trained microscopists computed the results of light microscopy and HRP-2 RDT, with the results of each being blinded from the other. Anthropometric measurements {#sec2-6} --------------------------- All patients were directed to be in light clothing and barefooted for their body weight and height measurement using an electronic weighing and a meter rule. The body weight and height were measured to the nearest 0.1 kg and 0.1 cm, respectively. Normal BMI was defined as 18.5--24.9 kg/m^2^, underweight as \<18.5 kg/m^2^ and overweight as 25--29.9 kg/m^2^. Blood glucose estimation {#sec2-7} ------------------------ FBG was estimated from plasma recovered by centrifugation of whole blood samples collected into fluoride oxalate using a calibrated glucometer (Roche Diagnostics, Indianapolis, USA). Each assay was done in duplicates by a single trained medical laboratory scientist and the average of the two was computed. Patients with a FBG \>126 mg/dL in two consecutive measurements were regarded as having poor glycemic control. Packed cell volume determination {#sec2-8} -------------------------------- To determine PCV (%), heparinized tubes filled with whole blood were centrifuged at 10,000 rpm for 5 min in a microhematocrit centrifuge, followed by PCV measurement using a microhematocrit reader (Hawsley and Sons, London, UK). This was done by the same trained medical laboratory scientist. Participants with PCV values of \<36% were defined as anemic, assuming that 3% PCV is equivalent to 1 g/dL hemoglobin concentration in whole blood.\[[@ref22]\] In this study, mild, moderate and severe anemia is defined as PCV values of 33--36%, 30--33% and \<30%. Insulin resistance measurement {#sec2-9} ------------------------------ In a fasted state, plasma insulin concentration was measured by a solid-phase sandwich ELISA using the Insulin Human ELISA Kit (Invitrogen, Carlsbad, CA, USA) according to the manufacturer\'s protocol. HOMA-IR was calculated by multiplying insulin (U/mL) by FBG (mg/dL) and dividing by 405.\[[@ref23]\] Detection, identification and quantitation of malaria parasites {#sec2-10} --------------------------------------------------------------- ### Rapid diagnostic test {#sec3-1} SD-Bioline Malaria-Ag-Pf/Pan™ (Batch no. 05CDD006A, Standard Diagnostic Inc. Suwon City, South Korea) test kit targeting the HRP2 of P. falciparum was used to detect its presence in the ethylenediaminetetraacetic acid (EDTA) blood samples of the patients. Same as light microscopy, this test was also carried out within 30 min of sample collection, according to the manufacturer\'s protocol. Results were interpreted following World Health Organization guidelines.\[[@ref24]\] ### Light microscopy {#sec3-2} Within 30 min of blood collection in EDTA tubes, thick and thin blood films were prepared on two grease-free slides per sample, followed by staining with 10% Giemsa stain (Capitol Scientific Inc. Austin Texas, USA). Slides were examined under oil immersion by two trained microscopists blinded to HRP2-RDT results by counting the number of parasites against 500 leukocytes in the thick film and quantifying parasitemia as parasites per microliter of whole blood. This quantification assumed that there are 8000 leukocytes per microliter of blood. A slide was considered negative after examining 100 high-power fields without finding a malaria parasite.\[[@ref25]\] The thin film was examined for malaria parasite speciation. All malaria parasites were identified as P. falciparum, and no discordance of \>15% count were recorded to warrant examination by a third expert microscopist. For every positive slide, an average count of two readings was computed and recorded by the single trained microscopist. ### Malaria diagnosis by polymerase chain reaction {#sec3-3} Parasite genomic DNA extraction: For extracting the parasite genomic DNA, blood spot samples in 3 mM Whatman filter paper were hemolyzed by soaking in a 1.5-mL Eppendorf tube containing 0.5% saponin in phosphate-buffered saline (PBS) (pH 7.2), followed by incubation for 10 min at room temperature. After centrifugation at 14,000 rpm for 10 min, the supernatant was discarded and blood spots were washed with 1 mL of PBS. Then, 150 μL of 2% Chelex-100 (BioRad, Hercules, CA, USA) and 50 μL of sterile water were added to the blood spot, followed by incubation at 100°C for 10 min. After centrifugation at 10,000 rpm for 1 min, the supernatant was collected into a fresh Eppendorf tube and stored at 4°C prior to PCR assay within 24 h of preparation.PCR: The nested PCR protocol of Fuehrer et al.\[[@ref26]\] and Snounou et al.\[[@ref27]\] was adopted to amplify the 18s rRNA gene of the malaria parasite using species-specific primers; cycling conditions and annealing temperatures are shown in [Table 1](#T1){ref-type="table"}. All PCR amplifications (Techne TC-312 thermocycler; Techne, Burlington, NJ, USA) were performed in 20 μL volume containing 1.5 mM MgCl~2~, 200 nM of each of the dNTPs, 20 picomole of each primer and 1. 25 U of Taq polymerase (Invitrogen^®^, Carlsbad, CA, USA). Electrophoresis was performed on 2.5% agarose gel to analyze the PCR products. A sample was considered positive if a a 500-, 125- or 105-bp product for P. falciparum, Plasmodium malariae or Plasmodium ovale was detected. In this study, all the ASM cases detected were caused by P. falciparum \[[Figure 2](#F2){ref-type="fig"}\]. ###### Primers, sequences, cycling conditions and annealing temperatures for the amplification of 18s rRNA genes of Plasmodium species ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Gene PCR round Primer name Primer sequence (5′-3′) Cycling conditions Annealing condition (°C) Positive control ------------- --------------------------------------------------------------- ------------- --------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------ -------------------------- ------------------ 18s rRNA Primary rPLU1 rPLU5 TTAAAATTGTGCAGTTAAAAGGA\ Initial denaturation: 95°C for 5 min; PCR: 25 cycles of 94°C for 1 min, 58°C for 2 min, 72°C for 2 min; final elongation: 72°C for 5 min 58 3D7 CCTGTTGTTGCCTTAAACTTC 18s rRNA Secondary rFALf rFALr TTAAACTGGTTTGGGAAAACCAAATATATT ACACAATGAACTCAATCATGACTACCCGTC Initial denaturation: 95°C for 5 min; PCR: 30 cycles of 94°C for 1 min, 65°C for 2 min, 72°C for 2 min; final elongation: 72°C for 5 min 65 3D7 rMALf\ ATAACATAGTTGTACGTTAAGAATAACCGC AAAATTCCCATGCATAAAAAATTATACAAA rMALr rVIVf\ CGCTTCTAGCTTAATCCACATAACTGATAC ACTTCCAAGCCGAAGCAAAGAAAGTCCTTA rVIVr rOVAf rOVAr ATCTCTTTTGCTATTTTTTAGTATTGGAGA GGAAAAGGACACATTAATTGTATCCTAGTG ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- PCR -- Polymerase chain reaction ![Agarose gel electrophoresis of the polymerase chain reaction amplified 18s rRNA gene of Plasmodium falciparum from the clinical samples of the type 2 diabetes mellitus patients. M = 100 base pair DNA ladder markers; Lane 1 = Positive control, 3D7 Plasmodium falciparum; Lanes 2--4, 6--8, 11, 12, 14--22 = Plasmodium falciparum- infected blood samples; Lanes 5, 9, 10 and 13 = non-Plasmodium falciparum-infected blood samples. Faint bands on lanes 2, 3, 7, 14--19 were artifacts](SJMMS-8-32-g008){#F2} Data management and analysis {#sec2-11} ---------------------------- Data were entered into Microsoft Excel, checked for entry errors, edited and transferred to SPSS version 17.0 (SPSS, Inc., Chicago, IL, USA) for analysis. Categorical variables were expressed as frequency and percentages and continuous variables as mean ± standard error of mean and median. Medications taken by the patients were categorized into two groups: single therapy, if treatment involved only one hypoglycemic agent, or combination therapy, if treatment involved two or more oral hypoglycemic agents. Data of T2DM with and without ASM (as frequency and percentages) were compared using chi-square test or Fisher\'s exact test. The measures of performance of HRP2-RDT and light microscopy, such as sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV), were benchmarked against that of PCR and expressed as percentages (95% confidence interval \[CI\]). RESULTS {#sec1-3} ======= Occurrence and sociodemographic determinants of ASM, anemia and poor glycemic control are summarized in [Table 2](#T2){ref-type="table"}. Of the 208 T2DM patients included in this study, ASM was diagnosed in 35 (16.8%), 15 (7.2%) and 9 (4.3%) patients by PCR, light microscopy and HRP2-RDT, respectively. The PCR-positive patients also comprised all patients diagnosed by light microscopy and HRP2-RDT, thereby validating the use of PCR for benchmarking. ###### Sociodemographic determinants of asymptomatic malaria, anemia and poor glycemic control among the type 2 diabetes patients Characteristics n ASM\, n* (%) P Anemia, n (%) P PGC, n (%) P P\^ ------------------------------------------------------------------------------------------------------------ ----- ---------------- ----------- ----------------- ------------ -------------- ------------- -------------- Age (years) ≤50 87 25 (28.7) 0.001 17 (19.5) 0.84 47 (54) 0.001 \<0.001 \>50 121 10 (10.7) 25 (20.7) 94 (78.5) Sex Male 93 21 (22.6) 0.046 20 (21.5) 0.67 55 (59.1) 0.047 \<0.001 Female 115 14 (12.2) 22 (19.1) 83 (72.2) Education No education 31 5 (16.1) 0.78 7 (22.6) 0.95 10 (32.2) \<0.001 \<0.0001 Primary 103 18 (17.5) 0.92 20 (19.4) 0.78 73 (70.9) 0.33 0.3 ≥Secondary 74 12 (16.9) Reference 15 (21.1) Reference 55 (77.5) Reference Reference Marital status Married 112 18 (16.1) 0.75 19 (17) 0.21 64 (57.1) 0.02 \<0.001 Not married 96 17 (17.7) 23 (24) 74 (77.1) Occupation Working 125 20 (16) 0.69 20 (16) 0.06 75 (60) 0.017 0.017 Not working 83 15 (18.1) 22 (26.5) 63 (75.9) Household, number of children aged \<5 years 0 132 18 (13.6) Reference 17 (12.9) Reference 88 (66.7) Reference 1-2 58 15 (25.9) 0.04 16 (27.6) 0.014 42 (72.4) 0.43 0.9 ≥3 18 5 (27.8) 0.11 9 (50) \<0.01 8 (44.4) 0.065 0.007 Use of LLIN the previous night among children aged \<5 years living with type 2 diabetes patients (n=76) Yes 30 7 (23.3) 0.78 9 (30) 0.56 19 (63.3) 0.44 0.29 No 46 12 (25) 11 (25.9) 25 (54.3) P^\^^ -- P value adjusted by parasitemia for PGC. Significant P or P\^ values are in bold. PCR -- Polymerase chain reaction; ASM -- Asymptomatic malaria by PCR; PGC -- Poor glycemic control; LLIN -- Long-lasting insecticidal net. \*P\<0.05 was considered significant Forty-two (20.2%) patients had mild to moderate anemia and 138 (66.3%) had poor glycemic control, and ASM was significantly associated with both (P \< 0.04). In addition, ASM was found to be significantly (P \< 0.05) associated with age, male gender and living with 1--2 children aged \<5 years \[Tables [2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}\]. Further, the overall parasitemia ranged from 85 to 3789 parasites/μL. The median of parasitemia was 1580 parasites/μL, FBG was 134.3 mg/dL and PCV was 37.8%. ###### Assessment of asymptomatic malaria by glycemic control and packed cell volume among the type 2 diabetes patients Parasite indices n PGC, n (%) FBG (mg/dL), mean ± SEM Anemia, n (%) PCV (%), mean ± SEM --------------------------------- ----- ----------------------------- -------------------------- ----------------- --------------------- ASM^\^^ Yes 35 29 (82.9) 139.8 ± 5.9 13 (37.1) 35.3 ± 0.6 No 173 109 (63) 124.7 ± 7.2 29 (16.8) 39.5 ± 1.1 P \<0.05 \<0.05 \<0.05 \<0.05 Parasitemia\* (parasites/μL) \<100 2 1 (50) 126 ± 4.4 1 (50) 35.8 ± 0.7 100-1000 5 2 (40) 134.3 ± 5.1 2 (50) 35.1 ± 0.5 \>1000 8 6 (75) 145.2 ± 5.9 6 (66.7) 34.2 ± 0.4 P (ANOVA) \>0.05 \<0.05 \>0.05 \>0.05 Parasitemia, range (median) Range^@^ (median) mg/dL Range^@^ (median), % 85-3780 (1580) 120-158 (134.3) 32-41.3 (37.8) ^\^^ASM -- ASM cases detected by PCR; \Parasitemia based on light microscopy; ^@^Range (median) of ASM cases only. PGC -- Poor glycemic control; Significant P* values were bolded. PCR -- Polymerase chain reaction; ASM -- Asymptomatic malaria by PCR; FBG -- Fasting blood glucose; PCV -- Packed cell volume; SEM -- Standard error of mean Of all the clinical and behavioral factors analyzed, only HOMA-IR was found to be associated with ASM (P \< 0.05). However, three of these factors, including a T2DM duration of \>5 years and HOMA-IR \>2.5, were significantly (P \< 0.05) associated with poor glycemic control \[[Table 4](#T4){ref-type="table"}\], ruling out the influence of cofounders such as overweight and obesity. ###### Clinical and self-management behavioral characteristics of the type 2 diabetes patients with asymptomatic malaria and poor glycemic control Characteristics N ASM\^, n (%) P Poor glycemic control, n (%) P ----------------------------------------------------------- ----- ---------------- ------------ -------------------------------- -------------- Duration of diabetes, years ≤5 116 19 (16.4) 0.84 63 (54.3) \<0.001 \>5 92 16 (17.4) 75 (81.5) HOMA-IR ≥2.5 No 77 7 (9.1) 0.02 38 (40.3) \<0.0001 Yes 131 28 (21.8) 100 (81.7) Medication Monotherapy 103 19 (18.4) 0.46 62 (60.2) 0.06 Combination therapy 105 16 (15.2) 76 (72.4) Frequency of exercise in the previous week, days \<3 135 20 (14.8) 0.29 91 (67.4) 0.66 ≥3 73 15 (20.5) 47 (64.4) Frequency of fruit and vegetable servings per day \<3 139 27 (19.4) 0.16 95 (68.3) 0.39 ≥3 69 8 (11.6) 43 (62.3) Frequency of self-blood glucose monitoring per week, days ≤5 142 23 (16.2) 0.72 88 (62) 0.05 \>5 66 12 (33.3) 50 (75.8) Cigarette intake Yes 23 5 (21.7) 0.5 19 (82.6) 0.08 No 185 30 (16.2) 119 (64.3) \^ASM -- ASM cases detected by PCR; Significant P values are in bold. HOMA-IR -- Homeostasis model assessment for insulin resistance; PCR -- Polymerase chain reaction; ASM -- Asymptomatic malaria by PCR Compared to PCR, the number of false negatives by light microscopy and HRP2-RDT were 20 (57.1%) and 27 (77.1%), respectively; interestingly, HRP2-RDT detected one ASM case not identified by PCR \[[Table 5](#T5){ref-type="table"}\]. Benchmarking against PCR, light microscopy was found to have a sensitivity (95%CI) of 42.9% (26.5--59.3), specificity of 100%, NPV of 89.6% (82.3--93.9) and PPV of 100% for diagnosing ASM, while those for HRP2-RDT were 22.9%(12.1--39), 99.4% (96.8--99.9), 86.4% (81--90.5) and 88.9% (56.5--98), respectively. The proportion of cases that were correctly diagnosed by HRP2-RDT and light microscopy were 86.5% and 90.4%, respectively, indicating that light microscopy is more accurate in diagnosing cases of ASM \[[Table 6](#T6){ref-type="table"}\]. ###### Performance evaluation of light microscopy and histidine-rich protein 2 rapid diagnostic test using polymerase chain reaction as a reference for the diagnosis of asymptomatic malaria among the type-2 diabetes patients Diagnostic test PCR results Total Performances, % (95% CI) ----------------------- ------------- ------- -------------------------- ------------------ ------------------ ---------------- ------------------ Light microscopy Positive 15 0 15 42.9 (26.9-59.3) 100 100 89.6 (82.3-93.9) Negative 20 173 193 Rapid diagnostic test Positive 8 1 9 22.9 (12.1-39) 99.4 (96.8-99.9) 88.9 (56.5-98) 86.4 (81-90.5) Negative 27 172 199 P\<0.05 was considered significant. PCR -- Polymerase chain reaction; CI -- Confidence interval; PPV -- Positive predictive value; NPV -- Negative predictive value ###### Diagnostic effectiveness analysis between light microscopy and histidine-rich protein 2 rapid diagnostic test for asymptomatic malaria in the studied patients Indicator Rapid diagnostic test Microscopy ------------------------------------- ----------------------- ------------ Number of cases correctly diagnosed 180 188 Proportion correctly diagnosed (%) 86.50 90.40\* \*P (χ^2^)=0.1 DISCUSSION {#sec1-4} ========== Nigeria\'s adoption of the global malaria elimination target by 2030 through its National Malaria Strategic Plan 2014--2020 has necessitated the need for new surveillance approaches aimed at identifying potential reservoirs of malaria parasites and development of novel interventions to interrupt transmission. In this study, 35 of the 208 afebrile T2DM patients were found to harbor P. falciparum by PCR. This finding agrees with a previous study conducted in Ghana by Danquah et al.,\[[@ref16]\] who found that T2DM patients harbored P. falciparum asymptomatically. Therefore, this collective evidence suggests that adult T2DM patients are potential reservoirs of P. falciparum. In malaria-endemic settings, it is well-known that adults are better protected from uncomplicated and severe malaria compared with children aged \<5 years, as they are likely to have acquired antimalarial immunity over time from constant exposure to malaria parasites.\[[@ref28]\] However, it is now important to investigate high-risk adult populations for asymptomatic carriage of malaria parasites in a malaria-endemic setting like Nigeria for ensuring malaria elimination. The ASM diagnosed among the adult T2DM cohort studied may also be attributed to antiparasitic immunity, which prevents parasitemia from reaching a clinical level called pyrogenic threshold and premunition for preventing secondary infections in adults with primary asymptomatic infection.\[[@ref28][@ref29][@ref30][@ref31][@ref32]\] In this study, the rate of ASM in T2DM patients according to PCR was 16.8%, higher than that found by light microscopy (7.2%) and HRP-2 RDT (4.3%). Danquah et al.,\[[@ref16]\] in their study of 675 T2DM Ghanaians patients, also found higher ASM rate with PCR (14.1%) than light microscopy (0.9%). Therefore, our findings further add evidence that PCR is more sensitive in identifying P. falciparum reservoirs than light microscopy and HRP2-RDT, and thus its use would ensure that even cases of low parasitemia are detected. It also implies that in malaria endemic areas, if T2DM patients have poor glycemic control despite adherence to medication, PCR should be carried out despite negative malaria test by HRP-2 RDT or light microscopy to completely rule out ASM.\[[@ref31]\] The single positive case detected by HRP2-RDT for which PCR was negative suggests a false positive result due to persistent antigenemia because light microscopy was also negative. It has been shown that P. falciparum HP2 antigen may persist for weeks after successful clearance of the parasites by an effective antimalarial therapy.\[[@ref33]\] Our findings also support previous findings of ASM being common in malaria-endemic settings and that light microscopy detects only a small proportion of these ASM cases.\[[@ref34]\] In the current study, the median parasitemia level was 1580/μL (range: 85--3780/μL), which differs from the 880/μL (range: 80--4950/μL) found by Danquah et al.\[[@ref16]\] in a Ghanaian population. However, such differences can be attributed to different sample size, geographical location and level of malaria elimination activities between the studied settings. Although HRP2-RDT is considered an essential tool for enhancing the diagnosis of malaria in sub-Saharan Africa, especially in areas where access to light microscopy and PCR are limited or lacking, its performance in this study was not encouraging. In fact, of the three methods used, HRP2-RDT had the lowest sensitivity (only 22.9% compared with PCR). Benchmarking against PCR results, HRP2-RDT made correct diagnosis in only 86.5% of the cases compared with 90.4% by light microscopy. This can be attributed to our study having a number of cases with \<100 parasites/μL, which is below the detection threshold of RDT.\[[@ref34]\] In terms of light microscopy, its low sensitivity in this study may be attributed to the number of patients with submicroscopic ASM infections (n = 20). It should be noted that other studies have also recently reported submicroscopic malaria infection in T2DM patients and other carriage populations from sub-Saharan Africa.\[[@ref16][@ref34][@ref35]\] In this study, although HRP2-RDT diagnosed a lower proportion of ASM cases in T2DM patients than light microscopy, these results should be interpreted with caution. This is because previous studies have found inconsistent results while comparing these methods in different populations such as pregnant women, apparently healthy blood donors and other geographical locations in sub-Saharan Africa.\[[@ref35][@ref36][@ref37]\] Therefore, to validate the findings of this study, the authors recommend large-scale studies from other parts of Nigeria where the efficiency of other diagnostic methods such as loop-mediated isothermal PCR may also be evaluated. In this study, ASM (both microscopic and submicroscopic) was found to be significantly associated with poor glycemic control. As ASM was also significantly associated with HOMA-IR \>2.5, insulin resistance may also be an interlinking factor between ASM and poor glycemic control. Further, two recent studies on T2DM patients from Central Africa and Sweden reported a significant relation between insulin resistance and malaria infection.\[[@ref13][@ref38]\] The finding of the current study could connote serious clinical implications for Nigerian patients, as it suggests that ASM can potentially compromise the management of T2DM and may cause anemia of unknown etiology. However, the mild anaemia cases seen in this study may also be due to chronic hydrolysis of erythrocytes by the malaria parasites in the infected patients despite low parasitemia. Therefore, the authors recommend integration of ASM screening into the care of Nigerian patients with T2DM, especially among those with anemia and poor glycemic control despite \>95% treatment adherence. The present study also confirmed previously established factors for poor glycemic control in T2DM patients such as age, level of education, duration of disease and frequency of self-blood glucose monitoring.\[[@ref39][@ref40]\] Therefore, intensified and sustained awareness development through education, training and engagement of diabetes patients is also needed as a measure to control the rising trend of T2DM in the study area and Nigeria. It is important to note that the present study is limited by its cross-sectional design and its sampling coverage. These factors hindered the determination of protective immunity against clinical malaria offered by ASM in the infected patients. This also prevents generalization of the findings to all T2DM patients in the country. Therefore, future longitudinal studies would not only increase sampling coverage but also characterize the genetic diversity and complexity of ASM infection among T2DM patients. CONCLUSION {#sec1-5} ========== This study found that in Lagos, Nigeria, T2DM patients are potential reservoirs of asymptomatic P. falciparum and that ASM has a significantly negative effect on glycemic control. The authors recommend that in T2DM patients with poor glycemic control despite optimal treatment adherence, malaria screening should be carried out using PCR to detect submicroscopic ASM. Ethical considerations {#sec2-12} ---------------------- This study was approved by the Ethics Committee of Lagos State Hospital (Ref. no.: LSHMB 027) on February 4, 2015, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki, as revised in 2013. All the participants provided written informed consent before their inclusion in this study. Peer review {#sec2-13} ----------- This article was peer reviewed by three independent and anonymous reviewers. Financial support and sponsorship {#sec2-14} --------------------------------- Nil. Conflicts of interest {#sec2-15} --------------------- There are no conflicts of interest. The authors thank the expert microscopists at Nigerian Institute of Medical Research for validating speciation and parasitemia data obtained in this study. The authors would also like to thank the physicians, nurses and patients at the six diabetes care health facilities in Alimosho where this study was carried out.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1} =============== REM sleep behavior disorder (RBD) is characterized by dream-enacting behaviors and unpleasant dreams and presents a risk for self-injury and harm to others due to abnormal REM sleep during which control of muscle tonus is lacking (REM sleep without atonia) \[[@B1], [@B2]\]. RBD is a heterogeneous disease entity consisting of a variety of manifestations \[[@B1]\]. Idiopathic RBD (iRBD), which develops in middle age or later and progresses chronically, in particular is a common clinical manifestation of Lewy body-related syndrome and is regarded as a clinical entity from the pathological aspect. For example, it has been elucidated that iRBD is often accompanied by soft motor signs, olfactory and color identification deficits, decreased cardiovascular and respiratory changes between REM and NREM sleep, reduced cardiac ^123^I-metaiodobenzylguanidine (^123^I-MIBG) uptake, impairment of visual memory and visuospatial construction on neuropsychological testing, EEG slowing during wakefulness or sleep, and decreased strial dopaminergic innervations, and reduced presynaptic strial dopamine transporter binding on SPECT or positron emission topography (PET) scans, which are considered as nonmotor symptoms of Parkinson\'s disease (PD) \[[@B3]--[@B5]\]. Furthermore, despite the limited number of pathological reports on iRBD, the characteristics of iRBD have been supported to have a close relationship with Lewy body pathology \[[@B6]\]. Therefore, additional clinical features that could distinguish iRBD with Lewy body-related α-synucleinopathies from iRBD from other causes would be helpful in clinical practice. Recently, an association between loss of olfactory function and loss of cardiac noradrenergic innervation in PD has been noted \[[@B7]\]. Moreover, both loss of sense of smell and cardiac sympathetic denervation can precede the onset of motor symptoms, suggesting that the combination might provide a biomarker for the risk of Lewy body disease \[[@B8]\]. In this study, by the combination of tests to detect odor identification \[[@B9]\] and to determine cardiac ^123^I-MIBG cardiac uptake \[[@B4], [@B5]\] that were selected from among various tests to examine nonmotor symptoms of PD, we investigated if there was an association between olfaction and cardiac ^123^I-MIBG uptake in patients with iRBD and PD to evaluate the correlation between results of these two tests and to determine which would more enhance the detection of pathogenesis resembling that of Lewy body-related α-synucleinopathies in patients clinically diagnosed as having iRBD. 2. Methods {#sec2} ========== 2.1. Patient Selection {#sec2.1} ---------------------- This study was performed in accordance with the Declaration of Helsinki. Procedures were approved by the Ethics Review Committee of Dokkyo Medical University, and informed consent was obtained from each subject. Subjects were 82 patients matched according to age group: 30 had iRBD (iRBD group; 66.3 ± 5.7 years; 25 males, 5 females), 38 had PD (PD group; 65.4 ± 9.2 years; 24 males, 14 females), and 20 had obstructive sleep apnea (OSA) without RBD (control group; 62.3 ± 6.8 years; 18 males, 2 females; pretreatment AHI = 38.5 ± 31.3 events/h) ([Table 1](#tab1){ref-type="table"}). RBD and OSAS were defined according to the International Classification of Sleep Disorders, second edition \[[@B2]\]. Exclusion criteria were an abnormal neurologic examination in subjects with iRBD and OSAS. PD patients fulfilled the United Kingdom Parkinson\'s Disease Society Brain Bank criteria for idiopathic PD \[[@B10]\]. Parkinsonism was assessed and classified according to the Hoehn and Yahr stage \[[@B11]\] and the motor subset of Unified Parkinson\'s Disease Rating Scale (UPDRS-III), which was administered to the PD and iRBD groups \[[@B12]\]. The mean UPDRS-III score was 14.4 ± 7.8 points for PD patients and 0.6 ± 1.0 points for the iRBD subjects. There were two phenotypes of PD among the PD group: tremor-dominant and postural instability gait difficulty-(PIGD-) dominant based on UPDRS components. The mean PIGD score was defined as the sum of an individual\'s baseline falling, freezing, walking, gait, and postural stability scores divided by 5. Patients were categorized as having tremor-dominant PD if the ratio of the mean tremor score to the mean PIGD score was 1.5 or higher and as having PIGD-dominant PD if that ratio was 1.00 or lower \[[@B13]\]. According to that categorization, 13 patients in the PD group were placed in the tremor-dominant subgroup and 21 in PIGD-dominant subgroup. For this particular study, all patients with Mini-Mental State Examination (MMSE) scores \<24 were excluded \[[@B14]\]. 2.2. Polysomnographic Evaluation and Definition of iRBD {#sec2.2} ------------------------------------------------------- Polysomnographic (PSG) monitoring included electroencephalography (C3, C4, O1, and O2), electrooculography, chin muscle electromyography (EMG), electrocardiography, detection of airflow by thermistor, plethysmography for ribcage and abdominal wall motion, oximetry for measurement of arterial oxyhemoglobin saturation, detection of changes in sleeping position, and bilateral EMG of the tibialis anterior muscles. PSG monitoring was performed for at least 8 h. Sleep stages were manually scored according to criteria of Rechtschaffen and Kales \[[@B15]\]. Apnea was defined as the absence of breathing for more than 10 s. Hypopnea was defined as a reduction of more than 50% in breathing; when the reduction in breathing was less than 50%, more than 3% oxygen desaturation or arousal for more than 10 s was defined as hypopnea. The apnea-hypopnea index was calculated as the average of the total number of apnea and hypopnea episodes experienced per hour of sleep. For the iRBD group, the presence of REM sleep without atonia was based on EMG findings of excessive sustained or intermittent elevation of submental EMG tone or excessive phasic submental or lower limb EMG twitching according to the ICSD-2ed \[[@B2]\]. 2.3. Cardiac ^123^I-MIBG Scintigraphy {#sec2.3} ------------------------------------- ^123^I-MIBG planar images of the chest were obtained using a triple-headed gamma camera (GCA-9300A-HG, Toshiba Co, Tokyo, Japan). ^123^I-MIBG (Fujifilm RI Pharma Co., Tokyo, Japan) accumulations were recorded at the early (15 min) and delayed phases (4 h). The heart-to-mediastum (H/M) ratio was calculated by dividing the count density of the left ventricular region of interest (ROI) by that of the mediastinal ROI \[[@B16]--[@B18]\]. None of the subjects took medications that could influence the ^123^I-MIBG examination, and none had a history of cardiac disease. 2.4. Odor Stick Identification Test for Japanese (OSIT-J) {#sec2.4} --------------------------------------------------------- The Odor Stick Identification Test for Japanese (OSIT-J) (Daiichi Yakuhin, Co., Ltd., Tokyo, Japan) is composed of 12 different odorants familiar to the Japanese population \[[@B9]\]. All subjects were free from other conditions that can affect olfactory function such as usage of certain medications, chronic nasal infection, chronic sinonasal disease, head trauma, and abuse of drugs or alcohol by medical history. Subjects with an infection of the upper airways at the time of the investigation were also not allowed to participate. OSIT-J was performed as previously described. In this study, functional hyposmia was defined by an OSIT-J score ≤8 \[[@B9]\]. 2.5. Statistical Analysis {#sec2.5} ------------------------- Values are expressed as mean ± SD. Age, MMSE, OSIT-J scores and degrees of accumulation based on the H/M ratio (early and delayed phase) in cardiac muscle obtained were compared among the three groups by the one-way analysis of variance (ANOVA) followed by post-hoc Bonferroni correction. The Mann-Whitney U-test was applied for statistical comparisons of disease duration and scores of the motor subsets of the UPDRS between the iRBD and PD groups. Group comparisons of the frequency of genders and smokers were performed by the Chi-square test for categorical variables. The correlations of OSIT-J and early or delayed ^123^I-MIBG uptake were assessed by the Spearman correlation coefficient for iRBD and PD. To explore the most influential factor on the OSIT-J score, we included age, sex, disease duration, MMSE, Hoehn and Yahr stage, UPDRS motor subsets, motor phenotypes, levodopa-equivalent unit (LEU), smoking status, and delayed ^123^I-MIBG uptake in a multiple regression analysis for PD. Receiver operating characteristic (ROC) curves were analyzed. The significance level was set at P \< .05. Statistical analyses were performed with the Statistical Package for Social Science Software (Graphpad Prism, San Diego, CA and SPSS II Windows Ver 11.0, Japan). 3. Results {#sec3} ========== A total of 30 iRBD patients (mean symptom duration: 5.7 ± 8.8), 38 PD patients, and 20 control subjects with a similar age distribution were evaluated. Demographic information is presented in [Table 1](#tab1){ref-type="table"}. Interval between ^123^I-MIBG examinations and OSIT-J was a mean of 72 days. Cardiac ^123^I-MIBG accumulation based on H/M (early phase, delayed phase) was significantly decreased in the iRBD group (1.85 ± 0.33, 1.46 ± 0.29) and the PD group (2.14 ± 0.52, 1.83 ± 0.60) compared with the control group (2.74 ± 0.33, 2.82 ± 0.47) (P \< .000). OSIT-J scores were significantly lower in the iRBD group (5.2 ± 3.0) and the PD group (4.4 ± 2.4) than in the control group (10.3 ± 1.3) (P \< .000) ([Table 1](#tab1){ref-type="table"}). As shown in [Table 2](#tab2){ref-type="table"}, results of multiple regression analysis of OSIT-J-related parameters in patients with PD, which included age, sex, disease duration, MMSE, Hoehn and Yahr stage, UPDRS motor subset, motor phenotype, LEU, smoking status, and delayed H/M ratio, showed that age, sex, and delayed H/M ratio were the significant factors in the model for the OSIT-J score. There was a significant positive correlation between the OSIT-J score and delayed H/M ratio (r = 0.337, P = .039, Spearman\'s correlation). On the other hand, results of multiple regression analysis of OSIT-J-related parameters in patients with iRBD, which included age, sex, disease duration, MMSE, UPDRS motor subset, smoking status, and delayed H/M ratio, showed that there were no significant factors in the model for the OSIT-J score. Among the iRBD group, there was no significant correlation between the OSIT-J score and delayed H/M ratio (r = −0.195, P = .302, Spearman\'s correlation) or the duration of RBD (r = −0.217, P = .249, Spearman\'s correlation). ROC analysis of the early or delayed H/M ratio and odor identification showed an area under the curve of 0.96 (95% CI = 0.92--1.00, P = .000), 0.99 (95% CI = 0.96--1.00, P = .000) and 0.96 (95% CI = 0.91--1.00, P = .000) for iRBD ([Figure 1(a)](#fig1){ref-type="fig"}) and of 0.84 (95% CI = 0.74--0.94, P = .000), 0.88 (95% CI = 0.79--0.97, P = .000) and 0.98 (95% CI = 0.96--1.00, P = .000) for PD ([Figure 1(b)](#fig1){ref-type="fig"}). For differentiating the iRBD patients from control subjects, the sensitivity and specificity of the OSIT-J score with a cut-off value at 8.5 were 90.0% and 83.3%, respectively; those of the early H/M ratio with a cut-off value at 2.30 were 95.0% and 86.7%, respectively, and those of a delayed H/M ratio with a cut-off value at 2.03 were 95.0% and 96.7%, respectively ([Table 3](#tab3){ref-type="table"}). For differentiating PD patients from control subjects, the sensitivity and specificity of the OSIT-J score with a cut-off value at 8.5 were 90.0% and 92.1%, respectively; those of the early H/M ratio with a cut-off value at 2.30 were 95.0% and 71.1%, respectively; and those of the delayed H/M ratio with a cut-off value at 2.03 were 95.0% and 71.1%, respectively ([Table 3](#tab3){ref-type="table"}). With optimal cut-off values set at 2.03 for the delayed H/M ratio and at 8.5 for OSIT-J score for iRBD patients ([Table 3](#tab3){ref-type="table"}, [Figure 2](#fig2){ref-type="fig"}), identification of iRBD was confirmed in 23 (76.7%) patients in the iRBD group through abnormal results of both of these tests. Optimal cut-off values were set at 2.03 for delayed H/M ratio and at 8.5 for OSIT-J score for PD patients ([Table 3](#tab3){ref-type="table"}, [Figure 2](#fig2){ref-type="fig"}); as a result, 25 (65.8%) were confirmed through abnormal results of both of these tests to have PD. However, results of both of these tests were normal in the control group. 4. Discussion {#sec4} ============= In this study, the olfactory dysfunction and reduced cardiac ^123^I-MIBG uptake that were found in most iRBD patients share similarities with those described in PD. Moreover, delayed cardiac ^123^I-MIBG uptake was more severe in the iRBD group than in the PD group ([Figure 1(a)](#fig1){ref-type="fig"}) while reduced odor identification was more severe in those with PD than with iRBD ([Figure1(b)](#fig1){ref-type="fig"}). Also, the degree of olfactory identification impairment correlated significantly with the degree of delayed cardiac ^123^I-MIBG uptake only in the PD group. Age, sex, and delayed cardiac ^123^I-MIBG uptake were significant factors with regard to PD patients in the model using the OSIT-J score. The degree of reduction of MIBG uptake does not match the degree of reduction in odor identification in iRBD patients but does match in PD patients. Results of a recent study indicated that the cardiac sympathetic nervous system might degenerate in parallel with the olfactory system in patients with early PD and that these two systems might degenerate at different rates of speed in advanced PD \[[@B16]\]. 4.1. Cardiac Sympathetic Denervation Is Related to Lewy Body Pathology {#sec4.1} ---------------------------------------------------------------------- There is evidence that ^123^I-MIBG cardiac uptake is markedly reduced in patients with Lewy body diseases such as PD, dementia with Lewy bodies (DLB), and pure autonomic failure (PAF) \[[@B17]--[@B19]\]. In was reported that cardiac ^123^I-MIBG imaging could distinguish between clinically diagnosed DLB and Alzheimer\'s disease (AD) with high levels of sensitivity and specificity \[[@B19]\]. Interestingly, pathological findings occur even in patients with DLB who have no parkinsonism. Since PAF patients do not have parkinsonism or decreased striatal dopaminergic innervation and since cardiac noradrenergic denervation occurs in both diseases, the pathogenetic mechanisms of cardiac noradrenergic denervation in Lewy body diseases differ from those producing parkinsonism and nigrostriatal dopaminergic denervation \[[@B8]\]. As pathological evidence \[[@B20]--[@B22]\], it has been proven that the Lewy body is present in cardiac sympathetic nerve postganglionic fibers and it has been suggested that Lewy body-related pathology potentially causes severe denervation and reduced ^123^I-MIBG uptake in the cardiac postganglionic sympathetic nerve. The reduction in ^123^I-MIBG uptake in sympathetic terminals was observed in cases with early-phase PD and incidental Lewy body disease (ILBD) irrespective of the presence or absence of remarkable autonomic nerve injury \[[@B17]\]. 4.2. Olfactory Dysfunction Is Related to Lewy Body Pathology {#sec4.2} ------------------------------------------------------------ Olfactory disturbance is present in most PD or PAF cases \[[@B8], [@B23], [@B24]\]. The deficit in olfaction in PD contrasts with previous reports of preserved or only mildly reduced olfaction in patients with atypical parkinsonism such as a tauopathy or multiple system atrophy \[[@B25], [@B26]\]. In dementia patients, neuropathologic studies reported neuronal alterations in several subcortical structures such as the olfactory tract/bulb, anterior olfactory nucleus, orbito-frontal cortex, hippocampus, and amygdala in the olfactory system \[[@B27]\]. Olfactory abnormalities have been reported in AD, but anosmia appears to be common in DLB but not in pure AD \[[@B28]\]. Interestingly, a Lewy body variant of AD had an increased frequency of anosmia compared with "pure" AD \[[@B29]\]. Furthermore, olfactory impairment is more marked in patients with mild dementia with Lewy bodies than in those with mild AD \[[@B30]\]. In addition to reduced odor identification prior to the onset of PD, olfactory dysfunction also has been frequently recognized in patients with ILBD \[[@B31]\]. Also, neuropathological olfactory bulb alpha-synclein has high specificity and sensitivity for Lewy body formation in confirmed cases of PD and DLB \[[@B32]\]. On the basis of pathological studies of a large number of autopsy cases, Braak et al. proposed a hypothesis as to the onset and advancement pattern of PD in that the disease developed from the medulla and olfactory bulb and extended to the pons and substantia nigra (SN) \[[@B33]\]. In the Honolulu-Asia Aging Study on Japanese Americans, 2,267 males without PD and dementia at the time of olfaction testing were followed up and significantly more subjects who developed olfactory dysfunction in the first 4-year follow-up period also developed PD \[[@B34]\]. Haehner\'s study was to clinically follow up patients with idiopathic hyposmia to determine the percentage of patients who developed idiopathic PD after a 4-year interval \[[@B35]\]. In Ponsen\'s prospective study involving first-degree relatives of PD patients, a low score on three olfactory processing tasks was associated with an increased risk of developing PD within 5 years \[[@B36]\]. The point of view that reduced odor identification is manifested at the very beginning of the development of PD has been supported. These reports indicate that the rhinencephalon may be an area of selective vulnerability for α-synuclein accumulation \[[@B37]\] and iRBD that develops in middle age or older progresses mostly to Lewy body diseases among synucleinopathies such as PD and DLB, which have some pathological features in common. 4.3. RBD and Relevance to Lewy Body Pathology {#sec4.3} --------------------------------------------- Men over the age of 50 years who have iRBD are at very high risk for future PD or DLB several years after the onset of RBD \[[@B38], [@B39]\]. Neuropathologic studies at autopsy of cases that had been diagnosed with RBD while alive showed that every case had Lewy bodies \[[@B6]\]. For example, cognitive abnormalities appeared 15 years after the onset of RBD and probable DLB was diagnosed, and eventually the presence of Lewy bodies was confirmed pathologically \[[@B40]\]. In PD, approximately 60% of the nigrostriatal neurons of the substantia nigra are degenerated before patients fulfill the clinical criteria of PD \[[@B41]\]. In some cases of PD, the patient appears to develop cortical disease before the motor sign of "stage 3" disease, whereas, iRBD patients with ILBD could be diagnosed after long-standing disease with no evidence of motor or cognitive abnormalities \[[@B3], [@B4]\]. If progression of syncleinopathies is not universal, it is essential to understand the reason. Since a variety of symptoms of PD and disorders resembling PD have been elucidated, Langston \[[@B42]\] proposed a "Parkinson\'s complex" because parkinsonism would represent only the tip of the iceberg as typically viewed by both clinicians and researchers. However, when the disease process is measured by neuronal degeneration, the presence of Lewy bodies and neuritic pathology are widespread in the central and peripheral nervous systems. From this point of view, iRBD can be positioned as an earlier preclinical stage of PD or DLB, or a variant of Lewy body-related α-synucleinopathies. To gain such an understanding, it is necessary to extract a group with abnormalities in a combination of markers from among iRBD patients and provide follow up, considering the possibility that some patients in that group may develop neurodegenerative disease. These steps may help elucidate the possibility that iRBD is the spectrum of manifestations or a subtype of Lewy body-related α-synucleinopathies. In conclusion, reduced cardiac ^123^I-MIBG uptake may be more closely associated with iRBD than olfactory impairment. Moreover, odor identification impairment or cardiac sympathetic function assessed by cardiac ^123^I-MIBG uptake and the nigrostriatal dopaminergic function would occur and progress independently in iRBD patients or PD patients. Limitations of this study include the small number of OSAS patients that comprise the control group. Untreated sleep apnea syndrome may be associated with physical movements during sleep \[[@B1]\], and, therefore, it is important to first differentiate between those with iRBD and OSAS with abnormal sleep behavior. A second limitation is that more than 80% of patients presenting at sleep centers with iRBD are men \[[@B1], [@B3], [@B39]\] so that the male-to-female ratio was not matched in the iRBD, PD, and control groups. Conflict of Interests {#sec5} ===================== The authors report no financial conflict of interest. This work is not an industry supported study. The authors are indebted to their colleagues with whom they work on the RBD, namely, Y. Inoue (Japan Somnology Center, Neuropsychiatry Research Institute and Department of Somnology, Tokyo Medical University). The authors thank T. Hashimoto (Department of Radiology, Dokkyo Medical University School of Medicine) for his helpful comments. ![(a) ROC analysis of the early or delayed H/M ratio and odor identification showed an area under the curve of 0.96 (95% CI = 0.92--1.00, P = .000), 0.99 (95% CI = 0.96--1.00, P = .000), and 0.96 (95% CI = 0.91--1.00, P = .000) for iRBD. (b) ROC analysis of the early or delayed H/M ratio and odor identification showed an area under the curve of 0.84 (95% CI = 0.74--0.94, P = .000), 0.88 (95% CI = 0.79--0.97, P = .000), and 0.98 (95% CI = 0.96--1.00, P = .000) for PD.](PD2011-941268.001){#fig1} ![The relation between the OSIT-J score and the delayed H/M ratio of ^123^I-MIBG uptake in patients with iRBD (the red circle), PD (the green square), and controls (the blue triangle). Broken line (vertical): cut-off value set at 8.5 by OSIT-J to distinguish between iRBD or PD and controls. Broken line (horizontal): cut-off value set at 2.03 by ^123^I-MIBG to distinguish between iRBD or PD and controls. OSIT-J: Odor Stick Identification Test for Japanese; H/M: heart-to-mediastinum; ^123^I-MIBG:^123^I-metaiodobenzylguanidine; OSAS: obstructive sleep apnea syndrome; iRBD: idiopathic rapid eye movement behavior disorder; PD: Parkinson\'s disease.](PD2011-941268.002){#fig2} ###### Clinical characteristics of patients in the iRBD, PD, and control groups. iRBD PD Controls P value --------------------------- ------------- --------------- ------------- ----------- Number of patients, (n) 30 38 20 --- Age, years 65.8 ± 9.1 65.4 ± 9.2 62.3 ± 6.8 0.210^a^ Sex (male, %) 83.3 63.2 90.0 0.102^c^ Disease duration, years 5.7 ± 8.8 4.8 ± 4.8 N/A 0.599^b^ MMSE (\>24) 27.7 ± 2.3 27.9 ± 1.9 28.8 ± 1.4 0.096^a^ Hoehn and Yahr stage N/A 2.4 ± 0.8 N/A N/A UPDRS motor subset, score 0.6 ± 1.0 14.4 ± 7.8 N/A 0.000^b^ LEU, mg/day N/A 238.6 ± 261.9 N/A N/A Smoker, (%) 50.0 39.5 5.0 0.464^c^ OSIT-J, score 5.2 ± 3.0 4.4 ± 2.4 10.3 ± 1.3 0.000^a^ ^123^I-MIBG Early H/M ratio 1.85 ± 0.33 2.14 ± 0.52 2.74 ± 0.33 0.000^a^ Delayed H/M ratio 1.46 ± 0.29 1.83 ± 0.60 2.82 ± 0.47 0.000^a^ Data are mean ± SD. PD: Parkinson\'s disease, iRBD: idiopathic REM sleep behavior disorder, OSAS: obstructive sleep apnea syndrome, MMSE: Mini-Mental State Examination, UPDRS: Unified Parkinson\'s Disease Rating Scale, LEU: daily levodopa-equivalent unit, OSIT-J: Odor Stick Identification Test for Japanese, H/M: heart-to-mediastinum, and ^123^I-MIBG: ^123^I-metaiodobenzylguanidine. ^a^ P value was determined by one-way ANOVA. ^b^ P value was determined by Mann-Whitney U test. ^c^Chi-square test for categorical: PD versus iRBD. ###### Multiple regression analysis of OSIT-J related parameters for Parkinson\'s disease (N = 38) and iRBD (N = 30). PD iRBD --------------------------------- ------------ ----------- -------- ------- Age −0.348 0.037 0.016 0.936 Sex 0.404 0.033 0.080 0.701 Disease duration, month 0.147 0.591 −0.314 0.377 MMSE (\>24) 0.189 0.230 0.184 0.377 Hoehn and Yahr stage −0.013 0.940 N/A N/A UPDRS motor subset 0.079 0.674 −0.153 0.517 Motor phenotype −0.239 0.177 N/A N/A LEU 0.074 0.969 N/A N/A Smoking atatus 0.007 0.798 −0.040 0.834 Delayed H/M ratio (^123^I-MIBG) 0.399 0.015 −0.274 0.183 R^2b^ 0.493 0.184 iRBD: idiopathic REM sleep behavior disorder, PD: Parkinson\'s disease, OSIT-J: Odor Stick Identification Test for Japanese, MMSE: Mini-Mental State Examination, UPDRS: Unified Parkinson\'s Disease Rating Scale, LEU: daily levodopa-equivalent unit, H/M: heart-to-mediastinum, and ^123^I-MIBG: ^123^I-metaiodobenzylguanidine. ^a^Standardized regression coefficients. ^b^Coefficient of determination. Bold values indicate statistically significant values. N/A: not aplicable. ###### Sensitivity and specificity of OSIT-J and cardiac ^123^I-MIBG scintigraphy. ----------------------------------------------------------------------- OSIT-J score Early H/M ratio Delayed\ H/M ratio -------------------------- -------------- ----------------- ----------- iRBD patients (n = 30) Cut-off 8.5 2.30 2.03 Sensitivity 90.0% 95.0% 95.0% Specificity 83.3% 86.7% 96.7% PD patients (n = 38) Cut-off 8.5 2.30 2.03 Sensitivity 90.0% 95.0% 95.0% Specificity 92.1% 71.1% 71.1% ----------------------------------------------------------------------- iRBD: idiopathic REM sleep behavior disorder, PD: Parkinson\'s disease, OSIT-J: Odor Stick Identification Test for Japanese, ^123^I-MIBG: ^123^I-metaiodobenzylguanidine, H/M: heart-to-mediastinum. [^1]: Academic Editor: Irena Rektorova	{ "pile_set_name": "PubMed Central" }
Transfer of a signal from one site in a protein to another is a pervasive theme in protein regulation. Although our understanding of this process has advanced[@b1][@b2][@b3][@b4][@b5] our ability to identify allosteric sites and the pathways between sites remains incomplete. This is largely due to difficulties in observing imperceptible conformational changes and associated altered dynamics that frequently accompany allosteric signaling[@b2][@b4][@b6][@b7]. Identifying allosteric sites is therapeutically desirable since these sites can provide greater specificity and potency compared to orthosteric sites, which are often spatially generic across functionally diverse proteins[@b8][@b9]. High throughput experimental[@b10] and computational methods[@b11][@b12][@b13] have recently been developed to identify allosteric binding sites. Characterizing the pathways of physically and dynamically coupled residues between these sites provides insight into the mechanism of allostery and can even lead to identification of novel allosteric sites[@b12]. The ability to identify signaling pathways is made difficult due to prerequisites of detailed structural and functional knowledge as well as limitations in longer timescales of simulated motions. We present an NMR based method to observe allosteric information flow independent of structural and functional knowledge. In contrast to NMR methods that identify a functional correlation[@b14][@b15][@b16][@b17][@b18][@b19] the method presented here provides a description of the intrinsic relationship between conformationally and dynamically coupled residues. NMR chemical shifts report on the time-averaged local chemical environment and therefore reflect conformation and dynamic processes on a broad timescale (ps-ms). Thus the chemical shift is ideally suited to detect forms of allostery that may be difficult or impossible to detect by other methods. Here we show how site-directed mutagenesis of methyl containing side chains can be used to mildly perturb the allosteric network at various locations throughout the protein. The effects of each of these perturbations propagate through allosteric networks and are measured as changes in ^1^H-^13^C methyl chemical shifts. Communication pathways are identified by mediation analysis of the chemical shift perturbations and are used in 2^nd^ order Markov model network analysis[@b20] to characterize overlapping allosteric communities and flow between communities. Using a network analysis approach with chemical shift perturbations provides characterization of the network structure, reveals the overall flow of communication within that structure, and identifies critical nodes within the pathways. In the present work, chemical shift network analysis was used to investigate the allosteric network in a eukaryotic kinase, p38γ MAPK. Eukaryotic kinases are a large family of phosphotransferases that regulate diverse cellular functions and yet share a common architectural core[@b21] composed of an active site and an ATP binding site at the interface between the N-terminal and C-terminal lobes (N and C lobes). The spatially conserved core includes catalytic residues surrounding and contributing to contiguous networks of hydrophobic residues in "spines" that span across the two lobes[@b21]. The dynamic formation or breakage of the spines determines the activation state of the kinase[@b21]. It is likely that allosteric signals from core regulatory motifs, such as the activation loop, gatekeeper residue, DFG loop, catalytic loop, substrate binding site, and the hinge between lobes, are communicated through the completed spines. Experimental approaches have provided evidence for allostery between the conserved motifs in various kinases[@b22][@b23][@b24][@b25] and molecular dynamic simulations have suggested correlated motions across the spines and regulatory motifs in Protein kinase A (PKA) and c-Src kinase[@b26][@b27][@b28]. However, the nature of the communication between conserved elements is yet to be completely characterized. By network analysis of chemical shift perturbations caused by methyl mutagenesis of p38γ we observe the pathways of communication between the various regulatory elements of p38γ and how they respond to changes in catalytic state. The network structure reveals how these elements are connected to each other and how they communicate. In the inactive apo, ATP bound, inhibitor bound (DFG-out), and phosphorylated forms of p38γ we find that the structure of the allosteric network reflects the conserved kinase infrastructure, including distinction between N and C lobes, hydrophobic spines, as well as MAPK specific regulatory elements. The flow of information revealed by the analysis identifies novel mechanisms for allosteric effects in docking site interactions, auto-activation, and DFG-out inhibition. Critical nodes in these networks - those that have the highest number of connections and greatest amount of communication flowing through them - are found to correspond very well to known allosteric sites. We characterize the pathways connecting these regulatory nodes to the active site. The results presented demonstrate the method's effectiveness to characterize network structure and identify allosteric sites by improving our understanding of the allosteric pathways within p38γ. Results ======= Identification of long-range perturbations ------------------------------------------ A set of 20 conservative methyl mutants was used in conjunction with triple resonance NMR experiments to assign 73% of methyl resonances for Ile(δ1), Leu(δ1/δ2), Val(γ1/γ2), and Met(ε) in p38γ (see Methods and [Supplementary text](#S1){ref-type="supplementary-material"}). The residues for which methyls have been assigned are well distributed throughout the protein, including all regulatory elements. The single-site mutations used for assignments were found to cause long-range perturbations, at distances up to 40 Å or greater, in the local environment of methyl groups ([Fig. 1a](#f1){ref-type="fig"}). Long range chemical shift perturbations caused by binding of substrate, inhibitor, and nucleotide to kinases have been observed previously, but have been described only in qualitative terms[@b23][@b24][@b29][@b30]. By using conservative mutagenesis, all intrinsic pathways present in a single state of p38γ are probed. In addition, we considered the full range of chemical shift perturbations down to very small differences, \~1 Hz, revealing subtle and underappreciated redistribution of sampled states in the conformational ensemble. The most obvious pathways for propagation of chemical shift changes follow known conserved hydrophobic spines. For example, mutations such as L268V in the MAPK insert, exhibit chemical shift perturbations that extend from the site of mutation to the αF-helix and residues within and surrounding the 'regulatory' spine (R-spine), a distance of \>40 Å ([Fig. 1b](#f1){ref-type="fig"}). This supports the hydrophobic spine model of allosteric communication and agrees well with statistical sequence comparison that implicates a linkage between the MAPK insert and active site[@b31]. To characterize the communication pathways along the spines and other previously unknown routes, rank correlation coefficients between methyl groups were determined across the set of mutation-induced chemical shift perturbations. For example, the rank correlation between M112 and V53 is shown in [Fig. 1c](#f1){ref-type="fig"}, wherein mutations were given integer ranks based on the magnitude of induced chemical shift perturbations for each methyl resonance. We ranked perturbations in order to 1) allow identification of monotonic relationships, 2) reduce the effect of errors in chemical shift measurement, and 3) allow inclusion of perturbations near mutation sites (see [Supplementary text](#S1){ref-type="supplementary-material"}). In addition, the mutations are distributed throughout the protein ([Supplementary Fig. S2](#S1){ref-type="supplementary-material"}) so that analysis of the chemical shift changes should reflect an unbiased view of the network. As expected, we found correlations of perturbations between residues not only local to one another but distant as well. This led us to investigate the communication network reflected by these correlations. A network of overlapping communities ------------------------------------ Uncovering the underlying communication network of a system reveals its inherent structure and interdependency, which can be as valuable in understanding the function of that system as its tertiary structure. Since multiple communication pathways have been proposed to connect regulatory elements to effector sites within proteins[@b32], including MAPKs[@b25], and since a residue may be involved in more than one pathway, it is intuitive to consider allosteric networks as overlapping communities of interacting residues. There are several network analysis methods to characterize overlapping communities that provide information on the network hierarchy[@b33][@b34], as well as the behavior within the network[@b20]. Since we are interested in how the communities of correlated residues interact with one another within the context of the intramolecular communication network, we chose to use the Mapequation method[@b20]. This method, which optimizes the minimum length of a random walk on the network, identifies communities as patterns of flow between residues and, among popular methods, provides one of the most accurate descriptions of networks[@b35]. This treatment is ideal for analysis of the p38γ data since each mutation causes chemical shift perturbations that flow in a spatially directed manner from the site of mutation. To convert chemical shift perturbation pathways into quantifiable flow we utilized the correlations of perturbations across the entire set of mutations, where a stronger correlation between residues corresponds to stronger flow of information between those residues. Mediated effects of these perturbations were analyzed in order to identify and quantify directed flow along 3-residue paths ([Fig. 2](#f2){ref-type="fig"}). These 3-residue paths provide insight into how the flow of chemical shift perturbation depends on where it originated, in other words the memory effects of the network. The mediation analysis identified 875 pairwise correlations and 1304 sets of 3-residue flow in inactive p38γ ([Supplementary Table S8](#S1){ref-type="supplementary-material"}). In the Mapequation analysis, random walks on the network were weighted by the directed flow for every 3-residue path in order to model the spread of perturbations across the protein. The use of memory in the Mapequation has been shown to produce a richer and more accurate network description than analysis without memory effects, which utilizes pairwise flow[@b20]. Network structure of inactive apo p38γ -------------------------------------- To ensure that the novel data handling used in our method did not adversely impact the networking results, we compared the results with those obtained using simpler pairwise ranked correlations. We found that the clustering results from the Mapequation network analysis agrees well with results from two point correlations using the link community method[@b34] ([Supplementary Fig. S3](#S1){ref-type="supplementary-material"}) and agglomerative clustering ([Supplementary Fig. S4](#S1){ref-type="supplementary-material"}). For all of these methods, despite the fact that no structural information was used, the communities identified from chemical shift perturbations correspond extremely well to structural features of p38γ ([Fig. 3](#f3){ref-type="fig"}, see [Supplementary Table S9](#S1){ref-type="supplementary-material"} for community membership). Communities can be conceptualized as clusters of residues that react similarly to each mutation and have significant communication with one another. The modules with the highest amount of internal communication from the Mapequation analysis, represented by module diameter in [Fig. 3a](#f3){ref-type="fig"}, correspond to the N-lobe (yellow) and C-lobe (purple) of p38γ. Although the network structure nicely reflects the tertiary structure, it also identifies communities that span various structural elements within the architectural core, demonstrating that communication networks cannot be readily inferred by protein structure alone. The two major communities in inactive apo p38γ are connected to each other by two overlapping communities composed of 1) the active site, at the interface between lobes, and docking site (green) and 2) the MAPK-insert and GHI subdomain in the C-lobe (light purple) ([Fig. 3a](#f3){ref-type="fig"} box, [Fig. 3d](#f3){ref-type="fig"}). The GHI subdomain, composed of the α-helices G, H, and I, is a conserved feature of eukaryotic kinases that has been proposed to integrate signals originating in the C-lobe with the active-site[@b21][@b26]. Thus the communities of chemical shift perturbation reflect the conserved and MAPK-specific infrastructure. Network flow reveals relationship between communities ----------------------------------------------------- The network analysis of inactive apo p38γ also reveals the degree of information flow, which corresponds to the propagation of chemical shift perturbations, between communities of related residues. This novel nuanced understanding of the p38γ communication network is represented on the map in [Fig. 3a](#f3){ref-type="fig"} as the thickness of the arrows connecting communities. Mapping flow reveals that the active site community has stronger connections to the N-lobe than to the C-lobe and MAPK-insert, therefore discounting a role of the C-lobe in the regulation of inactive apo p38γ. Interestingly, a community that corresponds well to the spatially conserved 'catalytic spine' (C-spine) is detected and displays weak flow to the rest of the protein ([Figs. 3a,b](#f3){ref-type="fig"}; blue). The weak information flow between the C-spine and other parts of the protein likely reflects the lack of bound ATP. ATP binding has been proposed to complete the C-spine and prime the active kinase for activity by stabilizing compaction of the N and C lobes[@b21]. The compaction of the lobes concomitant with ATP binding will likely increase the C-spine module's connections and flow to other communities in the protein. In contrast, a community that corresponds closely to the spatially conserved hydrophobic R-spine ([Fig. 3a,c](#f3){ref-type="fig"}; red) is strongly connected to the rest of the protein, reflected by high information flow to other communities. This suggests, unexpectedly, that the R-spine is at least partially preformed and transmits information within inactive p38γ. A preformed R-spine is unexpected in an inactive kinase, since the hydrophobic packing of the R-spine typically indicates that the active-site residues are preorganized for catalysis[@b21]. However, it is possible that the R-spine residues form a dynamic contiguous network yet are not properly organized for phospho-transferase activity. The physically connected but likely ill-formed spine in p38γ may have evolved to allow allosteric signal propagation along the conserved architecture while remaining inactive. Our results offer a broadened understanding of hydrophobic spines, outside of their fundamental role in facilitating the active state: these pathways dynamically pre-exist in the inactive state and enable allosteric communication between lobes. ATP binding to inactive p38γ reconfigures the network structure --------------------------------------------------------------- To investigate how completion of the C-spine affects the network, we used the same mutagenesis strategy to produce 13 mutants of p38γ in complex with ATP. These mutants were chosen quasi-randomly, to ensure that mutations sampled the entire protein. Simulations using chemical shift data for the apo enzyme, for which 20 mutations are available, show that data obtained from as few as 11 mutations are sufficient to define the network structure, demonstrating the robustness of the method ([Supplementary text and Fig. S5](#S1){ref-type="supplementary-material"}). Mediation analysis was performed on the chemical shift changes caused by the 13 mutations in the ATP complex and 719 significant pairwise correlations and 1066 three amino acid paths were identified ([Supplementary Table S10](#S1){ref-type="supplementary-material"}). Although the adenine ring of the bound ATP completes the p38γ C-spine, the unphosphorylated kinase is catalytically inert. The network analysis reveals that, in the presence of bound ATP, the C-spine's connections are strengthened to the extent that the C-spine becomes incorporated into the N-lobe community ([Fig. 4a,b](#f4){ref-type="fig"}, see [Supplementary Table S11](#S1){ref-type="supplementary-material"} for community membership). This is illustrated by the expansion of the N-lobe community further into the C-lobe, in comparison to the apo state network (compare [Figs 4a](#f4){ref-type="fig"} and [3a](#f3){ref-type="fig"}). In addition, the relative flow into and out of the R-spine is decreased compared to the inactive apo state. This may reflect the improper compaction of the R-spine caused by ATP binding[@b30][@b36], which likely reduces aberrant hydrolysis of ATP by inactive p38γ. The completion of the C-spine by ATP reconfigures p38γ's network by creating two discrete pathways between the N and C-lobes via 1) methyls surrounding the R-spine and 2) through methyls in the αF-helix, docking-site, and C-spine ([Fig. 4a,c,d](#f4){ref-type="fig"}). Activation reinforces existing pathways --------------------------------------- Activation of MAPKs involves dual phosphorylation of the activation loop, which leads to salt bridge formation between the phosphorylated residues and the N and C-lobes, and causes the lobes to close towards each other accompanied by a concerted rearrangement of secondary structural elements around the active site. This compaction and reorganization aligns the catalytic residues for activity and completes the R-spine[@b21]. Since our chemical shift network analysis finds evidence for the pre-existence of the hydrophobic R-spine pathway in the inactive apo state of p38γ, we were motivated to investigate how this network responds to activation. The same set of 13 conservative p38γ mutants used in the analysis of the ATP-bound state were dual phosphorylated by MKK6 and chemical shift perturbations due to the mutations were used to analyze the network structure. 625 significant pairwise correlations and 460 sets of 3-residue pathways were identified in active p38γ ([Supplementary Table S12](#S1){ref-type="supplementary-material"}). Chemical shift based network analysis reveals that upon activation, the R-spine community, as expected, forms more extensive connections between the N and C-lobes ([Fig. 5a,c](#f5){ref-type="fig"}; red, see [Supplementary Table S13](#S1){ref-type="supplementary-material"} for community membership) than in the inactive state, which only displays connections from the R-spine to the N-lobe. The residues in this hydrophobic spine have membership in multiple overlapping communities, including those containing the ATP binding site, active site, GHI subdomain, and MAPK insert. This suggests the R-spine is properly formed by activation and makes more extensive connections with the rest of the protein when in the activated state. The N and C-lobe compaction is also reflected in an extension of a pathway between lobes, through the docking site ([Fig. 5a,b](#f5){ref-type="fig"}; blue). This highlights the importance of the docking site, which has been observed to allosterically enhance substrate binding, nucleotide binding, and the catalytic rate of active p38α[@b30][@b37]. Concomitant with activation, the regulatory GHI subdomain[@b21][@b26] becomes a member of the active site community ([Fig. 5a,d](#f5){ref-type="fig"}; orange), which is also the community with the second highest amount of internal communication. Together, this suggests that the GHI subdomain has more influence on the active site in the activated state of p38γ than in the inactivated state. A comparison of the network maps of inactive and activated apo p38γ reveals that activation reduces overall flow across the network but extends existing communication pathways involved in regulation ([Figs 3a](#f3){ref-type="fig"} and [5a](#f5){ref-type="fig"}). It is likely that binding of substrate and nucleotide, which is accompanied by further conformational rearrangements that lead towards catalytic competence[@b26][@b30], will further enhance flow over the pathways. BIRB796 binding to inactive p38 causes network fragmentation ------------------------------------------------------------ Kinases can also be artificially modulated towards dysfunctional states. In particular, DFG-out inhibitors bind in the active site of kinases and disrupt the DFG-motif that constitutes a part of the R-spine[@b38][@b39][@b40]. This disruption impedes catalysis and alters the dynamics and conformation of the activation loop to prevent phosphorylation by upstream kinases. To gain insight into how the communication network in p38γ behaves following breakage of the conserved infrastructure, we bound the DFG-out inhibitor BIRB796 to the inactive kinase and performed network analysis using the chemical shift perturbations associated with 12 methyl mutations. 589 significant pairwise correlations and 500 sets of three residue paths were identified ([Supplementary Table S14](#S1){ref-type="supplementary-material"}). The resulting community map reveals that the DFG-out inhibitor severely reduces connection and flow between the N and C-lobes ([Fig. 6a,b](#f6){ref-type="fig"}, see [Supplementary Table S15](#S1){ref-type="supplementary-material"} for community membership). The disruption of the R-spine is reflected in the network by the loss of communities spanning across the active site. This includes the loss of the R-spine community as well as the loss of the active-site community, which acts as an interface between the N and C-lobes in the inactive apo state. Within the C-lobe of the inhibitor-bound complex, regulatory elements are identified in separate overlapping communities. Information flow between the module containing the docking-site and the overlapping module containing only the MAPK insert and GHI subdomain is relatively weak, indicating that the docking site is connected with the rest of the C-lobe but is fairly isolated in terms of communication ([Fig. 6a,b](#f6){ref-type="fig"}). The loss of communication between the docking site and C-lobe, as well as the loss of communication between N- and C-lobes, likely contributes to the inability of upstream kinases to phosphorylate BIRB796-bound p38, since docking site interactions will no longer cause the allosteric conformational change in the activation loop required for phosphorylation[@b25]. Identification of critical nodes and pathways in allostery ---------------------------------------------------------- The network structure maps described above provide novel insight into pathways and nodes in allosteric communication. For example, in inactive apo p38γ, the N-lobe community not only has high internal communication between residues within the community (module diameter) but also has more connections to other modules, four, than does the C-lobe community, which connects to only one module, suggesting that the N-lobe has an important role in inactive apo p38γ ([Fig. 3a](#f3){ref-type="fig"}). Inactive p38γ complexed with ATP exhibits increased flow between the C-lobe community and overlapping communities consisting of the MAPK-insert and GHI subdomain ([Fig. 4a,c,d](#f4){ref-type="fig"}). Activation also causes an inversion of the major community membership and communication networks, where more internal connectivity and flow is now present in the C-lobe ([Fig. 5a](#f5){ref-type="fig"}) compared to the inactive state ([Fig. 3a](#f3){ref-type="fig"}). This suggests regulatory importance of C-lobe elements, such as the MAPK insert and GHI subdomain, in the compact conformation of the activated and ATP-bound states. A more direct approach to identifying critical residues and regions is to consider residues that have the highest number of network connections. The changes in the localization of residues with the highest number of connections highlight the reorganization of the allosteric network in the different states of p38γ as well as the importance of regulatory elements in various states ([Supplementary text](#S1){ref-type="supplementary-material"}, [Supplementary Fig. S6](#S1){ref-type="supplementary-material"}). Network flow identifies critical pathways ----------------------------------------- Examination of residues with high connectivity and high information flow identifies undiscovered critical pathways. We ranked the magnitude of flow of chemical shift perturbations across all pairs of correlated residues (memory nodes) to and from other memory nodes. Flow is explicitly attributable to a series of 3-residue paths over each memory node. Thus we can identify critical memory nodes, in terms of flow, and the communication pathways of that flow. In the inactive apo form of p38γ, a series of highly ranked memory nodes form pathways from the docking site to the activation loop and active site residues, over a distance of \>20 Å ([Fig. 7a](#f7){ref-type="fig"}). A highly ranked memory node representing flow through L170, within the DFG loop, to M112, a hinge residue, exhibits incoming flow from docking site residues L116, L119, and V161 ([Fig. 7a](#f7){ref-type="fig"}-left). Information is transferred across the L170-M112 memory node to the so-called 'gatekeeper' residue (M109) and the N-terminal portion of the conserved R-spine (L89 and L78). The gatekeeper residue, which is between hydrophobic spines and has been proposed to interface with the R-spine to constrain DFG loop and activation loop flexibility[@b41], also has connections to the R-spine memory node (L89 --L78) ([Fig. 7a](#f7){ref-type="fig"}-middle). The R-spine memory node takes additional input from the DFG loop (L170) and directly from the docking site (L116) with flow continuing to the catalytic loop (L149), the L16 loop (L343), and the αC-helix (L77). Finally, information flow across the memory node L77-L174, which involves the αC-helix (L77) and the DFG motif/activation loop (L174), spreads further into the activation loop (M182) and into the P + 1 loop (V186, V187) ([Fig. 7a](#f7){ref-type="fig"}-right). Thus the highly ranked memory nodes and their communication pathways reveal a novel mechanism for the allosteric communication related to activation between docking site, activation loop, and P + 1 loop. In addition, these pathways may be involved in docking site mediated auto-activation (see [Supplementary text](#S1){ref-type="supplementary-material"}). In apo forms of p38γ, the significance of the αC-helix is highlighted by the M81-L77 highly ranked memory node. The memory node displays flow from the αC-helix (L77, M81) to active site residues ([Fig. 7b](#f7){ref-type="fig"}). Communication is observed across the M81-L77 memory node to the DFG motif (L170), αC-helix/R spine (L78), and activation loop (M182). Significantly, the side chain of L77 makes contact with the side chain of Y326 in the L16 loop in crystal structures of p38. The phosphorylation of Y326 in p38α/β in T-cells induces activation of p38 via auto-phosphorylation of T183 in the activation loop[@b42][@b43]. Thus, changes in interaction with the M81-L77 memory node following Y326 phosphorylation may be communicated by the pathways detected in our network analysis and stabilize an active conformation of the R-spine and DFG loop. A final example of the network structure providing a mechanistic understanding of allosteric pathways is given by a memory node in the MAPK-insert of activated p38γ ([Fig. 7c](#f7){ref-type="fig"}). The M262-L198 memory node, representing flow from M262 in the MAPK-insert to L198 in the αEF-helix, reveals information flow to the P + 1/activation loop (V187) and HRD motif (I150). The HRD motif includes the catalytic base (D153). Thus, the flow over this memory node suggests that changes in interaction between L198 and M262 in the MAPK insert leads to perturbation of the P + 1 loop, activation loop, and catalytic loop, direct participants in p38γ activity. Interestingly, this memory node is only highly ranked in the network analysis of the activated state and is not present in the inactive states. It is possible that the formation of this pathway is concomitant with stabilization of the active state. Indeed, lipid molecules have been found to bind in the hydrophobic groove formed between the MAPK-insert and αEF-helix, between L198 and M262, in the inactive form of p38α leading to auto-activation[@b44]. Discussion ========== We have demonstrated the utility of a simple memory network analysis of NMR chemical shift perturbations caused by conservative mutagenesis. Mutations of methyl-containing residues are commonly made to obtain methyl chemical shift assignments and thus these constructs can serve dual purpose. The method provides a highly detailed map of overlapping communities in allosteric networks, corresponding information flow, and ranks of residues in relation to the network. The overlapping communities found in p38γ correlate well with the 3D structure, despite no structural information being used as input. The analysis also confirms an underlying conserved kinase infrastructure, originally identified by local spatial alignment[@b21], as well as the MAPK specific infrastructure, including the MAPK insert and L16 loop. Consideration of the network structure of a protein and its flow provides insight into important structural regions and pathways of communication. Previous application of graph theory to computational simulations to identify the topology of non-overlapping communities of correlated motions has demonstrated the utility of revealing the network structure of a protein[@b26][@b45][@b46]. In particular, in silico network analysis of PKA by McClendon et al. showed that functionally significant structural features are partitioned into non-overlapping communities, effectively acting as a hub for these communities[@b26]. Similarly, our network analysis of p38γ finds that key regulatory elements are contained entirely in their own communities and that functional significance is indicated by a high degree of overlap and flow with other communities. For instance, communities that overlap a large number of neighboring communities likely act as nodes in communication. In addition high flux of information identifies communities that share relatively large amounts of information, again suggesting functional relevance. One example of this is the active site community, including docking site residues and the interface between N and C lobes, which has high overlap and high flux with neighboring communities in inactive p38γ. In the activated form of p38γ, the R-spine module is highly overlapped with the N-lobe, C-lobe, and MAPK insert but lacks high information flux traveling through it. The flow in and out of this community will likely increase upon ATP or substrate binding as the compact state becomes stabilized[@b30][@b36]. Likewise, the C-spine/N-lobe community in the ATP bound inactive form of p38γ is highly overlapped with other communities but lacks high flux. Again, activation is likely to increase the flow through this spine. Thus the network structure and flow not only reflects important regulatory regions but also reveals the functional state of the system. Finally, one of the more powerful aspects of the network analysis is the ability to identify critical network nodes and pathways that correspond to biologically relevant allosteric sites in the enzyme. We have demonstrated that our method correctly identifies previously known regions of regulatory function and also provides new understanding of the allosteric communication pathways between these regions, providing valuable mechanistic insights into allosteric pathways that can guide further experiments. In p38γ, for example, residues within key pathways could be mutated and the effect on cooperativity of docking peptide binding, activation, or catalysis could be characterized. In addition, solvent accessible pockets that are strongly connected to critical nodes and pathways may be more promising candidates for targeting by small molecule compounds than pockets that are only weakly connected to the network[@b8][@b9]. The network analysis method of chemical shift perturbations provides an information-rich and direct view of allosteric networks of proteins in solution. Methods ======= Minimally perturbing mutagenesis -------------------------------- Isotopically labeled His~6~ N-terminally tagged human MAPK p38γ was overexpressed in a PET15b (Invitrogen) plasmid in E. coli Rosetta plysS BL21(DE3) (Invitrogen) cells at 288 K. Site-directed mutagenesis was performed by Quikchange (Agilent) to generate the following conservative mutations: L77V, M81L, M109L, M112L, L116V, M120L, M137L, I150L, L159V, L167V, L174V, V186A, V187A, L198V, M216L, L225V, I238L, M239L, L268V, and M291L. Methyl mutants were expressed in cells grown in 50 ml of M9 minimal media containing ^13^C glucose as carbon source. Purification procedures were adapted from[@b47]. Briefly, cells were lysed by sonication in 50 mM Tris pH 8.0, 500 mM NaCl, 10 mM β-mercaptoethanol, 0.1% Triton X-100, and 1 mM PMSF and p38γ was purified by Ni-NTA affinity chromatography. Purified protein was exchanged into NMR buffer containing 25 mM deuterated Hepes pH 7.2, 250 mM NaCl, 1 mM TCEP, and 0.01% w/v NaN~3~ in 100% D~2~O. Final protein concentrations were between 100--400 μM. Sets of mutants were expressed simultaneously with a wildtype sample and purified identically with the same buffer stock. Activation and ligand binding ----------------------------- A constitutively active mutant of GST-tagged MKK6 (S207D, T211D) was overexpressed in Rosetta plysS BL21(DE3) (Invitrogen) E. coli cells at 293 K in LB media. Cells were lysed by sonication in 20 mM Tris pH 8.0, 100 mM NaCl, 5 mM DTT, 0.1% Triton X-100, 1 mM EDTA, and 10% glycerol. GST-MKK6 was purified by GST-Sepharose 4b (GE) affinity chromatography. Protein was dialyzed into 20 mM Hepes pH 7.6, 100 mM NaCl, and 2 mM DTT. p38γ was activated (dual phosphorylation on Thr183/Tyr185) by incubation with the mutant MKK6 at room temperature in 75 mM Hepes pH 7.5, 0.1 mM PMSF, 1 mM NaF, 0.1 mM NaVO~4~, 0.1 mM β-glycerol phosphate, 10 mM MgCl~2~, and 1 mM DTT. Completion of phosphorylation was monitored by a phosphorylation-dependent change in elution time over analytical HPLC (C4 reverse-phase) and by change in ^1^H-^13^C chemical shift of p38γ . Following activation of p38γ, MKK6 and ATP were removed by Ni-NTA affinity chromatography, as described above, with an added 10x column volume wash of column buffer with 1 M NaCl and additional dialysis against 25 mM Hepes pH 7.2, 250 mM NaCl, 5 mM DTT, and 1 mM EDTA. Doramapimod (BIRB796) from Selleck Chemicals was resuspended in deuterated DMSO and incubated at room temperature with purified inactive p38γ at 1:2 molar excess. The pH of the sample was readjusted, as necessary, to pH 7.2 and centrifuged to remove insoluble BIRB796. ATP and MgCl~2~ were prepared in NMR buffer and added in molar excess (1:50) to inactive p38γ. Complete binding of BIRB796 or ATP to p38γ was monitored by ^1^H-^13^C chemical shift. NMR spectroscopy ---------------- ^1^H-^13^C HMQC NMR experiments were performed in NMR buffer at 293 K on Bruker Avance 750 MHz or 900 MHz spectrometers equipped with triple resonance gradient probes. Data were acquired using optimized Poisson-gap non-uniform sampling (NUS)[@b48] in the indirect dimension. This allowed a higher number of transients to be collected over a given time for greater signal to noise. 40--100% of 512 real and imaginary points in the ^13^C dimension were collected over a spectral width of 40 ppm. Spectra of the mutants were acquired together with an identically prepared wildtype sample and with identical NMR experimental parameters (% NUS, number of increments, spectral width). Data were reconstructed and processed with compressed sensing in MDDNMR and NMRPipe[@b49][@b50][@b51]. Spectra were analyzed with NMRView5[@b52]. ^1^H-^13^C combined chemical shift differences were calculated in Hz at 750 MHz using the equation Δv = ((Δ^1^H\750)^2^ + (Δ^13^C\188.8)^2^)^1/2^[@b53]. Network analysis ---------------- Three residue pathways of information flow were identified by mediation analysis of correlated chemical shifts over the entire set of mutations. A modified method of Baron & Kenny[@b54] as described in[@b55] was used to identify mediated pathways ([Fig. 2](#f2){ref-type="fig"}). In brief, residues A, B, and C form a 'pathway' of A to B to C if B mediates the effect of A on C. We simplify analysis by only considering positive correlations ([Supplementary text](#S1){ref-type="supplementary-material"}). The tests used to identify mediation through B are: 1) A affects B, i.e. the two-point correlation between residues A and B (R~AB~) is significant; 2) B affects C when considering multiple regression of A, B, and C, i.e. the partial correlation between residues B and C (ρ~bc~) is significant; 3) there is an indirect effect, i.e. the partial correlation between A and C (direct effect, ρ~ac~) is less than the two point correlation between A and C (R~AC~). The mediated effect is then quantitated by R~AB~\ρ~bc~ ([Fig. 2](#f2){ref-type="fig"}). Uncorrected p-values from t-tests of two-point correlations and multiple-regressions were used to determine significance. In the inactive apo p38γ dataset, correlations and partial correlations were considered significant if p \< 0.02, for the inactive ATP complex p \< 0.03, for the inactive BIRB796 complex p \< 0.03, and for the active apo p38γ p \< 0.04. The p-value cutoffs were chosen for enrichment in positive correlations ([Supplementary Fig. S8](#S1){ref-type="supplementary-material"}). In order to minimize the effects of errors due to p-value cutoff and to create a continuum of information flow from 0 (no flow) to 1 (high flow), soft-thresholding was utilized[@b56]. This involves choosing a soft-threshold limit using the approximate scale-free topology criterion of [@b57] and raising correlation coefficients by the chosen soft-thresholding power. For all data, a soft-threshold of 8 was applied to the two-point correlation coefficients. The mediated or indirect effects were used as the weights in a random walk over the network and the two-point correlation coefficients were used as teleportation probabilities (jumps between nodes) in the MapEquation software with 2^nd^ order dynamics[@b20][@b58]. The perturbations for a common set of 78 methyl containing residues were included in the analysis of various states of p38γ. Other assigned methyls in the inactive apo state were discarded due to ambiguity in assignment in either ATP bound, BIRB796 bound, and activated states. Under these conditions we found that at least 11 mutants were required to reliably reproduce the network structure reported ([Supplementary text](#S1){ref-type="supplementary-material"}). For each state of p38γ, 20,000,000 runs in the MapEquation software were performed and the shortest walk length was chosen. All statistical analyses were conducted in the R software environment[@b59] utilizing various packages[@b57][@b60][@b61][@b62]. Python scripts for preparation of chemical shift data for R input, R scripts, and input for MapEquation are available for download at <http://www.scripps.edu/wright>. Homology models for p38γ were generated with the I-TASSER structure prediction method[@b63]. In brief, the inactive apo and BIRB796 bound models were predicted by I-TASSER using threading structures (p38α) with 62 to 67% sequence identity to aligned regions of p38γ. In addition activated structures of p38 were manually excluded as templates. The model for BIRB796 bound p38γ was predicted with the added restraint of a p38α DFG-out template (1WBT), which has 63% sequence identity to aligned regions in p38γ. The activated model was generated with the activated structure of p38γ (1CM8) as a template. Additional Information ====================== How to cite this article: Aoto, P. C. et al. NMR Characterization of Information Flow and Allosteric Communities in the MAP Kinase p38γ. Sci. Rep. 6, 28655; doi: 10.1038/srep28655 (2016). Supplementary Material {#S1} ====================== ###### Supplementary Information This work was supported by grant GM075995 from the National Institutes of Health and by the Skaggs Institute for Chemical Biology. We thank Jane Dyson, Bryn Fenwick, and David Oyen for insightful discussions; Sulakshana Mukherjee for assistance with protein purification; Gerard Kroon for assistance with NMR experiments; and Nathan Wineinger and Andrew Su for input on statistical data handling. Author Contributions* P.C.A. and P.E.W. designed research; P.C.A. and B.T.M. performed molecular biology and NMR experiments; P.C.A. and P.E.W. wrote the manuscript. All authors reviewed the manuscript. ![Long-range methyl chemical shift perturbations.\ (a) Heat-map of ^1^H-^13^C chemical shift perturbations (in Hz) for methyl resonances in p38γ (mutation-wildtype) caused by minimally disruptive mutations. Green represents mutation site. (b) ^1^H-^13^C chemical shift perturbations due to the substitution L268V in the MAPK insert mapped onto a homology model of the inactive apo p38γ structure. Colors correspond to shift differences in Hz, using the same scale in (a). (c) The ranks of each mutation induced chemical shift perturbation for V53 and M112. Each point corresponds to a mutant as ranked among the set of mutations for the specified residue.](srep28655-f1){#f1} ![Mediation analysis.\ Effects of residue A on C are mediated by B if 1) A affects mediator or residue B (2 point correlation R~AB~); 2) B affects C (multiple regression coefficient ρ~bc~); 3) Positive correlations; 4) Indirect effect (ρ~ac~ \< R~AC~). If these criteria are met then flow travels from A to B to C with the mediated effect of R~AB~\ρ~bc~.](srep28655-f2){#f2} ![Network structure and flow of inactive apo p38γ.\ The method identified 15 communities with membership greater than 5 residues in inactive p38γ, but only the largest 8 communities are shown for clarity. Identified communities from chemical shift perturbation networks are colored based on tertiary structure and regulatory element: N-lobe, yellow; C-lobe, purple; active-site, green; C-spine, blue; R-spine, red; MAPK-insert, lavender. Modules depicted with similar shades of color overlap the same structural element. (a) Network map of inactive p38γ. The size of the modules represents the amount of flux and connections between residues within the module. Thickness of arrows between communities represents the amount of flux between communities. The C-spine (b) and R-spine (c) and connected communities are mapped on the homology model of inactive apo p38γ. Residues in multiple communities are depicted as multicolor spheres. The boxed region of the network map (panel a) is mapped on the structure in (d), illustrating the communities and flow linking C-lobe and N-lobe.](srep28655-f3){#f3} ![Response of p38γ network to ATP binding.\ (a) Network structure of ATP + p38γ, communities are colored based on tertiary structure and regulatory element: N-lobe/C-spine, yellow; C-lobe, purple; docking site, blue; R-spine, red; MAPK-insert, lavender. (b) Comparison of N-lobe communities in inactive apo and ATP bound p38γ. Completion of the C-spine by ATP extends the N-lobe community further into the C-lobe. (c) Communities representing a pathway (left black arrow in (a)) between C-lobe (purple) and N-lobe (yellow) through the docking site (blue) and C-spine (yellow). The pathway through the C-spine is not present in apo inactive p38γ. (d) Communities involved in the R-spine (red) pathway (right black arrow in a) between N (yellow) and C (purple) lobes.](srep28655-f4){#f4} ![Network response to activation.\ Network analysis identified 16 communities with membership greater than 5 residues in active p38γ. Communities are represented in shades of color based on tertiary structure and regulatory element: N-lobe/active-site, yellow; C-lobe/MAPK-insert/GHI subdomain, purple; active-site/GHI subdomain, orange; C-spine/hinge, blue; R-spine, red. The network map is depicted in (a) with flow and membership represented as in [Fig. 3](#f3){ref-type="fig"}. Flux through communities connecting N and C-lobes are depicted: through the C-spine hinge (b), through the R-spine (c), and through the active site residues (d).](srep28655-f5){#f5} ![Network disruption by BIRB-796.\ (a) The method identified 17 communities with membership greater than 5 residues in DFG-out p38γ, of which only 10 are shown for clarity. Communities are represented in shades of color based on tertiary structure and regulatory element: N-lobe, yellow; C-lobe, purple; docking site, blue. Communities in the C-lobe (purple) are heavily overlapped and include the MAPK insert, portions of the GHI subdomain, and F-helix. Darker shades of purple indicate communities that include the F-helix, lighter shades of purple indicate communities that include the H-helix. (b) Communities are mapped on the DFG-out p38γ homology model, illustrating the disruption in communication between N and C-lobes. The docking site has become insular, represented by weak flow to other communities and membership only in the C-lobe.](srep28655-f6){#f6} ![Critical network nodes reveal regulatory sites and pathways.\ (a) A series of highly ranked memory nodes in inactive p38γ reveal multiple pathways that extend greater than 20 Å from the docking site to the activation loop. Large circles represent residues with a direct connection to the memory node of interest and small circles represent connected memory nodes. Semi-circles depict residues with an indirect connection to the memory node. The relative amount of flow between residues is indicated by the thickness of solid arrows. Dashed arrows indicate the common residue belonging to the next memory node in the path. Colors correspond to regulatory elements, as labeled. Memory nodes (left to right) include M109-M112, L89-L78, and L77-L174. (b) A significant memory node in the αC-helix of apo p38γ suggests a route for auto-activation by Y326 phosphorylation. (c*) A highly ranked memory node in the MAPK insert of activated p38γ reveals a communication pathway to catalytic residues.](srep28655-f7){#f7}	{ "pile_set_name": "PubMed Central" }
Introduction {#s0010} ============ Plant biotechnology has experienced exponential development during the past 30 years, facilitated in large part by the capacity of a clever soil bacterium (Agrobacterium tumefaciens) to transfer DNA segments into plant cells ([@bib23]). This powerful function, harnessed by clever plant biologists during the 1980s, now enables routine delivery of recombinant DNA into plant cells for, among other things, production of recombinant proteins in plants. The use of plants for expression of subunit vaccine antigens, first pursued in the early 1990s ([@bib16], [@bib68]), was a natural outgrowth of the emerging technology, coming soon after the groundbreaking demonstration that immunoglobulins could be produced in transgenic tobacco ([@bib35]). Interested readers are referred to recent review articles that describe advances in the use of plants for production and delivery of vaccine antigens ([@bib67], [@bib82], [@bib87], [@bib88], [@bib99], [@bib102], [@bib127], [@bib133], [@bib107], [@bib131]). Unfortunately, commercial development of plant-derived vaccine products has been very slow, perhaps due to a very conservative pharmaceutical industry that is hesitant to try new technologies. A hypothesis that drove much of the early work on plant-derived vaccines held that cheap and orally delivered "edible" subunit vaccines could provide protection against infectious diseases in developing countries with limited medical infrastructures. Thus, ingestion of plant material would stimulate protective immunity, mediated by uptake and processing of antigens by the gut-associated lymphoid tissue (GALT) ([@bib99], [@bib107]). A number of hurdles can interfere with the effectiveness of this approach, including poor stability of proteins in the gut, substantial variability of antigen content in plant tissues, and regulatory concerns for the containment of transgenic crop plants and their use exclusively as pharmaceutical products. These and other issues are discussed in greater detail in other papers ([@bib67], [@bib87], [@bib89]). Such challenges notwithstanding, most workers and watchers of the field continue to see great value in the use of plants for production of vaccine antigens. Importantly, compelling data on the protective efficacy of orally delivered plant-derived vaccine antigens continue to accumulate, as discussed below in detail in the section on animal vaccines. The approaches for plant-based protein expression include stably integrated transgenes into chromosomes in either the nucleus or the chloroplast, and transient expression using vectors that may use plant virus replication elements to amplify the transgene. We present only a very brief discussion here, and interested readers are referred to excellent recent review articles ([@bib88], [@bib109]). Nuclear genes are inherited in a Mendelian fashion, whereas chloroplast genes are passed through progeny only from the maternal parent, since pollen cells are devoid of chloroplasts. Tissue-specific promoters allow foreign proteins to accumulate in seeds ([@bib79], [@bib99], [@bib127]; and see discussion of MucoRice below). Proteins that accumulate in seeds are very stable to temperature extremes and can be stored for months to years with little loss of protein activity. During the past decade, the use of rapid transient expression in plants (rather than the slow production of plants with stably integrated transgenes) has increased greatly and represents an ever-growing proportion of plant-based protein expression work. The development of viral vectors has enabled very fast and very high levels of protein expression in plants ([@bib109], [@bib25], [@bib132]). Nicotiana benthamiana, a tobacco relative ([@bib27]), is the typical plant host, due to its ease of culture and effective support of plant virus replication. Protein expression levels at 1--2 mg per kilogram of leaf mass are often ≥10-fold higher than those obtained with stably transgenic plants, and indicate the potential for purification of plant-produced proteins at costs that are economically competitive with microbial fermentation systems. Moreover, culture of N. benthamiana can be readily scaled up to massive levels ([@bib84]). Kentucky Bioprocessing, Inc. (Owensboro, KY, <http://www.kbpllc.com/>) has developed a large-scale system for plant growth, infiltration with Agrobacterium, and purification of recombinant proteins ([Figure 1](#f0010){ref-type="fig"} ). The convenience of transient plant-based expression enables rapid response to emerging viruses such as influenza and norovirus, as described by Charles Arntzen in his address to the 238th National Meeting of the American Chemical Society ([@bib21]). These developments make plant-based expression and purification of vaccine antigens, including virus-like particles (VLP) ([@bib37], [@bib90]), for mucosal delivery commercially viable.Figure 1Large-scale process for transient expression of recombinant proteins at Kentucky Bioprocessing (Owensboro, KY).A plant expression vector, often a viral replicon, is constructed and mobilized into Agrobacterium tumefaciens. The Agrobacterium cultures are used for massive infiltration of whole Nicotiana benthamiana plants in a large vacuum chamber. The plants are then cultured for 4--10 days, depending on the recombinant protein and expression vector, before they are harvested and extracted to purify the recombinant protein. We can also utilize purified antigens for intranasal delivery, which requires more concentrated antigen at lower doses than oral delivery. Nasal immunization induced antigen-specific mucosal IgA responses in salivary glands, respiratory tract, intestines, and reproductive tracts in mammals ([@bib31], [@bib40], [@bib53], [@bib86], [@bib96]). Moreover, mucosal immunization can prime both mucosal and systemic immune responses ([@bib54]). Thus, intranasal delivery of plant-derived vaccine antigens, especially VLP, is a promising strategy, as we discuss further below. In this chapter, we will describe the progress in selected areas with mucosal vaccines in human and animal hosts. We focus on studies that involved humans or have shown strong indication of efficacy in preclinical studies. Mucosal Vaccines for Humans {#s0015} =========================== Norovirus Vaccine and Mucosal Adjuvants {#s0020} --------------------------------------- Noroviruses are causal agents of epidemic gastroenteritis ([@bib33]), sometimes called "stomach flu," and when occurring on ocean liners, "cruise ship virus" ([@bib21]). Other than foundering or loss of power on the ship, a norovirus epidemic with projectile vomiting and persistent diarrhea must be one of the most disappointing vacations imaginable. There is yet no vaccine available for norovirus, but much work during the past 20 years has focused on production and immunogenicity testing of norovirus VLP (NVLP) ([@bib33]). The capsid protein (NVCP) assembles VLP when expressed in insect cells ([@bib42]) or plants ([@bib66]), and the VLP are substantially acid stable and thus amenable to oral delivery. Preclinical studies of transgenic tobacco and potato-derived NVLP delivered orally to mice showed efficient production of antigen-specific serum IgG and fecal IgA ([@bib66]). A clinical trial tested feeding human volunteers 150 g raw potato tuber tissue containing ∼500 μg NVCP in two or three doses over a period of 3 weeks indicated promising results ([@bib103]). Nearly all of the subjects who ingested NVCP potato developed antigen-specific serum IgG and/or IgA responses, but the titers were modest compared to those obtained in a later study with 250 μg/dose of purified insect cell-derived NVLP ([@bib104]). A difficulty with the raw potato vehicle was the variable amounts of NVCP among different tuber samples and limited 50--60% efficiency of VLP assembly. Thus, purified NVLP is the preferred material for oral vaccine delivery. Transgenic tomato fruit expressing NVCP proved to be a more tractable system ([@bib135]). Freeze-dried tissues of transgenic tomato fruit and potato tuber were fed to mice, with tomato shown to be a superior vehicle. Four doses of 0.4 g dried tomato (64 μg NVCP, 40 μg NVLP) caused seroconversion in 80% of mice, and double those amounts resulted in 100%. Furthermore, excellent levels of anti-NVCP fecal IgA were induced in all seroconverted mice. Conversely, similar doses of freeze-dried potato produced substantially poorer results, probably due to oxidation of phenolic compounds that are abundant in potato tuber. Since tomato fruit is low in phenolics, and high in the antioxidant ascorbic acid, such oxidation effects were minimal. Although the tomato delivery strategy seemed "ripe" for further development, the winds of change were blowing and bringing faster and more robust transient expression systems. When it became apparent that NVLP from transient expression in leaves could be purified in consistently high yield and quality ([@bib37], [@bib90]), the stable transgenic systems were largely left behind. Description of a process to produce NVLP under "current good manufacturing practice" (cGMP) conditions ([@bib55]) paved the way for future clinical studies that are likely to occur soon (Charles Arntzen, personal communication). Meanwhile, several interesting studies investigated intranasal (IN) delivery of plant-derived NVLP in mice, and showed strong effects of Toll-like receptor agonists as mucosal adjuvants. CpG oligodeoxynucleotides (CpG ODN) are known to interact with TLR9 to stimulate maturation and proliferation of multiple cell types of the immune system, and provide strong vaccine adjuvant effects ([@bib1]). When 25 μg purified plant-derived NVLP was delivered IN to mice in two doses given 3 weeks apart, much stronger serum and mucosal antibody responses were obtained when 10 μg CpG ODN was codelivered with NVLP ([@bib67]). The mucosal anti-NVLP IgA responses measured at multiple epithelial sites, including intestinal and vaginal secretions, were strongly enhanced by the adjuvant effect. Moreover, IN delivery of 25 μg NVLP without adjuvant was vastly superior to oral delivery of 100 μg NVLP for production of antigen-specific serum IgG. These data make a compelling case that IN delivery of NVLP could make an efficacious vaccine, and support the case that IN delivery of other VLP vaccines against sexually transmitted diseases ([@bib6]) is a viable strategy. Herbst-Kralovetz and coworkers studied the effects of IN immunization of mice with NVLP and other adjuvants, including TLR7 agonists ([@bib113]). They tested resiquimod (R848), a TLR7/8 agonist that is FDA-approved for clinical trials, and a related midazoquinoline-based TLR7 agonist, gardiquimod, compared with cholera toxin (CT), a strong but risky mucosal adjuvant that is an experimental "gold standard." In experiments similar to those described above with CpG adjuvant, gardiquimod provoked anti-NVCP serum IgG and mucosal IgA responses (intestinal, respiratory, and reproductive mucosa) on par with CT and superior to R848. Thus, a TLR7 agonist produces excellent mucosal immune responses at intestinal and vaginal sites distal from the nasal delivery site. Herbst-Kralovetz and coworkers also studied murabutide as a vaccine adjuvant for plant-derived NVLP delivered IN in mice ([@bib41]). Murabutide is a synthetic immunomodulator derived from muramyl dipeptide (MDP), the smallest bioactive unit of bacterial peptidoglycan. Unlike MDP, murabutide has no pyrogenic activity and is thus safer. Murabutide is recognized by the nucleotide binding oligomerization domain-containing protein 2 receptor on mammalian cells, and was approved for testing in several clinical trials as an injectable vaccine adjuvant. [@bib41] showed that it is a potent mucosal adjuvant as well, inducing serum and mucosal anti-NVLP antibodies after IN delivery. Murabutide was tested at 25, 100, and 250 μg/dose delivered with 25 μg NVLP/dose, and showed adjuvant activity at all doses, with 100 μg being optimal. Moreover, its mucosal adjuvant activity was comparable with gardiquimod (TLR7 agonist) and CT, both of which suffer some safety concerns. Thus, a murabutide adjuvanted IN-delivered NVLP is an excellent candidate for clinical evaluation. Further studies with plant-derived NVLP tested the adjuvant effect of a gelling polysaccharide derived from the plant Aloe vera ([@bib114]). The product is called GelSite^®^, manufactured under cGMP by Nanotherapeutics (<http://www.nanotherapeutics.com/products_pipeline_gelvac.php>) and is a high-molecular-weight anionic polysaccharide (sodium polygalacturonate). The preparation is liquid but can be processed to make a dry powder formulation called GelVac™, and creates a gel whether in liquid or powder form upon contact with body fluids at the site of administration. The GelVac dry formulation with NVLP delivered IN to mice produced anti-NVLP serum and mucosal antibodies at titers higher than the liquid formulation. The latter could be improved by addition of gardiquimod adjuvant, but was still similar to GelVac. Presumably, the mucosal adjuvant activity of the polygalacturonate derives from a depot effect, which prevents draining or ciliary clearance of the IN-delivered material away from the lymphoid effector sites in the nasal mucosa. Charles Arntzen is collaborating with Advanced Bionutrition (ABN) (<http://www.advancedbionutrition.com/default.aspx>) to evaluate norovirus oral vaccine formulations, with a goal of entering human clinical trials. The oral delivery formulation is supported by US Patent No. 7,998,502, which makes use of bioadhesive materials in order to enhance the efficiency of oral delivery. Human Immunodeficiency Virus Vaccines {#s0025} ------------------------------------- Human immunodeficiency virus (HIV) is now a well-known and feared infectious agent that targets CD4^+^ lymphoctyes and ultimately results in loss of adaptive immune system functions. After monumental attempts over the past 30 years by researchers worldwide to develop a vaccine, the results to date are modest, at best. A number of studies have used plants for recombinant expression of HIV antigens. A consensus of opinion among plant-based vaccinologists holds that mucosal immune responses are important, if not crucial, to protective efficacy, and thus often focus on delivery of antigens at mucosal sites. We will review selected works here, and readers are directed to more extensive review articles ([@bib74], [@bib133]) for further information. Many plant-based efforts at HIV vaccines have utilized VLP as a means to increase the antigen density and provide a repeating antigen array that can more effectively engage B-cell receptors. An interesting early example is the use of potato virus X (PVX) to produce chimeric plant virus particles that display the highly conserved "ELDKWA" epitope from the HIV glycoprotein gp41 ([@bib65]). The epitope coding sequence was fused to the N-terminus of the PVX coat protein, and the recombinant virus used to infect N. benthamiana plants, resulting in robust production of chimeric virus particles. Purified virus was used to immunize mice either intraperitonially (IP) or mucosally IN without adjuvant, resulting in serum IgG and IgA specific for ELDKWA. The authors also used a mouse model for human immunity: severe combined immunodeficient mice reconstituted with human peripheral blood lymphocytes (hu-PBL-SCID), immunized with human autologous monocyte-derived dendritic cells pulsed with chimeric ELDKWA virus particles. Serum from both normal and hu-PBL-SCID mice neutralized HIV-1, demonstrating the potential for protection against HIV infection. The authors suggest that the PVX chimeric particle system could also be used for T-cell epitopes. However, it is interesting to note that even low-titer antibodies could protect nonhuman primates against repeated low-dose mucosal exposure to SHIV ([@bib34]), which suggests that neutralizing antibodies alone may be at least partially protective. Mor and his colleagues have also pursued the HIV gp41 and peptides derived from it. They fused the ELDKWA epitope to cholera toxin B subunit (CTB) in order to achieve targeting to the mucosal epithelium ([@bib70]). Mice immunized IN with CTB- ELDKWA produced in Escherichia coli developed serum and mucosal (fecal) antibodies that inhibited transcytosis of HIV-1 in vitro. The authors demonstrated expression of the fusion protein in N. benthamiana, but did not test the immunogenicity of that product. Later, Mor and colleagues created stable transgenic plant lines that expressed a CTB fusion that contained a longer segment of the gp41 membrane proximal region, called CTB-MPR(649-684) ([@bib69]). They performed extensive characterization of the purified fusion protein, including glycan analysis, as they found asparagine-linked glycosylation of a consensus site in the CTB sequence. The glycan, not found in bacterially expressed CTB, did not interfere with ganglioside binding or display of the HIV epitope. Mice immunized IN with 35 μg of liposome-conjugated CTB-MPR(649-684) and 1 μg cholera toxin for 5 consecutive weeks and boosted IP with 3 μg of CTB-MPR(649-684) at week 9 produced substantial serum IgG and vaginal IgA specific for MPR(649-684). Thus, one can expect a need to balance the advantage of mucosal targeting afforded by CTB with its overpowering immunogenicity. Mor and colleagues recently turned to a different strategy for plant-based HIV vaccine: a VLP comprising gag p55 and gp41 ([@bib46]; [Figure 2](#f0015){ref-type="fig"} ). They created stably transgenic N. benthamiana expressing HIV gag p55 and showed that gag VLP accumulated in the plants. The p55 was not processed and remained as mostly 55 kDa protein. They then used an ICON TMV-based replicon to express "deconstructed gp41" (dgp41), comprising the membrane proximal external region (MPER), transmembrane domain, and cytoplasmic tail, in transgenic gag plants. The result was accumulation in leaves of VLP containing both gag p55 and dgp41. The evidence for VLP formation included electron micrographs showing particles forming at the plasma membrane, in endosomes, and in the extracellular space. Moreover, both antigens cosedimented in rate zonal density gradients. These workers immunized mice with particle preparations and with CTB-MPR, priming IP with one antigen (weeks 0, 2, and 4) and boosting IP with the same or with the other at weeks 8 and 10 ([Figure 3](#f0020){ref-type="fig"} ). Mice primed and boosted with VLP showed strong anti-gag serum IgG, with all mice responding by week 6. Anti-gp41 responses were less impressive, but seven of eight mice seroconverted by week 12 after VLP prime and boost. All mice primed with either VLP or CTB-MPR and boosted with the other antigen seroconverted for gp41 by week 10. All immunizations were adjuvanted with Ribi adjuvant (Sigma) to a final concentration of 2% oil, which suggests that a good mucosal adjuvant will be needed if the VLP are used for IN delivery.Figure 2Models of HIV virion and deconstructed gp41 virus-like particles (VLP).The HIV virion (left) contains a core of gag capsid enclosing the RNA genome segments, gag matrix protein associated with a lipid membrane (derived from host cell plasma membrane), and envelope glycoproteins gp120/gp41. [@bib46] describe a VLP (right) made in plants by coexpression of gag p55 and "deconstructed" gp41, which lacks the gp120 component, thus exposing the membrane proximal external epitopes that are conserved and capable of inducing HIV-neutralizing antibodies.Figure 3Seroconversion of mice immunized with various combinations of HIV-derived CTB-MPR (C) and VLP (V).Groups of eight mice were primed IP with either CTB-MPR or VLP at weeks 0, 2, and 4, and subsequently boosted IP with either CTB-MPR or VLP at weeks 8 and 10. The number of mice in each group that seroconverted with antibody responses to gag protein (a) or gp41 (b) are plotted vs. the week number. Another VLP approach for plant-based HIV vaccines used membrane particles generated by fusions with hepatitis B surface antigen (HBsAg). HBsAg expressed in plant cells inserts into the ER membrane, which invaginates to form particles within the ER lumen ([@bib95]). Transgenic potato tuber expressing HBsAg was orally immunogenic in mice ([@bib85]) and humans ([@bib108]; and see below). Thus, HBsAg fusions seem a reasonable strategy. One fusion (TBI-HBsAg) encodes an artificial polypeptide composed of several immunogenic T- and B-cell epitopes of HIV-1 env and gag proteins fused to the N-terminus of HBsAg small protein ([@bib93]). These workers generated stable transgenic tomato plants, and freeze-dried the fruits for oral delivery to mice. Tomato powder was suspended in water and gavaged into mice at a dose of 30 ng antigen in 1 mL, three times over a 4-week period. The mice were then boosted at day 42 using plasmid DNA encoding TBI-HBsAg. Over the time course, anti-HBsAg antibodies rapidly accumulated in serum and fecal samples, which increased after boosting with the DNA vaccine. Anti-HIV responses were more modest, but measureable in both serum and feces, although no endpoint titers were performed. Nonetheless, the results looked promising. Sala and colleagues aimed to elicit cellular immune responses against HIV-1, using a polyepitope comprising eight HLA-A∗0201 restricted class I epitopes identified in five major HIV-1 proteins (Gag p17 and p24, protease, reverse transcriptase, and integrase), fused to the N-terminus of HBsAg small protein and the C-terminus of the HBsAg preS2 domain ([@bib28], [@bib29]). They generated stably transgenic tobacco plants expressing the fusion protein "HIV-1/HBV" and fed HSB humanized transgenic mice, in which HLA-A2.1 restricted epitopes are presented by the human class I MHC allele on a mouse knockout background. Ten days before ingesting the freeze-dried transgenic tobacco (eight doses over 2 weeks), the mice were primed with a plasmid for expression of the same fusion protein. The treatment produced anti-HIV-1-specific CD8^+^ T-cell activation in mesenteric lymph nodes, indicating the potential to develop specific cytotoxic T-cell-mediated immunity. However, the authors also found significant activation of regulatory T cells in the same animals, suggesting the potential for development of tolerance. These conflicting results indicate that one must be very careful when interpreting results of feeding studies. Hepatitis B Vaccines {#s0030} -------------------- Hepatitis B virus (HBV) infection may lead to liver cancer, with a million deaths per year worldwide. Because HBV can be transmitted by sexual contact and births to infected mothers, mucosal immunity would enhance the systemic protection provided by the injectable yeast-derived (BsAg vaccine ([@bib72])). A large volume of work over the past 20 years has demonstrated expression of HBsAg in a variety of plant systems including potato, lettuce, tomato, tomatillo, corn, and banana ([@bib97]). A recent review article provides a very good survey of this field ([@bib94]). HBsAg expression in plants was first demonstrated in transgenic tobacco ([@bib68]), and then transgenic potato tubers ([@bib85]), which were used to demonstrate oral immunogenicity in mice ([@bib51]). The antibody titers without cholera toxin adjuvant were modest, and a "subimmunogenic" priming dose with yeast-derived HBsAg was required to enable demonstration of a boosting effect by fed HBsAg potatoes. A subsequent clinical trial showed that 19 of 33 previously vaccinated subjects who ate 100 g raw potato doses (∼850 μg HBsAg) had substantial boosting of serum anti-HBsAg titers ([@bib108]). These data indicate that oral delivery of HBsAg could be used as a booster vaccination, since some of the subjects experienced the boosting effect 10--15 years after their intramuscular vaccinations. A smaller study showed that two naïve human subjects were seroconverted after eating transgenic lettuce leaves that contained submicrogram quantities of HBsAg ([@bib43]). Although neither of these studies used an adjuvant (unless one considers the possibility that compounds found in either potato or lettuce could function so), an interesting study of oral immunogenicity in mice of alum-adsorbed yeast-derived vaccine at low doses over longer intervals showed efficient seroconversion ([@bib44]). The suggestion from that study was that low doses and longer intervals would yield better results for oral delivery of HBsAg. Immune tolerance is always a potential concern when antigens are delivered orally, and needs further study. In humanized transgenic mice (HLA-A2.1 and HLADR1 alleles), oral delivery of different doses of transgenic HBsAg tobacco resulted in increasing levels of regulatory T cells (Tregs, associated with tolerance) with increasing HBsAg dose ([@bib52]). Serum and fecal IgG and IgA correlated inversely with antigen dose, suggesting that oral tolerance occurred at higher doses. However, the relatively poor antibody responses made the correlations somewhat tenuous, and Tregs increased substantially in mice immunized with wild-type tobacco. Nonetheless, the differences in Tregs in HBsAg tobacco-immunized and in wild-type tobacco-immunized mice were statistically significant. Consistent with these findings, Pniewski and coworkers ([@bib83]) immunized mice with low doses of freeze-dried transgenic lettuce containing 0.1 μg HBsAg without adjuvant at a 60-day interval between prime and boost immunizations, and observed mucosal and humoral anti-HBsAg antibodies. These data coupled with those described above ([@bib44], [@bib52]) argue for lower doses in order to avoid development of Treg-mediated tolerance. Different findings were obtained by feeding mice HBsAg transgenic potatoes at different doses (0.1--30 μg/dose) with 10 μg CT adjuvant, three times weekly followed by a parenteral boost of 0.5 μg HBsAg at week 8 ([@bib126]). In this study, anti-HBsAg antibodies in both serum and fecal samples increased in direct relation to the oral dose. The difference in these and the results discussed above ([@bib44], [@bib52]) may relate to the use of CT adjuvant in the oral doses and the parenteral boosting after oral dosing, both of which could suppress Treg activation. In summary, much work using oral delivery of plant-derived HBsAg shows substantial potential for vaccine development. It seems likely that for human use, a purified preparation will be preferred, in order to accurately adjust the dosage to optimal levels, and to avoid possible and variable interfering effects due to other components in crude samples. In this regard, high-level expression will facilitate purification, and the use of a viral replicon system will benefit these pursuits ([@bib38]). Rice-Based Oral Vaccine (MucoRice) {#s0035} ---------------------------------- ### Attractiveness of Oral Vaccine for the Healthy Life {#s0040} The mucosal immune system is capable of simultaneously executing two opposing immune responses including active (e.g., antigen-specific neutralizing SIgA antibodies) and quiescent ones (e.g., oral tolerance; Chapter 42). For example, oral immunization of protein antigens with an appropriate mucosal delivery vehicle and/or mucosal adjuvant has been shown to induce antigen-specific SIgA and serum IgG antibody responses, leading to a double layer of protection against mucosally invading pathogens. On the other hand, injectable vaccines, which are currently and most commonly used in the field, are designed to induce protective immunity in the systemic compartment, but are not aimed at inducing mucosal immunity. Since most pathogens related to emerging and reemerging infectious diseases (e.g., Vibrio cholera, rotavirus, HIV, influenza virus, and enterotoxigenic E. coli) infect the host at the mucosal surface of respiratory or digestive tracts, mucosal vaccination will most likely improve the efficacy of protective immunity offered by the current parenterally administered vaccines and could potentially prevent invasion of the mucosal surface by pathogens. We recognize that cooking the rice (or other edible plant material) is very likely to denature the vaccine antigens, thus rendering them poorly immunogenic. In order to overcome this challenge, Kiyono and coworkers are investigating the oral delivery of rice flour derived from transgenic seeds. A clinical examination of the safety of powdered MucoRice-CTB (dissolved in water) in healthy volunteers showed no adverse reactions (Kiyono et al., unpublished data). A phase I clinical trial is being planned for 2015. ### Development of Rice-Based Oral Vaccines from Edible Grains {#s0045} To overcome critical concerns of the production, storage, and delivery of vaccines, an adaptation of the transgenic rice system has been used for the development of rice-based vaccines (or MucoRice system; [Figure 4](#f0025){ref-type="fig"} ). The MucoRice system has recently attracted interest as an alternative approach for the development of oral vaccines ([@bib79], [@bib80], [@bib110], [@bib127], [@bib128]).Figure 4Rice-based oral vaccine induces protective immunity against diarrheal diseases.The MucoRice system using the rice protein body (PB)-specific promoter containing the B subunit of cholera toxin (CTB) resulted in the generation of rice seeds expressing CTB. The CTB expressed in the rice seeds was protected from digestive enzyme and stable over a prolonged time (e.g., 36 months) without refrigeration storage. Oral vaccination of mice with MucoRice-CTB resulted in the induction of protective immunity (e.g., CTB-specific SIgA antibodies) against CT-induced diarrhea. Lower right shows intestines from mice after treatment with cholera toxin (CT). Rice seeds are edible tissues and capable of producing and storing relatively large amounts of transgenic recombinant proteins ([@bib8], [@bib100]). Recombinant proteins expressed and accumulated in rice seeds are stable for prolonged periods ([@bib18]). Another advantage of rice plants is the presence of two protein storage vacuoles called protein bodies (PB) in the starchy endosperm of seeds, which provide a greater space for the accumulation of recombinant proteins ([@bib45]). PB-I and PB-II are distinguished by differences in shape, density, and protein composition ([@bib122]). The main storage proteins for PB-I are the alcohol-soluble prolamins (e.g., 13K prolamin) and the water-soluble glutelins (e.g., glutelin B1), which accumulate preferentially in PB-II ([@bib45], [@bib122]). Due to their water solubility, the proteins in PB-II are more vulnerable to digestion in the gastrointestinal tract than the proteins stored in PB-I. The PB system is thought to be a great advantage of rice for its applicability to oral vaccine development, since an effective oral vaccine would benefit from a built-in safeguard against digestion, particularly against the harsh acidic environment in the stomach. Further, the unique nature of protein storage systems of PB-I and PB-II could be used as a natural way to provide either slow or immediate release of vaccine antigens in the gut. These novel characteristics of the rice seed protein expression and storage system can be used for the development of oral vaccine. ### Rice-Based Oral Vaccine for Infectious Diseases {#s0050} For the proof of concept of MucoRice vaccines, the B subunit of cholera toxin (CTB), an epithelial GM1 ganglioside receptor-binding molecule, was purposely selected as a prototype antigen to demonstrate: (1) the capacity of the rice-based mucosal vaccine to induce systemic as well as mucosal immunity, and (2) to showcase the practicality of using the rice transgenic expression system ([Figure 4](#f0025){ref-type="fig"}). Using the transgenic rice system, the accumulation of CTB was achieved at approximately 2% of total seed protein (0.15% of seed weight), which was higher than the other stable transgeneic plant expression systems (e.g., potato tubers and tobacco leaves) ([@bib4], [@bib17], [@bib79]). Transgenic rice is stable in the harsh environment of the gastrointestinal tract because rice seed possess protein storage organelles (PB) that can be used as natural antigen-containing capsules. To examine the ability of CTB in rice PB (or MucoRice-CTB) to withstand digestive enzymes in the intestinal tract, total seed proteins were subjected to pepsin treatment in vitro. Most of the CTB accumulated in PB-I of MucoRice seeds remained intact after pepsin treatment, whereas most of the antigen accumulated in PB-II was digested by pepsin under the same conditions ([@bib79]). It was further shown that the vaccine antigens expressed in the rice seed PB system are stable at room temperature for several years without loss of immunogenicity ([@bib79], [@bib110]). The MucoRice system is thus qualified as a suitable system for the expression, accumulation, and storage of vaccine antigens. To examine the ability of MucoRice-CTB to induce antigen-specific immune responses, MucoRice-CTB was delivered to the intestinal immune system in mice. When the murine intestinal tract containing Peyer's patches (PP) was exposed to MucoRice-CTB, the rice-based vaccine was taken up by professional antigen-sampling M cells located in the follicle-associated epithelium of PP. Successful oral vaccination of mice with MucoRice-CTB thus induced antigen-specific mucosal IgA and serum IgG antibody responses that showed toxin-neutralizing activity and thus prevented toxin-induced diarrhea in the murine model ([@bib79], [@bib129], [@bib110]). Furthermore, the oral MucoRice-CTB vaccine induced cholera toxin-neutralizing antibodies in nonhuman primates ([@bib80]). The MucoRice system has further advanced to the expression of the nontoxic form of the entire CT toxin consisting of the mutant A and B subunits ([@bib129]).To demonstrate the versatility of the MucoRice system, the other toxin-related antigen, a nontoxic fragment of the C-terminal half of the heavy chain from botulinum neurotoxin type A (BoHc), was expressed and produced by the seeds of MucoRice and induced neutralizing antigen-specific antibody responses in mice orally immunized with MucoRice-BoHc ([@bib128]). In another adaptation of the transgenic rice seed system, researchers showed the expression of the major outer membrane protein (MOMP) of Chlamydophila psittaci, a causative agent for avian chlamydosis, with or without genetic fusion to the B subunit of E. coli heat-labile enterotoxin (LTB) ([@bib136], [@bib138]). Oral immunization of experimental animals with the rice expressing MOMP or MOMP-LTB resulted in the induction of antigen-specific protective immunity. Further, the rice-based vaccine development was extended to the expression of virus-associated antigens. A neutralizing epitope of porcine epidemic diarrhea virus, the capsid precursor polypeptide of foot-and mouth-disease virus, and the glycoprotein of Newcastle disease virus have been expressed in the rice seeds and showed oral immunogenicity for the induction of virus antigen-specific immune responses ([@bib39], [@bib117], [@bib125]) These findings clearly demonstrate the usefulness and attractiveness of the rice transgenic system for the development of cold chain-, syringe, and needle-free vaccines. ### Seed-Based Oral Immunotherapy for the Control of Allergy {#s0055} The concept of oral tolerance (or systemic unresponsiveness to orally exposed antigen) has been considered as an attractive and logical strategy for the development of immunotherapy for the prevention and/or treatment of allergic diseases. The cellular and molecular mechanisms of oral tolerance have not yet been precisely clarified; however, it is well known that prolonged oral administration of allergens can induce a state of systemic unresponsiveness to the mucosally exposed allergens, perhaps via the generation of regulatory T cells in the GALT. For the purpose of continuous delivery of allergen via an oral route for the induction of oral tolerance, the transgenic rice system has been adopted for the generation of rice seeds expressing allergenic proteins ([@bib105]). To minimize undesired anaphylactic reactions that might be elicited during the oral desensitization process using allergens, the transgenic rice plants expressing T-cell epitope peptides from pollen allergens of Japanese cedar (e.g., Cry j I and Cry j II) were generated. Oral administration to mice of the transgenic rice seeds expressing the Cry j I and Cry j II T-cell peptides prior to systemic challenge with total cedar pollen allergens resulted in the induction of oral tolerance with the inhibition of allergen-specific allergy-inducible T helper 2 (Th2) responses and their supported allergen-specific IgE responses ([@bib105]). It was also important to note that the tolerance induced by the rice expressing the T-cell peptides of Cry j I and Cry j II resulted in the inhibition of pollen-induced clinical symptoms of nasal sneezing in a murine inhalation allergic model. These data supported a possibility for the adaptation of transgenic rice seeds expressing allergen-specific T-cell epitopes for oral delivery and induction of oral tolerance against pollen allergen-specific responses. The transgenic rice system for the expression of allergens was extended to allergic asthma caused by house dust mites (HDM) ([@bib101], [@bib124]). A major HDM allergen, Der p1, and Der p1-associated immunodominant human and mouse T-cell epitopes have been expressed and accumulated in the rice PB system and used for the induction of oral tolerance. Oral administration of mice with the rice seed expressing these HDM antigens resulted in the inhibition of allergen-specific pathological responses including bronchial hyperresponsiveness ([@bib101]). These evidences suggest that the rice transgenic system can be a useful tool for the development of allergen-specific mucosal immunotherapy. Plant-Based Mucosal Animal Vaccines {#s0060} =================================== Over 75% of newly emerging human diseases of the past 10 years have been zoonotic, i.e., of animal origin, and some seasonal diseases, such as influenza, can infect both humans and livestock. In addition to this direct link between animal and human health, veterinary diseases indirectly affect public health and well-being through their potentially devastating impact on food production and trade. There is a growing need for the development of safe, effective, and affordable veterinary vaccines. Plant-based production platforms can help to meet some of the specific requirements for the development of successful and economically viable animal vaccines: The cost of production in the case of veterinary vaccines must be very low in order to allow the vaccine to find broad use in the farming community, particularly in developing countries where these vaccines are often most needed ([@bib89]). Production in transgenic plants is scalable and relatively inexpensive. Veterinary vaccines will ideally be stable and not require cold chain storage and transport; expression of vaccine antigens in plant storage organs such as seeds and tubers provides a protective storage vehicle. An easy route of administration will allow the vaccine to be used more broadly in settings where trained veterinary professionals are not at hand and for administration to large populations: Production of an edible vaccine in a fodder crop or reconstituted in feed pellets allows for easy administration by feeding. Most importantly, any vaccine must be safe and effective at protecting the immunized animal from the target disease: Plant-production systems have been used to express protective vaccine candidates, without the risk of contamination with animal pathogens. Since the transmission routes of many animal pathogens are fecal-oral, respiratory, or through direct contact resulting in infection via mucosal surfaces, vaccines targeting these diseases must elicit a strong mucosal immune response in order to be efficacious. Plants particularly lend themselves to the development of oral vaccine candidates for stimulating mucosal immunity. Since the report of the first potential plant-based animal vaccine in 1993, a chimeric plant-virus particle displaying an epitope of foot-and-mouth disease virus (FMDV) VP1 ([@bib112]), many groups have successfully used plants as a production platform for the development of veterinary vaccines (comprehensively reviewed by [@bib98], [@bib22], [@bib30], [@bib88], [@bib89]). In 2006, the United States Department of Agriculture (USDA) licensed the first plant-produced vaccine, a veterinary vaccine against Newcastle disease virus in chickens to Dow AgroSciences ([@bib115]), setting a precedence for regulatory approval of plant-made vaccines. Here we will review some of the recent developments in the field of plant-produced mucosal animal vaccines. Particularly, we will focus on vaccines for four target animal groups: bovine/ovine, porcine, poultry, and aquaculture. Bovine and Ovine Vaccines {#s0065} ------------------------- Early efforts in the development of plant-made veterinary vaccines have focused on FMDV ([@bib9], [@bib112]). Foot-and-mouth disease is to this day one of the most devastating diseases affecting farmed animals and wild ruminants worldwide, causing economic losses associated with a high morbidity rate and containment procedures involving culling and movement restrictions. Current vaccines are based on inactivated virus, with safety risks associated with its production. In fact, a vaccine manufacturing facility and associated research laboratories were linked to the 2007 outbreak of FMDV in the United Kingdom ([@bib15]). Efforts are therefore focused on developing an inherently safe subunit vaccine. All plant-produced FMD vaccine candidates are based on all or parts of the polyprotein precursor P1, the products of which (VP1, VP2, VP3, and VP4) form the FMDV viral capsid. Various strategies have been followed to express P1 or epitopes thereof, ranging from display of VP1 epitopes on chimeric plant virus particles ([@bib112], [@bib123], [@bib139]), expression of a hybrid protein containing FMDV B- and T-cell epitopes ([@bib3]), fusion of VP1 to reporter proteins ([@bib61]), and expression of full-length P1 to allow formation of FMDV-like particles ([@bib19]). Chimeric VLPs have been produced by virus inoculation of Chenopodium species with an epitope of VP1 displayed on tobacco necrosis virus A (TNV-A) ([@bib139]) and bamboo mosaic virus (BaMV) ([@bib123]). Similar yields of 0.33 mg/g fresh weight tissue (FWT) and 0.5 mg/g FWT were achieved, respectively. The efficacy of the BaMV-VP1 preparation with adjuvant was tested by intramuscular (i.m.) and subcutaneous injection of swine, followed by challenge with FMDV. The results indicated stimulation of high levels of anti-VP1 IgG and neutralizing antibodies, production of IFN-γ, and full protection against challenge, even at a single dose of 1 mg BaMV-VP1 ([@bib123]). Testing of the TNV-VP1 preparation was performed in mice via the intranasal route without adjuvant. While animals developed high titers of serum IgG and IgA as well as fecal and vaginal IgA, much of the immune response was raised against the carrier TNV, with lower levels of anti-VP1 IgG and IgA detected ([@bib139]). No challenge experiment was performed. Nevertheless, the results indicate that a mucosal immune response was induced. Following another strategy, the full precursor polypeptide P1 of FMDV has been expressed in transgenic alfalfa ([@bib19]), rice leaves ([@bib117]), and transiently in leaves of N. benthamiana ([@bib109]). The strategy of Dus Santos et al. was to express P1 together with the 3C protease, which cleaves this polyprotein during natural FMDV replication and assembly, with the aim of producing virus-like particles. Relatively low expression levels of 0.005--0.01% total soluble protein (TSP) were achieved in transgenic alfalfa leaves; however, IP immunization of mice with crude leaf extract (total of four injections with adjuvant) induced an FMDV-specific immune response and successfully protected mice against challenge ([@bib19]). Production of FMDV VLPs was also attempted by transient coexpression of a modified FMDV VP1 with protease cleavage sites recognized by cowpea mosaic virus 24K protease, and the protease itself ([@bib109]). Though expression and processing of P1 was detected, no VLP formation was observed. Recently, Wang et al. demonstrated the oral immunogenicity in mice of P1 expressed in transgenic rice leaves at up to 0.13% TSP ([@bib117]). Serum IgG and mucosal IgA were detected after the first 2 weeks of a 4-week feeding regimen of five weekly doses of 10 μg P1 antigen. However, only 20--40% of animals showed virus clearance after challenge. One reason for the low efficacy of orally delivered FMDV vaccine from plants may be the low expression levels achieved. This low expression level was addressed by the transplastomic expression of FMDV VP1 cotranslationally fused to the β-glucuronidase (β-GUS) reporter ([@bib61]). Expression levels of up to 51% TSP were achieved and the fusion protein was able to induce serum IgG production after IP injection in mice. Further investigations are necessary to determine whether the described FMDV vaccine candidates are efficacious in the target animals, cattle and sheep. Another fusion protein consisting of B- and T-cell epitopes of FMDV VP1, VP4, 2C, and 3D has been expressed transiently in N. benthamiana at levels of 0.5--1% TSP ([@bib3]). Immunogenicity studies in guinea pigs showed the formation of neutralizing antibodies (NA) and protection from challenge with FMDV. Recently, a VLP-based strategy has been employed for the production of a bovine papillomavirus (BPV) vaccine candidate by transient expression in N. benthamiana ([@bib62]). Bovine papillomavirus is tumorigenic in cattle and horses and traditional vaccines cannot be produced because BPV cannot be grown in culture. Virus-like particles of BPV were produced by expression of BPV L1 protein, and yields of 183 μg/g FWT were obtained after purification. Immunogenicity of the particles was shown by parenteral administration in rabbits. A group in India has shown the potential of plants as a production platform for oral rinderpest virus (RPV) and peste des petits ruminants virus (PPRV) by performing immunogenicity studies in the target animals. Both viruses are members of the Morbillivirus genus of the Paramyxoviridae and can cause severe diseases in cattle and small ruminants, respectively. Khandelwal et al. have expressed the hemagglutinin (H) protein of RPV and hemagglutinin-neuraminidase (HN) protein of PPRV in transgenic peanut leaves. Cattle were fed with 5--7.5 g of peanut leaves expressing RPV H protein at a level of 0.5% TSP. Single doses administered once a week for 3 weeks elicited a humoral anti-RPV response and NA were detected ([@bib49]). Later immunogenicity tests in mice revealed that oral vaccination induced serum IgG and IgA ([@bib47]). In the case of PPRV HN protein, sheep were fed in a similar way with 5 g of transgenic peanut leaves once a week for 5 weeks. High levels of NA were reported after the first week of feeding and a cell-mediated response was detected ([@bib48]). These results demonstrate that plant-produced edible vaccines can elicit strong immune responses in large animals. However, in the case of rinderpest a successful Global Rinderpest Eradication Programme started in the 1990s has resulted in the World Organization for Animal Health (OIE) declaring the world free of rinderpest in 2011, therefore eliminating the need for an edible RPV vaccine. In contrast, an increasingly economically important disease of cattle is bovine viral diarrhea, caused by bovine viral diarrhea virus (BVDV). The disease is mainly associated with fertility problems in infected cattle and can lead to abortion and malformations of calves. In a quest to develop a safer alternative to the commercially available inactivated vaccine, recent progress has been made by two groups in Argentina using plant expression systems. Dus Santos and Wigdorovitz reported the development of transgenic alfalfa plants expressing BVDV structural protein E2, which encodes the main antigenic sites ([@bib20]). The same group has further developed this system by fusing a truncated version of E2 (tE2) to an APCH molecule, which targets antigen presenting cells ([@bib2]). Expression levels were relatively low at 1 μg/g FWT; however i.m. immunization of cattle with plant extract containing 3 μg APCH-tE2, followed by a booster shot, was able to afford protection from challenge with BVDV, with no virus detectable in these animals. Another group achieved higher expression levels of 20 μg/g FWT by transient expression of truncated E2 (not as a fusion) in Nicotiana tabacum ([@bib78]). Guinea pigs were immunized subcutaneously with 20 μg of E2 antigen that was shown to stimulate production of NA in six out of seven animals. It will be interesting to see whether these vaccine candidates can stimulate mucosal responses in cattle when applied directly to the mucosa. The research groups in Argentina have also investigated the plant-based production of bovine rotavirus (BRV), which is the leading cause of diarrhea in neonatal calves in that country. Calves are susceptible to the disease within the first 3 weeks after birth; therefore induction of lactogenic immunity is a preferred method of protection. In 2004, Wigdorovitz et al. reported on the efficacy of a plant-produced fusion protein of eBRV4, a previously identified peptide capable of eliciting a protective response, to β-GUS ([@bib118]). This fusion protein was expressed in transgenic alfalfa at levels up to 0.09% TSP. When tested in a suckling mouse model, pups born to dams fed weekly for 2 months with 6 μg eBRV4 were protected against BRV challenge. The same group recently produced BRV VP8∗, the major immunogenicity determinant, in transplastomic N. tabacum at levels of 600 μg/g FWT (150 μg/g FWT soluble fraction) and tested this antigen in the suckling mouse model ([@bib60]). Dams were IP immunized with 2 μg VP8∗ and boosted once. Dams showed serum IgG and NA responses. Pups were challenged 5 days after birth and were 80--100% protected compared to control pups, 100% of which developed diarrhea. These results indicate that a plant-based vaccine conferring lactogenic immunity to BRV in calves may be possible. Some early successes in the production of vaccine candidates in plants were achieved with the development of recombinant rabies subunit vaccines in tomato, spinach, and tobacco ([@bib73], [@bib76], [@bib130]). Oral immunogenicity and parenteral protective efficacy of these vaccines were shown in mice. Rabies is a zoonotic disease affecting mammals worldwide, spread by bites of infected animals. In developing countries, rabies represents a particular health threat due to widespread infection in the stray dog population. In Latin America, another key reservoir of the rabies virus is the bat population that can spread the disease to livestock. Recently, Loza-Rubio et al. have demonstrated the oral efficacy of a maize-based rabies vaccine in sheep ([@bib63]). Sheep in different groups were fed with a single dose of ground maize kernels containing 0.5--2.0 mg of rabies virus G protein (25 μg/g seeds). All doses were shown to elicit NA, with the 2.0-mg dose producing NA titers comparable to those elicited by a commercial inactivated rabies vaccine. After challenge with a lethal dose of rabies virus, all vaccinated groups achieved survival rates of at least 50%, with the highest dose of 2.0 mg recombinant G-protein achieving the same survival rate as the inactivated vaccine (83%). These results show that a plant-produced subunit vaccine can be effective at protecting large livestock animals against rabies. In addition to these virus vaccines, plants have also been investigated as vaccine production platforms to protect livestock from bacterial pathogens. Mannheimia haemolytica is one of the main causal agents of pneumonic pasteurellosis in cattle, a disease with worldwide economic impact. Current treatment involves the use of antibiotics, but prevention with vaccines would be preferable. Lee et al. have expressed the GS60 outer membrane lipoprotein of M. haemolytica in transgenic alfalfa at levels of 0.02% TSP ([@bib58]). An immunogenicity study in rabbits was performed to compare oral to i.m. application. Whereas injection efficiently induced anti-GS60 antibody production, oral administration only elicited an immune response in one of six animals. This may be due in part to the low expression levels achieved. Further investigations in cattle should be undertaken. Porcine Vaccines {#s0070} ---------------- Bacterial and parasitic diseases, in addition to several viral diseases, have also been targeted by researchers investigating plants as production platforms for porcine vaccines. In 2012, researchers reported the oral efficacy of plant-produced vaccines against porcine cysticercosis ([@bib7]) and Toxoplasma gondii ([@bib121]), as well as the production of a mucosal immunogen against postweaning diarrhea ([@bib50]). Porcine cysticercosis, or pig tapeworm, is caused by the parasite Taenia solium. A vaccine consisting of three peptides has been produced in transgenic papaya callus (S3Pvac-papaya) ([@bib140]). Rabbits received oral administration of 20 mg of callus, boosted on day 7, and were challenged (with Taenia pisiformis, closely related to T. solium) on day 15. Upon sacrifice of the animals, researchers found a 90% reduction in hepatic lesions and 94% reduction in cysticerci recovered, thereby showing efficient protection of rabbits ([@bib7]). Another economically important parasite of pigs, as well as humans and other warm-blooded animals and livestock, is T. gondii, which causes toxoplasmosis and infects its host through gut mucosa. The T. gondii GR4 antigen was expressed in transplastomic N. tabacum at levels of 6 μg/g FWT, and suspensions of lyophilized leaf extract were used to orally immunize mice by gavage ([@bib121]). Mice were orally challenged and later sacrificed to enable brain examination. The experiment showed induction of both systemic and mucosal immune responses and a 59% reduction in brain cyst load compared to that of a control. In addition to these parasitic diseases, the bacterial postweaning diarrhea caused by enterotoxic E. coli (ETEC) has recently been targeted. Kolotilin et al. have expressed ETEC F4 fimbrial adhesin (FaeG) to high levels of 11.3% TSP in transplastomic tobacco, and up to 20% TSP by transient chloroplast-targeted expression in N. benthamiana ([@bib50]). The authors showed that the antigen could survive extended incubation in simulated gastric and intestinal fluids, indicating its stability in the gastrointestinal tract and suitability as an oral immunogen. Porcine edema disease in also caused by E. coli and a shiga toxin B subunit has recently been expressed in transgenic lettuce ([@bib71]). Oral immunogenicity experiments in swine are currently in progress. Most of the porcine diseases currently targeted by plant-based vaccines are caused by viruses. One of the earliest examples was the production of glycoprotein S of porcine transmissible gastroenteritis virus (TGEV) in Arabidopsis thaliana and its immunogenicity in mice following i.m. injection ([@bib26]). This host plant is not compatible with large-scale production of antigens; therefore other groups have expressed the S protein in transgenic N. tabacum and maize ([@bib111], [@bib57]). Maize-produced TGEV S protein (13 mg/kg seed) was shown to induce NA after oral immunization of sows, and NA were also detected in milk after farrowing ([@bib56]), showing the potential of lactogenic immunity in piglets. Furthermore, passive immunization of piglets was achieved by oral application of a plant-produced small immune protein (SIP) targeting TGEV ([@bib77]). Transient expression was used to produce SIP at levels of 2% TSP in cowpea, and animals receiving daily oral doses of resuspended lyophilized plant extract showed a reduction of TGEV virus titer in gut and lung as well as reduced pulmonary lesions compared with the control group. Several groups have recently investigated the efficacy of plant-produced subunit vaccines against porcine reproductive and respiratory syndrome (PRRS). This disease, first recognized in 1987, has become one of the most important disease of farmed pigs in North America and Europe, and is becoming a problem in other parts of the world, particularly Asia. Current commercial vaccines targeting the causative agent, porcine reproductive and respiratory syndrome virus (PRRSV), either have limited efficacy (killed virus vaccines) or carry inherent risks of reversion (modified-live virus vaccines); therefore an alternative safe, efficacious vaccine is desperately needed. A group in Taiwan first expressed PRRSV envelope glycoprotein GP5, the major target of neutralizing antibodies, as well as an LTB-GP5 fusion protein in transgenic N. tabacum ([@bib12], [@bib13]). Oral immunogenicity and efficacy studies in pigs demonstrated induction of serum IgG and salivary IgA after the second administration, as well as a reduced viral load in organs and reduced clinical disease signs after intranasal challenge with PRRSV. However, yields in these studies were very low at 108--155 ng/g FWT. Improved yields were achieved by expression of GP5 in transgenic potato at up to 4.7 μg/g FWT in leaves and 1.2 μg/g FWT in tubers, with immunogenicity confirmed by intragastric immunization of mice ([@bib11]). Other groups have recently reported expression and immunogenicity of the nucleocapsid protein (N) produced in transgenic soybean ([@bib116]) and the matrix protein (M) produced in transgenic maize callus ([@bib36]). Production of antigen in soybean and maize has the benefit of easy storage without cold chain; however, further experiments will be needed to show the efficacy of these vaccine candidates. Vaccines against other viral diseases relevant to the swine industry have been investigated for production in plants with limited success. One example is that of classical swine fever virus (CSFV) which is related to BVDV in cattle and can cause severe acute illness in pigs resulting in fever, loss of appetite, diarrhea, and a high mortality rate. Strategies for vaccine production in plants have focused on CSFV glycoprotein E2 either as a subunit vaccine ([@bib32], [@bib59]) or as an epitope displayed on a chimeric VLP of potato virus X ([@bib64]). Of these studies, only one was able to show induction of serum IgG and fecal IgA after oral immunization of mice ([@bib59]). Another example is porcine epidemic diarrhea virus (PEDV), a coronavirus related to TGEV but causing less severe symptoms. In 2003, a subunit vaccine candidate based on the neutralizing epitope of the TGEV spike protein (COE) was expressed in transgenic N. tabacum and shown to stimulate production of specific serum IgG and fecal IgA when administered orally to mice ([@bib5]). Similar approaches have involved the rice-based expression fusion proteins of COE to LTB ([@bib81]), and COE to the M cell-targeting ligand Co1 ([@bib39]). The latter COE-Co1 showed weak induction of serum IgG and fecal IgG when administered to mice by gavage. Challenge experiments with PEDV and testing in the target animal will need to be performed to determine efficacy of these vaccine candidates. Avian Vaccines {#s0075} -------------- There has been a recent surge of interest in plants as a production platform for poultry vaccines. This may in part be due to the USDA approval of a plant cell culture-based vaccine for Newcastle disease virus (NDV) in chickens in 2006 ([@bib115]). Though the vaccine has not been marketed, this was a significant milestone for the regulatory approval of plant-produced vaccines and pharmaceuticals in general. In addition to NDV, targets of recently developed plant-based poultry vaccine candidates include infectious bursal disease virus (IBDV), avian reovirus, C. psittaci, and the causal agent of chicken coccidiosis. Infectious bursal disease is an economically important disease worldwide, which can cause either subclinical infections leading to long-lasting immunosuppression, or acute clinical infection that can have a high mortality rate. The main immunogenicity determinant IBDV VP2 has been expressed in A. thaliana and rice ([@bib119], [@bib120]). Feeding of chickens with 5 g of transgenic rice grains on four occasions induced production of IgG and NA and after challenge with IBDV afforded better protection against formation of bursal lesions than a live-attenuated commercial vaccine administered intranasally. Recently, a different strategy has seen the presentation of a VP2 antigen on bamboo mosaic virus particles, and chimeric VLPs were shown to induce IBDV-specific serum IgG after i.m. injection in chicken ([@bib10]). These results indicate that a plant-based mucosal vaccine against IBDV may provide a good alternative to commercial vaccines. Chlamydophila psittaci is the causal agent of avian chlamydiosis, a zoonotic disease that causes respiratory symptoms in birds as well as humans. Transgenic rice seed has been used for the expression of C. psittaci MOMP as well as fusions of MOMP to LTB ([@bib137]). Though expression levels were very low (approaching 0.01% TSP), MOMP-specific humoral and mucosal immune responses were detected after oral application of plant-produced MOMP in mice ([@bib136], [@bib138]). In addition to serum IgG and fecal IgA, the induction IFN-γ and other cytokines was reported. Despite these positive results, only approximately 50% of animals were protected against challenge with aerosol-contained C. psittaci. Chicken coccidiosis is caused by protozoans of the Eimeria genus that can cause various species-dependent symptoms associated with damaged intestinal mucosa. The widespread use of anticoccidial drugs has led to the development of resistance; therefore safe, effective, and low-cost vaccines are needed to control the disease. Recently, Sathish et al. have reported the plant-based transient expression of Eimeria tenella microneme-1 and microneme-2 by vacuum infiltration of N. tabacum ([@bib91], [@bib92]). Chickens were immunized by i.m. injection and challenged by oral inoculation with sporulated oocycsts. Researchers found induction of serum IgG, a 60% reduction in oocyst shedding, and significantly higher weight gain compared to unvaccinated challenged animals. It will be interesting to see whether administration of these vaccine candidates via mucosal routes further enhances the protective efficacy. Recent progress has also been made in the development of plant-made natural and recombinant adjuvants and immune stimulators for use in the poultry sector. Natural ginseng stem-and-leaf saponins, a traditional Chinese herbal medicine, have been found to enhance the immunogenicity of a commercial mucosal NDV vaccine when administered orally with water for 7 days before vaccination ([@bib134]). Another group has shown the safety and immunogenicity of a plant-produced mutant E. coli heat labile holotoxin (LTA-K63/LTB) for use as an adjuvant in the poultry industry ([@bib75]). The combination of these adjuvants and stimulators with the plant-made poultry vaccine candidates described above may help in the formulation of more efficacious vaccines in this sector. Aquaculture Vaccines {#s0080} -------------------- A relatively new area of application for plant-produced vaccines is in the aquaculture industry. Farming of fish and other aquatic organisms is gaining increasing importance as many seas become overfished. Intensive fish farming brings increased risks of disease that must be dealt with in an efficient and cost-effective manner. In-feed delivery of vaccines is one potential method of protecting fish stock without labor-intensive application. In 2006, a publication provided the first indication that oral immunization of fish with plant-produced antigens could work. Companjen et al. used transgenic potato plants to express two model fusion protein antigens: E. coli heat-labile toxin B (LTB) fused to either GFP (LTB-GFP) or epitopes of influenza HA and canine parvovirus VP2 (LTB-p) ([@bib14]). Oral administration by incorporation of LTB-GFP potato into food pellets induced an anti-GFP antibody response in carp. Anal intubation of carp with extract from LTB-p potato induced an antiparvovirus response after the first administration. These results provide a good indication that oral immunization of fish is possible using plant-produced antigen. However, further studies using antigens derived from relevant fish pathogens, as well as challenge experiments, will need to be performed. Recently, a group in Thailand reported the protection of shrimp from infection with white spot syndrome virus (WSSV) by i.m. coinjection of WSSV with a viral binding protein produced in A. thaliana suspension cultures ([@bib106]). Shrimp do not possess a developed immune system as such; therefore this cannot be classified as a vaccine. However, it does indicate the growing acceptance of plant-produced pharmaceuticals for use in aquaculture. Conclusions {#s0085} =========== The use of plants for production of recombinant proteins has substantially evolved over the past 25 years. The first plant-based vaccines were expressed in stably transgenic plants, with the idea to conveniently deliver "edible vaccines" by ingestion of the antigen-containing plant material. These systems provided a proof of concept that oral delivery of vaccines in crude plant material could stimulate antigen-specific serum and mucosal antibodies. Several clinical trials have shown the potential of plant-derived orally delivered vaccines in humans using raw potato tubers as a convenient model system. Difficulties with implementation of the potato tuber system led to the use of transgenic grains such as rice and corn (maize). "MucoRice" expressing CT-B emerged as a strong candidate for clinical trials, providing a stable and robust vehicle for oral antigen delivery. Moreover, transgenic grains have strong potential for use as edible vaccines for agricultural animals (swine, poultry), and several studies support the idea. Some challenging issues exist with stably transgenic plants, including relatively low expression levels and regulatory issues related to the use of human food crops as hosts. Thus, many recent studies used transient expression with plant viral vectors to achieve rapid, high-level expression in a nonfood plant, N. benthamiana, followed by purification of the antigenic proteins. Delivery of purified proteins by oral or intranasal routes can result in robust stimulation of both systemic and mucosal immune responses. A few studies examined the potential of novel adjuvant, muco-adhesive, or immune-modulating molecules for mucosal delivery with very promising results. In conclusion, the use of plants for production and mucosal delivery of vaccine antigens has grown tremendously in research laboratories around the world. The acceptance of plant-based technologies by the vaccine industry may yet be some years away, but is likely to occur when further safety and efficacy studies prove their worth.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-micromachines-07-00143} =============== The standby power consumption of conventional complementary metal-oxide semiconductor (CMOS) circuits increases to the dynamic ON state level as they are scaled down to the scale of a few tens of nanometers \[[@B1-micromachines-07-00143]\]. On the other hand, nanoelectromechanical (NEM) switches are being investigated because of their promise for future low-power-consumption applications \[[@B2-micromachines-07-00143],[@B3-micromachines-07-00143]\]. The switching operation of such devices is mainly based on electrostatic actuation, which leads to a very low leakage current and high ON/OFF ratios. They are also expected to achieve abrupt switching with subthreshold swing values less than 60 mV/dec \[[@B4-micromachines-07-00143]\]. Furthermore, NEM switches are expected to be robust against external disturbances such as radiation and temperature fluctuations, which makes them ideal for inhospitable environments \[[@B5-micromachines-07-00143]\]. On the other hand, the inherently low ON current in comparison with that of CMOS devices and the high pull-in voltage of conventional NEM switches restrict their use as a viable alternative for CMOS circuits and low-power applications \[[@B6-micromachines-07-00143],[@B7-micromachines-07-00143]\]. Graphene is a two-dimensional material with excellent mechanical stability and electrical conductivity, and a high Young's modulus of \~1 TPa \[[@B8-micromachines-07-00143]\]. These outstanding properties of graphene make it a very promising material for high-performance NEM contact switches. Using graphene as a material for such devices can address some of the problems of conventional NEM switches by providing high reliability and a low pull-in voltage \[[@B9-micromachines-07-00143],[@B10-micromachines-07-00143],[@B11-micromachines-07-00143]\]. In this work, we present a three dimensional (3D) finite element method (FEM) simulation of double-clamped nanocrystalline graphene beam NEM switches. We focus on designing the graphene NEM switch in line with the experimental work \[[@B12-micromachines-07-00143]\]. The pull-in and pull-out characteristics from the simulation results are consistent with the experimental results. These models were used to study the double-clamped beam scaling on switching characteristics. After validating the simulation results for the pull-in voltage of the double-clamped graphene beam NEM switch with the experiment results, we studied the von Mises stress to evaluate the reliability of the switch. These results indicated that a longer and thinner graphene beam is more reliable. In addition, we also analyzed the effect of the applied electric field on the double-clamped graphene beam NEM switch. 2. Description of the Device Geometry {#sec2-micromachines-07-00143} ===================================== In this section, we briefly describe the geometry and operation principles of NEM switches. In our earlier work \[[@B12-micromachines-07-00143]\], we experimentally studied the switching operation of nanocrystalline graphene (NCG) beam NEM switches. The NCG was synthesized by direct deposition of NCG on an Si/SiO~2~ substrate using plasma-enhanced chemical vapor deposition (PECVD) \[[@B13-micromachines-07-00143]\]. The NCG deposited by PECVD contains both sp^2^- and sp^3^-hybridized carbon atoms. The deposited NCG film is polycrystalline in nature \[[@B12-micromachines-07-00143]\]. The polycrystalline nanographene has randomly distributed grain orientation and size \[[@B14-micromachines-07-00143]\]. The mechanical behavior of NCG depends on both the grain misorientation and the grain boundary rotation \[[@B15-micromachines-07-00143]\]. Moreover, the mechanical strength of NCG depends on the arrangement of the defects in the NCG film \[[@B16-micromachines-07-00143]\]. The NCG sheets have almost constant fracture stress and strain, and the fracture strength is independent of the grain size \[[@B17-micromachines-07-00143]\]. The polycrystalline NCG sheets have a flaw-insensitive fracture mechanism \[[@B18-micromachines-07-00143]\]. Usually, fractures in NCG film originate from the grain boundary and propagate to the rest of the polycrystal \[[@B19-micromachines-07-00143]\]. The propagation of the fracture also depends on the grain orientation. Moreover, microcracks in the surfaces are initiated from topological defects in the NCG polycrystal, which coalesce to form a big crack, eventually leading to the breakdown of the NCG film. As NCG is a polycrystalline material, we considered it as an isotropic material in the FEM simulation of the nanocrystalline NEM switches. The device structure and dimensions are adopted from our previous experimental work \[[@B12-micromachines-07-00143]\]. We used a Young's modulus of 860 GPa for graphene in all the FEM simulations as reported by this experimental work. This value is comparable to the reported Young's modulus of 500 GPa for layered suspended graphene sheets of between 2 and 8 nm thickness \[[@B20-micromachines-07-00143]\]. For the FEM simulation of the experimental device, we considered a graphene beam of length L, width W, and thickness t, and a top metal electrode. The schematic representation of the device is shown in [Figure 1](#micromachines-07-00143-f001){ref-type="fig"}. The double-clamped graphene beam NEM switch dimensions are detailed in [Table 1](#micromachines-07-00143-t001){ref-type="table"}. For each double-clamped graphene beam NEM switch, the air gap thickness changes owing to the natural buckling of the suspended graphene beam. The initial air gap of the device is g~0~. As the voltage applied between the suspended graphene beam and the top electrode is increased, the resulting electrostatic force balances with the elastic restoring force of the deformed graphene. When the applied voltage reaches a critical point, the electrostatic force overwhelms the restoring force, which causes the graphene beam to be pulled towards the top electrode/fixed element, resulting in closing the switch and thereby leading to a sharp rise in the current flow through the device \[[@B21-micromachines-07-00143]\]. When the applied bias voltage is sufficiently reduced, the elastic restoring forces in the deformed active element pull the switch open, and thus the device operates as a volatile switch. A mechanical hysteresis is formed between the pull-in voltage and the pull-out voltage. 3. Finite Element Method Simulation Results and Discussion {#sec3-micromachines-07-00143} ========================================================== 3.1. Graphene NEM Switch Pull-In and Pull-Out Characteristics {#sec3dot1-micromachines-07-00143} ------------------------------------------------------------- The static electrical and mechanical characteristics of the NEM switch were simulated using the FEM-based CAD tool IntelliSuite (8.8.5.1, IntelliSense, Lynnfield, MA, USA) \[[@B22-micromachines-07-00143]\]. [Figure 1](#micromachines-07-00143-f001){ref-type="fig"} shows a schematic of our graphene beam NEM switch with a graphene beam connected to electrodes at each end. This graphene beam NEM switch features a metal top gate (actuation electrode), which enables the graphene beam to be pulled onto the gate when a voltage is applied, and then pulled away, disconnecting from the channel when voltage is no longer applied. In order to be consistent with our experimental device structure, we used the device dimensions mentioned in [Table 1](#micromachines-07-00143-t001){ref-type="table"}. [Figure 2](#micromachines-07-00143-f002){ref-type="fig"}a shows the initial geometry of NEM switch A used in the FEM simulation. To analyze the pull-in and pull-out characteristics of the NEM switch, the voltage applied between the top electrode and the graphene beam was first increased until pull-in was confirmed and then decreased back to 0 V. [Figure 2](#micromachines-07-00143-f002){ref-type="fig"}b illustrates the geometry of the pull-in state of the double-clamped graphene beam NEM switch. The color bar shows the displacement of the graphene beam with respect to the initial position. [Figure 3](#micromachines-07-00143-f003){ref-type="fig"}a shows the pull-in/pull-out characteristics obtained for the different graphene beam NEM switches mentioned in [Table 1](#micromachines-07-00143-t001){ref-type="table"}. The pull-in voltages for the graphene beam NEM switches A, B, and C are approximately 8.6, 13.2, and 20.8 V, respectively. The obtained pull-in voltage shown in [Figure 3](#micromachines-07-00143-f003){ref-type="fig"}a is in good agreement with the experimentally reported pull-in voltages \[[@B13-micromachines-07-00143]\]. In order to clarify the impact of the thickness of the graphene beam, we conducted the FEM simulation for one of our experimental device structures (NEM switch A) with different graphene thicknesses of t = 3, 5, and 9 nm. [Figure 3](#micromachines-07-00143-f003){ref-type="fig"}b shows the pull-in and pull-out characteristics for the graphene beam with different thicknesses for the voltage applied between the top gate and the graphene beam. When the thickness is reduced, the pull-in voltages are evaluated to be 8.6 V, 4.3 V, and 2.1 V for t = 9 nm, 5 nm, and 3 nm, respectively. This result shows a clear dependence of the pull-in voltage on the scaling of the thickness of the suspended graphene beam. Thus, we confirm that the introduction of thickness scaling leads to a reduction of the pull-in voltage. 3.2. Von Mises Stress Analysis {#sec3dot2-micromachines-07-00143} ------------------------------ The von Mises yield criterion is a general way to estimate the yield of any ductile material, such as metals \[[@B23-micromachines-07-00143],[@B24-micromachines-07-00143]\]. The mechanical reliability of the graphene beam NEM switch can potentially be improved by properly choosing the switch dimensions. To quantitatively demonstrate the mechanical reliability of the double-clamped graphene beam NEM switch, we compared the maximum von Mises stress exerted along the length of the graphene beam \[[@B25-micromachines-07-00143],[@B26-micromachines-07-00143]\]. The von Mises stress profile analysis is essential to comprehend the spatial variation of the stress generated on the suspended graphene owing to the applied voltage. A Cartesian coordinate system is used to represent the numerical coordinates on the suspended graphene beam. The stress profile was obtained after the pull-in state was achieved, giving the three-dimensional stress profile for the deformed graphene beam. However, the stress variation along the thickness is constant. [Figure 4](#micromachines-07-00143-f004){ref-type="fig"} shows the von Mises stress for the different graphene beam NEM switches. The von Mises stress reaches the maximum value towards the ends of the graphene beam. When the length of the graphene beam is reduced, the von Mises stress is increased to the maximum value. As evident from [Figure 4](#micromachines-07-00143-f004){ref-type="fig"}a, the device with the shortest graphene beam length has the maximum probability of failure. When the thickness of the graphene beam is scaled for the fixed length of the beam, the von Mises stress is reduced as the thickness is reduced. The results suggest that NEM switch A is at least as reliable as NEM switches B and C. NEM switch C has a maximum von Mises stress of 6.2 GPa. When the length of the graphene beam is increased to 1 μm and 1.5 μm, it leads to a decrease in the von Mises stress of 3.8 GPa and 2.3 GPa, respectively. Furthermore, the maximum stress for NEM switch A is 2.5 GPa; when the thickness of the graphene is reduced to 5 nm and 3 nm, the stress is decreased to 1.4 GPa and 0.9 GPa, respectively. [Figure 5](#micromachines-07-00143-f005){ref-type="fig"} illustrates the top view of the von Mises stress contour plot of the graphene beam. It is evident from the contour plot that the von Mises stress is highest nearer to both fixed ends of the beam. The von Mises stress reaches the minimum value between the fixed end of the beam and the center of the beam in the pull-in state. If we examine the von Mises stress across the beam carefully, then we can observe higher stresses at the edges of the graphene beam compared to those at the center of the beam. In order to clarify this point, 3D electric field distributions in the NEM switch were carried out in the pull-in state. 3.3. Three-Dimensional Electric Field Distribution and Its Role in Graphene Beam NEM Switch Operation {#sec3dot3-micromachines-07-00143} ----------------------------------------------------------------------------------------------------- To analyze the impact of the applied electric field on the double-clamped graphene beam NEM switch, we made the same model in COMSOL Multiphysics (5.1, COMSOL Inc., Burlington, MA, USA) \[[@B27-micromachines-07-00143]\]. The NEM switch was built inside a vacuum environment. The model was meshed with triangular mesh elements to reduce the computational complexity. The density of the mesh was varied adaptively in order to study the structural displacement of the graphene beam. In this simulation, the actuation electrode was kept at the bottom and the graphene beam was placed at the top. For the electric field analysis at different voltages, a constant bias of 0 V was applied to the bottom electrode (Au) and the voltage applied at the top electrode (graphene beam) was swept. The potential, V, and the electric field, E, in the free space can be obtained by solving Poisson's equation \[[@B28-micromachines-07-00143]\]. [Figure 6](#micromachines-07-00143-f006){ref-type="fig"}a shows the cross-sectional view of the electric field distribution across the center of the NEM switch for the applied voltage of 1 V to the bottom electrode. The dimensions of the graphene beam are equivalent to those of NEM switch A. Arrows in this plot show the electric field lines directions. At the center of the beam, the electric field lines are distributed vertically. The orientation of the electric field distribution is gradually changed to the horizontal direction towards the edges of the beam. At both edges of the beam, the electric field is distributed more horizontally in the outward direction from the center of the beam. [Figure 6](#micromachines-07-00143-f006){ref-type="fig"}b illustrates the one dimensional (1D) electric field strength in the Z direction at 5 nm above the bottom electrode for the different applied voltages. Consistent with [Figure 6](#micromachines-07-00143-f006){ref-type="fig"}a, the electric field strength is highly concentrated at the edges of the graphene beam. These results demonstrate that the downward component of the electrostatic force acting on the edges of the graphene beam is higher than that at the center of the beam. In order to analyze this edge field termination effect, mechanical deflection analysis of NEM switch A with a 3 nm graphene thickness was done. [Figure 7](#micromachines-07-00143-f007){ref-type="fig"} shows the displacement of the graphene beam nearer to the pull-in state. If we consider the edge of the graphene beam, then the downward bend of the beam edges is apparent. This is also consistent with a higher von Mises stress at the beam edges. 4. Conclusions {#sec4-micromachines-07-00143} ============== In this paper, we have studied the electro-mechanical switching and mechanical reliability of graphene beam NEM switches by FEM simulations. To evaluate the mechanical reliability of a graphene beam NEM switch, we scaled the length and thickness of the graphene beam and studied the von Mises stress for each structure. This analysis showed that the graphene beam NEM switch with a longer length of 1.5 μm and a thickness of 3 nm has a pull-in voltage of 2 V. The electrostatic force concentration at the edges of the graphene beam leads to more mechanical deflection at the edges than at the center of the beam. This work was supported by the Grant-in-Aid for Scientific Research No. 25220904 from the Japan Society for the Promotion of Science. Manoharan Muruganathan conceived the concept; Jothiramalingam Kulothungan performed the simulations; Jothiramalingam Kulothungan, Manoharan Muruganathan, and Hiroshi Mizuta analyzed the data and wrote the paper; Hiroshi Mizuta supervised the project. All authors equally contributed to the results analysis and preparation of the manuscript. The authors declare no conflict of interest. ![Schematic of double-clamped graphene beam NEM switch with top metal actuation electrode.](micromachines-07-00143-g001){#micromachines-07-00143-f001} ![The geometry of NEM switch A. (a) Initial structure of double-clamped graphene beam NEM switch with a top metal electrode; (b) Pull-in state of the graphene beam; color bar indicates the relative displacement with respect to the initial condition.](micromachines-07-00143-g002){#micromachines-07-00143-f002} ![Pull-in and pull-out switching characteristics of the double-clamped graphene beam NEM switches. (a) Switching characteristics of graphene beam NEM switches A, B, and C; (b) Switching characteristics of graphene beam NEM switch A, with different graphene thicknesses of t = 3, 5, and 9 nm.](micromachines-07-00143-g003){#micromachines-07-00143-f003} ![The von Mises stress of double-clamped graphene beam NEM switches. (a) von Mises stress of graphene NEM switches A, B, and C; (b) von Mises stress of graphene beam NEM switch A, with different graphene thicknesses of t = 3, 5, and 9 nm.](micromachines-07-00143-g004){#micromachines-07-00143-f004} ![Contour plot (top view of graphene beam) of von Mises stress for the NEM switch A, with different graphene beam thicknesses of t = 3, 5, and 9 nm.](micromachines-07-00143-g005){#micromachines-07-00143-f005} ![Two-dimensional electric field distribution across the center of the NEM switch. Dimensions of the graphene beam (top electrode) are equivalent to those of NEM switch A. (a) The electric field distribution across the switch at the center of the beam. Arrows indicate the electric field direction; (b) The electric field strength in the Z direction at 5 nm above the bottom electrode at different voltages applied between the bottom and top electrodes. Inset shows zoomed-in version of electric field distribution for lower actuation voltages as indicated by the dashed box; the scale is the same as (b).](micromachines-07-00143-g006){#micromachines-07-00143-f006} ![Effect of the beam edge electric field termination at 8.3 V. (a) The displacement of the beam at 8.3 V shown as side view; (b) The displacement of the beam from initial position at different applied voltages. Birds-eye cross-sectional view of the graphene beam at (c) 0.1 V; (d) 1 V; and (e) 8.3 V.](micromachines-07-00143-g007){#micromachines-07-00143-f007} micromachines-07-00143-t001_Table 1 ###### NEM switch dimensions. NEM Switch Dimension Air Gap Thickness (nm) ------------ ----------- ------------------------ --- ---- A 1.5 0.5 9 75 B 1 0.5 9 50 C 0.8 0.5 9 45	{ "pile_set_name": "PubMed Central" }
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus. Introduction ============ Emphysematous pyelitis (EP) is a subclass of a necrotizing infection of the urinary system called emphysematous pyelonephritis (EPN). EPN is a life-threatening medical emergency occurring mostly in women with diabetes mellitus (DM). We report a case of an 81-year-old nondiabetic man who developed EP after urinary tract manipulation. Case presentation ================= An 81-year-old man presented to our hospital for infected lower extremity ulcers and worsening kidney function. His past medical history was significant for benign prostatic hyperplasia, chronic kidney disease stage IIIb, hypertension, and peripheral artery disease. On physical examination, he had a temperature of 98.7F, heart rate of 85 beats per minute, respiratory rate of 16 breaths per minute, blood pressure of 138/72 mm Hg, and oxygen saturation of 98% when breathing ambient air. The patient had a 1+ pitting edema in both lower extremities and infected ulcers on the right hallux and the right calf. The rest of his physical examination was normal. The patient\'s serum creatinine (sCr) was 3.12 milligrams per deciliter (mg/dL), which was elevated from his outpatient baseline of 1.69 mg/dL, and urea nitrogen was 32 mg/dL. His urinalysis showed small blood, large leukocytes, and 30 mg/dL of protein on urine dipstick, 20 to 30 white blood cells per high power field (p/hpf), three to six red blood cells p/hpf, few hyaline casts p/hpf, and occasional mucous threads on microscopic examination. His serum total protein, albumin, complement, low-density lipoprotein, and high-density lipoprotein levels were normal. Antiproteinase 3 antibodies, anti-myeloperoxidase antibodies, and antinuclear antibodies were negative. A renal ultrasound showed bilateral hydroureteronephrosis and a postvoid residual urine volume of 350 milliliters. An indwelling urinary catheter was placed, and intravenous fluids were administered. The patient's sCr level improved 24 hours after introducing the urinary catheter. A repeat ultrasound was done 48 hours after the urinary catheter placement, which showed bilateral renal collecting shadowing echogenic foci, consistent with possible EP. A computed tomography (CT) scan of the abdomen and pelvis without contrast showed mild bilateral hydronephrosis with air seen throughout the collecting systems and ureters (Figure [1](#FIG1){ref-type="fig"}). The urine gram stain and culture were negative and the patient's antibiotic regimen was changed from ampicillin-sulbactam to vancomycin and ertapenem empirically due to a recent hospitalization and antibiotics use. ![Coronal CT abdomen and pelvis\ A coronal view illustrating air in the bilateral renal pelvises with extension into the ureters (arrows) without parenchymal renal involvement. The fat-stranding of the bilateral kidneys is illustrated (asterisks). There is also air at the left ureterovesicular junction (arrowhead).\ CT = computed tomography](cureus-0009-00000001612-i01){#FIG1} Throughout his hospitalization, the patient denied abdominal pain, flank pain, suprapubic pain, dysuria, or urinary urgency, and he remained afebrile, with stable vital signs. His renal function improved to its pre-admission baseline, so a percutaneous catheter drainage (PCD) of the renal pelvises was not performed. The patient was sent home with the urinary catheter in place on an antibiotics regimen of vancomycin and ertapenem for a total of 14 days. He was instructed to have a close follow-up with his primary care physician, urologist, and nephrologist. Follow-up details were obtained from the patient's primary care physician, who informed us that a subsequent CT of the patient's abdomen and pelvis demonstrated complete resolution of his EP. Discussion ========== EPN is a rare, but life-threatening, acute necrotizing infection of the kidney, first clinically described by Kelly and MacCallum in 1898. EPN is defined as gas within the renal parenchyma, collecting system, or perinephric tissues \[[@REF1]\]. EP is a subclass of EPN, in which gas is confined to the renal collecting system. EPN is a disease more commonly seen in women than in men, with a female-to-male ratio of 3 to 1, presumably because of women\'s increased susceptibility to urinary tract infections \[[@REF1]\]. Over 90% of patients with EPN have diabetes mellitus (DM) \[[@REF2]\], whereas EP is usually associated with urinary tract obstruction \[[@REF3]\]. Only 50% of patients with EP have DM, compared to the \> 90% of patients with EPN class II or higher. The most common cause of urinary tract obstruction that leads to EP is renal calculi \[[@REF4]\]. Urinary tract obstruction causes EPN by impairing the transportation of formed gas, which increases local pressure in the tissue and impairs circulation. Impaired circulation leads to infarction, which provides a culture medium for gas-forming pathogens, thereby creating a vicious cycle \[[@REF3]\]. Gas appearing in the urinary system may also be caused by fistulae originating in the gastrointestinal tract, gas reflux from the urinary bladder, trauma, and urinary system interventional procedures \[[@REF4]\]. EPN is caused by organisms that ferment glucose and lactate into carbon dioxide and hydrogen gas. The organisms involved are usually enteric facultative anaerobic bacteria, such as Escherichia coli, Proteus mirabilis, Klebsiella pneumoniae, Enterococcus species, and Pseudomonas aeruginosa; occasionally, gram positive bacteria, such asStreptococci species; and fungi such as Candida albicans\[[@REF1]\]. Patients with DM provide a favorable environment for gas-forming microbes because of the high levels of glucose in their tissues and urine, which is used as a substrate for fermentation. The common clinical features of EPN include fever, chills, costovertebral angle tenderness, vomiting, dysuria, and, rarely, crepitus in the lumbar region \[[@REF1]\]. The risk factors associated with high mortality rates from EPN are thrombocytopenia, altered mental status, severe proteinuria, shock, extension of the infection to the perinephric space, renal replacement therapy dependence, hypoalbuminemia, and polymicrobial infections \[[@REF1],[@REF5]-[@REF6]\]. The diagnosis of EPN is established by demonstrating gas in the urinary system, renal parenchyma, or perinephric tissue by a plain abdominal radiograph or a renal ultrasound. A CT scan of the abdomen and pelvis can confirm the diagnosis and show the extent of the disease, thus being the ideal test for the diagnosis of EPN \[[@REF5]\]. A poor response to antibiotic therapy in a patient with DM, with a seemingly uncomplicated pyelonephritis, should raise suspicion of EPN and warrant a CT scan of the abdomen and pelvis to rule out this life-threatening condition \[[@REF1]\]. EPN is divided into four classes \[[@REF5]\], which are determined according to the extent of the gas expansion (Table [1](#TAB1){ref-type="table"}). The gas may be limited to the collecting system or expand through the renal parenchyma and retroperitoneal space. Miller et al. described a severe case of EPN, in which gas extended through the retroperitoneal space and invaded the inferior vena cava \[[@REF2]\]. The treatment of EPN remains controversial. Early nephrectomy was the treatment of choice in the 1980s, with a general consensus that medical therapy without surgical intervention has higher rates of mortality \[[@REF7]\]. Studies in the 1990s proposed antibiotic therapy combined with CT-guided percutaneous drainage (PCD) as an acceptable alternative to nephrectomy \[[@REF8]\], which lowered mortality rates by as much as 21% \[[@REF6]\]. Further studies in the last two decades continue to demonstrate a significantly lower mortality rate with PCD alongside medical management compared to early nephrectomy in most classes of EPN \[[@REF6],[@REF8]\]. Empiric antimicrobials should primarily target gram-negative bacteria. Third-generation cephalosporins are recommended as the initial treatment of EPN. Carbapenems, frequently used in combination with vancomycin, are the empiric antibiotic of choice for patients with prior hospitalization and antibiotics use within the last year, patients requiring hemodialysis, and patients who develop disseminated intravascular coagulation. PCD and the placement of double-J catheters are often performed in conjunction with medical treatment to maximize nephron sparing, which significantly lowers the rate of mortality. Salvage nephrectomy or open drainage is usually performed when PCD or conservative treatments fail or have a high probability of failing \[[@REF5],[@REF9]\]. A 25% mortality rate has been observed in patients undergoing an emergency nephrectomy, 50% with medical management, and only 13% with medical management and PCD \[[@REF10]\]. Huang and Tseng proposed common guidelines for the treatment of EPN (Table [1](#TAB1){ref-type="table"}). Most classes of EPN require antibiotics and PCD. Nephrectomy is currently reserved patients with class III EPN with multiple risk factors and class IV if antibiotics and PCD fail. ###### EPN classification and management ^\^For all classes: relief of any existing urinary tract obstruction ^\\^Emphysematous pyelitis \\\That is, thrombocytopenia, acute kidney injury, altered mental status, shock PCD = Percutaneous drainage --------- ------------------------------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------- Class Gas Location Management^\^ I^\\^ collecting system antibiotics and PCD II renal parenchyma antibiotics and PCD IIIA extension to the space between the fibrous renal capsule and the renal fascia (perinephric space) \< 2 risk factors^\\\^, PCD and antibiotics; 2 or more risk factors, early nephrectomy IIIB extension to the space beyond the renal fascia and/or extension to adjacent tissues (pararenal space) \< 2 risk factors^\\\*^, PCD and antibiotics; 2 or more risk factors, early nephrectomy IV bilateral kidneys or a solitary functioning kidney trial of antibiotics and PCD; nephrectomy if antibiotics and PCD fail --------- ------------------------------------------------------------------------------------------------------- --------------------------------------------------------------------------------------------- Conclusions =========== Our patient had a negative urine culture, however, his concomitant antibiotic coverage for a coexisting infection may have altered the validity of the urine cultures. Nevertheless, the differential diagnosis of air in the collecting ducts of the urinary system is not confined to infectious causes and the placement of a urinary catheter in the setting of a postrenal obstruction may have caused an iatrogenic EP. Individual analyses of the risks and benefits of choosing surgical intervention in addition to medical management should be evaluated on a case-by-case basis. The authors have declared that no competing interests exist. Consent was obtained by all participants in this study	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== Continuous renal replacement therapy (CRRT) was developed originally as an alternative for hemodynamically unstable acute renal failure (ARF) patients who could not tolerate conventional hemodialysis \[[@CR1], [@CR2]\]. The early application of CRRT largely involved technology adapted from the maintenance dialysis setting and almost exclusively occurred as a salvage therapy, typically in hypercatabolic patients with severe, diuretic-resistant fluid overload. As CRRT technology evolved over the past four decades to produce devices specifically designed for the critically ill population, utilization of the therapy as a first-line treatment for acute kidney injury (AKI) and the patient populations treated have expanded substantially \[[@CR3]\]. Moreover, despite a lack of data definitively showing outcome benefits for CRRT, consensus statements now suggest its use, rather than conventional hemodialysis, for hemodynamically unstable AKI patients \[[@CR4]\]. While CRRT is now a mainstay therapy in the vast majority of large ICUs around the world, this modality can be challenging to implement at some institutions and significant opportunity for improvement exists \[[@CR5]\]. Clinicians' uncertainty about numerous aspects of CRRT, including therapy dosing, timing of initiation and termination, fluid management, anticoagulation, drug dosing, and data analytics, may lead to inconsistent delivery of the therapy and even reluctance to prescribe it. In this review, we perform a critical assessment of CRRT, discussing current limitations and potential solutions over the next decade from both a patient management and a technology perspective. In addition, we address the issue of sustainability for CRRT and related therapies beyond 2027 and raise several points for consideration. Renal support in 2027: addressing CRRT's current limitations {#Sec2} ============================================================ Adoption of precision CRRT {#Sec3} -------------------------- In the most recent Acute Dialysis Quality Initiative (ADQI) consensus conference, the participants rightly identified the need for CRRT patient management to align with the current focus on personalized medicine. In this regard, the ADQI participants proposed the term "precision CRRT" in calling for technology to be applied on an individualized basis \[[@CR6]\] rather than a "one size fits all" approach that is all too common in current practice. An important component of this individualized approach has been termed "dynamic CRRT", in which the treatment is adapted to the constantly changing clinical status of the critically ill AKI patient \[[@CR7]\]. In addition to ongoing clinical assessment of the patient, important technical components of dynamic CRRT include solute clearance, delivered/prescribed dose, effective treatment time, solute control indicators, circuit/filter pressure trends, fluid and hemodynamic management, and anticoagulation. Also assuming important roles in the implementation of precision CRRT are quality metrics \[[@CR8]\], biofeedback \[[@CR9]\], and data analytics \[[@CR10]\], all of which are discussed further in the following. Dosing of CRRT {#Sec4} -------------- Based on the landmark study performed by Ronco et al. \[[@CR11]\] and several other prospective trials \[[@CR12]--[@CR16]\], the use of effluent-based dosing to guide CRRT prescription and delivery is established firmly in clinical practice. Nevertheless, the effluent dose (expressed as ml/kg/hr) does not provide an accurate estimation of actual solute clearance and considerable confusion exists among clinicians, especially those familiar with urea-based dose measurements in the maintenance dialysis setting. Indeed, substantial differences between the effluent dose and actual solute clearance may exist under many CRRT operating conditions \[[@CR17]\]. Therefore, we recently reappraised dose prescription and delivery for CRRT \[[@CR18]\], and proposed an adaptation of a chronic dialysis parameter (standard Kt/V) \[[@CR19]\] as a benchmark to supplement effluent-based dosing. Our proposal allows for the target standard Kt/V to vary on a patient-to-patient basis according to clinical circumstances and can be modified in an individual patient, depending on the clinical course (e.g., a hypercatabolic, septic patient in need of higher dose to control azotemia). These dosage adjustments are entirely consistent with the concept of dynamic CRRT. The clinical relevance of urea as a toxin per se is very much an open question, especially in light of large prospective studies performed in patients with end-stage renal disease (ESRD), and many experts give credence to the potential importance of other uremic toxin classes \[[@CR20], [@CR21]\]. However, identification of a larger molecular weight toxin which is easily measured in clinical practice and has well understood kinetic properties for different renal replacement therapies has been elusive \[[@CR22]\]. Because the current reality is that urea is the only surrogate molecule whose kinetics during renal replacement therapy are well understood, we believe our proposal for application of standard Kt/V to CRRT is rational. For the future, we believe the incorporation of effluent urea nitrogen measurements, first through clinical protocols \[[@CR23]\] and then CRRT machines equipped with online sensors \[[@CR24]\], will occur in clinical practice. In addition, machines will provide clinicians with automated alerts when therapy trends suggest filter clotting, based on changes in effluent measurements or circuit pressures. Moreover, we predict that additional molecules, having specific relevance for AKI pathophysiology \[[@CR25], [@CR26]\], will be validated by 2027 as CRRT dose surrogates for patients with AKI and other disorders. Timing of CRRT initiation {#Sec5} ------------------------- Recent data have cast doubt on the use of conventional ESRD-based criteria for RRT initiation in AKI patients \[[@CR27]\]. Nevertheless, decisions about initiation of RRT for AKI continue to be difficult due to the lack of a clinically relevant parameter that has been validated in prospective trials. Moreover, recent prospective trials employing different initiation criteria have provided conflicting results \[[@CR28], [@CR29]\]. In a recently completed pilot trial from Canada, Wald et al. \[[@CR30]\] demonstrated the validity of changes in urine output, whole-blood NGAL concentration, and serum creatinine as initiation criteria---a full-scale RCT is now being performed. We believe the concept of demand/capacity imbalance, proposed recently by Mehta and colleagues \[[@CR31], [@CR32]\], will be validated in clinical trials as a useful parameter guiding decisions about CRRT initiation and incorporated into clinical practice by 2027. The components of renal demand include AKI disease burden, solute load, and fluid load. A significant imbalance between this demand and diminished renal function in AKI patients should prompt serious consideration of RRT initiation. The concept of demand/capacity balance will also be useful to guide decisions about renal recovery and RRT cessation. The ADQI group has recommended explicitly that RRT should be discontinued if kidney function has recovered sufficiently to reduce the demand--capacity imbalance (current and expected) to acceptable levels \[[@CR32]\]. We believe the preliminary work defining the important considerations with respect to renal recovery \[[@CR33], [@CR34]\] will be refined over the next decade, allowing clinicians to make more informed decisions about RRT termination. While awaiting the results from clinical trials, we also believe further progress will be made by 2027 in the clinical application of biomarkers \[[@CR35], [@CR36]\], not only for the initial diagnosis of AKI but also for decisions about CRRT initiation and termination. In addition, progress will be made in validating real-time GFR measurements for both of these applications \[[@CR37]\]. We predict that these technologies will be used routinely in conjunction with established clinical criteria, especially the extent of fluid overload (see later), to guide CRRT initiation. Likewise, these technologies will be useful in decisions for CRRT discontinuation or transition to another modality. Management of fluid overload {#Sec6} ---------------------------- Severe fluid overload continues to be a common trigger for CRRT initiation, especially in septic shock patients who have received aggressive volume resuscitation in the face of worsening renal function \[[@CR38]\]. A quantitative parameter of fluid accumulation, percent fluid overload (%FO), has been employed in many recent clinical trials \[[@CR39], [@CR40]\] and positive values have been associated with increased mortality, especially when \>10% at RRT initiation. In addition to septic AKI patients, postsurgical patients are also at high risk of developing severe fluid overload once AKI develops. Xu et al. \[[@CR41]\] found that a cumulative %FO ≥ 7.2% had a significant impact on 90-day outcome in critically ill AKI patients after cardiac surgery. Even while approaches designed to assess fluid overload and quantify fluid responsiveness are refined \[[@CR42]\], we predict that the occurrence of fluid overload as the primary trigger for CRRT initiation will increase over the next decade. The basis for this belief relates to the demographics of severe AKI, for which the primary cause will increasingly be severe sepsis and septic shock during this time. Thus, in conjunction with %FO (or similar measurement) and other clinical parameters, technologies providing real-time fluid assessment capabilities, including bioimpedance and ultrasound, will be used routinely by 2027 \[[@CR43]\]. Antibiotic dosing during CRRT {#Sec7} ----------------------------- The increased prevalence of sepsis-associated severe AKI over the next decade will result in the need for antibiotic therapy in an increasingly greater percentage of CRRT patients \[[@CR44]\]. The lack of reliable clinical data to guide antibiotic use and the associated risk of under-dosing during CRRT have been identified as major problems---dosing continues to be done largely on an empiric basis \[[@CR45]\] (Table [1](#Tab1){ref-type="table"}). Over the next decade, multiple clinical trials evaluating the antibiotics most commonly prescribed to CRRT patients will be performed. The typical range of CRRT flow parameters will be evaluated in these trials, along with commonly used filters, so that the contributions of diffusive, convective, and adsorptive clearance can be ascertained. These trials will provide relatively precise dosing recommendations for a series of widely used antibiotics, leading to the routine incorporation of this information into the CRRT prescription by clinicians.Table 1Proposed elements for CRRT pharmacokinetic assessment• Estimation of pharmacokinetic parameters, including variability• Comparison of pharmacokinetic parameters with those of typical patients with normal kidney function (literature or sponsor data) or the appropriate reference population• Quantification of the impact of changes in the prescribed [Q]{.smallcaps} ~[e]{.smallcaps}~ on the pharmacokinetic parameters of interest, and interpolation for flow rates not evaluated in this study• Assessment of whether dosage adjustment is warranted in CRRT recipients• If dosage adjustment is warranted, derivation of specific dosing recommendations for the studied conditionsReprinted with permission from \[[@CR45]\]CRRT continuous renal replacement therapy[Q]{.smallcaps} ~[e]{.smallcaps}~ - effluent rate (ml/hr) Anticoagulation {#Sec8} --------------- A series of recent prospective trials have demonstrated that regional citrate anticoagulation (RCA) significantly reduces the risk of hemorrhage for patients treated with CRRT (in comparison with heparin) \[[@CR46]\]. The majority of these recent studies have involved physiologic (as opposed to hypertonic) citrate solutions, allowing them to serve as both anticoagulant and replacement solutions. Moreover, some of these studies involved machines capable of semi-automated RCA delivery in which citrate infusion rates are modulated by device software, at least to some extent \[[@CR47]\]. We predict that CRRT machines will provide more fully automated RCA by 2027, as has been proposed for other acute RRT modalities \[[@CR48]\]. Finally, we foresee that the use of heparin as an anticoagulant for CRRT will be markedly reduced by 2027, due to its hemorrhagic risks in the CRRT population. In parallel with advances in RCA, we believe manufacturers will continue in the pursuit of developing antithrombogenic membranes that either minimize or obviate the requirement for anticoagulation during CRRT. Surface-modified versions of the AN69 membrane \[[@CR49]\] have been developed but clinical data demonstrating acceptable circuit lives during CRRT performed with no anticoagulation are currently lacking. Finally, another issue related to thrombogenicity during CRRT is from the perspective of catheter use. Catheter thrombosis is a very common treatment complication, resulting in decreased therapy delivery and contributing to significant morbidity and cost. Recent data suggest that use of a surface-modified catheter (in comparison with a standard unmodified polyurethane catheter) results in a longer catheter life and less dysfunction (as measured by blood flow rate) \[[@CR50]\]. We believe further progress in the development of surface-modified catheters will occur over the next decade, leading to less catheter-related dysfunction and higher blood flow capabilities. Quality metrics {#Sec9} --------------- One of the current factors potentially limiting outcome improvements and further dissemination of the therapy is the lack of standardization for CRRT. A specific limitation that contributes to this lack of standardization is an insufficient evidence base---the current confusion regarding the timing of CRRT initiation is a good example. As such, we predict that both randomized and pragmatic clinical trials performed over the next decade will address such critical issues and improve therapy standardization. Another major factor limiting therapy standardization is the lack of consensus CRRT quality metrics. Rewa et al. \[[@CR8]\] are currently evaluating which aspects of CRRT prescription and delivery should be targets for quality metric development, namely dose (including treatment downtime), anticoagulation, vascular access, and circuit-related issues (Table [2](#Tab2){ref-type="table"}). As is the case in maintenance dialysis, we predict that a number of these quality metrics will be established by consensus initiatives and be part of routine clinical practice in 2027. Another recent development which will support therapy standardization is the use of simulation-based CRRT training, which has demonstrated tangible improvements in the delivery of CRRT \[[@CR51]\].Table 2Proposed quality metrics for CRRTThemeMeasuresDose prescriptionHigh vs low doseDose deliveryPercentage of prescribed dose deliveredAnticoagulation selectionHeparin vs citrate vs noneAnticoagulation monitoringPTT monitoring, citrate monitoringAnticoagulation complicationsBleeding, hypocalcemia, incidence of HITTreatment interruptionNumber of interruptions and duration of interruptions; time to establish new circuitCatheter-related issuesInfections, bleeding, obstruction/thrombosisCircuit-related issuesHiter clotting, pressure alarmingReprinted with permission from \[[@CR8]\]CRRT continuous renal replacement therapy, PTT partial thromboplastin time, HIT heparin-induced thrombocytopenia CRRT data analytics and biofeedback {#Sec10} ----------------------------------- The technical limitations of current CRRT machines make efficient management of patient and treatment data difficult in some respects \[[@CR10]\]. As opposed to the automated, real-time data capture that characterizes many interventions in the ICU, CRRT machine data generally are collected and analyzed manually at present. This is a laborious, time-consuming process that frequently delays necessary treatment intervention and is a barrier to providing dynamic CRRT. A desired technical aspect of dynamic CRRT is the availability of real-time CRRT machine data as part of a biofeedback system. While any prescription changes needed to close a biofeedback loop have to be made manually by the clinical team at present, we predict that such changes will be made automatically by the CRRT machine in 2027 \[[@CR52]\] (Fig. [1](#Fig1){ref-type="fig"}). This will be accomplished by the incorporation of online tools for continuous, real-time measurement of dose delivery and fluid overload. Moreover, in addition to treatment data from the CRRT machine, patient-level data from the electronic medical record (EMR) will play a critical role in these biofeedback loops \[[@CR53]\].Fig. 1Various approaches for biofeedback in CRRT. Reprinted with permission from \[[@CR52]\]. CRRT continuous renal replacement therapy A dynamic CRRT program also implies the ability to use information technology beyond the real-time phase for longer-term purposes. At present, CRRT machine data are not routinely stored in an accessible warehouse, rendering impossible the systematic generation of reports for review by the clinical team. We predict that by 2027 clinicians will routinely be able to assess historical trends on a facility-level basis, especially those related to the basic quality metrics mentioned earlier, or to use these data for quality assurance purposes. Moreover, these data will facilitate design and implementation of pragmatic trials, including registries. Again, patient-level data from the EMR will supplement technical data. Extracorporeal multiorgan support {#Sec11} --------------------------------- While several extracorporeal approaches have been used as adjunctive therapies for organ failures beyond AKI \[[@CR49], [@CR54]--[@CR60]\], additional clinical outcome data are clearly needed. We believe prospective trials performed over the next decade will demonstrate outcome benefits for such approaches as adjunctive therapies (Fig. [2](#Fig2){ref-type="fig"}). We feel that extracorporeal systems designed to eliminate CO~2~ (as an adjunct to low tidal volume ventilation) \[[@CR54]\] and modulate inflammatory mediators (as an adjunct to sepsis therapy) have the greatest likelihood of showing these benefits. With regard to sepsis adjunctive therapies, we predict that mediator modulation will be achieved through both filter-based \[[@CR57], [@CR58]\] and hemoperfusion \[[@CR59], [@CR60]\] techniques.Fig. 2Components of extracorporeal multiorgan support Miniaturization of technology {#Sec12} ----------------------------- As is the case with data analytics and information management, CRRT is lagging behind many other therapies with respect to technology "down-sizing". The incorporation of microfluidics, micromechanics, and nanotechnology is driven by the desire not only to decrease the physical footprint of medical technologies (thus enhancing portability) but also to extend their applicability to greater numbers and subsets of patients \[[@CR61]\]. We predict that the enhanced portability of future devices will allow for a set of similar devices to be used over the entire RRT spectrum (ICU, ward, and even home) in a given patient. In turn, this will allow for more seamless transitions in care, leading to increased simplicity and possibly lower costs. One clear example of this trend within the renal replacement field is the development of wearable dialysis and ultrafiltration devices, on which several investigative groups have made significant progress during the past decade \[[@CR62], [@CR63]\]. While the initial application of these devices has largely been focused on ESRD patients, they may yet prove useful in the management of fluid overload, especially in the setting of heart failure. Another potential application in the future is their use for severe AKI survivors who need supplementation of kidney function in the recovery phase. The recent development of a CRRT device specifically designed to treat pediatric AKI patients is a more immediate-term example of technology miniaturization. While treatment of pediatric AKI with CRRT has grown substantially over the past decade \[[@CR64]\], pediatricians have been forced to use equipment designed primarily for adult patients. The design features of traditional CRRT equipment, especially with regard to fluid accuracy of the machine and extracorporeal blood volume of the circuit, have rendered pediatric treatments problematic---this is especially true for neonatal patients, who typically weigh \<3 kg. Based on the recent success reported by Ronco et al. in the management of neonatal AKI with the CARPEDIEM device \[[@CR65]\], we predict further growth in the utilization of CRRT for pediatric AKI will occur over the next decade. Furthermore, we believe the technology advances that made the CARPEDIEM device possible will accelerate the application of miniaturization principles to other aspects of both pediatric and adult CRRT. Nondialytic management of AKI {#Sec13} ----------------------------- The need for efficient processing of data in the management of critically ill AKI patients has been highlighted during a recent ADQI conference focused on the implications of "big data" for this population \[[@CR66]\]. While the ability to utilize data efficiently is a challenge during CRRT, limitations also exist upstream with regard to AKI diagnosis. Moreover, recent data also demonstrate that postdischarge management of patients who have had an AKI episode is fragmented and unpredictable \[[@CR67]\]. While early studies evaluating "sniffers" and alerts designed to facilitate the diagnosis of AKI have provided conflicting results \[[@CR68]\], we predict that their utility will be demonstrated conclusively in prospective trials and they will become part of standard clinical practice by 2027. Likewise, over the next decade, web-based algorithms will be developed to triage post-AKI follow-up to nephrologists or other medical specialties according to AKI severity, extent of CKD, and comorbidities. These algorithms will be used routinely in clinical practice to minimize the risk of progression to ESRD. Health economics {#Sec14} ---------------- The cost of medical care for both chronic diseases and acute conditions is increasingly being scrutinized by government agencies and payers, in both the developed and the developing world. Because the societal costs of treating relatively small numbers of patients with ESRD are disproportionately high, health economic analyses and cost-effectiveness studies are now performed frequently upon the introduction of new chronic dialysis therapies. While visibility for the costs of treating AKI seems to be low and few health economic assessments have been performed, it is a costly disorder \[[@CR69], [@CR70]\]. Recent health economic assessments have demonstrated that simplistic comparisons of only product costs for different AKI renal replacement modalities are not informative, because they capture only a very short-term time window. Valid comparisons instead have to incorporate the costs of long-term outcomes potentially related to modality choice, including ESRD \[[@CR71]\]. We predict that more rigorous health economic analyses will be performed in the AKI setting over the next decade and robust assessments will be required on a routine basis in 2027 for new technologies \[[@CR72]\]. Another important consideration is the general affordability of acute RRT in the future. In the developed world, acute RRT costs typically are not an important factor due to widespread health insurance coverage and relatively generous reimbursement policies. On the other hand, hospital reimbursement and patient self-payment policies vary considerably across the developing world \[[@CR73]\], resulting in the need for some patients and their families to make very difficult decisions about potentially life-saving medical technology. We believe an increased demand for CRRT in the developing world will be satisfied by a combination of expanded insurance coverage and lower overall cost of CRRT delivery in the future. Renal support in 2027: sustainability of dialysis techniques {#Sec15} ============================================================ A final consideration over the next decade is the need to begin developing acute dialysis techniques designed to provide sustainability beyond 2027. One of the major environmental footprints associated with most extracorporeal dialysis modalities is the generation of large masses of plastic disposables, including filters and tubing sets. The development of miniaturized technologies, including wearable devices, would be an important advance in addressing this problem. Another important dialysis-related environmental footprint is the large volume of fluid generated as waste \[[@CR74]\]. Both sorbent-based techniques \[[@CR61], [@CR62], [@CR75]\] and membrane-based reclamation of spent fluid \[[@CR76]\] are potential approaches for reducing the volume of effluent generated during CRRT and other dialysis modalities. We predict that progress in addressing long-term sustainability for acute renal support modalities will begin to be made over the next decade. Conclusion {#Sec16} ========== As the utilization of CRRT and ancillary therapies in the management of critically ill patients increases, their limitations have become more evident to many clinicians. These limitations apply both to the manner in which patients are clinically managed and how the technology is used. We have critically assessed these problems and rendered predictions about the manner in which they will be solved over the next decade. While we have tried to prognosticate on several topics, we have not attempted to address certain contentious issues that we believe will continue to be debated even in 2027. For example, we believe questions about the optimal CRRT mode \[[@CR77]\] and nutritional regimen \[[@CR78]\] will not be resolved completely. We have also attempted to raise awareness about the issue of sustainability for CRRT beyond 2027 and have raised several points for consideration. While a number of problems currently exists, we believe the future of extracorporeal therapy for critically ill patients is very bright and its use in this patient population will continue to grow. %FO : Percent fluid overload ADQI : Acute Dialysis Quality Initiative AKI : Acute kidney injury CKD : Chronic kidney disease CO~2~ : Carbon dioxide CRRT : Continuous renal replacement therapy EMR : Electronic medical record ESRD : End-stage renal disease GFR : Glomerular filtration rate ICU : Intensive care unit RCA : Regional citrate anticoagulation RCT : Randomized controlled trial RRT : Renal replacement therapy Not applicable. Funding {#FPar1} ======= Not applicable. Availability of data and materials {#FPar2} ================================== Not applicable. Authors' contributions {#FPar3} ====================== WRC and CR were responsible for preparing the initial draft, ensuring that comments from the other authors were incorporated appropriately, and finalizing the manuscript. Based on their original published research, all authors contributed to different sections in the preparation of the manuscript, according to their expertise. In particular, MN and FG primarily developed the aspects involving current and future technologies, SLG and ZR provided the pediatric perspective in several respects, and XD and JX were the major contributors to the topics of timing of treatment initiation, management of fluid overload, and biomarkers. All authors reviewed each version of the draft manuscript during preparation and approved the final version. Competing interests {#FPar5} =================== WRC was formerly employed by Baxter Healthcare, has received consulting fees and honoraria from Baxter, and owns Baxter stock. WRC is also a consultant for Astute Medical. CR is an Associate Editor of Critical Care. Consent for publication {#FPar6} ======================= Not applicable. Ethical approval and consent to participate {#FPar7} =========================================== Not applicable. Publisher's Note {#FPar8} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION ============ In eukaryotes, three of the four cytoplasmic rRNAs are transcribed from polycistronic coding units. The nascent transcripts contain 18S, 5.8S and 25S/28S rRNAs flanked by 5′ and 3′ external transcribed spacers (ETS) and two internal transcribed spacers (ITS1 and ITS2). This precursor undergoes nucleotide modification at scores of sites and cleavage and trimming reactions (processing) to form the mature rRNAs ([Figure 1](#F1){ref-type="fig"}) ([@B1; @B2; @B3; @B4]). Both modification and processing of rRNA require small nucleolar RNAs (snoRNAs) that act through base-pairing with pre-rRNA ([@B5; @B6; @B7; @B8; @B9]). Most snoRNAs, functioning as snoRNPs, guide pseudouridine (Ψ) or 2′-O-methylation (Nm) modifications; only a few are required for the cleavage reactions ([@B9]). For the 76 known yeast snoRNAs, 45 have one modification guide domain, 25 have two different guide domains and two mediate both processing and modification reactions ([@B10]). These functions are targeted by complementarities to the various, specific rRNA regions. Figure 1.Yeast pre-rRNA processing pathway. The major rRNA precursors and final products are shown. Cleavage sites relevant to the present study are indicated, as well as the approximate location of the Ψ guided by snR10 in 25S rRNA. While it is generally accepted that the snoRNA modification and processing reactions do not occur randomly, the basis and extent of coordination of these functions remain unknown. Binding of snoRNPs in a coordinated manner has implications for possible chaperone-like functions for snoRNAs in ribosome assembly, and for coupling modification and cleavage events. To gain insight into these important questions, we have carried out functional interference mapping of a yeast snoRNA (snR10) that is known to be required for both rRNA modification and normal processing. Our aim was to determine if the different interactions with pre-rRNA occur independently or are coordinated. Loss of this snoRNA causes slow growth and processing defects and the Ψ modification has been shown to affect ribosome translation activity in vivo ([@B11],[@B12]). This knowledge facilitated our functional analysis of the snoRNA by mutagenesis. The results reveal a cooperative interference effect between the two functional domains that is consistent with coordinated binding to pre-rRNA. Most nucleotide modifications in rRNA appear to occur before processing is completed; among these alterations the Ψ and Nm modifications are dominant. In eukaryotes, box H/ACA snoRNPs carry out pseudouridylation and box C/D snoRNPs mediate Nm modification ([@B6],[@B9],[@B13; @B14; @B15; @B16; @B17]). Each snoRNP contains a single snoRNA that selects the rRNA site to be modified, and a set of four family-specific core proteins, one of which catalyzes the modification reaction ([@B5],[@B18]). The consensus structure of the H/ACA snoRNAs, the type featured here, contains two hairpin regions, linked by an H box (hinge region) and an ACA box at the 3′-end ([@B19],[@B20]) ([Figure 2](#F2){ref-type="fig"}A). In Saccharomyces cerevisiae, 29 H/ACA snoRNAs have been identified and all but one have been assigned to specific Ψs in rRNA. For 14 of these snoRNAs both hairpin structures target Ψ, to sites in the same or different rRNA species ([@B10]). The dual-function snoRNA featured in the present study has both Ψ guide and processing functions, with Ψ formation targeted to a single site. Figure 2.A 7-nt sequence element in snR10 is required for normal cell growth and pre-rRNA processing. (A) Schematic structure of the snR10 snoRNA. Key features are identified, including the H and ACA boxes, the Ψ guide region and five segments complementary to pre-rRNA. The complementary segments are numbered and highlighted with lines outside the snoRNA structure. Segment 2, which is complementary to the 5′ETS (thick line), is revealed in the present study to be essential for the processing function of snR10. Sites within and neighboring this segment were subjected to mutational analysis; the flanking sites are identified with dashed lines ([Supplementary Table S1](http://nar.oxfordjournals.org/cgi/content/full/gkq043/DC1)). A lower panel shows complementarity between segment 2 and the 5′ETS of pre-rRNA. The 7-nt element required for processing is boxed. (B) Mutation of segment 2 impairs cell growth. Test cells expressing snR10 with mutations were serially diluted (1:5), dotted on solid medium lacking Trp, and incubated at 30°C for 30 h. Control cells include: a wild-type strain (YS602) with an empty vector (Con.), and a parental test strain depleted of endogenous snR10 that contains either an empty vector (−snR10) or a plasmid that expresses wild-type snR10 (+WT). Growth properties of cells expressing snR10 variants with mutations in segment 2 or 4, (M2 or M4), are shown. The arrow above the colony patterns indicates decreasing numbers of cells. (C) Mutation of segment 2 causes accumulation of a 23S precursor to 18S rRNA: this product is defined by the 5′-end of the transcript and the A~3~ cleavage. Total RNA was analyzed by northern hybridization. U2 snRNA is shown as a loading control. (D) Polysome patterns are altered by mutation of segment 2. Whole-cell extracts containing similar amounts of RNA were separated by sucrose gradient fractionation and ribosomal complexes detected by UV absorbance. The identities of ribosomal complexes are indicated for control cells (Con.). Reductions in the level of free 40S subunits are marked by open arrowheads and increases in the level of free 60S subunits by asterisks. More than 70% of the 45 Ψ and 54 Nm modifications in yeast rRNA are conserved in humans, and the functionally important regions of the ribosome are especially rich in these modifications, arguing that they have important roles in ribosome production or function. Structural studies with small model RNAs containing Ψ and Nm nucleotides have shown that these modifications stabilize RNA structure in different ways ([@B21; @B22; @B23]). Considerable genetic and biochemical evidence is now available showing that these modifications are indeed important, based on examining effects of blocking modification formation by deleting specific guide snoRNAs or mutating guide sequences. These results, many from our laboratory, have revealed that depleting small subsets of modifications from different rRNA regions can significantly impair ribosome biogenesis and activity ([@B12],[@B24; @B25; @B26]). Modification is accepted to be the major function of the snoRNPs, however, since all snoRNPs appear to act through base-pairing with pre-rRNA, it is also possible that some modifying snoRNPs could affect pre-rRNA folding, distinct from the effect of modification itself ([@B27]). In the case of the processing snoRNPs, remodeling pre-rRNA seems highly likely (see below) ([@B28; @B29; @B30; @B31; @B32; @B33; @B34]). In the yeast rRNA processing pathway, the longest detectable pre-rRNA---a 35S species, is cleaved sequentially to generate the immediate precursor of 18S rRNA (20S) and a 27S precursor that contains the 5.8S and 25S rRNAs ([Figure 1](#F1){ref-type="fig"}). The 20S pre-rRNA, which is pertinent to the current study, is exported to the cytoplasm where it is further cleaved (at site D) to create 18S rRNA ([@B35]). A few snoRNPs are also required for specific early cleavages. In yeast, the processing snoRNPs, i.e. U3, U14, U17/snR30 and snR10, are required for 18S rRNA maturation. The first three are thought to be common to all eukaryotes, whereas snR10 is thus far yeast-specific. U3 and U14 are C/D snoRNAs and snR30 and snR10 are H/ACA species. The universal processing snoRNAs contain sequence elements that base-pair with pre-rRNA without guiding modification, however, the roles of these elements and, indeed, the snoRNPs themselves remain to be defined. Speculation includes remodeling of pre-rRNA for cleavage by cis- or trans-acting components ([@B28],[@B31],[@B34],[@B36],[@B37]). The U3 C/D snoRNP is the first to associate with full-length 35S pre-rRNA ([@B38]). In mediating the cleavages at sites A~0~, A~1~ and A~2~, a 10-nt element of U3 base-pairs with a 5′ ETS segment ([@B39; @B40; @B41; @B42; @B43; @B44]). In addition, a 7-nt segment in U3 base-pairs with the 5′ region of 18S rRNA, which otherwise forms a pseudo-knot structure ([@B31]), inferring the U3 snoRNP has a role in pre-rRNA folding. The U14 C/D snoRNA contains two sequence elements (13 and 14 nucleotides) complementary to pre-rRNA. One base-pairs with the 5′-end of 18S rRNA and is essential for pre-rRNA cleavages ([@B28],[@B45]). The other is a guide sequence for a methylation site in 18S rRNA (Cm414), and is dispensable for processing ([@B28],[@B46]). Finally, production of 18S rRNA has been shown to also depend on two short complementary elements in the H/ACA snoRNA snR30/U17, which base-pair with 18S rRNA ([@B34],[@B47]). Determining how these processing snoRNAs affect pre-rRNA cleavages remains an important challenge. The yeast U3, U14 and snR30 snoRNAs are all essential for cell growth, whereas deletion of snR10, the snoRNA examined here, has a less severe effect, resulting in a moderately reduced growth rate and a cold-sensitive phenotype ([@B11]). Although not severe, this phenotype can be used to screen for functional requirements of the snoRNA sequence elements. In the absence of snR10, processing of 18S rRNA is moderately impaired as evidenced by accumulation of full-length pre-rRNA (35S) and an intermediate that is normally of very low-abundance (23S); the latter species is generated by cleavage at site A~3~ prior to removal of the 5′ ETS segment ([Figure 1](#F1){ref-type="fig"}) ([@B48],[@B49]). The snR10 snoRNA also directs Ψ modification at site U2923 in the A-loop of the peptidyl transferase center (PTC) region in 25S rRNA ([Fig 1](#F1){ref-type="fig"}, Ψ) ([@B14],[@B15]). This Ψ is not required for pre-rRNA processing, as revealed by a point deletion in the guide domain of snR10 that disrupts Ψ formation, yet has no effect on cell growth or pre-rRNA cleavages ([@B12]). This observation suggests that the processing function depends on structural elements outside the guide domain. Thus far, no processing elements have been identified in snR10, and it is not known if the modification guide domain communicates with its processing determinant(s). In this study, we identify a sequence element that is necessary for normal pre-rRNA processing. Functional mapping also revealed that the processing and modification domains are interconnected functionally. MATERIALS AND METHODS ===================== Plasmids -------- Mutations in snoRNA coding units were created using a PCR approach ([@B24],[@B50]). All constructs are listed in [Supplementary Table S1](http://nar.oxfordjournals.org/cgi/content/full/gkq043/DC1). Strains and cell growth ----------------------- A yeast strain (YS602) with an empty vector was used as a positive control throughout the study. The parental test strain lacks the snR10 snoRNA gene, which was replaced with a marker gene using targeted gene disruption. The strain was transformed with plasmids that encode wild-type or mutant snR10 snoRNAs. All experimental strains are listed in [Supplementary Table S2](http://nar.oxfordjournals.org/cgi/content/full/gkq043/DC1). For growth comparisons, yeast cultures prepared in synthetic medium lacking Trp were adjusted to 2 OD~600~/ml, diluted 1:5 serially, dotted on plates with synthetic medium lacking Trp, and incubated at 30°C for 2--3 days ([@B24]). RNA preparation and analysis ---------------------------- Total RNA was isolated from cells using the Tri-reagent procedure (Sigma) following the manufacturer's instructions. The presence and abundance of snoRNAs and rRNA species were determined by northern blotting ([@B51]). To examine the rRNA components in an abnormal 50S pre-rRNP complex detected in a C-insertion mutant of snR10 (M + C), RNA was prepared from sucrose gradient fractions containing this complex and subjected to northern analysis ([@B52]). The Ψ and Nm modifications were detected by primer extension analysis ([@B53],[@B54]). Analysis of polysome patterns ----------------------------- Whole-cell extract was fractionated by sucrose gradient (7--47%) centrifugation and examined as described previously ([@B24]). In vivo pulse-chase labeling of rRNA -------------------------------------- Test cells were grown in synthetic medium lacking methionine to an OD~600~ of 0.8--1.0, and ∼3 OD~600~ units of cells were diluted to 3 ml with the same, prewarmed medium. Next, 0.1 mCi of \[methyl-^3^H\]methionine (1 mCi/ml, PerkinElmer) was added and the cells pulse-labeled at room temperature for 3 min. Labeling was terminated by addition of 200 μl of 60 mM unlabeled ([d, l]{.smallcaps})-methionine. Subsequently, 800 μl samples were taken at time points 0, 2, 5 and 15 min after the chase was initiated, and immediately frozen in liquid nitrogen. Total RNA was prepared and dissolved in 10 μl water. Three microliters of RNA was fractionated on 1.2% agarose gels, transferred to a membrane and visualized with a Fuji PhosphorImager ([@B24]). In vivo chemical probing of RNA structure ------------------------------------------- Chemical modification with dimethyl sulfate (DMS) and primer extension probing of snoRNA and rRNA structure were carried out as described earlier ([@B24]). RESULTS ======= A 7-nt sequence element in the 5′ hairpin of snR10 is required for pre-rRNA processing -------------------------------------------------------------------------------------- Our functional study of snR10 proceeded in three stages: (i) identifying and testing the importance of elements complementary to pre-rRNA for rRNA processing activity; (ii) examining effects of guide domain mutations on Ψ formation and rRNA processing; and (iii) evaluating the relationship between the processing and modification domains. Because other processing snoRNAs act through base-pairing with pre-rRNA, we reasoned this is likely the case for snR10. Screening of the snR10 sequence with bioinformatics tools LALIGN Server (<http://www.ch.embnet.org/software/LALIGN_form.html>) identified five segments of at least 8 nt that are complementary to different regions of pre-rRNA ([Figure 2](#F2){ref-type="fig"}A). The importance of these sequences was tested by replacement with complementary nucleotides and screening for effects on growth and pre-rRNA processing. We anticipated that functional defects in processing should impair growth rate, as seen earlier for deletion of snR10 ([@B11]). The mutant snoRNAs were characterized in a test strain depleted of wild-type snR10, and determined to be produced at levels comparable to wild-type snR10 expressed in the same way (data not shown). As expected, mutating the Ψ guide domain disrupted modification at U2923 in 25S rRNA (segment 5), as shown by primer extension analysis (data not shown). The growth data revealed that only one substitution mutation caused slower growth---in particular, mutation of segment 2 in the 5′ half of snR10 (U~78~GAAGAAU~85~). Based on growth on solid medium, the mutant strain has a growth rate that is ∼20% of normal, essentially the same as the effect of snR10 depletion (M2, [Figure 2](#F2){ref-type="fig"}B, and data not shown). This observation suggests that segment 2, which is complementary to the 5′ ETS of pre-rRNA (−327 to −320), is involved in processing. Indeed, the mutation caused significant accumulation of the normally low-abundance 23S pre-rRNA ([Figure 2](#F2){ref-type="fig"}C). Subunit levels were also affected, with strong accumulation of free 60S subunit (∼5-fold) and dramatic reduction in the level of free 40S subunit (∼4-fold), as determined by analysis of polysome profiles by density gradient fractionation ([Figure 2](#F2){ref-type="fig"}D). These defects mimic exactly those that accompany loss of wild-type snR10, indicating that the complementary segment is required for its processing activity. To better define the sequence and folding requirement of this particular snoRNA region, we analyzed 10 additional substitution mutations of 1--8 nt ([Supplementary Table S1](http://nar.oxfordjournals.org/cgi/content/full/gkq043/DC1)). These included changes in the flanking sequences and sequences predicted to be single and double stranded ([Figure 2](#F2){ref-type="fig"}A, indicated as dashed lines). Together, the new data defined a 7-nt sequence required for normal processing function (C~75~CAUGAA~81~). Two mutants, C~75~CAU~78~→GGUA and U~78~GAA~81~→ACUU, exhibited growth defects similar to that caused by loss of snR10 itself ([Figure 2](#F2){ref-type="fig"}A, lower panel, and data not shown). The important 7-nt sequence partially overlaps the complementary sequence examined initially, with the 3′ 4 nt located in the complementary region ([Figure 2](#F2){ref-type="fig"}A, lower panel). In addition, mutations designed to disrupt the predicted secondary structure of the upper stem had no effect on cell growth (data not shown). These results show the 7-nt element required for processing likely functions in a sequence-dependent manner. Next, we carried out an in vivo chemical probing analysis to determine if the 7-nt processing element is single-stranded and available for base-pairing. The structure probe was DMS, which methylates accessible A and C nucleotides. Following DMS treatment, total cell RNA was prepared and the relevant snoRNA region was examined by primer extension analysis. The results show that the 7-nt region is clearly modified by DMS (C~76~ to A~84~), indicating that this segment is single-stranded as predicted in the folding model ([Figure 3](#F3){ref-type="fig"}A and B). The results also suggest that the novel processing element is not covered by protein. Figure 3.The 7-nt processing element in snR10 is single-stranded. (A) Secondary structure of the 5′-hairpin region in snR10 that contains the 7-nt processing element. The structure was predicted with the folding program Mfold, confirmed and refined using the in vivo chemical probing data in (B) below. The relative extent of DMS modification is indicated by the number of asterisks. The seven nucleotides required for pre-rRNA processing are circled. The potential H box is indicated. (B) The 7-nt element is present in an internal loop of the 5′ hairpin. Following DMS modification in vivo, the structure of snR10 was probed by primer extension analysis with extension products separated on an 8% sequencing gel, next to sequencing products created with the same primer. Nucleotides accessible to DMS are identified. The snoRNA information required for processing is located exclusively in the 5′ hairpin --------------------------------------------------------------------------------------- The mutational results available for the processing and modification functions of snR10 suggest that the information for processing of 18S rRNA is located in the 5′ hairpin of the snoRNA ([Figure 2](#F2){ref-type="fig"}), and that for modification of Ψ2923 in 25S rRNA occurs solely in the 3′ hairpin ([@B12]). To test this model directly, we attempted without success to separate these functions by expressing the individual hairpins alone. While single hairpin snoRNAs occur in trypanosomes and euglena ([@B55],[@B56]), to our knowledge, none have yet been stably expressed in S. cerevisiae. Thus, we turned to the reliable strategy of expressing hybrid snoRNAs created by combining the 5′ or 3′ hairpin of snR10 with the appropriate hairpin from another yeast H/ACA snoRNA. In these experiments, the second snoRNA was snR36, which guides Ψ modification in 18S rRNA. Depletion or over-expression of snR36 does not affect cell growth or pre-rRNA processing (our unpublished data). The origins of the hairpins are indicated in the hybrid names, for example, a hybrid (H) containing the 5′ hairpin of snR10 and 3′ hairpin of snR36 is designated H10/36, and the alternate combination is H36/10 ([Figure 4](#F4){ref-type="fig"}A). Figure 4.The processing function of snR10 is mediated by its 5′ hairpin. (A) The roles of the 5′ and 3′ hairpins in snR10 function were examined with hybrid snoRNAs containing hairpins from snR10 and snR36 in test cells lacking snR10. The structures of the hybrid snoRNAs are depicted schematically, with the snR10 portion shown with thick lines and the snR36 portion with thin lines. (B) The hybrid snoRNAs are expressed at normal levels as evident in northern hybridization results. (C) Normal growth occurred with the hybrid containing the 5′ hairpin of snR10 (H10/36), but not the hybrid containing the 3′ hairpin of snR10 (H36/10). The snoRNA snR38 was used as a control for loading. (D) A normal ribosome--polysome pattern was restored by the hybrid snoRNA with the 5′ hairpin from snR10 (H10/36). Half-mer polysomes are evident in test cells that express a snR10 mutant lacking a C nucleotide (M-C) in the guide domain; this deletion blocks Ψ2923 formation ([@B12]). Accumulation of half-mer polysomes is indicated by an arrow. (E) The Ψ2923 modification is formed with a hybrid snoRNA containing the 3′ hairpin of snR10 (H36/10). Total RNA was treated with CMC and Ψ detected by primer extension analysis. The extension stops representing Ψ2923 and Ψ2880 (guided by snR10 and snR34, respectively) are identified. The hybrid snoRNAs were expressed at normal levels in the test strain lacking snR10, and as anticipated, the hybrid containing the 5′ hairpin of snR10, but not the 3′ hairpin, restored normal growth for cells depleted of snR10 ([Figure 4](#F4){ref-type="fig"}B and C). The inference that the 5′ hairpin of snR10 is sufficient for its processing function was borne out by the finding that this hybrid also restored a normal polysome profile ([Figure 4](#F4){ref-type="fig"}D). Also, as expected, the hybrid containing the 3′ hairpin from snR10 restored the Ψ2923 modification ([Figure 4](#F4){ref-type="fig"}E). Together, these data demonstrate that the information required for normal rRNA processing is exclusively located in the 5′ hairpin of snR10, and that the processing and modification functions are mediated by the 5′ and 3′ hairpins, respectively. Earlier, we showed that the rate of rRNA processing is normal for a snR10 mutant with only a point deletion in the guide domain that disrupts its modification function (M−C) ([Figure 5](#F5){ref-type="fig"}A) ([@B12]). The same cells also accumulate half-mer polysomes, whereas cells expressing the present hybrid with the 5′ hairpin of snR10 do not ([Figure 4](#F4){ref-type="fig"}D); Ψ2923 formation was blocked in both strains. The new data argue that loss of the Ψ modification is not the basis of the half-mer polysome effect. A noteworthy difference between the two mutant snoRNAs is that the snoRNA with the point deletion has the potential to base pair with its normal target sequence in pre-rRNA, whereas the hybrid snoRNA with the 3′ hairpin of snR36 (H10/36) does not. These observations infer that altering the base-pairing pattern between the guide domain of snR10 and pre-rRNA may affect pre-rRNA folding or pre-rRNP assembly (see below). Figure 5.A 1-nt insertion mutation in the Ψ guide domain severely impairs cell growth. (A) Sequences of three mutant variants of snR10 snoRNA. The uridine normally targeted in 25S rRNA (U2923) is circled and an adjacent uridine that is modified with the two insertion variants of snR10 is marked with an arrow (U2924). In one mutant, a single C (italicized) has been deleted (M−C). Two other mutants contain C and CA insertions at the same position, as indicated by an arrow (M + C, M + CA). (B) Northern hybridization results show that the two insertion mutant snoRNAs are expressed at comparable levels. (C) The shift in Ψ modification site for the insertion mutations is shown by primer extension data, as in [Figure 4](#F4){ref-type="fig"}, and the positions of the natural and novel Ψs are indicated. (D) The C-insertion mutation severely impairs cell growth (M + C). Growth was examined in a serial dilution assay as in [Figure 2](#F2){ref-type="fig"}B. A mutation in the guide domain that shifts Ψ modification to an adjacent site severely impairs growth ----------------------------------------------------------------------------------------------------- Our functional mapping included mutating the guide domain of snR10 to identify potential effects on rRNA modification and processing. We were especially interested in the possibility of shifting Ψ formation to an adjacent uridine and determining if this shift influences ribosome biogenesis or function. With the knowledge that the distance between the ACA box and the target uridine is an important determinant in pseudouridylation ([@B14],[@B15]), we created two small insertion mutations predicted to alter modification specificity. The mutations include C and CA insertions in the 3′ element of the guide motif (M + C and M + CA) ([Figure 5](#F5){ref-type="fig"}A). Both mutant snoRNAs were expressed at nearly normal levels in the test strain, and each caused Ψ modification to be shifted from U2923 to U2924 ([Figure 5](#F5){ref-type="fig"}B, C). No modification appears to occur at the natural site. The modification level with the C-insertion mutation is similar to that observed with wild-type snR10, whereas that with the CA-insertion is somewhat weaker than normal ([Figure 5](#F5){ref-type="fig"}C). These results confirm the importance and relationships of the snoRNA--substrate base-pairing interactions in modification specificity and efficiency. Surprisingly, the C-insertion mutation causes a near-lethal phenotype in the snR10 depleted cells. Only moderate effects on growth rate occur with deletion of snR10 or with expression of snR10 bearing the C-insertion mutation in control cells (data not shown). However, expression of the C-insertion variant of snR10 in cells depleted of wild-type snR10 impairs cell growth severely, to a much greater extent than simple loss of the snR10 snoRNA ([Figure 5](#F5){ref-type="fig"}D). The CA-insertion also causes slower growth; however, the effect is similar to that caused by snR10 depletion. These results suggest that the deleterious effects of the insertion mutations on cell growth are caused by either the shift in modification location, or the altered base-pairing between the mutant snoRNAs and pre-rRNA, or a combination of these effects. The steady-state levels of rRNA are dramatically reduced by the C-insertion mutation ------------------------------------------------------------------------------------ With a view to determining the basis of the severe growth defect caused by the C-insertion, we examined the rRNA levels in the mutant cells, which reflect the level of mature ribosomes. In cells expressing the C-insertion variant of snR10, the levels of mature 18S, 5.8S and 25S rRNAs were reduced by ∼60--80%, whereas the levels of the 35S full-length precursor and pre-18S rRNA (20S) were increased by ∼2-fold, as compared with control cells ([Figure 6](#F6){ref-type="fig"}A). No significant difference was found for the 27S pre-rRNA, a precursor of the large subunit rRNAs, or for 5S rRNA, which is transcribed by RNA pol III. The levels of two control RNAs, tRNA^Leu^ and U2 snRNA, remained unchanged in the C-insertion mutant cells (not shown). Notably, for the CA-insertion mutation, no significant reduction in the levels of 18S and 25S rRNA was observed (data not shown). These results indicate that the C-insertion mutation causes a strong reduction in the level of RNAs generated by RNA pol I, presumably due to impaired processing since 35S and 20S pre-rRNAs accumulate in the mutant cells. Figure 6.The single nucleotide insertion in the guide domain causes a strong decrease in rRNA levels. (A) Northern hybridization probing of rRNA species reveals sharp reductions of ∼60--80% in the level of mature rRNAs, and significant accumulation of 35S and 20S pre-rRNAs. (B) The steady-state levels of 18S, 5.8S and 25S rRNAs are dramatically reduced as indicated by measuring the hybridization signals in (A), and normalizing to the amount of 5S rRNA. (C) The 20S pre-rRNA with Ψ at the novel site accumulates normally in the cytoplasm, as indicated by the presence of dimethylation modifications, which are formed only in cytoplasmic 20S pre-rRNA. The methylation modifications were detected by primer extension analysis. In yeast, the 20S precursor to 18S rRNA is formed in the nucleolus and is then exported to the cytoplasm where it is cleaved to form mature 18S rRNA ([@B4]). To determine if the C-insertion mutation affects export of 20S pre-rRNA, we examined its cellular distribution. This was done by assaying the presence of a unique pair of dimethylation modifications (m^6^~2~A1781m^6^~2~A1782) that are catalyzed by the enzyme Dim1p after 20S pre-rRNA has been exported to the cytoplasm ([@B57],[@B58]). The presence of the dimethyl modifications was probed by primer extension analysis and is reflected by strong extension stops ([Figure 6](#F6){ref-type="fig"}C). Dimethylations were detected in both mutant and control cells showing that 20S pre-rRNA containing the mis-targeted Ψ is exported to the cytoplasm. The C-insertion mutation severely impairs rRNA processing and causes accumulation of an abnormal rRNP complex ------------------------------------------------------------------------------------------------------------- We determined that pre-rRNA processing is strongly impaired by the C-insertion mutation in the guide domain, as examined by in vivo pulse-chase labeling with \[^3^H\]-methionine and gel electrophoresis ([Figure 7](#F7){ref-type="fig"}A; strain M + C). Production of 18S rRNA was severely inhibited, as reflected by accumulation of 20S precursor at a high level after 15 min into the chase and a strong delay in the appearance of mature 18S rRNA. In addition, two low abundance species often undetected, 23S and 21S pre-rRNAs, also accumulated. The 35S and 32S pre-rRNAs accumulated and persisted at least 5 min into the chase, whereas in control cells, these species were quickly processed and disappeared by 2 min after the chase was initiated. Thus, as a consequence of the C-insertion mutation, cleavage is delayed at the site that separates precursors for the small and large subunit rRNAs (site A~2~). Cleavages at sites that generate the 32S pre-rRNA (A~0~ and A~1~) and conversion of 27S precursor to 25S rRNA were also moderately delayed. Interestingly, the CA insertion mutation causes only minor processing defects, as evidenced by slightly slower maturation rates. Taken together, the results indicate that although the guide domain of snR10 and the Ψ that it normally targets are not required for pre-rRNA processing, a point mutation in the guide sequence can lead to severe impairment of rRNA processing, especially for production of 18S rRNA. Figure 7.rRNA processing is severely impaired by the single nucleotide insertion in the Ψ guide domain. (A) rRNA processing was examined by in vivo pulse-chase labeling. Test cells were labeled with \[methyl-^3^H\]methionine, and total RNA was prepared during a chase with unlabeled methionine. RNA was separated in a 1.2% agarose gel, transferred to a membrane, and the radioactive bands visualized with a phosphorimager. Normal rRNA species are identified by arrows and unusual pre-rRNAs are marked with arrowheads. (B) The C-insertion mutation causes an abnormal polysome pattern (M + C). Polysome profiles were analyzed as in [Figure 2](#F2){ref-type="fig"}. The abnormal 50S complex in the mutant cells is indicated by an arrowhead. To determine if the C-insertion mutation affects production of ribosome particles, the patterns of these complexes were analyzed for equivalent amounts of rRNA ([Figure 7](#F7){ref-type="fig"}B). Interestingly, in the mutant cells, the level of free 40S complexes was 2- to 3-fold higher than for cells depleted of snR10, although the C-insertion mutation causes much stronger processing defects in 18S rRNA production than simple loss of snR10. The C-insertion mutant also exhibits an increase in the level of free 60S subunits (1.5-fold higher), but to a lower extent than the snR10 depletion condition, where an increase of \>3-fold was seen ([Figure 7](#F7){ref-type="fig"}B). In the C-insertion mutant, the levels of 80S ribosomes and polysomes are reduced by ∼50%, compared with control cells expressing wild type snR10. An abnormal complex of ∼50S was observed in the C-insertion mutant cells. This complex is most likely an intermediate of the small subunit, since northern probing of gradient fractions containing the complex revealed increased accumulation of 20S pre-rRNA, but not 27S pre-rRNA (data not shown). In the CA insertion mutant (M + CA), the polysome pattern is also altered, but to a much smaller extent. Here, the level of free 40S complex was reduced by ∼50% with a concomitant increase of ∼80% in free 60S complexes; levels of 80S ribosomes and polysomes were nearly normal. Thus, the data for this mutant suggest that subunit association and ribosome function are impaired moderately. Together, the results show that the C-insertion mutation has a severe negative impact on pre-rRNA processing and ribosome formation. The snoRNA insertion mutations cause changes in ribosome structure ------------------------------------------------------------------ Both the C and CA insertion mutations shift the targeted Ψ to an adjacent site. The patterns of complementarity between the two mutant snoRNAs and pre-rRNA are different from each other, and also different from that of wild-type snR10 ([Figure 5](#F5){ref-type="fig"}A). Because altered base-pairing of snoRNA with pre-rRNA could, in principle, influence pre-rRNA folding, and because Ψ modifications are already known to stabilize RNA duplexes, we next asked if the two mutant snoRNAs cause detectible changes in ribosome structure. Screening was by chemical probing of rRNA in vivo with DMS, followed by primer extension analysis of the PTC region in 25S rRNA that includes the Ψ sites of interest. One nucleotide in this region showed a qualitative change in DMS reactivity (C2889): it is strongly modified by DMS in control cells, but not in cells expressing either mutant snoRNA ([Figure 8](#F8){ref-type="fig"}A and B). This result indicates that ribosome structure is indeed changed by the mutant snoRNAs, due either to the altered interaction between the mutant snoRNAs and pre-rRNA or the altered Ψ modification pattern. Figure 8.A change in rRNA structure accompanies the shift in Ψ modification. (A) Secondary structure of the A-loop region of the PTC region in 25S rRNA. The arrow identifies a nucleotide (C2889) with reduced DMS reactivity, as shown in (B). The normally modified Ψ2923 site and new site of modification (U2294) are circled. (B) Primer extension probing of rRNA structure in the A-loop region. Total RNA was prepared from cells treated with DMS and subjected to primer extension mapping. A nucleotide (C2889) showing altered DMS accessibility in test cells is marked with an arrow. A control nucleotide (A2847) sensitive to DMS treatment in all test strains is marked with an arrowhead. The deleterious effect of the C-insertion mutation in the guide region depends on the presence of the processing element in the 5′ hairpin ------------------------------------------------------------------------------------------------------------------------------------------ While both insertion mutations cause the same shift in Ψ targeting, the C-insertion mutation results in a substantially higher modification yield (full versus ∼60%) and much stronger defects in growth and processing. These observations suggest that the much stronger detrimental effects of the C-insertion mutation are due to the different base-pairing patterns between the two insertion mutant snoRNAs with pre-rRNA, or seemingly less likely, the different levels of Ψ modification at the same novel site. Since the 3′ hairpin of snR10 can be deleted without affecting processing, an additional interesting possibility is suggested as the basis of the strong processing defects caused by the C-insertion mutation. In particular, the defects could result from a cooperative negative interference effect between the processing and modification domains of snR10. Such an effect could occur directly or indirectly. For example, an altered sequence in the guide domain could cause a change in the structure of the snoRNP that affects its interactions with pre-rRNA. We reasoned that the importance of the modification level at the novel sites could be evaluated using new hybrid guide snoRNAs that guide modification at the novel site with the same level as the C-insertion mutation, but should not cause the detrimental effects seen for the C-insertion mutation. In this context, we examined the effects of expressing a hybrid snoRNA that contains the 3′ hairpin of snR10 with the C-insertion mutation joined with the 5′ half of snR36 (H36/10 + C) ([Figure 9](#F9){ref-type="fig"}A). This hybrid snoRNA was expected to give the same modification specificity and yield as the C-insertion mutation. If this is indeed the case, and growth and processing are not affected or only moderately affected, those results would be strong evidence that the severe interference effects are not due to the change in modification specificity or yield. Figure 9.The severe growth defect caused by the C-insertion mutation is not due to mis-targeting of the Ψ modification. (A) Construction of a hybrid snoRNA with the mis-targeting guide domain from snR10 (H36/10 + C). The mutant snoRNA includes the 5′ hairpin of snR36 (thin line) and 3′ hairpin of snR10 (thick line) with the C-insertion mutation. (B) The levels of Ψ modification at the novel site are similar for both the novel hybrid and the initial snR10 variant with the C-insertion mutation (H36/10 + C and M + C), as shown by primer extension data. (C) The hybrid snoRNA with the C-insertion in the guide domain causes moderate growth defects similar to those seen for snR10 depletion. (D) Similar moderate defects in processing are also apparent for the hybrid snoRNA with the C-insertion mutation and for the snR10-depletion condition. (E) Structure of a double mutant of snR10 that includes changes in both the processing and modification hairpin domains (M2 + C). The changes are: substitution mutations in segment 2 of the 5′ hairpin, and the C-insertion in the guide region in the 3′ hairpin. (F) A 4-nt substitution in the 7-nt processing element suppresses the growth defect caused by the 3′ C-insertion mutation. (G) Co-expression of hybrid snoRNAs H10/36 and H36/10 + C in snR10-depleted cells. Total RNA was analyzed by northern hybridization using probes specific to the 5′ half (upper panel), or the 3′ half (middle panel). 5S rRNA was used as a control for loading. (H) No growth defect resulted from co-expression of the H10/36 and H36/10 + C snoRNAs. The new hybrid snoRNA accumulated at a level comparable to the original C-insertion mutant snoRNA (data not shown), and as expected, Ψ modification was also shifted to U2924 and at a similar level ([Figure 9](#F9){ref-type="fig"}B). Cell growth rate with this hybrid was moderately impaired (20% of the normal), but to a very much lower extent than the snR10 mutant with same 3′ domain (20% versus 1% of normal growth rate) ([Figure 9](#F9){ref-type="fig"}C). Notably, the effect of the hybrid snoRNA on growth was comparable to loss of snR10 itself. In cells containing this hybrid snoRNA, pre-rRNA processing and ribosome fractionation patterns were also similar to those seen for snR10 depletion ([Figure 9](#F9){ref-type="fig"}D and data not shown). The moderate defect in processing is most likely due to loss of the 5′ hairpin from snR10, which is required for normal processing. Thus, the defects on cell growth and pre-rRNA processing observed for the new hybrid snoRNA mimic those of snR10 depletion. Importantly, the defects are much less severe than those seen for the C-insertion mutation, which reduced the growth rate by ∼99% and strongly impaired processing ([Figure 7](#F7){ref-type="fig"}). Together, these results show that the strong deleterious effects of the C-insertion mutation are not due to the change in Ψ location or level of modification. Rather these severe effects require the presence of the 5′ hairpin of snR10. Our prediction that the deleterious effects of the C-insertion mutation require the presence of the 7-nt processing element in the 5′ half was tested by creating another snR10 double mutant. This variant contains both the C-insertion in the 3′ domain and the four nucleotide substitution mutation in the 7-nt element described above (U~78~GAA~81~) ([Figure 9](#F9){ref-type="fig"}E). This mutant snoRNA (M2 + C) accumulated normally (data not shown), and caused a decrease in growth rate similar to that seen for snR10 depletion (20% of normal), but not the very severe growth defect seen for the C-insertion mutation alone ([Figure 9](#F9){ref-type="fig"}F). Thus, the strong detrimental effects of the C-insertion mutation in the 3′ hairpin of snR10 is suppressed by introducing a mutation into the 7-nt processing element in the 5′ hairpin that blocks the normal processing function of snR10. These results indicate that the strong deleterious effects on growth and processing caused by the C-insertion mutation depend on the presence of the wild type processing element in the 5′ hairpin. The conclusion that the processing element and modification guide motif act in cis is further supported by the finding that co-expression of H10/36 and H36/10 + C mutants in snR10 depleted cells did not cause growth defects ([Figure 9](#F9){ref-type="fig"}G and H). The nature of the interference effects indicates cooperative interactions between the processing and modification domains of this snoRNA. DISCUSSION ========== The novel processing element ---------------------------- The 7-nt element in the 5′ hairpin required for normal rRNA processing is present in a single-stranded region and processing depends on the sequence of the element, but not the nearby secondary structure features. This element partially overlaps (4 nt) an 8-nt sequence complementary to pre-rRNA, suggesting that this portion and the adjoining non-essential complementary nucleotides interact with pre-rRNA. Although not all 7 nt are complementary to pre-rRNA, the element could mediate processing through partial binding to rRNA or other means, such as interacting with a protein or another RNA sequence required for activity. The DMS chemical probing data seem to argue against stable binding of the 7-nt element with a partner molecule(s), since it is accessible to DMS modification. A transient interaction for this region of the snoRNP is also possible. Presently, we favor the possibility that the 7-nt element mediates processing by direct contact with pre-rRNA, like other processing snoRNAs. Among the snoRNAs with this function, three others bind to rRNA through sequence elements required for processing activity. These include: (i) the H/ACA processing snoRNA snR30/U17 with two essential conserved motifs of 9 and 7 nt ([@B34],[@B47]), (ii) the C/D snoRNA U3 with two essential sequences (10 and (iii) 7 nt) ([@B3],[@B31],[@B40]) the C/D snoRNA U14 with an essential 13-nt element ([@B59]). Mutational results for U14 strongly infer that this element and a 14 nt methylation guide element function together in rRNA folding (see below) ([@B28]). Any one of these snoRNAs, including snR10, could mediate processing through base-pairing with pre-rRNA, for example, by assisting in an autocatalytic cleavage reaction, or recruiting a trans-acting nuclease. The present mutational analysis also shows that loss of the 3′ hairpin or the Ψ modification it guides does not affect rRNA processing. However, a C-insertion mutation in the guide domain that causes Ψ formation to be targeted to an adjacent rRNA site strongly impairs processing, especially for 18S rRNA. This same mutation also causes accumulation of an abnormal, 50S pre-rRNP complex that is likely a precursor of the 40S subunit. These findings suggest the structure of the snoRNP--pre-rRNP complex is altered by the C-insertion mutation and that these changes are the likely basis of the interference in processing (see below). The processing defect caused by the C-insertion mutation does not result from the modification shift ---------------------------------------------------------------------------------------------------- The results argue that the shift in Ψ position caused by the C-insertion is not the basis of the strong processing defects. In particular, two hybrid snoRNAs that cause the same shift do not have strong processing and growth defects (M36/10 + C and M2 + C). Notably, the levels of modification at the novel site are comparable for the three mutant snoRNAs with the same C-insertion. This situation argues that the processing defect stems from altered interaction of the dual-function snR10 snoRNP with pre-rRNA, rather than the change in modification site or yield. For example, altering the sequence in the guide domain could alter (or create) binding of a protein or interacting RNA (rRNA or other). Such an effect could influence the processing activity of the snoRNP, perhaps by interfering with the function of the 5′ processing domain. It seems unlikely, however, that the interference in processing is due solely to altered binding properties of the snoRNA guide region. Arguing against that possibility is the finding that two mutant snoRNAs with the C-insertion and a second mutation do not exhibit the strong processing defects seen for the C-insertion mutation alone. More likely, we propose, is that the severe defect is caused by a strong, negative cooperative effect involving interactions of the mutated modification region and the 5′ processing region with pre-rRNA, directly or indirectly. Functional links between the processing and modification domains ---------------------------------------------------------------- Our functional mapping shows that the processing and modification functions of snR10 reside in distinct structural domains that can be separated genetically. However, we also show that a point insertion mutation in the 3′ hairpin severely impairs processing and that this effect depends on the presence of the 7-nt processing element in the 5′ hairpin. Mutating the processing element suppresses the severe growth defects caused by the C-insertion mutation in the guide domain, revealing that the modification and processing domains have a cooperative relationship. In addition, co-expression of hybrids H10/36 and H36/10 + C did not cause growth defects, suggesting that the two domains act in cis. An especially interesting possibility is that the two domains contact pre-rRNA simultaneously or in a tightly coupled fashion. In this context, snR10 could serve as a bridging factor to bring together or stabilize precursor forms of the small and large ribosomal subunits in a common complex. Also attractive is the possibility that such interactions could coordinate the modification and processing reactions to some benefit. A cooperative relationship between snoRNA processing and modification elements has been observed for the universal U14 snoRNA. This snoRNA (a C/D species) contains two long sequences (13 and 14 nt) complementary to 18S rRNA; one required for rRNA processing and the other for 2′-O-methylation ([@B46],[@B59]). Our laboratory has shown that a point mutation either in the processing element or in the modification guide domain has no effect on cell growth, but when combined cell growth is severely impaired ([@B28]). This situation suggests that the functions of the two snoRNA domains are coupled. That interpretation is supported by the fact that the complementary sequences in rRNA are in close proximity in an internal loop in the folded secondary structure, but well separated in the linear sequence. Simultaneous or tightly coupled binding to these sites was suggested as a possible basis for the cooperativity between these binding sites, although other interpretations are also possible. Coordination of snoRNA functions at different rRNA sites also seems to be formally possible for snoRNAs with two modification guide domains and no known involvement in processing. In S. cerevisiae, 76 snoRNAs that act on rRNA have been identified and this is thought to be the full complement of such snoRNAs. About 30% of these snoRNAs contain two guide domains that target modifications at more than one site, in the same or different rRNAs ([@B10]). While the guide regions in these snoRNAs may act independently, it is also possible that the modification events are linked in some fashion and that such coordination could be an important feature in ribosome synthesis. The strong interference effects described here for the snR10 snoRNA suggest that this species may have special promise in dissecting the mechanism of action for a snoRNA that mediates both processing and modification. SUPPLEMENTARY DATA ================== [Supplementary Data](http://nar.oxfordjournals.org/cgi/content/full/gkq043/DC1) are available at NAR Online. FUNDING ======= National Institutes of Health (GM19351 to M.J.F.). Funding for open access charge: National Institutes of Health (GM19351). Conflict of interest statement. None declared. Supplementary Material ====================== ###### \[Supplementary Data\] We are grateful to Ben Liu for help with early experiments, Wayne Decatur for help with the article and all laboratory members for stimulating discussions. [^1]: Present addresses: Xue-hai Liang and Qing Liu, Core Antisense Research, ISIS Pharmaceuticals, 1896 Rutherford Road, Carlsbad, CA 92008, USA. [^2]: Quansheng Liu, Skirball Institute of Biomolecular Medicine, NYU School of Medicine, New York, NY 10016, USA. [^3]: Thomas H. King, GlobeImmune, Inc. 1450 Infinite Drive, Louisville, CO 80027, USA.	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ Vinyl chloride (VC) -- a proven human carcinogen [@pgen.1000714-Kielhorn1] -- and other chloroethenes, such as trichloroethene (TCE) and tetrachloroethene (PCE), are among the most frequently detected groundwater contaminants in the United States of America and other industrialized countries [@pgen.1000714-McCarty1]. Some members of a deeply branching Chloroflexi subphylum, the Dehalococcoides (Dhc), exhibit the unique ability to completely reduce these chloroethenes to ethene via VC as intermediate [@pgen.1000714-MaymGatell1], thereby mediating a critical step in bioremediation of contaminated aquifers and subsurface environments [@pgen.1000714-McCarty1]. Dhc are strictly anaerobic microorganisms with a highly specialized catabolism that is apparently restricted to organohalide respiration with molecular hydrogen as electron donor [@pgen.1000714-MaymGatell1]--[@pgen.1000714-He1]. Despite some successful exploitation of Dhc activity for bioremediation, exploration of Dhc biology has been limited due to slow growth (doubling times in the laboratory between 19 hours [@pgen.1000714-MaymGatell1] and 57 hours [@pgen.1000714-He2]), low per-cell biomass, as well as the absence of techniques for genetic manipulation. Organohalide respiration in Dhc is catalyzed by heterodimeric, membrane-bound enzymes of about 500 aa in length, known as 'reductive dehalogenases' (RDases). The catalytically active 'A' subunit is believed to be anchored to the outside of the cytoplasmic membrane by a small (∼100 aa) predicted integral membrane 'B' subunit [@pgen.1000714-Smidt1]. Two Dhc genome sequences of isolates that cannot respire VC revealed many full-length non-identical [r]{.ul}eductive [d]{.ul}ehalogenase [h]{.ul}omologous genes (rdhAB) per genome; 17 in D. ethenogenes strain 195 [@pgen.1000714-Seshadri1] and 32 in Dhc strain CBDB1 [@pgen.1000714-Kube1]. These genome sequences revealed a bias in the location of rdhAB and associated genes toward the origin of replication (Ori) and the leading strand [@pgen.1000714-Seshadri1],[@pgen.1000714-Kube1]. Several of these rdhAB were also found to be located nearby or within integrated elements (IEs) [@pgen.1000714-Regeard1] or have a highly unusual codon bias [@pgen.1000714-McMurdie1] indicating possible horizontal acquisition. Culture-based studies have shown that strains 195 and CBDB1 contribute to dechlorination of a variety of priority pollutants including polychlorinated ethenes [@pgen.1000714-Magnuson1], benzenes [@pgen.1000714-Fennell1], phenols [@pgen.1000714-Adrian2], dibenzo-p-dioxins [@pgen.1000714-Bunge1], dibenzofurans, biphenyls, and naphthalenes [@pgen.1000714-Fennell1]. Notably, strain CBDB1 does not respire chloroethenes [@pgen.1000714-Adrian1], and strain 195 cannot respire VC, instead exhibiting a slow cometabolic VC reduction activity [@pgen.1000714-Magnuson2]. Efforts to isolate microorganisms that can couple growth with VC dechlorination have resulted in the isolation of Dhc strains VS [@pgen.1000714-Rosner1] and BAV1 [@pgen.1000714-He3], both of which share highly similar (\>99% identical) 16S rRNA gene sequences with strains 195 and CBDB1. The VC RDases responsible for catalyzing VC transformation to ethene have been identified biochemically in strain VS [@pgen.1000714-Mller1], and are encoded by vcrAB in strain VS [@pgen.1000714-Mller1] and bvcAB in strain BAV1 [@pgen.1000714-KrajmalnikBrown1]. To further understand Dhc genome organization and the genetic adaptations that led to VC respiration, we determined the complete genome sequence of Dhc strains VS and BAV1. To our knowledge, these are the first genome sequences of microorganisms able to grow by reductive dehalogenation of VC, a critical step for bioremediation of sites impacted by chlorinated ethenes [@pgen.1000714-Kielhorn1]. Comparison between these VC respiring Dhc strains revealed that the VC RDase genes, vcrAB and bvcAB, are each located on distinct genomic islands (GEIs) at disparate locations of their respective genome, suggesting that the genetic basis for VC reduction was horizontally acquired in both strains through independent events. We also show that the integration site for the vcr-GEI is not unique to this GEI nor to strain VS, but instead appears to be an integration site associated with many other strain-specific rdhA in strains VS, 195, and CBDB1. Similar to comparative analyses between other closely related, free-living bacteria (for example, [@pgen.1000714-Mathee1],[@pgen.1000714-Coleman1]), this four-way comparison of Dhc strains reveals that many strain-specific genes occur in a limited number of continuous segments and GEIs. Unlike those genomes, however, the GEIs and other strain-specific segments of Dhc are further clustered mainly within two regions positionally analogous to the 'less-structured regions' that flank the Ori-macrodomain in Escherichia coli [@pgen.1000714-Valens1]. These [H]{.ul}igh [P]{.ul}lasticity [R]{.ul}egions include most examples of interruptions to core synteny between the four strains, including genomic rearrangements, IS elements and other repeated elements, insertions and deletions, as well as rdhAB. These HPRs reflect rapid evolutionary dynamics toward a presumed respiratory-niche diversification that contrasts an otherwise remarkably small and stable genome that, at 1.3--1.5 Mbp, is among the smallest of known free-living microorganisms [@pgen.1000714-Giovannoni1]. This compartmentalization of genome dynamics to catabolism-associated Ori-flanking HPRs indicates an unusual biological solution to the opposing evolutionary pressures for genome streamlining and respiratory diversification. Results/Discussion {#s2} ================== Features and Organization of Strains VS and BAV1 Genomes {#s2a} -------------------------------------------------------- The genomes of strains VS and BAV1 are similar in total length (1413462 and 1341892 bp, respectively), % (G+C) (47.3% and 47.2%, respectively), as well as overall structure ([Figure 1](#pgen-1000714-g001){ref-type="fig"}, [Figure S1](#pgen.1000714.s001){ref-type="supplementary-material"}) to those of strains 195 and CBDB1 [@pgen.1000714-Seshadri1],[@pgen.1000714-Kube1]. There are 1029 orthologous groups of protein encoding genes (CDS) that are conserved across all four Dhc genomes, henceforth referred to as the core genome. The core genome accounts for 68 to 77% of a strain\'s genome ([Table S1](#pgen.1000714.s010){ref-type="supplementary-material"}) and, generally speaking, these genes share the same order, orientation and genomic context (synteny) and encode the essential metabolic functions ([Table S2](#pgen.1000714.s011){ref-type="supplementary-material"}) previously described for strains 195 and CBDB1 [@pgen.1000714-Seshadri1],[@pgen.1000714-Kube1]. GC-skew maps ([Figure S2](#pgen.1000714.s002){ref-type="supplementary-material"}) identified a consistent predicted location for the origin and termini of replication of each Dhc strain and also revealed separate inversions in strains VS and BAV1 that are further supported by a corresponding disruption to gene synteny ([Figure 1](#pgen-1000714-g001){ref-type="fig"}, [Figure S1](#pgen.1000714.s001){ref-type="supplementary-material"}). Genome-level relationships were estimated by concatenation and multiple alignment of the core genome of each strain. The resulting phylogenetic tree is consistent with a previous 16S rRNA gene-based structure [@pgen.1000714-Hendrickson1] ([Figure S3](#pgen.1000714.s003){ref-type="supplementary-material"}). The core genomes of the two strains from the Pinellas phylogenetic subgroup, BAV1 and CBDB1, were found to be extremely similar with a median nt identity of core CDS \>99%. By contrast, strain 195 (Cornell subgroup), strain VS (Victoria subgroup), and the Pinellas subgroup are separated by comparable Jukes-Cantor genetic distances. Disruptions to gene synteny occur predominately within two regions on either side of the Ori ([Figure 1](#pgen-1000714-g001){ref-type="fig"}). These regions are variable in length (up to 200 kbp) and contain elevated occurrences of genomic islands (GEIs), insertion sequences (IS), other repeated elements, as well as apparent insertions, deletions and inversions. These regions account for less than one-fourth the cumulative length of the four genomes but contain 91 of 96 rdhA. We designated these two High Plasticity Regions, HPR1 and HPR2, and identified tRNA genes at their conserved boundaries ([Figure 1](#pgen-1000714-g001){ref-type="fig"}, [Figure 2](#pgen-1000714-g002){ref-type="fig"}). ![Circular map of Dehalococcoides (Dhc) genomes using strain VS as template.\ Genomic coordinates are shown in 100 kbp intervals beginning with the origin of replication. The rings depict from outside to inside: (Ring 1) Codon usage bias (CUB) [@pgen.1000714-Karlin1] of each gene larger than 200 codons, shaded in greyscale according to their CUB value for emphasis, with black indicating highest CUB value. (Ring 2) Location of genes of special interest, shaded according to the following legend: (red) rdhA and rdhB, (blue) other rdh-associated, (dark yellow) hydrogenase, (violet) rRNA, (dark green) ssrA, (dark orange) tRNA. (Rings 3--6) Ortholog multiplicity of genes. Genes with orthologs in 3, 2, 1, or 0 of the other strains are shaded green, orange, yellow and black, respectively. Genes are drawn in Rings 3--5 only if orthologs to strain VS were detected in strain BAV1, strain 195, or strain CBDB1, respectively. Ring 6 shows the position and orientation of all predicted CDS in strain VS, shaded according to ortholog multiplicity. Also shown in Ring 3 is the location of IS elements (purple transecting bar). Genes in the outer or inner half of Rings 3--6 are in the forward or reverse direction, respectively. Grey wedge highlights are layered behind Rings 2--6 to indicate the positions of the high plasticity regions, HPR1 and HPR2, with small axis breaks also denoting these boundaries. Repeated elements overlapping any part of the genes of special interest in Ring 2 (excluding H2ase genes) are drawn as connected bezier ribbons, shaded according to the same legend, and with their radial span at each end indicating the length of the respective element.](pgen.1000714.g001){#pgen-1000714-g001} ![Scaled map of homologous in both HPR regions of each strain.\ All but two rdhA (DET0876, DehaBAV1_0847 -- bvcA) are depicted here as black bottom-side tick marks. Large orthologous regions are shaded in the same color, in a gradient that indicates their relative orientation. Unfilled (white) boxes indicate strain-specific regions. Brown and black curves indicate less than 0.05 and 0.01 respective Jukes-Cantor distance for orthologous rdhA, or less than 0.10 and 0.01 Jukes-Cantor distance between rdhA paralogs, respectively (see [Figure S7](#pgen.1000714.s007){ref-type="supplementary-material"}). Unless otherwise noted, a commutative property is implied for rdhA orthologs/paralogs that are not on adjacent strains in the diagram. Small top-side green ticks indicate the location of intergenic ssrA (tmRNA gene) repeated fragments. A fatter green tick mark indicates ssrA, iteslf. tRNA genes are indicated with red top-side tick marks and are accompanied with the abbreviated labels, "tR-Aaa." Repeated fragments of tRNA genes are labeled 'fAaa' or connected to the presumed source gene with a red curve. A large black oval indicates the only CRISPR region detected in these genomes, in strain CBDB1 [@pgen.1000714-Kube1]. A triangle, oval, diamond, rectangle, and curved rectangle are used to indicate the location and size of the fdh, hym, hup, hyc, and trp operons, respectively. An upside-down triangle indicates a homologous recombinase-containing gene cluster apparently involved in ssrA-specific integration, and is present in VS, CBDB1, and BAV1 (see [Figure S9](#pgen.1000714.s009){ref-type="supplementary-material"}). A wide middle box labeled "Ori core" represents approximately 130 kbp of synteny-supported, core genome that separates the two HPRs and includes the Ori. The chromosomal location of the edges of these regions are indicated with flags.](pgen.1000714.g002){#pgen-1000714-g002} High-Plasticity Regions (HPRs) {#s2b} ------------------------------ HPR1 begins following tRNA-Ala-1 approximately 60 genes forward from the Ori and ends at tRNA-Ala-2 approximately 200 genes downstream ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). HPR1 of each strain contains examples of genomic rearrangements and GEIs that appear to have integrated at tRNA-Ala. Strain 195 also contains previously described GEIs that integrated at tRNA-Val as well as a repeated GEI that integrated at tRNA-Lys [@pgen.1000714-Seshadri1]. Recently repeated ISDsp2 and ISDsp1 or ISAli5 IS elements can be found in HPR1 of strains VS and BAV1, respectively. Additionally, two non-identical ISDet2 type IS elements are located in HPR1 of strain 195. Whole genome alignment, the order and orientation of orthologous genes ([Figure S1](#pgen.1000714.s001){ref-type="supplementary-material"}), as well as GC-skew analysis ([Figure S2](#pgen.1000714.s002){ref-type="supplementary-material"}) show that the majority of HPR1 in strain BAV1 was recently inverted, and also that at least one inversion has occurred in HPR1 of strain VS. These are the only inversions \>40 kbp in length detected in the four strains. HPR1 includes a total of 30 rdhA, 7 of which are strain specific. This region also includes ∼100 kbp of core genes with occasionally interrupted order, many of which are believed to be involved in key catabolic functions, including the hup operon (encoding \[NiFe-(Se)\] uptake hydrogenase), the hym operon (encoding \[Fe\] hydrogenase), and the fdh-like operon that is strongly expressed in dechlorinating Dhc cultures [@pgen.1000714-Morris1]. Also within this region is the only conserved full-length rdhA with an orthology status that is also supported by synteny in all four strains (DehaBAV1_0173, cbdb_A187, DET0180, DhcVS169, [Figure 2](#pgen-1000714-g002){ref-type="fig"}). HPR2 begins with a ∼30 kbp region of atypically high similarity (\>99% nt identity) between strain 195 and the Pinellas strains following a conserved set of three tRNA genes (tRNA-Leu,Arg,Val). A phylogram based on a multiple alignment of this region is incongruent with the phylogeny estimated for the rest of the genome ([Figure S3](#pgen.1000714.s003){ref-type="supplementary-material"}, [Figure S4](#pgen.1000714.s004){ref-type="supplementary-material"}), suggesting a xenologous displacement of this region in strain 195, sourced from a Pinellas strain. Probes for this region of strain 195 were among the only highly conserved oligonucleotides that failed to hybridize to genomic DNA from ANAS, a culture highly enriched in different Cornell strains [@pgen.1000714-West1],[@pgen.1000714-Holmes1]. The dubious evolutionary history of this region and its proximity to rdhAB in HPR2 is especially interesting because it includes the biosynthesis operon for tryptophan, a significantly enriched [@pgen.1000714-Kube1] and positionally conserved [@pgen.1000714-KrajmalnikBrown1] residue in RdhB sequences ([Figure S5](#pgen.1000714.s005){ref-type="supplementary-material"}). This ∼30 kbp region, potentially displaced in strain 195, is followed by a region with the highest density of rdhA in the genomes (discussed below). Following this, the terminal ∼40 kbp of HPR2 is a syntenic region shared between strains CBDB1 and VS, and includes six orthologous pairs of rdhA ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). The last 12.5 kbp of this region is also syntenic in strain 195, encompassing two rdhA orthologous triplets (DhcVS1430/cbdb_A1627/DET1538; DhcVS1436/cbdb_A1638/DET1545). The final rdhA gene of HPR2 has syntenic representatives in all four strains (DehaBAV1_1302, cbdb_A1638, DET1545, DhcVS1436), but this rdhA gene is present in BAV1 only as a N-terminal fragment that stops abruptly at the boundaries of an ISDsp1 type IS element. This genome truncation in strain BAV1 (CBDB1 equivalent locations 1174623--1326022, 151 kbp) accounts for its lower total number of rdhA and explains the shorter overall length of the BAV1 genome. Similarly, a deletion event appears to have taken place in the strain 195 genome between two rdhA (corresponding to DhcVS1399 and DhcVS1427), leaving behind the apparently chimeric rdhA, DET1535 ([Figure 2](#pgen-1000714-g002){ref-type="fig"}, [Figure S6](#pgen.1000714.s006){ref-type="supplementary-material"}). Reductive Dehalogenase Homologous Genes {#s2c} --------------------------------------- rdhAB and genes believed to be involved in assembly and maturation (rdhF-I) or regulation (rdhC, D, R) [@pgen.1000714-Kube1] comprise between 3.5 and 8.6% of these genomes by length. Strain VS contains 36 full-length rdhA, the most of any genome to date and the most unique (15 rdhA) among the four Dhc strains. While 32 of the 96 rdhA are unique to an individual strain, the remaining genes have at least one predicted Dhc ortholog. Most ortholog pairs are present in the same HPR and supported by local synteny ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). This emphasizes that many rdhA have been vertically inherited or horizontally acquired en bloc from another Dhc, consistent with a previous observation that many rdhA have a codon usage that is indistinguishable from the rest of the genome [@pgen.1000714-McMurdie1]. Furthermore, available rdhA cluster into two major phylogenetic clades, the largest of which (Cluster 1) contains only Dhc-derived rdhA. Cluster 1 also includes the VC RDase genes bvcA and vcrA, the TCE RDase gene tceA, the PCE RDase gene, pceA [@pgen.1000714-Magnuson1],[@pgen.1000714-Fung1], the chlorobenzene RDase gene, cbrA [@pgen.1000714-Adrian3], and a total of 85 rdhA. Notably, Cluster 2 contains two of the Dhc core rdhA groups, in addition to all presently available non-Dhc rdhA ([Figure 3](#pgen-1000714-g003){ref-type="fig"}, [Figure S7](#pgen.1000714.s007){ref-type="supplementary-material"}). Though obscured somewhat by rearrangements and HGT, as many as 19 contemporary rdhA orthologous groups may have been present in the most recent common ancestor ([Figure 2](#pgen-1000714-g002){ref-type="fig"}, [Figure S8](#pgen.1000714.s008){ref-type="supplementary-material"}). Overall, HGT between Dhc strains, horizontal transfer from non-Dhc, interruption, and deletion all appear to contribute to the evolution and availability of rdhA in Dhc. ![Phylogenetic overview of available RdhA sequences.\ Radial Neighbor-Joining consensus tree of the deduced amino acid sequence from all full-length rdhA available in public databases. Leaves represent RdhA derived from Dehalococcoides (Dhc) unless otherwise indicated. RdhA sequences from Dhc strain FL2 and Dhc-containing enrichment culture KB1 are included. Orthologous groups with a representative in all 4 presently sequenced Dhc strains (core) are labeled and shaded in light blue as in [Figure 1](#pgen-1000714-g001){ref-type="fig"}. RdhA from non-Dhc microorganisms are labeled with the following abbreviations: Dehalob Dehalobacter.; Desulfitob Desulfitobacterium; Photobac Photobacterium; Sulfurosp Sulfurospirillum; Shew Shewanella. A vertical tree with comprehensive leaf labels is provided in [Figure S7](#pgen.1000714.s007){ref-type="supplementary-material"}.](pgen.1000714.g003){#pgen-1000714-g003} Horizontal Transfer of Key RDase Genes {#s2d} -------------------------------------- In strains VS and BAV1, the respective VC RDase-encoding operons are located on distinct GEIs ([Figure S9](#pgen.1000714.s009){ref-type="supplementary-material"}). In BAV1, bvcAB is one of only two rdh operons outside of the HPRs, and it is embedded in a GEI that is flanked on either side by a tRNA-Arg and a 58 bp directly repeated fragment of its 3′ end. This GEI has an unusually low %(G+C) and also contains a putative integrase, DehaBAV1_0846 [@pgen.1000714-McMurdie1]. Similarly, vcrAB of strain VS is embedded within a low %(G+C) GEI in HPR2 that is flanked by ssrA -- a single-copy essential gene [@pgen.1000714-Mao1] encoding transfer messenger RNA (tmRNA) [@pgen.1000714-Keiler1] -- and its 20 bp [d]{.ul}irect [r]{.ul}epeat (DR). This GEI (positions 1177831--1189175) also contains a putative integrase (DhcVS1282) gene as well as two other genes associated with recombination (DhcVS1283, DhcVS1286), consistent with the canonical description of ssrA-specific integration of GEIs described for other bacterial genomes [@pgen.1000714-Williams1]. This vcrA-containing, ssrA GEI (heretofore vcr-GEI) is followed immediately by another apparently ssrA-specific GEI, the first six genes of which (DhcVS1292-1298) comprise an integration-associated block that is homologous to the six genes immediately adjacent to ssrA in vcr-GEI ([Figure S9](#pgen.1000714.s009){ref-type="supplementary-material"}). The distal boundary of this downstream tandem ssrA GEI is defined by a 19 bp ssrA DR, and the remaining genes have no detectable similarity to vcr-GEI. Syntenic homologs to these six genes are also adjacent to ssrA in CBDB1 (cbdb_A1480-A1486), and three of these are at the corresponding location in BAV1 as well (DehaBAV1_1297-1299). The ssrA DR is found a total of 11 times downstream of ssrA in strain VS, 5 times in strain CBDB1, and once in strain 195 ([Figure 1](#pgen-1000714-g001){ref-type="fig"}, [Figure 2](#pgen-1000714-g002){ref-type="fig"}), and is further supported by traces of sequence similarity in the 150 bp of intergenic sequence surrounding each DR. In total, 19 of the 32 strain-specific rdhA in these four genomes are colocated with these ssrA fragments in strains CBDB1 and VS ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). The relative abundance of rdhA in strains VS and CBDB1 is largely attributable to this segment of HPR2 (12 in CBDB1, 18 in VS), further implicating ssrA-specific integration as a key mechanism of novel rdhA acquisition in Dehalococcoides. The tRNA-Ala gene also appears to be a recombination site for one or more strain-specific GEIs that contain rdhA and phage-associated genes. HPR1 of BAV1 contains four repeated 3′ fragments of tRNA-Ala-1 in close proximity to rdhA, and two were identified in HPR1 of CBDB1. One of the fragments in CBDB1 flanks a GEI (previously entitled Region 1 [@pgen.1000714-Kube1]) that contains 4 rdhA, 2 of which are unique to CBDB1. In strain 195, both IE1 (which includes tceA) and IE9 also appear to have integrated at tRNA-Ala [@pgen.1000714-Seshadri1]. In strain VS, a 25 bp DR from tRNA-Ala-1 (DhcVS62) is accompanied by a highly similar truncated homolog (DhcVS107) of site-specific recombinase gene DhcVS63. This gene and tRNA-Ala DR flank an inversion within HPR1 (∼57500--103500) that includes 5 rdhA (1 unique, DhcVS82) and two ISDsp2 type IS elements. Similarly, two of the tRNA-Ala-1 repeated fragments in HPR1 of BAV1 form a pair of inverted repeats at the approximate boundary to the ∼100 kbp inversion in BAV1 ([Figure 1](#pgen-1000714-g001){ref-type="fig"}). The observation that tRNA-Ala genes are found at each boundary of HPR1, as well as the downstream boundary of HPR2, suggests that tRNA-Ala is also a key recombination site for acquisition of novel genetic elements in Dehalococcoides. In addition to the VC reductase genes, all other currently identified chloroethene reductive dehalogenase genes appear to occur on GEIs. Unlike the pceAB-containing catabolic transposon (Tn-Dha1) found in Desulfitobacterium [@pgen.1000714-Maillard1],[@pgen.1000714-Nonaka1], we did not detect flanking IS elements or any other flanking repeats at the island containing pceAB in HPR1 of strain 195. We further observed that this island is homologous to sequence located within a larger region in HPR2 that is contextually conserved between CBDB1 and VS ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). This pceAB island in strain 195 therefore appears to be the result of an intragenomic rearrangement between the HPRs, or a horizontal acquisition from another closely related microorganism. DNA microarray hybridization failed to detect pceA in three other Cornell strains, namely the two strains in the ANAS enrichment [@pgen.1000714-West1],[@pgen.1000714-Holmes1] as well as the recently described strain MB that nevertheless dechlorinates PCE [@pgen.1000714-Cheng1]. Although highly similar (aa ID, 94.5% DhcVS1393, 93.7% cbdb_A1588), the function of the corresponding homologs to pceA in CBDB1 and VS is unclear, as neither CBDB1 [@pgen.1000714-Adrian1] nor VS [@pgen.1000714-Cupples1] appears capable of PCE respiration, and PceA from strain 195 was implicated as a bifunctional enzyme that also catalyzes the reductive dehalogenation of 2,3-dichlorophenol [@pgen.1000714-Fung1]. The typical codon usage bias of all three pceA homologs [@pgen.1000714-McMurdie1], as well as the conserved context of these genes in strains VS and CBDB1, suggests that these genes are not a recent addition to Dhc. Despite the apparent specialization of Dhc to organohalide respiration, the question as to whether or not certain rdhA are essential for Dhc remains unanswered. This is partly due to the high number and diversity of Dhc rdhA, their rampant horizontal transfer, as well as the genetic intractability of Dhc [@pgen.1000714-Wagner1]. The three orthologous groups of rdhA with members in all four strains are obvious candidates, but only one of these groups is supported by synteny -- albeit local and inverted in BAV1. An additional 3 or more rdhA orthologous groups would have been classified as synteny-supported core genes, were it not for their deletion in HPR2 of strain BAV1, including the orthologous group containing DET1545 ([Figure 2](#pgen-1000714-g002){ref-type="fig"}). Interestingly, DET1545 was found to be among just four rdhA strongly upregulated during the transition from exponential growth to late stationary phase in strain 195 with TCE as sole depleting electron acceptor [@pgen.1000714-Johnson1]. Among the remaining 3 strongly upregulated rdhA upon entry to stationary phase, one is core supported by synteny (DET0180), and another (DET1535) is predicted to have been syntenic-core prior to the deletion of its ortholog in strain BAV1. The synteny-supported core rdhA in CBDB1, cbdbA187, was shown to be differentially upregulated during respiration of 1,2,3-trichlorobenzene (TCB) compared with 1,2,4-TCB, although transcripts of cbrA were most abundant in both conditions [@pgen.1000714-Wagner1]. While the specific activity of the small subset of core rdhAB is unknown, their elucidation promises key insight into the biology of Dehalococcoides. The viability of strain BAV1, in spite of its loss of many otherwise conserved rdhAB in HPR2, is further evidence of a modular character of RdhAB activity. Taken in combination with the observation of high genome-wide similarity between strains CBDB1 and BAV1, as well as the phenotypic observation that BAV1 respires vinyl chloride -- while CBDB1 does not -- suggests that acquisition of a VC RDase operon is sufficient to confer VC respiration in Dhc. Concluding Remarks {#s2e} ------------------ We describe the emergence of a genomic structure in which rdhA and associated genes, as well as other strain-specific genes are concentrated predominantly within two HPRs near the Ori. Analogous compartmentalization of the chromosome into HPRs has been reported for diverse microbes such as Streptomyces [@pgen.1000714-Choulet1], Borrelia [@pgen.1000714-Casjens1], and Haloquadratum [@pgen.1000714-CuadrosOrellana1], and may facilitate adaptation while maintaining stability of the core genome [@pgen.1000714-Rocha1],[@pgen.1000714-Rocha2]. Dhc HPRs are at least partially explainable by their colocation with common integration sites for genetic elements [@pgen.1000714-Williams2], the most prominent of which are the structural RNA genes for tmRNA and certain tRNAs. A locally elevated concentration of GEIs, repeated elements, homologous genes, and other features associated with recombination may further contribute to any intrinsic instability of these regions. The chromosomal clustering of rdhAB into domain-like HPR structures may also facilitate the observed coexpression of multiple rdhA in response to limited or single electron acceptors [@pgen.1000714-Fung1],[@pgen.1000714-Wagner1],[@pgen.1000714-Johnson1],[@pgen.1000714-Rahm1],[@pgen.1000714-Waller1] by allowing improved access of RNA polymerase to nearby exposed DNA -- as has been proposed as a general explanation for short (≤16 kbp) and medium (∼100 kbp) range expression correlation patterns in some bacteria [@pgen.1000714-Rocha1]. Any selective advantage by facilitated coexpression also contends with a higher likelihood of losing key rdhAB because they are in regions of high plasticity. The elucidation of extremely small genomes of marine bacterioplankton like Canditatus Pelagibacter ubique [@pgen.1000714-Giovannoni1] and Prochlorococcus [@pgen.1000714-Dufresne1] have led to the emergence of the genome streamlining hypothesis, attributed to purifying selection acting on very large and globally dispersed populations [@pgen.1000714-Lynch1]. Our analysis provides insight into genome streamlining of a free-living microbial subphylum that is specialized for a fundamentally different lifestyle and environment, namely organohalide respiration in anoxic zones of the terrestrial subsurface. At 1371 predicted ORFs, the strain BAV1 genome contains just 17 and 71 additional ORFs than the respective genomes of Cand. P. ubique and OM43 strain HTCC2181 [@pgen.1000714-Giovannoni2], the smallest of known free-living microorganisms. Unlike Cand. P. ubique however, BAV1 and the other Dhc strains do contain pseudogenes, transposons, and IS elements. Their modest % (G+C) and 10-fold larger median length of intergenic spacers also indicate that the selective pressures toward genome reduction have been somewhat different for Dhc, while still allowing a comparably small genome size. The presence of between 11 and 36 rdhAB per genome implies a respiratory flexibility that may allow Dhc to use a variety of halogenated compounds as terminal electron acceptors. With no apparent alternative energy conservation mechanism for Dhc other than organohalide respiration, niche specialization to low abundance, naturally occurring chloroorganic compounds seems to be the dominant ecological strategy of this unique group of microorganisms. The apparent emphasis of Dhc genome dynamics on rdhAB diversity -- compartmentalized within specialized regions near the Ori -- may enhance opportunistic adaptation to (new) respiratory niches while protecting a streamlined core genome that is highly adapted to life in the anoxic subsurface, as evidenced here by the recent site-specific acquisition of vinyl chloride reductase genes. Methods {#s3} ======= Cultivation of Strains and DNA Extraction {#s3a} ----------------------------------------- For isolation of genomic DNA, Dhc sp. strain VS and BAV1 were grown as previously described [@pgen.1000714-He1],[@pgen.1000714-Mller1]. Total DNA was extracted as described previously for strain VS by a freeze-thaw lysis, phenol-chloroform-isoamyl alcohol purification, and ethanol precipitation [@pgen.1000714-Behrens1]. Complete Genome Sequencing and Annotation {#s3b} ----------------------------------------- Whole genome shotgun sequencing (Sanger method) and automated assembly of the genomes of Dhc strains VS and BAV1 was performed at the U.S. Department of Energy\'s Joint Genome Institute (JGI) following their standard production sequencing protocols (<http://www.jgi.doe.gov/sequencing/protocols/>). A combination of randomly sheared libraries with inserts in the 3 kb and 8 kb size range was used for each strain, as well as some 40 kb inserts (fosmid) for strain BAV1. The initial assembly of each genome was constructed with the Paracel Genome Assembler (PGA), with manual correction of possible mis-assemblies by editing in Consed [@pgen.1000714-Gordon1] and gap closure by primer walking. In strain VS, sequences originating from contaminant genomic DNA were excluded by their high %(G+C) bias and low coverage. Scaffolding of strain VS was further enhanced by comparison with the previously sequenced Dhc strains 195 and CBDB1 (CP000027 [@pgen.1000714-Seshadri1] and AJ965256 [@pgen.1000714-Kube1], respectively), and verified with primers designed by Projector2 [@pgen.1000714-vanHijum1] or the Primer3 [@pgen.1000714-Rozen1] implementation in Geneious [@pgen.1000714-Drummond1]. The complete circular consensus sequences of VS and BAV1 achieve 23X and 20X coverage, respectively ([Table S1](#pgen.1000714.s010){ref-type="supplementary-material"}). They are available at JGI\'s Integrated Microbial Genomes (IMG) website (<http://img.jgi.doe.gov> [@pgen.1000714-Markowitz1]), with the respective taxon IDs 641380429 and 640427111 (Genbank CP000688). Computational prediction of open reading frames (ORFs) utilized the output of GLIMMER [@pgen.1000714-Delcher1] and CRITICA [@pgen.1000714-Badger1]. Identification of ORFs unnoticed during automated prediction was performed by manual inspection of intergenic regions using Artemis [@pgen.1000714-Berriman1] and Geneious. Overlapping ORFs without a functional assignment, significant BLASTP hit [@pgen.1000714-Altschul1], or orthologous annotation in the previously sequenced strains were discarded. Functional assignments were created using JGIs automated annotation pipeline, with extensive manual inspection supported by SMART [@pgen.1000714-Schultz1] and KEGG database [@pgen.1000714-Kanehisa1] analyses. Genes for tRNA and tmRNA were detected using tRNAscan-SE and ARAGORN, respectively [@pgen.1000714-Lowe1],[@pgen.1000714-Laslett1]. Identification of Orthologs {#s3c} --------------------------- We identified orthologous relationships between protein encoding genes of these four Dhc strains using the same five heuristic criteria described in the pair-wise comparison of Dhc strains 195 and CBDB1 [@pgen.1000714-Kube1], applied to an all-versus-all BLASTP search incorporating the genes from all four genomes [@pgen.1000714-Altschul1]. A 'greedy' commutative property of ortholog pairs was assumed to create ortholog groups that also contain putative paralogs. Core CDS were defined as those in groups with at least one representative from each of the four genomes. Larger orthologous regions spanning multiple genes were identified at the nucleotide level through manual inspection of multiple whole genome alignments generated by Mauve version 2.2.0 [@pgen.1000714-Darling1] and Muscle version 3.6 [@pgen.1000714-Edgar1] for refinement. Whole-Genome Strain Phylogeny and the Core Genome {#s3d} ------------------------------------------------- For each genome the core CDS were uniformly ordered, oriented, and concatenated as a single nucleotide sequence. A multiple alignment was created for the resulting concatenation of core genes using Muave. Jukes-Cantor phylogenetic distances and a Neighbor-Joining consensus tree were calculated from this multiple alignment using Geneious. Identification of Repeated Elements and IS Elements {#s3e} --------------------------------------------------- An exhaustive determination of repeated elements greater than or equal to 18 bp in length was performed on all four genomes using the repeat-match algorithm in MUMmer3 [@pgen.1000714-Kurtz1]. IS elements and IS-transposases were detected by BLAST and BLASTP searches based on those described for strains 195 and CBDB1 [@pgen.1000714-Seshadri1],[@pgen.1000714-Kube1], as well as manual inspection of the genomic context surrounding significant search hits to IS elements and transposases in the ISFinder database [@pgen.1000714-Siguier1]. Recently repeated IS elements (perfect and nearly-perfect copies) were discovered by manual inspection of all repeated elements of an appropriate size (0.7--2.5 kbp) and comparison to the ISFinder database. RdhA Phylogeny {#s3f} -------------- Tree reconstruction was performed with a total of 152 RdhA sequences, including the 96 RdhAs from the four complete Dhc genomes and 56 RdhAs from other Dhc strains and non-Dhc species available in public databases. Deduced amino acid sequences were aligned with Muscle 3.6. The phylogenetic trees were calculated using the neighbour joining and maximum likelihood methods of the MEGA 4.0 software package [@pgen.1000714-Kumar1] as well as the PHYML [@pgen.1000714-Guindon1] implementation in Geneious. The tree topology was tested by the application of a 20% positional conservatory filter. Stability of the tree topology was further refined by bootstrapping (1,000 replications). Only RdhAs ≥400 amino acids were included. Genome Graphics {#s3g} --------------- Scaled circular representation of the four genomes ([Figure 1](#pgen-1000714-g001){ref-type="fig"}), including Bezier ribbons indicating repeats, was created with the Perl-based package, Circos [@pgen.1000714-Krzywinski1]. Custom algorithms written in R [@pgen.1000714-R1] were used to distill redundant repeat information, and combine multiple repeated elements into a representation with a single (arbitrary) source and many sinks. The stylized representation of orthologous regions and rdhA present in the HPRs ([Figure 2](#pgen-1000714-g002){ref-type="fig"}) was drawn as vector graphics by manual overlay on the multiple whole genome alignment [@pgen.1000714-Darling1] of these regions. Nucleotide skews ([Figure S2](#pgen.1000714.s002){ref-type="supplementary-material"}, top) were calculated according to the Oriloc [@pgen.1000714-Frank1] implementation in the seqinR package [@pgen.1000714-Charif1] of the R language for statistical computing [@pgen.1000714-R1]. Supporting Information {#s4} ====================== ###### Pairwise position of orthologs shared between Dehalococcoides (Dhc) strains 195 (blue), CBDB1 (dark green), and BAV1 (light green) with Dhc strain VS (red; horizontal axis). Unique genes are plotted along the axis (at zero) to which they belong. All genes are shaded according to the respective vertical axis genome. The position of rdhA are indicated with unfilled triangles that point up (forward strand) or down (reverse strand). The location of HPRs is indicated with a box, shaded according to its respective genome. (8.50 MB EPS) ###### Click here for additional data file. ###### Cumulative nucleotide skews along the Dhc strains VS and BAV1 genomes and circular map of strain BAV1. (Top) Shading indicates the specific skew type of each line, provided in the legend. Cumulative Combined Skew (CCskew) is the linear combination of the T-A and C-G skews. The origin of replication (Ori) is at the left and right edges of each dataset because genomic coordinates begin with the predicted Ori. Terminus of replication (Ter) is predicted at the CCskew global minimum. Local (within-replichore) reversals in slope of CCskew indicate genomic inversions, which are further highlighted with grey background. (Bottom) Circular map of strain BAV1 according to the same legend as in [Figure 1](#pgen-1000714-g001){ref-type="fig"}, except Rings 3--6 now correspond to the VS, CBDB1, 195 and BAV1 genomes, respectively. A reversed order of certain orthologs predicted by CCskew is evident in this map as a reversed orientation of orthologs relative to BAV1 (inner or outer half-ring). (3.08 MB EPS) ###### Click here for additional data file. ###### Unrooted phylogram of Dhc strains based on whole genome or 16S rRNA gene data. (Top) Phylogram of the estimated strain-level phylogenomic relationship based on nucleotide sequence of core protein-encoding genes. Core genes were ordered, oriented, and concatenated into a single core sequence for each strain. A multiple alignment of these four concatenated sequences was used to construct the Neighbor-Joining tree. (Bottom) Phylogram based on the 16S rRNA gene from isolated, enriched, as well as environmental Dhc strains. Three previously described phylogenetic subgroups [@pgen.1000714-Hendrickson1] are shaded in blue, red, and green to indicate the Cornell, Victoria and Pinellas subgroups, respectively. These Dhc 16S rRNA genes are all ≥99% identical. (0.88 MB EPS) ###### Click here for additional data file. ###### Similarity distribution of core genes with strain BAV1 as reference. Among core genes only, aa% identity relative to BAV1 is plotted for strains VS and 195. Each point is shaded according to a continuous color map based on the location of the core gene in the genome of strain BAV1, as indicated in the color legend (top-left). Nearly all core genes in strains VS and BAV1 are comparably similar to strain BAV1, except for a small co-localized group of genes (rounded box) that are all \>98% identical between strains 195 and BAV1. These genes are part of an orthologous region containing the tryptophan synthesis operon (trp) that is located from the start of HPR2 to ssrA in all four strains (loci DET1465-1506, DehaBAV1_1266-1296, cbdb_A1437-1479, DhcVS1243-1280, see trp in [Figure 2](#pgen-1000714-g002){ref-type="fig"}). This entire region is unusually similar between strains 195 and the Pinellas strains (See [Table S1](#pgen.1000714.s010){ref-type="supplementary-material"} for coordinates used in alignment). (Inset) A phylogram based on a multiple alignment of this region in the four strains. This phylogram is incongruent with the whole-genome phylogram displayed in [Figure S3](#pgen.1000714.s003){ref-type="supplementary-material"}. (0.94 MB EPS) ###### Click here for additional data file. ###### Tryptophan (W) at conserved positions in Dehalococcoides RdhB sequences. (Top) Alpha helix prediction and transmembrane region prediction in the consensus RdhB. The red, green, and blue lines of the 'TM Prediction Plot' indicate the strength with which the corresponding residue is predicted to be located at the inner membrane, transmembrane, or outermembrane, respectively. These predictions were calculated using a Transmembrane Hidden Markov Model (TMHMM version 0.7) plugin to Geneious, contributed by Mark A. Suchard [@pgen.1000714-Drummond1]. Green horizontal cylinders indicate the location and prediction strength of presumed alpha helices, supporting the hypothesis that these are three distinct membrane-spanning helices. Though shown here only for the consensus sequence, this pattern of TM helices is consistently predicted for individual RdhB of Dehalococcoides. (Bottom) Sequence logo representing the alignment of all Dehalococcoides RdhB in the four genomes. Letter height represents the statistical support (in bits) for a residue at a given position in the sequence. Arrows (very top) emphasize the position of conserved tryptophan residues. (1.40 MB EPS) ###### Click here for additional data file. ###### Three strain multiple alignment of the region surrounding a putative deletion event in Dhc ethenogenes 195. The deletion appears to have occurred intragenically such that the 3′ 1400 bp of rdhA DET1535 are orthologous to the corresponding positions of DhcVS1399/cbdb_A1595, while the 5′ 100 bp are orthologous to corresponding positions in DhcVS1427/cbdb_A1624. The chimeric nature of DET1535 is further supported by local synteny, and the apparent deletion accounts for the absence of these genes in strain 195. Genes are shaded according to the following: rdhAB - red, other rdh associated genes - blue, hypothetical genes - white, other genes for which a functional annotation was assigned - light blue. For concision, annotations are often indicated at only one gene in a vertically aligned (orthologous) group. Locus ID are shown as space permits, with preference given to genes discussed in the main text. Genomic location is labeled above the black horizontal bar of each respective genome, with a much thinner horizontal line indicating a gap. A bar plot of local nucleotide identity is shown above the annotated multiple alignment and shaded to reinforce contrast in values. Green indicates windows of near perfect identity. BAV1 is not shown because this region is absent in BAV1, occurring within a larger apparent deletion in its genome. (0.30 MB PDF) ###### Click here for additional data file. ###### Neighbor Joining consensus tree of the deduced amino acid sequence from all full-length rdhA genes available from six Dehalococcoides (Dhc) strains as well as non-Dhc microorganisms. Genes are labeled by gene ID or accession number and shaded in red, light green, dark green, blue, grey or brown if they are derived from Dhc strains VS, BAV1, CBDB1, 195, other Dhc strains, or non-Dhc, respectively. Bootstrap values are given at the tree nodes. Thin purple, red, and grey vertical lines denote 0.01, 0.05, and 0.10 Jukes-Cantor distances, respectively, cutoff values used in the classification of orthologs, paralogs, and duplications shown in [Figure 2](#pgen-1000714-g002){ref-type="fig"}. Previously assigned orthologous groups [@pgen.1000714-Kube1] are indicated here with the same numeric assignments and expanded to include new members where appropriate. By this nomenclature the three core rdhA are members of groups 10, 11, and 12. These rdhA form two major phylogenetic clusters. Cluster 1 includes only Dhc rdhA and core rdhA group 10. Cluster 2 includes core groups 11 and 12, as well as DhcVS1349, cbdb_A1503, cbdb_A1539, and all non-Dhc rdhA available in the public database. Abbreviations: cp orthochlorophenol; dcp - 3,5-dichlorophenol; dca - dichloroethane; RD - reductive dehalogenase; Desulfitob - Desulfitobacterium; Sulfurosp - Sulfurospirillum; Dehalob - Dehalobacter; Photobac - Photobacterium. (0.60 MB EPS) ###### Click here for additional data file. ###### Approximate reconstruction of the HPRs in the most recent common ancestor (MRCA) of the four contemporary Dhc strains. The phylogenetic tree is taken from the core genome phylogeny shown in [Figure S3](#pgen.1000714.s003){ref-type="supplementary-material"}. The block-representation of the HPRs is taken from [Figure 2](#pgen-1000714-g002){ref-type="fig"}. Each unique color denotes an orthologous region shared between two or more strains. No-fill denotes a region unique to that strain. Black tick marks indicate the location of rdhA genes. A total of 19 rdhA are predicted in the MRCA, but it should be noted that this is the estimated maximum, excluding rdhA that have been lost in all four strains since divergence. (0.78 MB EPS) ###### Click here for additional data file. ###### Vinyl chloride reductase genomic islands (GEIs) of strains VS (bottom) and BAV1 (top). A mauve and green highlight indicates the respective bvcA and vcrA island boundaries with genomic coordinates shown at top-facing tick marks. Genes are shaded according to the following: rdhAB - red, other rdh associated genes - blue, tmRNA-enoding gene, ssrA, and its direct repeat (DR) - purple, tRNA genes and repeated fragments - green, hypothetical genes - white, genes associated with recombination - black, other genes for which a functional annotation was assigned - light blue. A cluster of six recombination associated genes (DhcVS1282-1287) are located immediately following ssrA at the beginning of vcrA-GEI. This cluster has syntenic homologs immediately following the repeated gene fragment at the downstream boundary of vcrA-GEI (DhcVS1292-1298), suggesting a separate, earlier integration of another ssrA GEI. An additional 10 ssrA DRs are interspersed among rdhA downstream (right) of vcrA-GEI. (0.43 MB EPS) ###### Click here for additional data file. ###### Comparison of genomic features for the four Dehalococcoides genomes and their HPRs. Values for strains 195 [@pgen.1000714-Seshadri1] and CBDB1 [@pgen.1000714-Kube1] are from their respective publications. Number of shotgun Sanger reads was obtained from the Ensembl Trace Server (<http://trace.ensembl.org/>). Raw sequencing data was not publicly available for strain CBDB1. However, it was sequenced to a reported coverage of 12× [@pgen.1000714-Kube1], providing for an estimate of total reads with an assumed average read length of 750 bp. "Number repeated elements" refers to perfectly repeated elements greater than 18 bp, detected by an exhaustive repeat search with the repeat-match algorithm in MUMmer3 [@pgen.1000714-Kurtz1]. (0.24 MB PDF) ###### Click here for additional data file. ###### Differences in predicted core metabolism between the four Dehalococcoides genomes. Each column provides the presence of a gene in the respective genome as its locus ID, or "ND" for Not Detected. A brief annotation summary is provided for each core metabolism gene group (row). (0.22 MB PDF) ###### Click here for additional data file. The authors have declared that no competing interests exist. This work was supported by the Strategic Environmental Research Defense Project (SERDP) to AMS and FL by grants ER-1588 and ER-1586, respectively. SH was partially funded by grant NSF-DMS-0241246, JAM was partially funded by grant MSU 52019621187, and PJM was partially funded by fellowship grant FP-91671901 from the U.S. Environmental Protection Agency\'s Science to Achieve Results (STAR) program. Genome sequencing and assembly was performed under the auspices of the U.S. Department of Energy\'s Office of Science, Biological and Environmental Research Program, and by the University of California, Lawrence Berkeley National Laboratory under contract No. DE-AC02-05CH11231, Lawrence Livermore National Laboratory under Contract No. DE-AC52-07NA27344, and Los Alamos National Laboratory under contract No. DE-AC02-06NA25396. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: PJM SFB FEL AMS. Performed the experiments: PJM SFB KMR. Analyzed the data: PJM SFB JAM JG KMR RW EG AL SH FEL AMS. Wrote the paper: PJM SFB JAM FEL AMS.	{ "pile_set_name": "PubMed Central" }
Published: September 11, 2019 Introduction {#sec1} ============ Kinases are key regulators of inter- and intracellular communication, and their inhibitors are critical in targeted therapy and precision medicine ([@bib6], [@bib17], [@bib27]). Therefore, the capability to monitor the dynamics of kinase activity is essential to deepen our understanding of cellular function and could greatly impact rational drug design. Since the complete cataloging of all human kinases by [@bib30], an increasing number of studies have focused on the so-called "kinome;" however, robust methods to determine kinase activation on a kinome-wide level are still lacking. Initial attempts to measure global kinase activation states exploited the interaction of kinases with immobilized unspecific multiplexed inhibitor beads (MIBs), which demonstrated altered binding affinities upon enzymatic activation ([@bib4]). This primed various studies to interpret increased MIB binding affinity as increased kinase activity on a kinome-wide scale ([@bib43], [@bib44]). However, it was later shown that MIBs bind kinases largely independent of their activation status ([@bib37]). In line with this approach, several studies reported the use of biotin-conjugated acyl-nucleotide probes for the enrichment of kinases via their ATP-binding pocket from complex backgrounds ([@bib35], [@bib49]). In combination with targeted mass spectrometry (MS), these methods allow quantification of the expression of \>200 kinases; however, they do not directly deduce kinase activity. Alternative approaches to assess kinome activity involve analyzing large-scale phosphoproteomics datasets for over- or underrepresented sequence motifs that can be linked to known kinase substrates or sequence specificities ([@bib26], [@bib31]). Indeed, we and others have utilized such prediction tools to infer kinase activity profiles ([@bib21], [@bib50]); however, these methods severely suffer in sensitivity due to lack of knowledge on the majority of kinase substrates and thus display a strong bias toward well-studied kinases ([@bib13]). Tackling these limitations requires the analysis of direct markers for kinase activation. The majority of protein kinases are regulated through phosphorylation of their activation loop (T-loop). The well-studied functionality of these phosphorylations, in combination with their high level of evolutionary conservation throughout the entire kinome, facilitates the use of T-loop phosphorylation as a direct probe to assess kinase activation states ([@bib33]). In principle, T-loop phosphorylation can be detected using phospho-specific antibodies, but their availability is limited to a few well-studied kinases. Furthermore, T-loop phosphorylations often remain undetected or cannot be accurately quantified in large-scale phosphoproteomics studies due to the low abundance of the corresponding peptides. Sensitive and accurate characterization of selected phosphorylation events may be achieved by targeted MS approaches. Previous studies have shown the potential of using targeted quantification of proteins or post-translational modifications (PTMs) as a highly specific readout for previously characterized biological functions in yeast ([@bib42]) or human samples ([@bib1]). However, both previous approaches heavily rely on diverse well-characterized or at least empirically established cellular characteristics (e.g., starvation and proliferation). Here, we present a targeted MS method for the system-wide quantification of kinase T-loop phosphorylation levels, providing a direct and unbiased readout of cellular kinase activation states. Specifically, our approach combines selected reaction monitoring (SRM)-MS with highly specific Fe(III)-immobilized metal affinity chromatography (IMAC) phosphopeptide enrichment on an automated platform with parallel phosphopeptide enrichment of up to 96 samples. We demonstrate the general applicability of our strategy by analyzing the activation profiles of 178 kinases in various cell lines, blood platelets, and a breast cancer biopsy sample. Our approach allows rapid assessment of both well-characterized and novel biological processes, ranging from high-throughput diagnosis up to hypothesis-free screening experiments. Results {#sec2} ======= Our method currently encompasses robust molecular assays to accurately quantify 221 phosphorylation sites in the T-loop region of 178 kinases ([Figure 1](#fig1){ref-type="fig"}A; [Table S1](#mmc2){ref-type="supplementary-material"}). This number is primarily based on accessibility of the T-loop phosphopeptides by trypsin and can be expanded by the use of alternative proteases. Here, the reported assays represent roughly one-third of the human kinome ([@bib30]) and will be made available via [Table S2](#mmc3){ref-type="supplementary-material"} and are uploaded in Peptide Atlas ([@bib12]). Key challenges associated with the detection of T-loop phosphorylation in shotgun data sets such as low abundance, unfavorable liquid chromatography (LC)-MS characteristics, and a high prevalence of tyrosine phosphorylations, have been largely solved by exploiting the combination of high-specificity phosphopeptide enrichment with the unparalleled sensitivity of nanoLC-SRM on a triple-quadrupole MS. All assays are available in a survey and a quantification mode ([Figures 1](#fig1){ref-type="fig"}B and 1C). The survey mode allows for a rapid screening for activation states of all 178 kinases in a 2 h analysis, while the quantification mode enables accurate quantification of those kinases observed in the survey scan and includes confident phosphosite localization and a high tolerance for interferences.Figure 1Targeted Quantification of T-Loop Phosphorylation to Determine Kinase Activation(A) Conserved activation sites were determined based on sequence homologies on a kinome-wide scale. A target list was developed containing phosphorylation sites accessible by tryptic digest covering 33% of the human kinome. Heavy-isotope-labeled peptides were synthesized for all entries of the target list and used for nanoLC-SRM assay development.(B and C) Samples were subjected to tryptic digest and automated phosphopeptide enrichment using Fe(III)-IMAC on an Agilent Bravo AssayMap. Heavy-isotope-labeled synthetic phosphopeptides were used as internal standards. (B) The survey mode allows screening for kinase activation states for all 178 kinases in one LC-MS run using a limited number of transitions per peptide. (C) The quantification mode employs more transitions per peptide allowing for accurate quantification of kinase activation states across different conditions, including confident phosphosites localization and high tolerance to interferences.(D) LOD and LOQ values were determined for a subset of phosphopeptides exemplified for the peptide GHL-pS-EGLVTK, representing ERK3. Response rates of the phosphopeptide were determined within a representative phosphopeptide background for known concentrations in triplicate. Variations in the linear regressions were then used to determine LOD and LOQ values according to the formulas LOD = 3S~a~/b and LOQ = 10S~a~/b (b = slope and S~a~ = standard deviation of the intercept). Error bars represent standard deviation observed for triplicate measurements. To gain insight into the sensitivity of the method, we measured a dilution series of \>40 synthetic T-loop peptides spiked into a background of phosphopeptides enriched from a tryptic whole-cell digest. The resulting linear calibrations curves were used to determine the limits of quantification (LOQ) and detection (LOD), ranging from 0.2 to 10 fmol for the LOD and 0.7 to 34 fmol for the LOQ ([Figure 1](#fig1){ref-type="fig"}D; [Table S3](#mmc4){ref-type="supplementary-material"}). Important to mention here is that several kinase T-loops can become doubly phosphorylated, resulting in doubly phosphorylated peptides after tryptic digestion. In current proteomics workflows, detection of these doubly phosphorylated peptides can be hampered because of poor ionization efficiency and inefficient elution after phosphopeptide enrichment. Our final list of targeted kinases comprise clinically relevant kinases with inhibitors approved by the Food and Drug Administration (FDA) such as Met, Abl, Src, BTK, Jak3, and Kit ([@bib16]) ([Table S4](#mmc5){ref-type="supplementary-material"}) as well as numerous kinases classified as understudied ([@bib8]) ([Figure S1](#mmc1){ref-type="supplementary-material"}). As a first assessment, we analyzed the baseline kinome activation state of three different human cell lines: (1) Jurkat cells, an immortalized line of T lymphocytes; (2) PC9 cells, a non-small cell lung cancer cell line; and (3) HEK293T cells, a cell line derived from human embryonic kidney cells. Without any form of stimulation, we were able to cumulatively detect 52 T-loop phosphorylation sites ([Figure S2](#mmc1){ref-type="supplementary-material"}). Because of the highly conserved nature of the kinases' T-loop sequence, the representative tryptic peptides are not always unique. To deal with this ambiguity, we followed the principle of protein grouping ([@bib32]) and refer to these instances as kinase groups throughout this study. For the 52 phosphorylation sites observed, this resulted in 48 kinase groups ([Table S5](#mmc6){ref-type="supplementary-material"}). Unsurprisingly, a large part of the 48 detected kinase groups represented kinases crucial for cell growth under typical culturing conditions, such as cyclin-dependent kinases (CDKs) and mitogen-activated protein kinases (MAPKs) as well as the two abundant kinases PDK1 and GSK3. It is noteworthy that various kinases show cell-type-specific activation states. For instance, both Jurkat and PC9 cells showed an increased activity of Ca^2+^ and diacylglycerol (DAG)-dependent signaling compared to HEK293T cells, with several kinases from the Ca^2+^/calmodulin-dependent protein kinase (CaMK) group and the protein kinase C (PKC) family being detected in their active state. PC9 cells show increased activation of tyrosine kinase family members such as FAK, Met, and the two kinase groups ephrin type-A receptors 3, 4, and 5 (EphA3-4-5) and HCK-Lyn, likely due to elevated tyrosine kinase signaling via increased EGFR activity ([@bib40]). Zap70, on the other hand represents a highly tissue-specific kinase, exclusively expressed in cell types associated with the immune system, including T cells. Accordingly, here, it was exclusively detected in Jurkat cells, confirming the specificity of our approach. To benchmark our approach against two established online tools to predict kinase activity, NetworKIN ([@bib31]) and KEA2 ([@bib26]), we performed a deep phosphoproteomic shotgun experiment on Jurkat cell lysates detecting a total of \>11,600 phosphorylation sites. Both algorithms rely on detecting substrate motifs or known kinase substrates in large shotgun phosphoproteomic experiments, and [Figure 2](#fig2){ref-type="fig"}A depicts NetworKIN scores and KEA2 enrichment results in comparison to abundance values obtained from our targeted T-loop assay ([Table S6](#mmc7){ref-type="supplementary-material"}). Combined, both algorithms predicted activity of 18 out of the 31 kinases measured in our targeted kinase assay, performed using a single nanoLC-MS run on Jurkat cells, confirming the activity of kinases such as GSK3, PKC, PAK4, CDK1, or ERK1/2. As expected, a large number of kinases were exclusive to our targeted approach demonstrating the higher sensitivity of our method. The absence of kinase activity prediction for numerous kinases by both algorithms can be attributed to the lack of knowledge on kinase-substrate relations. This causes the prediction tools to be intrinsically biased toward well-characterized kinases, neglecting the majority of phosphorylation sites (KEA2 incorporated only 800 out of 11,600 phosphosites for the enrichment analysis) ([Figure S3](#mmc1){ref-type="supplementary-material"}). In addition, both prediction tools often disagree, reducing the reliability of the predicted kinase activations.Figure 2Probing Kinase Activation in Varying Cellular Systems(A) Determining kinase activation by targeted measurement of T-loop phosphorylations versus modeling of kinase activities from substrate detection in large shotgun proteomic datasets: orange bars represent the relative abundance of T-loop phosphorylations in Jurkat cells estimated by their intensity ratio to the heavy-isotope-labeled standard peptides; green and purple bars represent kinase activity prediction by KEA2 (result as p values, green bars) and NetworKIN (depicted as highest individual score found for any substrate, purple bars) when applied to a fractionated Jurkat cell sample analyzed in shotgun MS (\>11,600 phosphopeptides).(B) Alterations in kinase activation upon TNF-α treatment: Jurkat cells were grown with and without TNF-α stimulation for 8 h, resulting in an increased rate of cell death as determined by caspase-3/7 green apoptosis reagent (n = 4 for both groups, error bars depicting standard deviation).(C) TNF-α induces the recruitment of receptor-interacting protein serine-threonine kinases (RIPKs) to the TNF-receptor complex, resulting in its activation and initiation of necroptotic signaling.(D) RIPK1 activation through phosphorylation at S161 measured by SRM under both conditions. Representative SRM traces are shown for unstimulated and stimulated cells with upper panels representing internal heavy-labeled standard peptides and lower panels representing signals from endogenous peptides (^∗^ indicates neutral loss of phosphoric acid on the fragment ions).(E) Immunoblot analysis demonstrates equal protein expression levels of RIPK1 under both conditions.(F) Kinase activity profiling in PAR1-activated human blood platelets: platelets were activated with SFLLRN-NH~2~ hexapeptide, mimicking thrombin activation for 1 and 5 min, respectively.(G) Volcano plots depict changes in overall kinase activation upon PAR1 activation of platelets for 1 and 5 min, respectively (significance cutoff p \< 0.05).(H) Signaling network depicts known downstream signaling pathways of PAR1, highlighting key players where increased T-loop phosphorylation could be monitored in our assay (solid bold green). Baseline PKC activity could be detected for various PKC isozymes but with no significant change in activation upon PAR1 activation (outlined in dashed bold green).(I) Bar graphs show representative dynamic regulation of kinase activation for key players in the pathway, i.e., p38α, CaMK2, RAF, ERK, BTK, and TEC (quantification based on heavy-to-light ratio, normalized on Ctrl sample, error bars depict standard deviation, ^∗^p ≤ 0.05; ^∗∗^p ≤ 0.01; ^∗∗∗^p ≤ 0.001). Through direct monitoring of T-loop phosphorylations, we were able to alleviate this bias resulting in the detection of numerous understudied kinases in their activated state. Of note, activity for several kinases predicted by KEA2 or NetworKIN are thus far not included in our T-loop assays, a shortcoming that can be overcome by expanding the set of kinases in our approach using alternative proteases. A few kinases reported by the prediction tools were included in our assays but not detected (e.g., RSK2, JNK1, JNK2, and Braf), which could be caused by differences in timing between T-loop versus substrate phosphorylation or, more likely, redundancies in kinase-substrate relations, resulting in false positive predictions as exemplified in [Figure S4](#mmc1){ref-type="supplementary-material"}. After the successful detection of several T-loop phosphorylations in unstimulated cells, we reasoned that our technique should be able to reveal activation of specific kinases from the steady-state background upon selected stimuli. To demonstrate this capacity, we treated Jurkat cells with TNF-α for 8 h, which resulted in increased cell death ([Figure 2](#fig2){ref-type="fig"}B). Upon TNF-α stimulation, the receptor-interacting protein serine-threonine kinase (RIPK) is recruited to the TNF-receptor complex and mediates apoptosis and/or necroptosis ([@bib2], [@bib22]) ([Figure 2](#fig2){ref-type="fig"}C). Indeed, our method was able to reproducibly detect RIPK1 phosphorylation at S161 already upon TNF-α treatment for 8 h, a phosphorylation not detectable in untreated Jurkat cells ([Figure 2](#fig2){ref-type="fig"}D). Western blot analysis for RIPK1 shows no discernible change across the two conditions at the protein expression level ([Figure 2](#fig2){ref-type="fig"}E), advocating RIPK1 activation via its T-loop phosphorylation to mediate apoptosis. Inhibition of RIPK1 activity by its inhibitor necrostatin-1 (Nec-1), resulted in reduction of the number of apoptotic cells after 36 h ([Figure S5](#mmc1){ref-type="supplementary-material"}), highlighting the functional relevance of the observed kinase activation. Remarkably, despite the well-characterized role of RIPK1 in cell death, to the best of our knowledge this represents the first direct detection by MS of RIPK1 T-loop phosphorylation from cell lysates. Thus far, the only report of successful MS detection of phosphorylated S161 in RIPK1 was described by [@bib11] using a protein expression system in combination with IP, in vitro kinase assay, and phosphopeptide enrichment. This lack of evidence for RIPK1 activation in the literature primed us to further investigate its detectability in shotgun MS. Indeed, even performing a large-scale phosphoproteomics experiment, including high-pH fractionation, did not enable detection of RIPK1 phosphorylation at S161 among the \>11,600 detected phosphopeptides. Hence, our targeted approach offers a so far unachieved sensitivity in measuring S161 RIPK1 activation upon TNF-α signaling, providing an additional valuable tool to monitor the complex regulation of cell death. Next, we wanted to exploit the sensitivity of our method, performing in-depth analyses of rapid kinome dynamics in primary human cells. We applied our technique to study the mechanism of PAR1-mediated activation of blood platelets ([Figure 2](#fig2){ref-type="fig"}F). Platelet activation involves various intracellular signaling events; however, the key step is activation of Phospholipase C (PLC), resulting in an increase in intracellular Ca^2+^. This in turn activates PKC and CaMK signaling and results in activation of RAS, via its translocation to the plasma membrane, which subsequently activates the MAPK cascade ([@bib19]). By performing PAR1 activation for 1 and 5 min, we were able to closely monitor changes in kinase activation states. Overall, we were able to detect and quantify 32 T-loop phosphorylations in 27 kinase groups ([Figure 2](#fig2){ref-type="fig"}G), including major players of both PKC and CaMK signaling and the MAPK cascade ([Figure 2](#fig2){ref-type="fig"}H; [Tables S5](#mmc6){ref-type="supplementary-material"} and [S7](#mmc8){ref-type="supplementary-material"}). The well-established nature of the signaling cascade in combination with the two time points additionally allowed us to determine interesting basic signaling kinetics ([Figure 2](#fig2){ref-type="fig"}I) hinting toward a rapid response by p38 and CaMK signaling upon PAR1 activation, compared to a slower response by the RAF-MEK-ERK cascade. Lastly, our assay allowed us to study activation dynamics of the two TEC family tyrosine kinases BTK and TEC, both known to act as major PLCγ2 activators upon platelet activation. Both show an increase in T-loop phosphorylation upon platelet activation; however, BTK seems to be activated faster and to a larger extend, corroborating its leading role over TEC established in the literature ([@bib3]). Since kinases are a major class of drug targets, especially in cancer where 25 kinase-targeting drugs have been approved and numerous candidates are under clinical evaluation ([@bib18]), we wanted to assess the usefulness of our technology to study unbalanced activity of kinases in disease. A major challenge in kinase inhibitor treatment is the (long-term or downstream) effect on the rest of the kinome, which consistently leads to therapy resistance due to adaptation of cellular signaling networks. To demonstrate the potential of our technology to shed light on such mechanisms, we next probed kinase activation upon acquired BRAF inhibitor (BRAFi) resistance in melanoma. Roughly half of all melanomas are driven by the BRAF^V600E^ mutation, resulting in constitutive activity of BRAF kinase activity. Patient treatment with BRAFi shows initial success, but commonly the clinical benefit is only transient because of rapid acquisition of drug resistance ([@bib47]). Here, we exploit matched patient-derived melanoma cell lines from treatment-naive, treatment-sensitive, and NRAS^Q61K^-based resistant tumor states established from patient-derived tumor xenografts to study BRAF-resistance-driven alterations in kinase activation states ([Figure 3](#fig3){ref-type="fig"}A) ([@bib24]). Across all three cell lines, we were able to detect and quantify 39 phosphosites representing T-loop phosphorylations of 37 kinase groups ([Table S5](#mmc6){ref-type="supplementary-material"}). Several of the quantified kinases showed increased activation in the resistant cell line compared to the treatment-naive and sensitive cells ([Figure 3](#fig3){ref-type="fig"}B; [Table S7](#mmc8){ref-type="supplementary-material"}). Many of them are known to be involved in growth and proliferation such as ERK2, MAP2K4, CDK2/CDK3 and NLK, and various members of the PKC kinase family. Surprisingly, some kinases specifically activated in the drug resistant cell line have thus far mainly been linked to tumor suppressing activities such as Chk2 and p38α ([@bib48]).Figure 3Probing Kinase Activation in Patient-Derived Samples(A) Reorganization of kinase activities upon acquired BRAF inhibitor resistance in melanoma: matched melanoma cell lines were established from patient-derived xenografts (PDXs) from the same patient before and after acquired resistance to the BRAF inhibitor PLX-4720, giving rise to a model system comprising treatment naive, treatment sensitive, and treatment resistant cell lines.(B) Pairwise comparison of kinase activation between the 3 conditions is depicted as volcano plot using an arbitrary significance cutoff (p \< 0.05).(C) Molecular function of ERK4, which is part of a highly confined molecular system, together with ERK3 and MK5. Bar plots show the detected abundance of the T-loop phosphorylations for ERK3, ERK4, and MK5 (normalized to values observed in naive cells, error bars depicting standard deviation, ^∗^p ≤ 0.05; ^∗∗^p ≤ 0.01; ^∗∗∗^p ≤ 0.001).(D) Determining kinase activation states in a breast cancer needle biopsy: proteins were extracted from half a needle biopsy (\<5 mg of tissue), followed by tryptic digest, phosphopeptide enrichment by Fe(III)-IMAC, and global kinase activation screening by nanoLC-SRM.(E) The detected kinases span a dynamic range of more than 3 orders of magnitude, as estimated by their intensity ratio to the heavy-isotope-labeled standard peptide (assuming roughly equimolar internal standard concentration).(F) Selected kinases detected in their activated state are shown in the context of their molecular signaling pathway according to the Kegg database (outlined in bold green). The most striking activation we observed was that of the kinase ERK4, which was around or below the detection limit in the sensitive and naive cell lines, however, was found to be strongly activated (∼30 times) upon acquired drug resistance ([Figure 3](#fig3){ref-type="fig"}C). This strong activation remained in resistant cells upon drug withdrawal. It has been previously observed that oncogenic RAS increases RNA levels of ERK4 ([@bib25]), although no effect on kinase activation level has been reported so far. Our results, however, convey the use of ERK4 activation as a potential marker for NRAS-mediated drug resistance. The clinical value of our technology ultimately depends on its direct applicability in primary tissue samples, such as tumor biopsies. Here, we demonstrate the sensitivity of our approach by quantifying kinase activation states in a Her2+ breast cancer patient, analyzing tissue material obtained from half a 14G needle biopsy taken prior to treatment ([Figure 3](#fig3){ref-type="fig"}D). Using a starting amount of merely 300 μg of protein, we were able to detect and quantify 46 phosphorylation sites in the T-loop region of 43 kinase groups, spanning a dynamic range of more than 3 orders of magnitude ([Figure 3](#fig3){ref-type="fig"}E; [Table S5](#mmc6){ref-type="supplementary-material"}). The detected kinases include, besides a substantial number of understudied kinases such as CDK11A, CRK7, DYRK1A, DYRK2, DYRK4, HIPK3, NEK6, and PKN2 ([@bib8]), various crucial players in control of growth and proliferation ([Figure 3](#fig3){ref-type="fig"}F), a variety of which are targets for potent novel inhibitors currently in clinical trials. This includes detection of cdc2/CDK1 and CDK2, which can be inhibited by Dinaciclib and ERK1 and ERK2, the phosphorylation of which can be inhibited by various MEK inhibitors such as Selumetinib. Both inhibitors are currently in phase III trials. Furthermore, activation is observed for the kinases p38α and PKA, which have been shown to play crucial roles in invasiveness and acquired Herceptin resistance ([@bib14], [@bib20]). The most intense level of T-loop phosphorylation was observed for PKC, which is required for HER2-mediated NF-κB activation, thus mediating properties of malignancy, such as proliferation, invasiveness, and avoidance of apoptosis ([@bib34]). These results demonstrate the clinical potential of our technology to molecularly profile diseases such as breast cancer, being the most frequent cause of cancer deaths among women worldwide. It is a complex heterogeneous disease, comprising various subtypes with distinct genetic and morphological features, leading to differences in treatment response and clinical outcome. Consequently, present standard therapies are insufficient to treat metastatic disease and rationally chosen drug combinations are needed ([@bib38], [@bib51]). We argue that monitoring kinase activation in such a setting could have a profound impact, uncovering potential novel pharmaceutical targets. Discussion {#sec3} ========== In this study, we provide carefully optimized assays for T-loop phosphorylation on 178 protein kinases, accounting for roughly one-third of the human kinome. The strength of the technology lies in the combination of sensitivity, enabling substantial kinome coverage even from limited starting material (around 150--250 μg protein per sample) and throughput, using targeted SRM assays on a relatively simple triple-quadrupole MS instrument. Our approach is highly adaptable to various research questions and sample types as it provides enough sensitivity to reliably quantify kinase activation from limited sample amounts as encountered when working with primary cells or even clinical samples. The unbiased nature of the approach facilitates the detection of numerous understudied kinases ([Table S8](#mmc9){ref-type="supplementary-material"}), alleviating the research bias that is still omnipresent in kinase research ([@bib15]). The use of sample fractionation or the enrichment for specific cell compartments such as nuclei or the cell membrane will further increase kinome coverage. Applying this approach, we were able to detect 73 T-loop phosphorylations for 63 kinase groups reflecting 90 individual kinases across numerous cell types (primary and patient derived) ([Table S5](#mmc6){ref-type="supplementary-material"}), which, to the best of our knowledge, presents the largest compendium of kinase activation sites reported to date. We demonstrate the utility of our technology for exclusive detection of activated kinases upon selected stimuli or specific treatment regimes. Our technology reveals, besides known signaling pathway activation, T-loop phosphorylations of the elusive RIPK1 S161, dynamic activation of primary blood platelets, rewiring of signal transduction in melanoma upon acquired drug resistance, and global kinase activity status in a breast cancer tumor biopsy sample. In conclusion, we present here a robust approach to quantify system-wide T-loop phosphorylations as a proxy for human protein kinase activity in various biological settings. STAR★Methods {#sec4} ============ Key Resources Table {#sec4.1} ------------------- REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesPurified Mouse Anti-RIPBD BiosciencesCat\#610458; RRID: [AB_397831](nif-antibody:AB_397831){#intref0010}GAPDH antibody \[GT239\] (mouse)GeneTexCat\# GTX627408, RRID: [AB_11174761](nif-antibody:AB_11174761){#intref0015}Goat Anti-Mouse Immunoglobulins/HRPAgilent TechnologiesP044701-2Biological SamplesPatient derived platelet cellsSanquin Blood Bank<https://www.sanquin.org/working-at/production/blood-bank>Patient derived biopsy from primary breast tumorsTRAIN2 clinical trial<https://clinicaltrials.gov/ct2/show/NCT01996267>Chemicals, Peptides, and Recombinant ProteinsSpikeMix™ Kinase Activation Loops (Human) - heavyJPT GermanySPT-KAL-POOL-L-100pmHRM Calibration KitBiognosysKi-3003TNF-αPeproTech300-01APLX-4720SelleckchemS1152PAR1 mimicking hexapeptide SFLLRN-NH2Peptides InternationalPAR-3676-PINec-1MerckCat\#480065TirofibanIroko cardio, UKCAS 0144494-65-5Critical Commercial AssaysCaspase-3/7 green apoptosis reagentEssen BioscienceCat\#4440Sep-Pak C18 1 cc Vac CartridgeWatersWAT023590Bio-Rad Protein Assay Kit IBio-Rad5000001PhosSTOP™Merck4906837001cOmplete™, Mini, EDTA-free Protease Inhibitor CocktailMerck11836170001AssayMap Cartridge Rack, Fe(III)-NTA 5 μLAgilent TechnologiesCat\#G5496-60085Pierce™ ECL Plus Western Blotting SubstrateThermo Scientific™32132Deposited Data.raw and .wiff data of all experimentsSRMAtlas<http://www.peptideatlas.org/PASS/PASS01234>Analyzed SRM data within Skyline daily frameworkSRMAtlas<http://www.peptideatlas.org/PASS/PASS01234>MaxQuant Database search output performed on fractionated Jurkat lysates for motiv analysisSRMAtlas<http://www.peptideatlas.org/PASS/PASS01234>NetworKIN kinase activity predictionThis paper[Table S6](#mmc7){ref-type="supplementary-material"}Quantitative and statistical analysis of SRM assays performed in MSstatsThis Paper[Table S5](#mmc6){ref-type="supplementary-material"}KEA2 kinase activity predictonThis Paper[Table S6](#mmc7){ref-type="supplementary-material"}Experimental Models: Cell LinesJurkatDSMZACC 282HEK293TATCCCat\#ATCC CRL-3216PC9Sigma-Aldrich90071810-1 VLM026Laboratory of Daniel Peeper NKI<https://www.nki.nl/divisions/molecular-oncology-immunology/peeper-d-group/>Software and AlgorithmsSkyline daily[@bib29]<https://skyline.ms/project/home/software/Skyline/begin.view>MaxQuant (version 1.6.1.0)[@bib53]<https://www.maxquant.org/>MSStats[@bib52]<http://msstats.org/>MASCOTMatrix Science<http://www.matrixscience.com/>Proteome DiscovererThermo Fisher[www.thermofisher.com](http://www.thermofisher.com){#intref0070}NetworKin[@bib31]<http://netphorest.info>KEA2[@bib26]<http://www.maayanlab.net/KEA2>Other25 cm, 75 μm ID PepMap RLSC, C18, 100 Å, 2 μm particle size columnThermo ScientificES802 Lead Contact and Materials Availability {#sec4.2} --------------------------------------- Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Maarten Altelaar (<m.altelaar@uu.nl>). Experimental Model and Subject Details {#sec4.3} -------------------------------------- ### Cell Models {#sec4.3.1} PC9 NSCLC cells (originating from human adenocarcinoma, male) were purchased from Sigma-Aldrich (cat\# 90071810-1 VL) and were cultured in standard Roswell Park Memorial Institute medium 1640 medium (Lonza), containing 10% FBS (Thermo), 2 mM L-glutamine and 1% penicillin/streptomycin (Lonza), at 37°C in a humidified atmosphere containing 5% CO~2~. Cells were detached from the culture surface using trypsin (Lonza), and washed three times with PBS before lysis. Hek293T cells (embryonic kidney of female fetus purchased from ATCC, cat\# ATCC CRL-3216) were seeded at 15% density in 15-cm plates, allowed to adhere in full DMEM (Lonza) containing 10% heat-inactivated fetal bovine serum (Gibco), 2 mM L-glutamine (Lonza) and 20 mM HEPES (Sigma-Aldrich), and cultured to ∼90% confluence over 2.5 days. Twelve hours prior to harvesting, growth medium was replaced with fresh pre-warmed full DMEM. At harvesting, no dead or floating cells were visible by microscopic examination. Cells were washed twice with ice-cold PBS on-plate, detached by trypsin (Lonza), and collected by low-speed centrifugation at 200g for 5 min. Jurkat cells (male, purchased from DSMZ, ACC 282) were cultured in RPMI-1640 media complemented with l-glutamine, 10% fetal bovine serum, and PenStrep at a density between 5x10^5^ and 5x10^6^ per ml. PDX melanoma cell lines (M026, M026R) were cultured in RPMI 1640 with HEPES and L-Glutamine supplemented with 10% fetal bovine serum and Penicillin-Streptomycin. The resistant cell lines were grown in presence of 1 μM PLX-4720 (Selleckchem). When specified, the cell lines were treated with an IC~50~ dose (1 μM) of the inhibitors PLX-4720. All cell pellets were stored at -80°C prior to cell lysis. Pooled platelet concentrates in plasma (platelet concentrates from 5 donors in ABO blood-group-matched plasma) were obtained from the Dutch Bloodbank (Sanquin). Multiple platelet concentrates were pooled in three separate pools and processed independently. Platelets were isolated as described before ([@bib45]). Briefly, platelet concentrates were centrifuged for 20 min at 120g, to remove remaining red and white blood cells. Next, platelets were washed twice with isolation buffer (36 mM citric acid, 103 mM NaCl, 5 mM KCl, 5 mM ethylenediaminetetraacetic acid (EDTA), 5.6 mM D-glucose, pH 6.5, containing 0.35% \[w/v\] bovine serum albumin (BSA)) by centrifugation for 10 min at 2,000g. Platelets were washed once with Tyrode's Solution (Sigma), centrifuged for 10 min at 2,000g and resuspended to a final concentration of 2×10^8^ platelets/mL in Tyrode's Solution supplemented with tirofiban (1:1000) (Iroko cardio, UK). ### Patients Samples {#sec4.3.2} Frozen biopsies were obtained from primary breast tumors of patients undergoing neoadjuvant chemotherapy and dual HER2 blockade in the TRAIN2 trial ([NCT01996267](NCT01996267){#intref0090}) ([@bib46]). The study had been approved by the ethical committee and informed consent was obtained from all patients. Approximately fifteen 30 μm frozen sections were prepared with a cryostat microtome for protein extraction. R/DNase free H~2~O was used as adhesive instead of a polymer adhesive to eliminate any traces of polymers in the protein extract. About halfway through the biopsy one or two sections of 6-8 μm were prepared for hematoxylin and eosin (HE) staining. The HE stained sections were reviewed by a pathologist to estimate the tumor cell percentage of the tissue. Method Details {#sec4.4} -------------- ### TNF-α Stimulation and RIPK1 Inhibition {#sec4.4.1} One day prior stimulation, Jurkat cells were seeded at a concentration of 1x10^6^ per ml in fresh media. TNF-α (PeproTech, USA) was added at a concentration of 100 U/ml for 8 h, after which the cells were washed with ice cold PBS and snap frozen. For the apoptosis assay the Jurkat cells were plated at a density of 3x10^4^ cells in a 96 well plate coated with 0.01% poly-l-ornithine. Cells were pre-incubated with 0.5 μM Necrostatin-1 (Merck), and then stimulated with TNF-α at a concentration of 100U/ml in the presence of 2.5 μM Caspase-3/7 green apoptosis reagent (Essen Bioscience). Apoptosis was monitored using the IncuCyte Zoom system (Essen Bioscience) every hour for 48 h. Each well was divided into four views, and the number of green fluorescent cells was counted using the IncuCyte Zoom software (Essen Bioscience). ### RIPK1 Immunoblot Analysis {#sec4.4.2} Cell pellets were lysed by sonication in Urea buffer (8 M Urea, 50 mM ammonium bicarbonate, protease inhibitor Cocktail Complete (Roche), phosphatase inhibitors PhosSTOP (Roche)) and protein concentration was determined using a Bradford assay (Bio-Rad). Immunoblotting was performed using 12% Criterion XT precast gel (Bio-Rad) and nitrocellulose membrane. All blocking steps were performed using 5% milk in TBS-Tween (Tris-buffered saline (pH 7.5) with 0.1% Tween (Biorad)). GAPDH (GeneTex, RRID: [AB_11174761](nif-antibody:AB_11174761){#intref0095}) and RIPK1 (BD Biosciences, RRID: [AB_397831](nif-antibody:AB_397831){#intref0100}) were used as primary antibodies, goat anti-mouse immunoglobulins/HRP (Agilent Technologies) was used as secondary antibody. Protein detection was performed using the ECL Plus Substrate (Pierce) and an Amersham Imager AI600 for chemiluminescence imaging. ### PAR1 Activation of Pooled Platelets {#sec4.4.3} Platelets were activated with 50 μM PAR1 peptide (SFLLRN-NH2, Peptides international) or negative control (Tyrode's solution) for 2 or 5 min at 37°C in a thermomixer (Eppendorf). Activation was stopped by lysis of the platelets by addition of 2 times (% v/v) 1.5x SDC lysis buffer (1.5% SDC, 15 mM TCEP, 60 mM Chloroactamide, 150 mM tris(hydroxymethyl)aminomethane (Tris) pH 8.5, 1.5% phosphatase and protease inhibitor cocktail (Thermo Scientific)). Platelet lysates were snap frozen and stored at -80°C. ### Sample Preparation and Phosphopeptide Enrichment {#sec4.4.4} Frozen cell pellets and patient biopsy were lysed, reduced and alkylated in lysis buffer (1% sodium deoxycholate (SDC), 10 mM tris(2-carboxyethyl)phosphinehydrochloride (TCEP)), 40 mM chloroacetamide (CAA), and 100 mM TRIS, pH 8.0 supplemented with phosphotase inhibitor (PhosSTOP, Roche) and protease inhibitor (cOmplete mini EDTA-free, Roche). Cells were heated at 95°C and sonicated with a Bioruptor Plus (Diagenode) for 15 cycles of 30 s. Bradford protein assay (Bio-Rad Protein Assay Kit I, Bio-Rad) was used to determine protein amount. To avoid digestion bias, samples were split into aliquots containing equal amount of protein. Proteins were digested overnight at 37°C with trypsin (Sigma-Aldrich) with an enzyme/substrate ratio of 1:50 and lysyl endopeptidase (Wako) with an enzyme/substrate ratio of 1:75. SDC was precipitated with 2% formic acid (FA) and samples were desalted using Sep-Pak C18 cartridges (Waters). Subsequently samples were dried in vacuo and stored at -80°C until further use. Phosphorylated peptides were enriched using Fe(III)-NTA cartridges 5 μL (Agilent technologies) in an automated fashion using the AssayMAP Bravo Platform (Agilent Technologies) as previously described ([@bib36]). For all samples, 250 μg of peptides were used as input for one cartridge, except for the biopsy sample where only approximately 150 μg of protein was available per cartride in order to allow for analyses in both, survey and quantification mode. In brief, Fe(III)-NTA cartridges were primed with 200 μL of 0.1% TFA in ACN and equilibrated with 250 μL of loading buffer (80% ACN/0.1% TFA). Samples were dissolved in 200 μL of loading buffer containing 100 fmol synthetic isotope labelled t-loop standard peptides and loaded onto the cartridge at a loading speed of 5 μL/min. Subsequently columns were washed with 250 μL loading buffer and eluted with 35 μL of 10% ammonia directly into 35 μL of 10% formic acid. Samples were dried down and stored at −80°C until LC--MS analysis. ### LC-MS/MS Setup {#sec4.4.5} Spectral libraries were partly acquired on a TripleTOF 5600 (Sciex) coupled to an Agilent 1290 Infinity System (Agilent Technologies) adapted to nanoflow conditions by using a split flow setup as described in [@bib9]. The system was operated with in-house packed trap column (Dr. Maisch Reprosil C18, 3 μm, 2 cm × 100 μm) and analytical column (Agilent Poroshell 120 EC-C18, 2.7 μm, 50 cm × 75 μm). The split flow was adapted to achieve 300 nl/min flow at the front end of the column upon applying a flow rate of 0.2 mL/min. 0.6% acetic acid in water (Milli-Q, Millipore) was used as buffer A an 0.6% acetic acid, 80% ACN was used as buffer B. Upon injection, peptides were trapped at 5 μL/min during 5 min with 100% solvent A (0.1% FA in water) before being separated on the analytical column. All remaining measurements were executed on a TSQ-Vantage (Thermo Fisher) coupled to an Easy-nLC 1000 (Proxeon, Odense, DK). LC configuration was in one-column setup (25 cm, 75 μm ID PepMap RLSC, C18, 100 Å, 2 μm particle size column (Thermo Scientific, Odense, DK)). Formic acid (0.1%, Merck, Darmstadt, Germany) in deionized water (Biosolve, Valkenswaard, NL, ULC/MS grade) was used as solvent A, 0.1% formic acid in acetonitrile (Biosolve, Valkenswaard, NL, ULC/MS grade) as solvent B. All measurements were performed at 200 nl/min flow rate and all samples were analyzed with injection volumes of 2 μL containing 10% FA. ### Spectral Library Generation {#sec4.4.6} SRM assay development was guided by a spectral library generated in house. This provided confirmation of the synthesized peptide sequence and essential information about LC and MS characteristics of each peptide. For this spectral library generation, the heavy labeled phosphopeptides were mixed with 1x HRM Calibration KIT and analyzed in data-dependent mode on two different LC-MS setups. (1) Crude peptides were analyzed on a 5600 TripleTOF (Sciex). LC-MS setup and data acquisition methods were used as previously described ([@bib39]). In brief, peptides were separated on a 2 h gradient analyzed in TOP20 mode (selection criteria: intensity \> 50 cps, charge state ≥ 2+, dynamic exclusion 15 s). (2) Crude peptides were analyzed on a TSQ Vantage (Thermo) in data-dependent acquisition mode operated at TOP2. Survey scans were acquired in Q3MS mode (1.5 s, 0.4 Da fwhm) spanning the 375-1350m/z range. The TOP2 most abundant ions were analyzed in MS/MS mode (selection criteria: intensity \> 1000 counts, 5 repeats at 3 s, 2 min dynamic exclusion). Results of both acquisition types were subjected to database search using Mascot accessed by Proteome Discoverer (version 1.4). Parameters were set to tryptic digest, allowing for up to three missed cleavages, using carbamidomethyl cysteine as fixed modification and allowing for serine/threonine/tyrosine phosphorylation methionine oxidation and C-terminal isotope labels. Precursor mass tolerance and MS/MS tolerance were set to 50 ppm and 0.15 Da respectively for TripleTOF files and to 0.9 Da for TSQ files. Results were filtered using Percolator ([@bib23]) to an FDR below 1%. Spectral libraries were built in Skyline from both searches. ### SRM Assay Development {#sec4.4.7} All assays were developed and optimized on a TSQ Vantage as previously described ([@bib10], [@bib42]). In brief, the most intense fragment ions found in the spectral libraries were directly used as initial transitions, multiplexing up to 3-10 transitions per precursor. Those initial SRM assays were applied to the synthetic peptide library enabling subsequent optimization of multiple parameters such as collision energy and RT scheduling. Initial assays for a few peptides not identified in either of the spectral library were constructed from theoretically possible y- and b-ions in combination with their most likely precursor charge states. Extensive manual validation of phosphosites localization isomers was performed, including the use of site determining ions as transitions. Collision energies were optimized for each transition individually in an empirical way assisted by Skyline ([@bib28]). Instrument specific CE parameters were used as a starting point (CE = 0.03m/z + 2.905 for doubly charged precursors and CE = 0.038m/z + 2.281 for precursor charges of three and higher) to scan through different normalized collision energy values using a step size of one. ### T-Loop Detection in Various Sample Types {#sec4.4.8} Dried samples were reconstituted in 3 μL of 10% formic acid, containing 0.1x/μL iRT peptides. Injection volumes for all analyses were kept at 2 μL. LC-MS analysis contained the following steps: Analytical column equilibration (3 μL 100% Buffer A at 600 bars) followed by sample loading onto the column (loading volume 6 μL at 600 bars). Phosphopeptides were separated on a gradient from 2% to 25% B in 100 min, followed by a column washing step ramping up from 25% to 100% B in 5 min followed by 100% B for 15 min. To avoid carryover and monitor LC performance at least 1 BSA run was scheduled after each analysis. Samples within one experiment were analyzed in randomized order by defining and injection order based on random numbers prior to the analysis. Retention time scheduling was dynamically adapted to reflect the instrument state prior to sample analysis. Low confidence survey runs applied low transition numbers per peptide (∼3) in short scheduling windows (4 min) applying long cycle times (4 s). High confidence quantification runs applied less target peptides with a higher number of transitions per peptide (up to 7, including phosphosite localization specific ions), longer scheduling windows (6-10 min) and shorter cycle times (2.5-3 s). ### Experimental Design and Replicates {#sec4.4.9} Proof of concept experiments for Jurkat, PC9 and Hek293T cells were based on a single cell culture experiment (n = 1) as they did not involve accurate quantification. Sample preparation was performed in at least 2 replicas to facilitate separation into survey and quantification runs. Quantitative experiments were performed in triplicates (n = 3) on a cell culture level including lysis and tryptic digestion. For each sample, two processing replicas were prepared on the level of phosphopeptide enrichment which were independently analyzed by LC-MS resulting in n = 6 injection replicas per condition. Additional phosphopeptide enrichments were performed for survey analyses when sufficient sample amount was available. To avoid missing condition-specific kinase activation states at least one sample per condition was measured in survey mode. Specific experimental designs were as follows: for TNF-α stimulation independent cell culture of n = 3 with TNF-α stimulation and n = 3 without any stimulation were performed. Likewise, PDX-derived cells were grown independently in triplicates (n = 3) for each of the 3 conditions respectively (n = 9). Analysis of the breast cancer needle biopsy was performed from a single sample (n = 1). The sample was split into 2 processing replicates prior to phosphopeptide enrichment to allow for analysis in survey and quantification mode. ### SRM Data Assessment {#sec4.4.10} All SRM experiments were analyzed using Skyline ([@bib29]). Signal quality was assessed visually, primarily relying on sequence similarity between heavy labeled standard peptides and endogenous peptides. Key points used were perfect co-elution of both peptide forms in terms of retention time and peak shape in combination with a high similarity of the relative intensities of transitions found in the heavy and the endogenous peptides (rdotp \> 0.9). Signals for endogenous peptide forms not matching these criteria were excluded, unless the result indicated a clear on/off state between different conditions (RIPK1 and ERK4) in which case noise levels were used as a quantitative readout. ### Determining LOD and LOQ Values {#sec4.4.11} LOD and LOQ values of selected peptides were determined using dilution series of stable isotope labeled peptides spiked into Jurkat matrix samples. Jurkat cells were cultured, lysed and digested as described above. 200 μg of cell digest were enriched for phosphopeptide by Fe(III)-NTA and spiked with increasing amounts of the heavy isotope labeled phosphopeptide library (10 amol, 100 amol, 1 fmol, 5 fmol, 10, fmol, 50 fmol, 100 fmol). Samples for each quantity of heavy peptide spike-ins were prepared in triplicates. A subset of 44 heavy isotope labeled peptides were quantified in all samples by scheduled SRM using 3-7 transitions per peptide. Summed areas under the curve of all transitions were used as quantitative readout as reported by Skyline. Peaks were assessed visually, and representative noise area values were selected for spike-in amounts below the LOD. Linear regressions were performed for each peptide individually per sample replicate as well as combined for all replicas. LOD and LOQ values were determined according to the following equations: LOD = 3S~a~/b and LOQ = 10S~a~/b using b as the sensitivity (slope of the linear regression of all replicas combined) and S~a~ as the standard deviation of the intercept (determined for each regression individually) ([@bib41]). To avoid skewing of the linear regressions, for each peptide only the highest spike-in amount below the LOD was included in the analysis. ### Jurkat-Cells Shotgun Analysis {#sec4.4.12} 2 mg of cell digest from TNF-α treated Jurkat cell lines was fractionated on a high-pH (HpH) reversed-phase C18 column (Gemini 3 μm C18 110 Å, 100 × 1.0 mm, Phenomenex) coupled to an Agilent 1100 series (Agilent Technologies) on a 60 min gradient. 67 fractions of 1 min each were collected and concatenated into five pools as previously described ([@bib5]). These were dried down in vacuo and subjected to phosphopeptide enrichment as described above. DDA analysis was performed on a Q Exactive HF (Thermo Scientific) coupled to an Easy-nLC 1000 (Proxeon, Odense, DK), configured as described above. All LC settings and methods including analysis time and gradient were identical to the ones used during the SRM analysis. The mass spectrometer was operated in data-dependent acquisition mode containing a survey scan from 375 to 1600m/z (resolution 60,000, max injection time 20 ms, AGC target 3e6) acquired in profile mode. MS/MS spectra (HCD, 27% normalized collision energy at a target value of 50,000 ions, resolution 30,000) were recorded for the 12 most intense peaks using a 1.4m/z isolation window. Data was analyzed by Maxquant (version 1.6.1.0) with the integrated Andromeda search engine. Trypsin was specified as enzyme and up to two missed cleavages were allowed. Cysteine carbamidomethylation was set as a fixed modification, while methionine oxidation and protein N-term acetylation were set as variable modifications. Phosphorylation on serine, threonine and tyrosine was also selected as variable modification. The mass tolerance was set to 4.5 ppm for precursor ions, and to 20 ppm (FTMS) for fragment ions. Peptide and protein identification were set to 1% FDR, and the minimum score for modified peptides was set to 40. ### Web-Based Kinase Activity Prediction {#sec4.4.13} KEA2 (<http://www.maayanlab.net/KEA2>) was used to predict kinase activity from shotgun data using enrichment analysis ([@bib26]). Phosphosites detected by MaxQuant were used as data input. Enrichment analysis was performed using Literature Based Kinase-Substrate Library with Phosphosites option. NetworKIN analysis was performed on the same dataset (<http://netphorest.info>) ([@bib31]). In the subsequent data visualization, only the highest score detected per kinase was considered. Quantification and Statistical Analysis {#sec4.5} --------------------------------------- For all SRM assays, the ratios between analyte and internal standard were used as quantitative readout. Unless otherwise specified significance analysis was performed with MSstats ([@bib7]). The analysis entailed log~2~-transformation of the intensity values, followed by testing for abundance differences between different conditions using a linear-mixed effects model. An arbitrary cutoff of p ≤ 0.05 was considered significant. For individual bar plots quantitative peptide abundance were exported from Skyline (normalized based on internal standards). Bars represent average and error bars represent standard deviation. Significance is indicated according to the adjusted p-value output provided by MSstats. Data and Code Availability {#sec4.6} -------------------------- The datasets generated during this study are available at Peptide Atlas PASS01234 (<http://www.peptideatlas.org/PASS/PASS01234>). Supplemental Information {#appsec2} ======================== Document S1. Figures S1--S5Table S1. List of Target Peptides and Availability of Phosphosite-Specific Antibodies, Related to Figure 1Table S2. Transition List, Related to Figure 1Table S3. Determined Limits of Detection and Quantification for Selected Phosphopeptides, Related to Figure 1Table S4. List of Current FDA-Approved Kinase Inhibitors and Their Targets, Related to Figure 1Table S5. List of Detected t-Loop Phosphorylations, Related to Figures 1, 2, and 3Table S6. Web-Based Kinase Activity Prediction, Related to Figure 2Table S7. Quantitative Analysis of Platelet Activation and Acquired Drug Resistance in Melanoma, Related to Figures 2 and 3Table S8. Activation of Understudied Kinases, Related to Figures 1, 2, and 3Document S2. Article plus Supplemental Information This work was supported by the Netherlands Organization for Scientific Research (NWO) through a VIDI grant for M.A. (723.012.102) and Proteins\@Work, a program of the National Roadmap Large-Scale Research Facilities of the Netherlands (project number 184.032.201). M.v.d.B. was supported by a fellowship from the Landsteiner Foundation for Blood Transfusion Research (LSBR 1517), K.K. and D.S.P. were supported by a grant from the Dutch Cancer Society (NKI-2013-5799) and a European Research Council grant under the European Union's Seventh Framework Programme (FP7/2007-2013)/ERC synergy grant agreement no. 319661 COMBATCANCER, and M.A. and D.S.P. were supported by a grant from the Dutch Cancer Society (10304). We would like to thank Dr. Wei Wu and T. Celine Mulder for providing cell pellets used in the proof of principle analysis and Lennart Mulder for preparation of the tumor biopsy sample. Author Contributions {#sec5} ==================== Conceptualization, T.S., A.J.R.H., and M.A.; Methodology, T.S. and M.A.; Investigation, T.S., D.O.D., C.A.G.H.v.G., K.E.S., S.R., B.L.v.d.E., and K.K.; Writing -- Original Draft, T.S. and M.A.; Writing -- Review & Editing, T.S., E.H.L., M.v.d.B., D.S.P., A.J.R.H., and M.A.; Visualization, T.S. and D.O.D.; Funding Acquisition, M.v.d.B., D.S.P., A.J.R.H., and M.A.; Resources, E.H.L., M.v.d.B., and D.S.P.; Supervision, E.H.L., M.v.d.B., D.S.P., A.J.R.H., and M.A. Declaration of Interests {#sec6} ======================== Patent applicant, Utrecht Holding; name of inventor(s), M.A. and T.S.; application number, EP18179104.7; status of application, pending; specific aspect of the manuscript covered in the patent application, method for monitoring kinase activity in a sample. M.A. is a scientific advisor for Pepscope. Supplemental Information can be found online at <https://doi.org/10.1016/j.cels.2019.08.005>. [^1]: Lead Contact	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec0005} =============== Breast cancer is the most common type of cancer in women, accounting for 28% of total cancer cases across Europe \[[@bib0005]\]. In 2014, 92,500 women died from breast cancer in the EU-28 \[[@bib0010]\]. Through better diagnostic techniques and advancement in surgical and pharmacological treatment, mortality rates from breast cancer are decreasing \[[@bib0005]\]. However, there are substantial socio-economic inequalities in breast cancer prevalence and mortality: women in more socio-economically deprived areas have lower breast cancer incidence, but higher mortality rates \[[@bib0015]\]. For example, in England, incidence rates for breast cancer are 14% lower for women living in the most deprived areas, compared to the least deprived, while people living in the most deprived areas have a significantly higher mortality rate -- with an estimated 350 yearly excess deaths \[[@bib0020]\]. This breast cancer paradox may be attributable to later diagnosis \[[@bib0025]\], suboptimal cancer care \[[@bib0030]\], co-morbidities that may limit treatment options or increase the possibility of developing treatment complications \[[@bib0035]\], and -- most notably in terms of prevention -- lower rates of breast cancer screening uptake. In Europe, most countries have implemented or are developing mammography-based universal breast cancer screening programmes. It has been widely acknowledged that breast cancer screening is not without disadvantages, including breast cancer over diagnosis, which can lead to unnecessary surgical and pharmacological intervention. In view of this, many experts believe that the risks of breast cancer screening can outweigh the benefits, and women should always be fully informed about the risks and benefits of screening. Despite these limitations, the literature has shown that breast cancer screening can positively impact on survival \[[@bib0040]\]. However, even with universal screening programmes available in many European countries, there are still distinct area-based inequalities in breast cancer screening uptake related to socio-economic deprivation \[[@bib0045], [@bib0050], [@bib0055], [@bib0060], [@bib0065]\]. A systematic review by Pruitt et al. \[[@bib0070]\] examined the association between socio-economic deprivation and breast, cervical and colorectal cancer screening. The work included 13 studies for breast cancer screening -- all of which were conducted in North America, and showed some positive associations between socio-economic status (SES) and screening uptake \[[@bib0070]\], indicating that as SES increased (i.e. increasing income and education, decreasing unemployment and poverty), the odds of attending cancer screening increased. However, no European studies were included in this systematic review and given the substantial differences in healthcare systems and screening coverage between the USA and Europe, it is important to examine if there is also an association in universal European health systems. Our systematic review, therefore, aimed to examine the association between area-level socio-economic deprivation and breast cancer screening uptake in Europe. 2. Materials and methods {#sec0010} ======================== 2.1. Study design and inclusion criteria {#sec0015} ---------------------------------------- The systematic review is registered with PROSPERO (CRD42018083703) and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines \[[@bib0075]\] (Appendix A). We included observational studies (cross-sectional; prospective and retrospective cohorts, time series, repeat cross-sectional) of adult women (\>18 years) in high-income European countries (defined as European countries in the Organisation of Economic Cooperation and Development). To be included in the review, studies had to compare at least two areas and have some area-level measure of socio-economic deprivation. Area-level socio-economic deprivation can be measured differently, but essentially involves ranking areas on the basis of relative local scores for factors such as income, employment and housing quality. Common measures include indices of multiple deprivation, percentage of poverty, or percentage unemployed. The primary outcome of interest was breast cancer screening participation. This included attending mammography, clinical breast examination (CBE), and ultrasound, but excluded self-breast examination. The screening could have been as either part of a screening programme, through referral by a healthcare professional, through health insurance or by any other certified method. In keeping with the previous systematic review by Pruitt et al. \[[@bib0070]\], studies had to be published in English and in peer-reviewed journals (to enhance quality); any studies only published as abstracts were excluded. 2.2. Search strategy {#sec0020} -------------------- Our search terms are detailed in [Table 1](#tbl0005){ref-type="table"}. The search strategy was adapted from Cairns et al. \[[@bib0080]\] who investigated the association of area-level socio-economic deprivation and suicidal behaviour, with additional breast cancer and screening terms. A pilot search was undertaken to identify three indicator papers \[[@bib0085], [@bib0090], [@bib0095]\]. Three electronic databases were searched between 1 st January 2008 and 28^th^ January 2019 (host sites given in parentheses): Medline (via OVID), Embase (via OVID) and PsycINFO (via OVID). The search start date of 2008 was chosen because our review updates the systematic review by Pruitt et al. \[[@bib0070]\]. Citation follow up of all included articles was also conducted.Table 1Search terms.Table 1\[(breast cancer AND mammo\) OR (breast cancer AND screen\) OR (mammo\) AND (socioeconomic OR SES OR education\ OR employment OR income OR occupation\* OR poverty OR class OR depriv\* OR disadvantage\* OR social class OR social factors OR economic OR unemployment) AND (area\* OR geo\* OR place OR neighbourhood OR neighbhorhood OR region\* OR county OR ward OR city OR district OR county OR census tract OR metropolitan OR zip code)\] 2.3. Data extraction and quality appraisal {#sec0025} ------------------------------------------ Screening, data extraction and quality appraisal were conducted by two reviewers (DS and KT). Study inclusion agreement between the reviewers was 95% with a kappa score of fair (κ = 0.48) \[[@bib0100]\], as the first reviewer was more inclusive (all full texts were identified by both reviewers). The methods and main study findings of each study were extracted using a bespoke data extraction form. Data extraction and quality appraisal were conducted by DS and checked in full by KT; any discrepancies were resolved through discussion between the reviewers and then if agreement could not be reached by consensus through consulting the project lead (AT). Quality appraisal was conducted using the checklist for analytical cross sectional studies produced by the Joanna Briggs Institute (JBI) \[[@bib0105]\]. The checklist includes questions relating to sampling, inclusion criteria, confounding, types of outcomes and statistical analyses (Appendix B). Arbitrary labels of low (1--4), moderate (5 and 6) and high (7 and 8) were assigned to each primary study for ease of reporting. 2.4. Analysis and synthesis {#sec0030} --------------------------- A narrative synthesis thematically describing studies by country was undertaken. Studies used heterogeneous measures so meta-analysis was not possible. We report on the overall association between area-level deprivation and breast cancer screening uptake in terms of (1) the gap (between the best and the worst off areas); and (2) the gradient (across the whole spectrum of deprivation) \[[@bib0110]\]. 3. Results {#sec0035} ========== A total of 2125 unique citations were retrieved from the searches. Reasons for exclusion at the full paper stage (n = 77) are detailed in [Fig. 1](#fig0005){ref-type="fig"} (with further details in Appendix C). In total, 13 unique primary studies (representing 14 articles) from seven countries were included in the review: England (n = 4, from 5 articles) \[[@bib0115], [@bib0120], [@bib0125], [@bib0130], [@bib0135]\], France (n = 3) \[[@bib0095],[@bib0140],[@bib0145]\], Germany (n = 1) \[[@bib0150]\], Italy (n = 1) \[[@bib0155]\], Sweden (n = 1) \[[@bib0090]\], the Netherlands (n = 1) \[[@bib0085]\] and Turkey (n = 2) \[[@bib0160],[@bib0165]\]. All countries ran screening programmes, although it is acknowledged that, in Turkey, breast cancer screening is still performed primarily on an opportunistic basis \[[@bib0160],[@bib0165]\]. The included studies examined screening uptake over different time intervals -- the shortest of which was in Sweden \[[@bib0090]\] (1.5--2 years), while the longest screening interval was in England (3 years) \[[@bib0115], [@bib0120], [@bib0125], [@bib0130], [@bib0135]\]. The majority of the studies used small neighbourhood areas containing approximately 1500 residents called Lower Super Output Areas \[[@bib0125],[@bib0135]\]; IRIS - French 'aggregated units for statistical information' \[[@bib0095],[@bib0145]\]; or SAMS - Small Areas for Market Statistics \[[@bib0090]\]), although some used larger districts containing several hundred thousand residents (e.g. using English Primary Care Trust geographies \[[@bib0115],[@bib0130]\]). Study size varied greatly: the largest comprised 4,805,390 women in all departments in mainland France \[[@bib0140]\], whilst the smallest was a single French department in the Normandy region of north-western France \[[@bib0145]\], involving 4940 women. Twelve of the thirteen studies examined mammography attendance \[[@bib0085], [@bib0090], [@bib0095],[@bib0115], [@bib0120], [@bib0125], [@bib0130], [@bib0135], [@bib0140], [@bib0145], [@bib0150],[@bib0160],[@bib0165]\], whilst the other reported mammography and clinical breast examination attendance \[[@bib0170]\]. A variety of methods were employed in the primary studies to measure deprivation: most of the papers used composite measures of deprivation; for example the Index of Multiple Deprivation \[[@bib0115], [@bib0120], [@bib0125], [@bib0130], [@bib0135]\], the French European Deprivation Index \[[@bib0095],[@bib0140]\], the Townsend Index \[[@bib0145],[@bib0175]\], or other composite indices \[[@bib0085],[@bib0090],[@bib0170]\]; the other studies used single indicators \[[@bib0150],[@bib0160],[@bib0165]\]. The methodological quality of nearly half (6/13) of the included studies was high -- with scores ranging from 2/8 (low) to 8/8 (high) (Appendix D).Fig. 1Prisma flow chart showing study selection.Fig. 1 3.1. Overall findings {#sec0040} --------------------- Regardless of indicator of area-level socio-economic deprivation, 11/13 studies (of which 10 were statistically significant) demonstrated a negative association between area-level deprivation and screening -- with women living in more socio-economically deprived neighbourhoods less likely to attend breast cancer screening. Eight studies also examined the gradient (comparing more than two areas), and six demonstrated a socio-spatial gradient (three of which were statistically significant). The results for each study are narratively synthesised by country. [Table 2](#tbl0010){ref-type="table"} gives an overview of study characteristics and main findings, while [Fig. 2](#fig0010){ref-type="fig"} summarises the results of the studies that included odds ratios for breast cancer screening uptake.Table 2Summary table of included studies.Table 2Study and source of fundingDesignYear(s)ScaleScreening intervalOutcomes(s)Measure of deprivationResults\ (Confidence intervals given in square brackets)Methodological qualityDeprivation effectGap (+/-)Gradient (✓/ x)EnglandMassat et al. \[[@bib0130]\]\ Funding: UK Department of HealthCross-sectionalApril 2011 -- March 2012District\ Unit of analysis: Primary Care Trusts3 years% of eligible women who had adequate mammography in the last 3 yearsPercentage deprivation using the Index of Multiple Deprivation (IMD) 2010Model: Population and general practice-level variables\ OR 0.991 \[0.985, 0.997\]8/8 (high)✓ -- \NRRenshaw et al. \[[@bib0135]\]\ Funding: Not listedCross-sectionalApril 2004 -- March 2007Neighbourhood\ Unit of analysis: LSOA3 yearsAttendance at breast screeningIncome quintile of the IMD 2004% Sceening attendance given in brackets\ 1. (least deprived): OR 1.00 (69.2%)\ 2. OR 0.92 \[0.91, 0.94\] (67.5%)\ 3. OR 0.82 \[0.81, 0.84\] (64.9%)\ 4. OR 0.70 \[0.69, 0.71\] (61.1%)\ 5. OR 0.52 \[0.51, 0.53\] (53.8%)6/8 (medium)✓ -- \✓ -- \Jack et al. \[[@bib0120],[@bib0125]\]\ Funding: The London Quality Assurance Centre and the Thames Cancer RegistryCross-sectionalMarch 2006 -- December 2009Neighbourhood\ Unit of analysis: LSOA3 yearsWomen's earliest invitation to screening was examined (split between first call invitation \[for those women aged 50-52\] and women aged 50-69 who had a routine recall invitation)Income domain of the IMD 20072014 article:\ First call (% attended):\ 1 (most affluent): 66%\ 2 67%\ 3 63%\ 4 60%\ 5 56%\ Routine recall (% attended):\ 1 (most affluent): 79%\ 2 79%\ 3 77%\ 4 74%\ 5 70%\ 2016 article:\ OR 0.95 p\<0.0016/8 (medium)\[[@bib0125]\]:\ First call:\ ✓ -- NR\ Routine recall:\ ✓ -- NR\ \[[@bib0120]\]:\ ✓ -- \\[[@bib0125]\]:\ First call:\ x\ Routine recall:\ ✓ NR\ \[[@bib0120]\]:\ xDouglas et al. \[[@bib0115]\]\ Funding: UK Department of HealthCross-sectional2007-2012District\ Unit of analysis: Primary Care Trusts3 years% of eligible women who had adequate mammography in the last 3 yearsIMD (2010) aggregated to PCT level2007-2008 (average percentage coverage in brackets):\ 1 (least deprived): RR 1.00 (79%)\ 2. RR 0.99 \[0.98, 1.00\] (77%)\ 3. RR 0.94 \[0.93, 0.95\] (73%)\ 4. RR 0.94 \[0.93, 0.95\] (74%)\ 5. RR 0.85 \[0.84, 0.86\] (66%)\ 2012-13:\ 1 (least deprived): RR 1.00 (78%)\ 2. RR 1.00 \[0.99, 1.01\] (78%)\ 3. RR 0.96 \[0.95, 0.97\] (75%)\ 4. RR 0.95 \[0.94, 0.96\] (74%)\ 5. RR 0.89 \[0.88, 0.90\] (70%)6/8 (medium)2007-8:\ ✓ -- \\ 2012-13:\ ✓ -- \2007-8:\ x\ 2012-13:\ xFrancePornet et al. \[[@bib0145]\]\ Funding: Not reportedCross-sectional2004 -- 2006Neighbourhood\ Unit of analysis: IRIS2 yearsScreening mammography within the duration of the studyThe Townsend IndexModel: 2\ 1 (least deprived): OR 1.00 (60.29%)\ 2. OR 0.94 \[0.74, 1.20\] (59.18%)\ 3. OR 0.90 \[0.70, 1.14\] (58.10%)\ 4. OR 0.82 \[0.66, 1.02\] (56.04%)\ 5. OR 0.71 \[0.59, 0.86\] (52.11%)8/8 (high)✓ -- \✓ NSOuedraogo et al. \[[@bib0095]\]\ Funding: 'La Ligue Contre le Cancer' and 'la 'Fondation de France'Cross-sectional2010 -- 2011Neighbourhood\ Unit of analysis: IRISNot reportedAttended invitation to mammography screening between 2010 and 2011The French European Deprivation IndexModel: Multilevel logistic regression analyses\ % Screening given in brackets\ 1 (least deprived): OR 1.00 (43.9%)\ 2. OR 0.94 \[0.87, 1.02\] (30.6%)\ 3. OR 0.84 \[0.78, 0.92\] (25.4%)8/8 (high)✓ -- \✓ NSDeborde et al. \[[@bib0140]\]\ Funding: Not reportedCross-sectional2013-2014District\ Unit of analysis: Municipality2 yearsAge-standardised participation rateThe French Deprivation Index% Attending screening:\ 1 (least deprived): 44.9%\ 2 54.2%\ 3 53.8%\ 4 54.8%\ 5 52.1%6/8 (medium)✓ + NRxGermanyLemke et al. \[[@bib0150]\]\ Funding: Not reportedCross-sectional2007 - 2008, 2009 - 2010, 2011-2012Neighbourhood\ Unit of analysis: Not specified2 yearsParticipation in screening for each two year periodUnemployment rate (foreigner %)Model: Multivariable analysis\ Unemployment rate (foreigner) OR 0.94 \[0.90, 0.98\]7/8 (high)✓ -- \NRItalyGiuliani et al. \[[@bib0170]\]\ Funding: San Paolo FoundationCross-sectional1990 -- 2000 (the time period patients were diagnosed with breast cancer), followed up until 2010Neighbourhood\ Unit of analysis: Level of census sectionScreening for breast cancer survivorsYearly mammogram and CBEDeprivation indexAffluent (reference) class: OR 1.00\ Deprived class: OR 0.81 \[0.65, 1.00\]8/8 (high)✓ -- NSNRThe NetherlandsAarts et al. \[[@bib0085]\]\ Funding: Bevolkings Onderzoek Borstkanker ZuidCross-sectional1998 -- 2005Neighbourhood\ Unit of analysis: Postcode2 yearsScreening mammography attendanceIndicator of SES (based on house value and income)Low SES: OR 1.00 (79%)\ Intermediate: OR 1.54 \[1.5, 1.6\] (85%)\ High: OR 1.75 \[1.7, 1.8\] (87%)6/8 (medium)✓ -- \✓ \*SwedenLagerlund et al. \[[@bib0090]\]\ Funding: Not reportedCross-sectional2005 -- 2009Neighbourhood\ Unit of analysis:\ SAMs1.5-2 yearsNon-attendance to most recent mammography screening during 2005 to 2009Neighbourhood sociodemographic indexModel: 3\ Lowest % neighbourhood sociodemographic index (decile group 1: least deprived): OR 1.00\ Decile group 5: OR 1.26 \[1.07, 1.51\]\ Highest % neighbourhood sociodemographic index (decile group 10): OR 1.92 \[1.63, 2.34\]8/8 (high)✓ -- \✓ \TurkeyDundar et al. \[[@bib0160]\]\ Funding: NoneCross-sectional2008 -- 2009Regions\ Unit of analysis:\ Regions of Mansia2 yearsAttended screening between 2008 -- 2009Education, perceived family incomeAttendees %\ Urban (70.5%), slum (81.6%)2/8 (low)✓ + \NROzmen et al. \[[@bib0165]\]\ Funding: Not listedCross-sectionalDates not statedCity\ Unit of analysis:\ City (Mus) and county (Bahcesehir) in IstanbulOpportunistic screeningScreening mammography in the last two yearsLiteracy, graduation, working status, monthly incomeAttendees %\ Mus -- more deprived (35%)\ Bahcesehir (49%)2/8 (low)✓ -- \NR[^1][^2][^3][^4][^5][^6][^7]Fig. 2The association between area level socio-economic deprivation and breast cancer screening uptake for studies reporting odds ratios.Fig. 2 3.2. Country specific findings {#sec0045} ------------------------------ ### 3.2.1. England (n = 4) {#sec0050} Four studies (from five articles) examined the association between area-level deprivation and uptake of the English Breast Screening Programme (one high methodological quality, three medium). All four studies \[[@bib0115], [@bib0120], [@bib0125], [@bib0130], [@bib0135]\] found a negative association between screening uptake and area-level deprivation. The papers by Jack et al. \[[@bib0120],[@bib0125]\] observed both a gap and a gradient for the proportion of women who attended a first call for screening (significance not reported). The work by Massat et al. \[[@bib0130]\] found a statistically significant deprivation gap in uptake. Renshaw et al. \[[@bib0135]\] showed a gradient, whereby screening attendance was highest amongst the most affluent and fell significantly as deprivation levels increased. Douglas et al. \[[@bib0115]\] observed screening rates over a five year period and demonstrated a gap in all years and a gradient in all but one of the years. These findings suggest that there is a reduction in area-based screening inequalities from the years 2007--2008 to 2012-2013. ### 3.2.2. France (n = 3) {#sec0055} Two of the three French studies (considered to have high methodological quality), found a negative association between deprivation and screening uptake. Ouedraogo et al. \[[@bib0095]\] and Pornet et al. \[[@bib0145]\] found a significant gap where women living in the most deprived neighbourhoods were significantly less likely than those living in the most affluent to participate in breast cancer screening. They also noted a gradient, but these were not statistically significant in either study. The study by Deborde et al. \[[@bib0140]\], using a larger unit of analysis (municipalities), found an inverted U-curve relationship between deprivation and screening participation, with highest screening found in the intermediate quintiles. ### 3.2.3. Germany (n = 1) {#sec0060} The one high-quality study by Lemke et al. \[[@bib0150]\] found that in areas with a higher employment rate, women were more likely to attend breast cancer screening. However, this association was only statistically significant for foreigner unemployment rate, and not for other types of unemployment. ### 3.2.4. Italy (n = 1) {#sec0065} A high-quality study by Giuliani et al. \[[@bib0170]\] only included participants who had been diagnosed with either in situ or invasive breast cancer and received surgical intervention. The results showed that those in the deprived group were less likely to attend yearly mammography screening and/or clinical breast examination compared to those in the reference group (but this gap was not statistically significant). ### 3.2.5. Netherlands (n = 1) {#sec0070} Aarts et al. \[[@bib0085]\] explored breast cancer screening uptake in southern Netherlands by tertiles of deprivation. This medium-quality study found a statistically significant gap and gradient in breast cancer screening uptake. ### 3.2.6. Sweden (n = 1) {#sec0075} Lagerlund et al. \[[@bib0090]\] explored non-attendance at mammography screening and neighbourhood sociodemographic characteristics in southwest Sweden. Both a gap and gradient effect was observed (significance not given) in this high-quality study: decreasing deprivation was associated with an increase in the proportion of women who attend screening appointments. ### 3.2.7. Turkey (n = 2) {#sec0080} Two studies conducted in Turkey were identified, both of which were rated as low in quality. Dundar et al. \[[@bib0160]\] found that women in a more affluent, urban district were significantly less likely to attend the breast cancer screening programmes compared with those in the deprived district. Another study, by Ozmen et al. \[[@bib0165]\], compared the mammography screening behaviours of women in two socio-economically contrasting cities -- with those in the more affluent city significantly more likely to have had a mammogram within the last two years. 4. Discussion {#sec0085} ============= This review has found consistent, medium-high quality evidence of a negative association between area-level socio-economic deprivation and breast cancer screening uptake in Europe (i.e. as area level socio-economic deprivation increases, breast cancer screening uptake decreases). This work updates a previous systematic review exploring the association between breast cancer, colorectal cancer, and cervical cancer screening and area-level deprivation \[[@bib0070]\]. In this systematic review, which only included studies from outside Europe, thirteen studies focused on breast cancer screening; eight of which showed significant positive associations between SES and screening uptake. Our work, which focused on studies in Europe, supports the findings of the previous review by Pruitt et al. and provides evidence that even in the more universal health systems of Europe, women living in the least deprived areas are more likely to attend breast cancer screening than women living in the most deprived areas. Our work also builds on previous findings that as individual level SES increases, the uptake of breast cancer screening also increases \[[@bib0180],[@bib0185]\], although the relationship between individual and area-level factors are thought to be complex with one not being a simple proxy for the other in terms of breast cancer risk \[[@bib0190]\]. Qualitative research provides contextual information to understand the reasons for low rates of screening in deprived areas. Evidence suggests that women living in areas of high deprivation had limited knowledge about mammography screening programmes or had misconceptions regarding cancer and mammography \[[@bib0060],[@bib0195]\]. Embarrassment, fear and inconvenience have also been cited as possible reasons for low screening rates in these communities \[[@bib0200]\]. Furthermore, passive and practical barriers have been highlighted by an Australian Government Report \[[@bib0205]\] suggesting women in deprived communities find it difficult to attend mammography, because of transport (cost, availability) or family commitments. It is possible that the reduced screening uptake in deprived areas contributes to the 'breast cancer paradox', whereby rates of breast cancer survival are reduced in deprived communities, despite there being lower disease incidence. Although there is debate about the role of breast cancer screening, screening does allow for early disease detection, which can result in earlier treatment initiation and better outcomes for patients. By not attending routine breast cancer screening, it is possible that the disease would be discovered at a more advanced stage, potentially resulting in poorer outcomes for patients. Although we acknowledge that reduced screening uptake in deprived communities is not solely responsible for the higher breast cancer mortality rate. With respect to disease incidence, high SES enhances certain behaviours that reduce breast cancer incidence, including reduced levels of smoking, lower alcohol consumption and increased physical activity. There are, however, other factors that increase breast cancer incidence in women of high SES, such as increased nulliparity, having children at an older age, and the use of the combined oral contraceptive pill \[[@bib0210], [@bib0215], [@bib0220], [@bib0225]\]. It is thought that these latter factors contribute to the increased incidence of breast cancer in women of higher SES. Our findings are in line with the fundamental cause theory \[[@bib0230]\], whereby people with greater access to resources, including knowledge, money, power, prestige, and beneficial social connections are better able to take advantage of effective innovations in disease prevention and treatment. The theory predicts that, in people of higher SES, whenever technology and medical knowledge are available to manage a particular disease, the mortality rates of that disease will be lower. It is imperative then, that national breast cancer screening strategies across Europe and elsewhere, recognise the negative association of breast cancer screening uptake with area-level deprivation, and, potentially, how this could affect breast cancer mortality. Increasing knowledge, by targeting women of low socio-economic status and informing them of the benefits (and risks) of breast cancer screening and also providing practical steps to ensure mammography screening is fully accessible in deprived areas (e.g. by the use of mobile clinics) are thus potential routes to increase uptake. Small scale studies have demonstrated it is possible to achieve this; for example, Chambers et al. showed that a brief telephone support intervention doubled breast cancer screening attendance in women from deprived areas in Scotland who had missed their initial screening appointment \[[@bib0235]\]. Another study, undertaken by Kerrison et al. \[[@bib0240]\], showed that a text message reminder increased the uptake of routine breast cancer screening appointments for people living in the London Borough of Hillingdon. The findings of also suggested that text message reminders are particularly effective for increasing screening attendance among patients from deprived areas. Moving forward, it is important for these interventions to be tested at a national level and, if successful, be implemented into breast cancer screening policy. This review is subject to some limitations. Firstly, we only included studies from European countries -- we did not include studies from other high-income countries, such as the USA and Canada, nor did we analyse our findings according to healthcare system type \[[@bib0245],[@bib0250]\]. Our findings, therefore, may not be generalisable to all types of healthcare systems. Secondly, this review is subject to the usual limitations of observational research whereby we cannot claim that there is a causal relationship between area-level disadvantage and breast cancer screening uptake. It is not clear how this association impacts on overall breast cancer survival, if at all -- this limitation is even more pertinent in view of the debate around the risks and benefits of breast cancer screening \[[@bib0255]\]. Thirdly, other limitations due to practical resources include not being able to search for grey literature and unpublished studies. The studies included were also limited to the English language. Finally, as an evidence base we have only found a limited number of studies, and therefore it is possible that there is publication bias (that negative results are less likely to be published) and selective reporting within studies which may impact the assessment of risk of bias. 5. Conclusion {#sec0090} ============= The findings from this review provide evidence that there is lower uptake of breast cancer screening in areas experiencing higher levels of socio-economic deprivation in Europe. Strategies to improve breast cancer screening attendance should be aimed at more deprived areas, to increase individual participation and decrease inequalities in the general population. Further research should explore how the reduced breast cancer screening uptake in areas of high socio-economic deprivation impacts on breast cancer survival and the breast cancer paradox. Funding {#sec0095} ======= The project is supported by Fuse (UKCRC Centre for Translational Research in Public Health) (MRC grant ref. no. MR/K02325X/1). CB is a Fuse Senior Investigator, AT and KT are Associates and KT has been part-funded as a Research Associate. Funding for Fuse comes from the British Heart Foundation, Cancer Research UK, Economic and Social Research Council, Medical Research Council, and the National Institute for Health Research, under the auspices of the UK Clinical Research Collaboration, and is gratefully acknowledged. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Declarations of interest {#sec0100} ======================== None. Authorship contribution {#sec0105} ======================= DS, AT and CB contributed substantially to the conception and design of the project. DS and KT were responsible for the acquisition of data and the analysis. All authors contributed to the interpretation of data. All authors contributed to drafting the article and revising it critically for important intellectual content. All authors have provided the final approval of the version to be published. Availability of data and materials {#sec0110} ================================== All data generated or analysed during this study are included in this published article (and its additional files). The extraction forms for all the individual primary studies are available from the corresponding author on reasonable request. Ethics approval and consent to participate {#sec0115} ========================================== Not applicable. Only published primary studies were included in this review. Appendix A. Supplementary data {#sec0125} ============================== The following is Supplementary data to this article: Supplementary material related to this article can be found, in the online version, at doi:<https://doi.org/10.1016/j.canep.2019.03.008>. [^1]: OR indicates odds ratio. [^2]: RR indicates relative risk. [^3]: \- indicates a negative association between area-level deprivation and breast cancer screening (i.e. the higher area-level deprivation, the lower the uptake of breast cancer screening). [^4]: \- indicates a positive association between area-level deprivation and breast cancer screening (i.e. the higher area-level deprivation, the higher the uptake of breast cancer screening). [^5]: NR indicates data not reported in study. [^6]: \*Significance at p \< 0.05. [^7]: NS indicates gap/gradient not significant.	{ "pile_set_name": "PubMed Central" }
Background ========== Tendon injuries are a significant cause of morbidity in both man and veterinary species and are reported to represent 30% of the musculoskeletal caseload in a one year study of human general practitioners \[[@B1]\]. Injury to the equine superficial digital flexor tendon (SDFT) poses a significant problem amongst racing Thoroughbreds with a reported incidence of 11--43% \[[@B2],[@B3]\]. The extracellular matrix (ECM) of flexor tendons has evolved not only to both transmit forces from muscle to bone but also as an elastic energy store for efficient locomotion \[[@B4]\]. However, once a tendon has been injured a fibrous repair response ensues which, whilst being very efficient at repairing the damaged tissue, does not regenerate the original tendon matrix \[[@B5]\]. The fibrous scar tissue does not recapitulate the unique parallel collagen fibre alignment found in normal tendon. Consequently, the healed tendon does not retain the biomechanical properties of the original tendon prior to injury \[[@B5]\] and re-injury rates in horses can be as high as 56% \[[@B6]\]. The poor clinical outcome associated with tendon injury and the limited capacity for regeneration of injured tendon have resulted in a growing interest in the use of tissue engineering approaches for tendon therapy in both man and animals \[[@B7]\]. Objective demonstration of successful regeneration requires the identification of markers of tenogenesis. However, currently there are no specific molecular markers that can be used to characterise tendon fibroblasts \[[@B8]\] and identify relevant differentiation and repair. Verification of the success of tendon tissue engineering interventions is currently based on histological analysis and mechanical testing \[[@B9]\]. Identification of pertinent gene expression would be beneficial in confirming cell differentiation to a relevant tendon cell phenotype. Expression of key matrix genes within tissue engineered constructs has also been used to identify such differentiation, although validation of the relevance of the candidate marker genes has not yet been performed \[[@B10]\]. Unfortunately many of these key matrix genes are expressed in a variety of mesenchymal tissues and therefore may not be sufficiently discriminatory. In contrast, the genes COL2A1, COL10A1 and SOX9 are well accepted as being representative of chondrogenic differentiation \[[@B11]\] and Runx2 and osteopontin are discriminating for osteogenic differentiation \[[@B11]\]. In the current study eleven genes were selected as representative of tendon, cartilage and bone and quantified in these musculoskeletal tissues to identify which of these genes were most discriminating for a tendon phenotype. Collagen type I forms 95% of the collagen content of normal adult tendons, the remaining 5% constitutes small amounts of collagen types III, V, VI, XII and XIV \[[@B12]\]. Following acute rupture of the human Achilles tendon or equine SDFT gene expression of collagen types I, III and V is increased \[[@B13]-[@B16]\]. Tenascin-C is also up regulated following tissue wounding \[[@B17]\] and in degenerate tendinopathy \[[@B18]\]. Other important tendon matrix components include the small leucine rich proteoglycans (lumican, decorin, biglycan and fibromodulin). The role of these smaller matrix components has been demonstrated to be important in collagen fibrillogenesis \[[@B19]\]. In addition the glycoproteins such as cartilage oligomeric matrix protein (COMP) and tenascin-C are understood to be involved in collagen fibrillogenesis \[[@B20]-[@B22]\]. COMP is an extracellular matrix glycoprotein that is abundant in tissues subjected to load. Levels of COMP increase until skeletal maturity is reached after which time levels gradually decline \[[@B20]\]. Tenomodulin is a type II transmembrane glycoprotein that is preferentially expressed in dense connective tissues. Mice lacking tenomodulin have tendons with a disrupted fibril structure and exhibit severely reduce tendon fibroblast proliferation \[[@B23]\]. Tenomodulin has been reported to be a good phenotype marker for tendon fibroblasts \[[@B24]\]. Decorin is a member of the SLRPs that have been shown to be important in control of collagen fibrillogenesis \[[@B25]\]. Recent investigations have highlighted scleraxis as a specific marker of tendon progenitor cells \[[@B26]-[@B28]\]. Furthermore, scleraxis null mutant mice have distinct tendon defects \[[@B26]\]. Runx2 is an osteogenic transcription factor \[[@B11]\]. Osteopontin is a glycoprotein found in bone and forms part of the inorganic component of bone. Osteomodulin is a keratan sulphate proteoglycan found in the ECM of bone and thought to be a marker of osteoblast maturation \[[@B29]\]. SOX9 is a transcription factor important in chondrocyte regulation and is frequently used to identify chondrogenic differentiation \[[@B11]\]. The study performed here was designed to identify the presence of the eleven genes in normal tendon and quantify their expression levels in tendon compared to other musculoskeletal tissues. The expression of key genes which could identify an adult tendon phenotype were then characterised both in clinical cases of tendinopathy of the equine SDFT, as well as during development. Furthermore the expression profile of these genes was characterised in-vitroin both monolayer and three dimensional (3D) cultures of tendon fibroblasts to identify whether such models fully recapitulate the tendon phenotype. Methods ======= Tissue samples -------------- Tissue samples were obtained from animals subjected to euthanasia for clinical reasons other than orthopaedic disease. Full informed written consent was obtained from animal owners for all tissue collection and tissue collection was subject to ethical review. Three samples (10 × 5 × 2 mm) of equine superficial digital flexor tendon were obtained from the mid-metacarpal region from skeletally mature (aged 4--10 years) horses free of clinical orthopaedic disease. Matched cartilage samples (10 × 3 × 0.5 mm) were harvested from the articular surface of the distal aspect of the third metacarpal bone of the same donors with bone (10 × 5 × 2 mm) being harvested from the distal metacarpal epiphysis. Further samples were collected from five skeletally immature animals (three mid-gestational foetuses and two yearlings). All tissue specimens were grossly normal on post mortem examination. Pathological tissue specimens were collected from the mid-metacarpal region of the SDFT of 3 horses with acute (\< 6 weeks duration) and 3 horses with chronic (\> 6 months duration) tendinopathy. All samples were obtained within 4 hours of euthanasia and collected into RNAlater™ (Ambion, Applied Biosystems). Samples were stored at 4°C for 24 hours and then frozen at -80°C until used for RNA extraction. Monolayer cell culture ---------------------- For the cell culture experiments tendon samples were collected from 4 adult (age 4--10 yrs), mixed breed horses subjected to euthanasia. All tendons were free of pathology on clinical and post mortem examination. Tendons were collected from the SDFT at the level of the mid-metacarpus (tensional), within 4 hours of euthanasia. Samples were cut into 2 × 2 × 2 mm^3^pieces and subjected to collagenase type II (0.1%) digestion overnight in DMEM (Sigma) containing 5% FCS, penicillin/streptomycin and amphotericin) on an orbital shaker at 37°C. The isolated tendon fibroblasts were cultured in 75 cm^2^flasks in DMEM containing 10% FCS, penicillin/streptomycin and fungizone at 5% CO~2~and 37°C until approximately 90% confluent before passaging (usually 4--6 days for each passage). Cells were released from flasks with trypsin (0.05%) and re-seeded into flasks at 5000 cells/cm^2^. Culture medium was changed every 3--4 days. One millilitre of Tri-Reagent (Sigma) was added to the remaining cells for subsequent RNA extraction, these samples were stored at -80°C. The cells were passaged 5 times for the 3 different horses. Three dimensional cell culture in collagen gels ----------------------------------------------- Tendon fibroblasts harvested from the tensional area of the SDFT as described above were cultured until confluent. Collagen gels were prepared following a method described previously \[[@B30]\]. Briefly, following trypsinisation at the end of passage one, cells were counted and suspended in 90% type I collagen gel (2 mg/ml) (Invitrogen), 10% 10 × DMEM (Sigma) using 1 × 10^6^cells/mL. Two hundred microlitres of the collagen cell suspension were cast in a trough mold (Trough loader, Flexcell International) centrally located in a Tissue train culture plate (Flexcell International). The collagen gels were cultured at 5% CO~2~in a humidified incubator at 37°C for 24 hours prior to the addition of DMEM containing 10% FCS, penicillin/streptomycin and amphotericin. Cell seeded collagen gels were then cultured without dynamic load for 5 days. RNA extraction and quantification --------------------------------- RNA was extracted using phenol/guanidine HCl reagents (TriReagent™, Sigma) and isolated using the published methods \[[@B31],[@B32]\]. Tissue samples were pulverised for 1 minute at 2000 oscillations/minute in a liquid nitrogen cooled dismembranator (Braun Mikro-Dismembrator Vessel, Braun Biotech International, Melsungen, Germany). A 1 mL aliquot of phenol/guanidine HCl reagents (TriReagent™, Sigma) was added to the powdered tendon, cartilage or bone. Two hundred microlitres of chloroform was added to the microcentrifuge tubes prior to centrifugation at 12,000 g for 10 minutes. RNA was extracted from the aqueous phase using a commercially available RNA purification kit (RNeasy Mini Kit, Qiagen). Digestion of DNA was carried out using a commercially available kit (RNase-Free DNase Set, Qiagen). RNA was stored at -80°C prior to reverse transcription. Reverse transcription --------------------- The purified RNAs were measured (ND-1000 spectrophotometer Nanodrop Technologies) and then 1 μg RNA was used to prepare cDNA using M-MLV reverse transcriptase and random primers according to the manufacturer\'s instructions (Promega). Samples were stored at -20°C prior to relative quantification of gene expression. Primer Design ------------- Transcript sequences were obtained from the National Centre for Biotechnology Information (Bethesda, MD, USA) (Tables [1](#T1){ref-type="table"} &[2](#T2){ref-type="table"}). Equine gene sequences were aligned to human, bovine and canine sequences using online software (<http://www.ebi.ac.uk/Tools/clustalw/>, <http://www.ensembl.org/>) to predict exon boundaries. Primers were designed using Primer Express (Applied Biosystems) software and selected to span predicted exon boundaries where possible. BLAST searches were performed for all sequences to confirm gene specificity. Target and reference gene primers were synthesized by Eurogentec. All primers were validated using a standard curve of five serial dilutions so that all primer efficiencies were between 95--105% (Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). ###### Reference gene primer sequences used for qPCR ---------------------------------------------------------------------- Gene SCS Efficiency (%) Sequence ---------- ------- ---------------- ---------------------------------- GapDH\ -3.32 100.2 5\'GCATCGTGGAGGGACTCA3\'\ AF157626 3\'GCCACATCTTCCCAGAGG5\' 18S\ -3.24 103.3 5\'GGCGTCCCCCAACTTCTTA3\'\ AJ311673 3\'GGGCATCACAGACCTGTTATTG5\' ACTB\ -3.33 99.9 (Bogaerts et al., 2006)\ AF035774 5\'CCAGCACGATGAAGATCAAG3\'\ 3\'GTGGACAATGAGGCCAGAAT5\' SDHA\ -3.29 101.5 5\'ACAGAGGAATGGTCTGGAATACTGA3\'\ DQ402987 3\' GTGAGCACCACGTGACTCCTT5\' ---------------------------------------------------------------------- SCS = Standard Curve Slope ###### Target gene primer sequences use for qPCR ----------------------------------------------------------------------------- Gene SCS Efficiency (%) Sequence -------------- ------- ---------------- ------------------------------------- COL1A2\ -3.31 100.5 5\'GCACATGCCGTGACTTGAGA3\'\ XM_001492962 3\'CATCCATAGTGCATCCTTGATTAGG5\' COL2A1\ -3.21 105.0 5\'TCAAGTCCCTCAACAACCAGATC3\'\ NM_001081764 3\'GTCAATCCAGTAGTCTCCGCTCTT5\' COL3A1\ -3.22 104.6 5\'ACGCAAGGCCGTGAGACTA3\'\ XM_001501719 3\'TGATCAGGACCACCAACATCA5\' COL10A1\ -3.39 97.4 5\'TGCCCAGTGGACAGGTTTCT 3\'\ XM_001504101 3\'GTCTTTTCGTTTCTAGTCAGATTTTGAA5\' COMP\ -3.42 95.7 5\'GGTGCGGCTGCTATGGAA3\'\ AF325902 3\'CCAGCTCAGGGCCCTCAT5\' DCN\ -3.31 100.1 5\'CATCCAGGTTGTCTACCTTCATAACA3\'\ AB106279 3\'CCAGGTGGGCAGAAGTCATT5\' TNC\ -3.36 98.6 5\'GGGCGGCCTGGAAATG3\'\ XR_035757 3\'CAGGCTCTAACTCCTGGATGATG5\' TMD\ -3.38 97.4 5\'ACGTGACCATGTATTGGATCAATC3\'\ AB059407 3\'CACCATCCTCCTCAAAGTCTTGT5\' SIX1\ -3.23 104.0 5\' GATGCCCCAATGTTTGTGATG3\'\ XM_001492786 3\' AGGAGGCATTGCTGACAATCTT5\' SCXB\ -3.35 98.8 5\'TCTGCCTCAGCAACCAGAGA3\'\ NM_001105150 3\'TCCGAATCGCCGTCTTTC5\' SOX9\ -3.33 99.9 5\' CTTTGGTTTGTGTTCGTGTTTTGT3\'\ XM_001498424 3\'AGAGAAAGAAAAAGGGAAAGGTAA GTTT5\' OPN\ -3.33 99.9 5\'CGCAGATCTGAAGACCAGTATCCT3\'\ XM_001496152 3\'TGCTTTCCACAGGTGATGTGA5\' RUNX2\ -3.36 98.4 5\'CTGGGCCATGTGTATGATTTGT3\'\ XM_001502519 3\'TTTTGACCTGATATAGAGTGCATGGT5\' OMD\ -3.26 103.4 5\'CAAATTCATCAACCCCTGAAA3\'\ XR_035840 3\'CTTCATCTGGCTCTTGGTCA5\' ----------------------------------------------------------------------------- Normalisation of reference genes -------------------------------- The most stable pair of reference gene primers were selected using the normalization strategy proposed by \[[@B33]\]. Verification of reference gene selection was carried out using Normfinder \[[@B34]\]. Reference gene stability packages were downloaded from geNorm: <http://medgen.ugent.be/~jvdesomp/genorm/> and Normfinder: <http://www.mdl.dk/publicationsnormfinder.htm/>. RT-PCR ------ Real time RT (qPCR) assays were performed in triplicate using the 7900 HT Fast Real-Time PCR System (Applied Biosystems; Warrington, UK) in 384 well plates. Reaction volume in each well was 10 μl (4.6 μl of cDNA, 5 μl of Power SYBR mastermix (Applied Biosystems), 0.1 μl of DEPC water 0.15 μl of 3 μM forward primer and 0.15 μl of 3 μM reverse primer). The cycling conditions comprised 10 min polymerase activation at 95°C and 40 cycles at 95°C for 15 sec and 60°C for 60 sec. Data was then analysed using Sequence Detection Systems Software v2.2.1 (Applied Biosystems; Warrington, UK). Statistical analyses -------------------- All data were presented as mean ± SE. All data departed from normality and were therefore log~10~transformed. Significant differences in gene expression between cartilage, tendon and bone matched samples were identified using a mixed effects linear regression model to allow for the clustering of samples within individual horses (S-Plus software). Comparisons between gene expression of normal tissue and developing or diseased or cultured cells were carried out using a two sample student\'s t-test where log transformed data were normally distributed. The level of significance was set at P \< 0.05. Results ======= Normalisation of reference genes -------------------------------- GeNorm and Normfinder identified Glyceraldehyde -3-phosphate dehydrogenase (GapDH) and succinate dehydrogenase subunit A (SDHA) to be the two most stably expressed reference genes across normal tendon, cartilage and bone samples, and both the 2D and 3D in vitrocultures (M = 0.045, V = 0.027). The most stably expressed reference genes in tendon development and disease were beta actin (ACTB) and GapDH (M = 0.056, V = 0.057) (Table [3](#T3){ref-type="table"}). When all samples were considered together GapDH was identified as the most stable gene (Normfinder ratio 0.050) and ACTB the second most stable gene (Normfinder ratio 0.072) (Table [3](#T3){ref-type="table"}). GeNorm identified GapDH and ACTB to be the most stable pair of reference genes (M = 0.127, V = 0.048) (Table [3](#T3){ref-type="table"}). As geNorm and Normfinder were in agreement, gene expression data was normalised to the average expression of GapDH and ACTB. The average Ct values used to identify the most stable reference genes can be seen in Additional file [1](#S1){ref-type="supplementary-material"}. ###### M and V values generated by geNorm and Normfinder --------------------------------------------------------------------------------- Sample Reference\ M Value\ V Value\ Normfinder Genes (Gene stability) (Pairwise stability) ------------- ------------ ------------------ ---------------------- ------------ Tissue GapDH\ 0.045 0.027 0.012\ SDHA 0.018 Developing\ GapDH\ 0.031 0.020 0.011\ Tissue ACTB 0.032 Diseased\ ACTB\ 0.056 0.057 0.019\ Tissue GapDH 0.068 Monolayer\ GapDH\ 0.059 0.085 0.020\ Culture SDHA 0.062 Collagen\ GapDH\ 0.017 0.013 0.006\ Gel SDHA 0.020 All Samples GapDH\ 0.127 0.048 0.050\ ACTB 0.072 --------------------------------------------------------------------------------- Coefficients for tissue samples (tendon, cartilage and bone), developing tendon, diseased tendon, monolayer culture and 3D collagen gels. Tissue samples ============== Adult mesenchymal tissues ------------------------- Figure [1](#F1){ref-type="fig"} shows the data obtained for expression levels of genes presumed to be associated with tendon (a), bone (b) and cartilage (c). Scleraxis showed significantly higher expression in tendon than in bone (P = 0.002) (Figure. [1a](#F1){ref-type="fig"}), and whilst higher levels of expression were identified in tendon than cartilage this did not reach statistical significance. As expected, significantly higher COL1A2 was expressed in both tendon and bone than in cartilage (P = 0.01 and P = 0.008, respectively) (Figure [1a](#F1){ref-type="fig"}). Tenomodulin was identified in both tendon and bone but could not be detected in cartilage. No significant difference could be identified in the levels tenomodulin expression in tendon and bone. Tenascin-C expression was significantly lower in tendon than in bone (P = 0.02). COMP expression was significantly higher in tendon than in bone (P = 0.02) but no difference in expression was identified between tendon and cartilage (Figure. [1a](#F1){ref-type="fig"}). Decorin expression was highest in cartilage (P = 0.03) and lowest in bone (P = 0.02). Osteopontin, Runx2 and osteomodulin were able to discriminate between tendon and bone (P = 0.003, P = 0.03 and P = 0.01, respectively) (Figure. [1b](#F1){ref-type="fig"}) but Runx2 was the only hypothesised bone marker that distinguished between bone and cartilage (P = 0.007). Of the cartilaginous markers, only COL2A1 was significantly lower in tendon than in both bone and cartilage (P = 0.005 and P = 0.0003). Cartilage showed higher expression of COL10A1 and SOX9 than tendon (P = 0.01 and P = 0.005 respectively) (Figure. [1c](#F1){ref-type="fig"}). From the matched normal tissue samples a panel of COL1A2, scleraxis and tenascin-C were selected as the most discriminating genes of tendon phenotype. ![Gene expression of normal tendon, bone and cartilage samples. (a) Gene expression of COL1A2, scleraxis, tenascin-C, COMP and decorin in matched samples of tendon, cartilage and bone harvested from normal adult horses. (b) Gene expression of COL2A1, COL10A1 and SOX9 in matched samples of tendon, cartilage and bone harvested from normal adult horses. Relative gene expression data is represented graphically as log transformed values. P values were generated using a mixed effects linear regression model to allow for clustering within individual donors. (c) Gene expression of osteopontin, osteonectin and osteomodulin in matched samples of tendon, cartilage and bone harvested from normal adult horses.](1471-2474-10-27-1){#F1} Tendon development ------------------ Expression of COL1A2 was higher in the foetal tissues compared to mature tendon relative to reference gene mRNA (P = 0.01). There was increased tenascin-C expression in skeletally immature tendon in comparison to adult tendon (P = 0.02) (Figure. [2a](#F2){ref-type="fig"}). ![*Gene expression of developing and diseased tendon and in vitro gene expression**. (a) Gene expression of COL1A2, scleraxis, tenascin-C in SDFT collected from foetuses, yearlings and adults. (b) Gene expression of COL1A2, scleraxis, tenascin-C in acute and chronic disease in comparison to normal adult tendon. (c) Gene expression of COL1A2, scleraxis, tenascin-C in 2D and 3D in vitroculture in comparison to normal adult tendon. (2D P1 = passage 1 monolayer culture, 2D P5 = passage 5 monolayer culture, 3D = tendon fibroblasts cultured in collagen gels). P values were generated using a two sample student\'s t-test for normally distributed log transformed data.](1471-2474-10-27-2){#F2} Tendon disease -------------- Acute tendinopathy produced significant changes in gene expression. Tenascin-C expression was greatly increased in acutely diseased tendon in comparison to normal tendon (P = 0.0001), although expression of scleraxis and COL1A1 did not vary. Chronic tendinopathy resulted in a gene expression profile similar to that of normal tissue with no significant difference in the expression of tenascin-C (Figure. [2b](#F2){ref-type="fig"}). Cell culture samples ==================== Monolayer cell culture ---------------------- Scleraxis gene expression was decreased compared to normal adult tendon at passage 1 of monolayer culture of tendon fibroblasts and this significant decrease was maintained at passage 5 (P = 0.002 and P = 0.004 respectively) (Figure. [2c](#F2){ref-type="fig"}). No significant differences in COL1A2 and tenascin-C expression were identified between normal tissue and the 2D in vitroculture systems. Tenomodulin gene expression levels were not detectable in the in vitrocells. Three dimensional cell culture in collagen gels ----------------------------------------------- Placing tendon fibroblasts into a three dimensional collagen matrix resulted in increased expression of type I collagen (P = 0.03) compared to passage one monolayer cells. No significant difference was identified between levels of COL1A2 and tenascin-C expression by 3D cultured tendon fibroblasts and normal adult tendon. However, 3D cultured tendon fibroblasts showed significantly reduced expression of scleraxis (P = 0.03) in comparison to normal adult tendon (Figure [2c](#F2){ref-type="fig"}). Discussion ========== This study has confirmed that a panel of \'marker\' genes are required to identify tendon cell phenotype from other mesenchymal tissues. The matched adult samples of tendon, cartilage and bone show the tensional SDFT to express high levels of COL1A2 and scleraxis and low levels of tenascin-C in comparison to the other tissues. Bone also expresses high levels of COL1A2 but contrastingly low levels of scleraxis and high levels of tenascin-C. In contrast, cartilage expresses moderate amounts of COL1A2, scleraxis and tenascin-C. Tendon regions with a fibrocartilagenous phenotype that are subjected to compressive loads have been shown to have to have higher levels of tenascin-C \[[@B35]-[@B37]\], for example at the myotendinous and osteotendinous junctions \[[@B38]\]. Low levels of tenascin-C within the mid-body of the normal equine SDFT identified in the current study are likely to be a reflection of the tensile loads placed on this region of the tendon. The presence of scleraxis within equine cartilage may be a consequence of the inclusion of the perichondrium in the samples cartilage, a tissue that is derived from scleraxis expressing cells \[[@B39]\]. Scleraxis has been described as being an important marker of tendon neoformation however, there is currently no evidence that scleraxis can induce tendon neoformation \[[@B40]\]. Murchison et al (2007) clearly demonstrated the importance of scleraxis as a transcription factor during tendon development with scleraxis null mutants exhibiting severe defects in the force transmitting tendons. It has been suggested that scleraxis is important in directing condensations of tendon progenitor cells to form the force transmitting tendons \[[@B26]\]. The current experiments did not identify any significant increase in levels of scleraxis expression in mid-gestational foetal tendons compared to normal adult tendons. It is possible that once the tendon phenotype is established during development there is no further change in scleraxis expression. Alternatively, translational control of the protein may differ between developing and adult tendons. Scleraxis expression showed no significant difference between normal and diseased tendons. Tendon samples collected from horses with naturally occurring acute tendinopathy of the SDFT did not show a significant increase in COL1A2, this may be a result of the large range in expression levels in the normal adult tendon samples, leading to large standard errors. In man, increased expression of type I collagen has been a consistent finding in acute tendinopathy \[[@B41]\]. Tenascin-C was significantly increased in the acutely diseased samples; this is consistent with the appearance of tenascin-C during the inflammatory phases of wound healing \[[@B17],[@B21]\]. Increased type-I collagen has been reported in chronic degenerate Achilles tendinopathy \[[@B42],[@B43]\] this is at variance with the findings of the current study that identified no significant difference between gene expression of COL1A2 in normal adult tendon and tendinopathy of the SDFT of more than 6 mths duration in the horse. This may reflect differences in clinical condition between painful degenerate Achilles tendons and healed non-painful chronic tendinopathy of the equine SDFT. Levels of tenascin-C mRNA from chronic tendinopathy were not significantly different from normal adult SDFT demonstrating that levels of tenascin-C may only increase transiently in acute injury then to return to normal levels. These findings are at variance with some of the human literature describing degenerate tendons \[[@B8],[@B18]\] where increased tenascin-C may be associated with round cells and a more fibrocartilaginous phenotype or a different disease state. Monolayer culture of tendon fibroblasts provides a simple method to study cell phenotype. Recent investigations have highlighted significant differences in cellular gene expression in monolayer compared with three dimensional cultures \[[@B44]\]. The gene expression profile of tendon fibroblasts in monolayer culture has recently been shown to alter with progressive passaging \[[@B45]\]. The current experiments have highlighted further differences between monolayer tendon fibroblasts and those found in adult tendon. Scleraxis expression is decreased at both passage one and five in monolayer cultures of tendon fibroblasts hence, neither have a gene expression profile that recapitulates what is found in normal tendon. Three dimensional cell culture methods are thought to more closely mimic the in vivocellular environment \[[@B30],[@B44]\]. Unfortunately whilst increasing the expression of COL1A2 in agreement with the work of others \[[@B46]\] expression of scleraxis was not retained when these cells were cultured in three dimensional collagen gels. The in vitromodels used in the current study did not fully recreate the adult tendon phenotype. Further work is warranted to identify a culture system that more closely resembles adult tendon. Recently embryonic tendon fibroblasts cultured in a three dimensional fibrin gel \[[@B47]\] were able to maintain a tendon developmental phenotype in the fibrin gel as defined by electron microscopy. Other three dimensional culturing systems have been extensively investigated biomechanically \[[@B9]\] however, the gene expression of these alternative systems has yet to be reported. Clearly, the importance of recreating the transcriptomic profile of normal tendon relative to the functional properties of the engineered tissue requires further evaluation. The two methods of reference gene stability assessment were in agreement for all experiments carried out in this study, this is concordant with previous findings \[[@B48]\]. Interestingly GapDH was identified as one of the most stable reference genes in vitrowhile Normfinder found ACTB to be the most stable reference gene when comparing normal and diseased tendons. Conclusion ========== As no single molecular marker was capable of discriminating tendon from both bone and cartilage in the current study it is recommended that multiple genes are used to identify tenogenic differentiation. The current experiments would suggest high expression of COL1A2 and scleraxis and low expression of tenascin-C are most representative of the normal adult tendon phenotype. Furthermore, this work refutes the use of tenomodulin as a good marker of equine tendon fibroblasts as similar levels were identified in both tendon and bone. A genome wide screen may in the future identify specific markers of tendon phenotype. Whether they are maintained in cultured fibroblasts will also need to be determined. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= SET carried out the experimental work, participated in the sequence alignment and statistical analyses and drafted the manuscript. LCM participated in the sequence alignment. DNC and AVT participated in reference gene normalisation. GP helped perform statistical analysis. PDC and RKWS conceived the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2474/10/27/prepub> Supplementary Material ====================== ###### Additional file 1 Raw Ct values of reference genes of all samples*. The data provided represent the cycle thresholds of the three reference genes used in the data analysis. ###### Click here for file Acknowledgements ================ S.E. Taylor\'s Research Training Scholarship is generously funded by the Horserace Betting Levy Board. Thank you to Elizabeth Barr for the design of the osteomodulin and Runx2 primers and to Dr Emma Humphries for the design of the SDHA primer.	{ "pile_set_name": "PubMed Central" }
The content published in Cureus is the result of clinical experience and/or research by independent individuals or organizations. Cureus is not responsible for the scientific accuracy or reliability of data or conclusions published herein. All content published within Cureus is intended only for educational, research and reference purposes. Additionally, articles published within Cureus should not be deemed a suitable substitute for the advice of a qualified health care professional. Do not disregard or avoid professional medical advice due to content published within Cureus. Introduction and background =========================== As of March 11, 2020, 118,598 cases of COVID-19 were reported worldwide by more than 100 countries. Since late February, the majority of cases reported are from outside China, with an increasing majority of these reported from European Union (EU)/European Economic Area (EEA) countries and the UK. The Director-General of the World Health Organization declared COVID-19 a global pandemic on March 11, 2020. All EU/EEA countries and the UK are affected, reporting a total of 17,413 cases as of March 11. Seven-hundred eleven cases reported by EU/EEA countries and the UK have died. Italy represents 58% of the cases and 88% of the fatalities. The current pace of the increase in cases in the EU/EEA and the UK mirrors trends seen in China in January to early February and trends seen in Italy in mid-February \[[@REF1]\]. The end of 2019 was marked by the emergence of a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), which caused an outbreak of viral pneumonia (COVID-19) in Wuhan, China \[[@REF2]\]. Coronaviruses are a family of single-stranded enveloped RNA viruses that are divided into four major genera. The genome sequence of SARS-CoV-2 is 82% similar to severe acute respiratory syndrome coronavirus (SARS-CoV) and both belong to the β-genus of the coronavirus family \[[@REF3]-[@REF4]\]. Human coronaviruses, such as SARS-CoV and Middle East respiratory syndrome coronavirus (MERS-CoV), are known to cause respiratory and enteric symptoms. In the SARS outbreak of 2002-03, 16%-73% of patients with SARS had diarrhea during the course of the disease, usually within the first week of illness \[[@REF5]\]. SARS-CoV ribonucleic acid (RNA) was only detected in stools from the fifth day of illness onwards, and the proportion of stool specimens positive for viral RNA progressively increased and peaked at day 11 of the illness, with viral RNA still present in the feces of a small proportion of patients even after 30 days of illness \[[@REF6]\]. Till now, infection control and surveillance focus on the respiratory system. The ignorance of SARS-CoV-2 in the digestive system may cause difficulty with disease control. Gastrointestinal symptoms seem to be uncommon in patients with COVID-19 when compared with SARS \[[@REF7]-[@REF8]\]. However, they should not be ignored as the increasing rate of diarrhea occurs in confirmed COVID-19 patients according to a recent report that 14 of 138 confirmed patients had diarrhea \[[@REF9]\]. Those early reports may not represent the actual rate of gastrointestinal symptoms caused by SARS-CoV-2 because in the early stages of the outbreak, the limited resources for detection were only provided to those patients with severe symptoms like respiratory distress syndrome. About 27% of SARS patients have diarrhea and since full-length genome sequences identified that SARS-CoV-2 is 79.5% identical to SARS-CoV and shares the same receptor angiotensin-converting enzyme 2 (ACE2), it is estimated that the rate of gastrointestinal symptoms would be higher in patients with COVID-19 \[[@REF10]\]. One possible route for the movement of SARS-CoV-2 into the digestive system may be the "trachea-esophagus-ileum-colon" as single-cell transcriptome analysis showed ACE2, the entry receptor for SARS-CoV-2, highly expressed in lung AT2 cells, esophagus upper and stratified epithelial cells, and enterocytes from the ileum and colon \[[@REF11]\]. Review ====== Data and methods Literature Search and Selection We conducted a comprehensive systematic literature search of online databases, including PubMed, Embase, Web of Science, WanFang Data, and CNKI, from December 2019 to February 2020, to identify all case studies. The search terms and relative variants were as follows: COVID-19; 2019-nCoV; SARS-CoV-2; clinical characteristics; discharge rate; gastrointestinal infection; review. The inclusion criteria for the articles are as follows: study population: patients diagnosed with COVID-19; study design: case studies; measurement of results: at least one result that reported gastrointestinal symptoms and the presence of viruses in the stool. Abstracts from conferences and commentary articles were excluded. Data extraction and quality assessment data extraction and the evaluation of literature quality were conducted independently by two investigators (R.E. and G.A.). Any disagreement was resolved by another investigator (C.M.) Results Natural History and Clinical Characteristics Huang et al. first reported the clinical features of 41 patients confirmed to be infected with COVID-19 on January 2, 2020, which include 13 intensive care unit (ICU) cases and 28 non-ICU cases \[[@REF7]\]. More than half of the cases (66%) had been exposed to the Huanan Seafood Wholesale Market. Almost all patients had bilateral lung ground-glass opacity on computed tomography imaging. The initial symptoms included fever (98%), cough (76%), dyspnea (55%), myalgia or fatigue (44%), sputum production (28%), headache (8%), hemoptysis (5%), and diarrhea (3%). Only one patient did not present fever in the early stage of the disease. Twelve (29%) cases progressed to acute respiratory distress syndrome (ARDS), five (12%) had acute cardiac injury, three (7%) had acute kidney injury (AKI), and three (7%) had shock. At the data cutoff date, 28 (68%) patients were discharged and six (15%) had died. On January 20, 2020, Chen et al. reported 99 cases with SARS-CoV-2-infected pneumonia \[[@REF8]\]. This case series revealed that older males with comorbidities as a result of weaker immune function were the most susceptible to COVID-19 incidence. The symptoms, complications, and treatments in this study were similar to the previously published study by Huang and colleagues. At the data cutoff date, 31 (31%) of the patients were discharged, 11 (11%) died, and 57 (58%) were still hospitalized. A study by Li et al. reported on 425 COVID-19 cases in Wuhan confirmed between January 1 and 22, 2020 \[[@REF12]\]. The mean incubation period was 5.2 days, with the 95th percentile of the distribution at 12.5 days, though uncertainty remains. Two subsequent studies confirmed the pattern of signs and symptoms \[[@REF13]-[@REF14]\]. At the time of this writing, the most recent published case series of 138 confirmed cases included 36 requiring intensive care by the data cutoff date of February 3, 2020 \[[@REF9]\]. It also found the common presenting symptoms to be fever (136), fatigue (96), and dry cough (82), though there were two patients who did not present any signs of fever at the onset of illness. A higher proportion of cases presented with gastrointestinal symptoms, including diarrhea and nausea (14) initially presented one to two days prior to the development of fever and dyspnea. Forty-seven patients (34%) were discharged while six (4%) died, while the remainder were still hospitalized. Organ failure complications were similar to the original studies. Taken together, these studies indicate the main clinical manifestations of COVID-19 are fever (90% or more), cough (around 75%), and dyspnea (up to 50%). A small but significant subset had gastrointestinal symptoms. Basic Virology Coronaviruses are widespread in humans and several other vertebrates and cause respiratory, enteric, hepatic, and neurologic diseases. Notably, the severe acute respiratory syndrome coronavirus (SARS-CoV) in 2003 and Middle East respiratory syndrome coronavirus (MERS-CoV) in 2012 have caused human epidemics. A comparison with the current virus shows several significant differences and similarities. Both MERS-CoV and SARS-CoV have much higher case fatality rates (40% and 10%, respectively). Though the current SARS-CoV-2 shares 79% of its genome with SARS-CoV, it appears to be much more transmissible \[[@REF15]\]. SARS-CoV-2 has been sequenced. A phylogenetic analysis found a bat origin for SARS-CoV-2. There is a diversity of possible intermediate hosts for SARS-CoV-2, including pangolins, but not mice and rats. There are many similarities of SARS-CoV-2 with the original SARS-CoV. Using computer modeling, researchers found that the spike proteins of SARS-CoV-2 and SARS-CoV have almost identical 3-D structures in the receptor-binding domain that maintains van der Waals forces. The SARS-CoV spike protein has a strong binding affinity to human ACE2 based on biochemical interaction studies and crystal structure analysis. SARS-CoV-2 and SARS-CoV spike proteins share 76.5% identity in amino acid sequences and, importantly, the SARS-CoV-2 and SARS-CoV spike proteins have a high degree of homology \[[@REF16]\]. Both SARS-CoVs enter the cell via the ACE2 receptor \[[@REF17]\]. The SARS-Cov-2 first predominantly infects lower airways and binds to ACE2 on alveolar epithelial cells. Both viruses are potent inducers of inflammatory cytokines. The "cytokine storm" or "cytokine cascade" is the postulated mechanism for organ damage. The virus activates immune cells and induces the secretion of inflammatory cytokines and chemokines into pulmonary vascular endothelial cells. Transmission Dynamics and Protective Measures The preliminary estimate of R0 (the expected number of cases directly produced by one person in a population susceptible to infection) for COVID-19 is 2.2 (95% CI, 1.4 to 3.9) \[[@REF18]\]. Fomites are suspected as the main source of infectious particles, though some uncertainty remains. Other coronaviruses have been shown to persist for days on uncleaned surfaces \[[@REF15]\]. Additionally, SARS-CoV-2 RNA was detected in the stool specimen in a person who had symptoms while the serum specimen tested negative \[[@REF19]\]. Recently, SARS-Cov-2 was isolated from a swab sample of a confirmed patient's feces by Chinese researchers, indicating the potential for fecal-oral transmission \[[@REF20]\]. Studies have shown effective person-to-person transmission of 2019-nCoV even in the presence of isolation efforts in medical facilities \[[@REF21]-[@REF22]\]. The recent case series reported 57 (41%) of 138 patients were infected in hospital settings, including 40 (29%) medical staff \[[@REF9]\]. Handwashing is the mainstay of viral control. Contact isolation gear, such as masks, gowns, and gloves, are also recommended. Transmission via the ocular surface is possible, so eye protection should also be used \[[@REF23]\]. Enteric Involvement of Coronavirus Novel coronavirus symptoms seem to be mostly focused on fever and cough, but gastrointestinal symptoms should be a new focus for clinicians, according to two new papers published online in Gastroenterology, Jinyang Gu et al. describe how investigators from Shanghai, China, sought to document the symptoms of the novel coronavirus. Although fever, dry cough, and dyspnea present in most cases, they wanted to understand what impact the virus had on symptoms such as diarrhea, nausea, vomiting, and abdominal discomfort. So far, those symptoms have varied among different study populations. Former studies on SARS, which is related to the coronavirus and can present with similar symptoms, showed that SARS was verified in patients after detection in biopsy specimens and stool. This was true even after the patients had been discharged from the hospital. The study authors noted that the first US patient admitted to a hospital with confirmed coronavirus had a loose bowel movement on hospital day two \[[@REF19]\]. Labs in China have been able to isolate the live coronavirus from the stool of patients, the authors said (unpublished). These factors brought the gastrointestinal tract to the forefront of investigators' minds and suggest that clinicians should identify patients with gastrointestinal symptoms and carefully monitor those patients \[[@REF24]\]. Another similarity that may be noted between SARS and COVID-19 is that mild to moderate liver injury has existed in patients \[[@REF25]\]. Xiao F et al. examined the viral RNA in feces from 71 patients with confirmed COVID-19 during their hospitalization from February 1-14, 2020 \[[@REF26]\]. They collected serum, nasopharyngeal and oropharyngeal swabs, urine, stool, and tissues (from endoscopy) from the patients. The age of the patients ranged from 10 months to 78 years. The duration of positive stool tests ranged from 1 to 12 days, they added, and patients remained positive via stool tests after showing negative in respiratory samples. The researchers found that 53.4% of patients had SARS-CoV-2 RNA in their stool and 23% of patients tested positive in their stool despite testing negative for the virus in respiratory samples. This finding indicates that viral gastrointestinal infection and the potential fecal-oral transmission can last even after viral clearance in the respiratory tract. The author strongly recommends that rRT-PCR testing for SARS-CoV-2 from feces should be performed routinely in SARS-CoV-2 patients, and transmission-based precautions for hospitalized SARS-CoV-2 patients should continue if the feces tests positive by rRT-PCR testing. Table [1](#TAB1){ref-type="table"} provides a brief review of the study by Yu He et al. of COVID-19 while Table [2](#TAB2){ref-type="table"} summarizes all the included clinical studies to date \[[@REF27]-[@REF28]\]. ###### COVID-19: Coronavirus Disease 2019 Yu He MD et al. \[[@REF27]\] SARS: Severe Acute Respiratory Syndrome ------------------------- ------------ ----------- ------------ Items COVID-19 COVID-19 COVID-19 Journal Lancet Lancet JAMA Cases 41 99 138 Published data 24-01-2020 2020/1/29 07-02-2020 Fever (%) 98 83 98.6 Cough (%) 76 82 82 Shortness of breath (%) 55 31 31.2 Sputum production (%) 28 NA 26.8 Diarrhea (%) 3 2 10.1 Death (%) 15 11 4.3 ------------------------- ------------ ----------- ------------ ###### Summary of included clinical studies to date CRRT: Continuous Renal Replacement Therapy; ARDS: Acute Respiratory Distress Syndrome; AKI: Acute Kidney Injury; NIV: Noninvasive Ventilation; IMV: Invasive Mechanical Ventilation; ECMO: Extracorporeal Membrane Oxygenation Jiang, F. et al. \[[@REF28]\] ------------------------------------- -------------------------------------------------------------- ------------------------------------------------------ --------------------------------------------- ---------------------------------------- ----------------------------------- Author Huang et al. \[[@REF7]\] Chen et al. \[[@REF8]\] Li et al. \[[@REF12]\] Song et al. \[[@REF13]\] Chen et al. \[[@REF14]\] Study setting Wuhan Jinyintan Wuhan Jinyintan Hospitals in Wuhan Shanghai Public Health Clinical Center Tongji Hospital Hospital from Dec 16, 2019, to Jan 2, 2020 Hospital from Jan 1 to Jan 20, 2020 on Jan 22, 2020 from Jan 20 to Jan 27, 2020 from Jan 14 to Jan 29, 2020 City Wuhan, China Wuhan, China Wuhan, China Shanghai, China Wuhan, China Total patients 41 99 425 51 29 Age, mean (IQR) or mean ± SD, years 49 (41--58) 55.5 ± 13.1 56 (26--82) 49 ± 16 56 (26--79) Gender, male 30 (73%) 67 (68%) 31 (66%) 25 (49%) 21 (72%) Exposure history, cases 27 (66%) exposed to 49 (49%) exposed to 26 (55%) exposed to 50 (98%) exposed to 2 (7%) exposed to Huanan Seafood Huanan Seafood Huanan Seafood Wuhan Huanan Seafood Wholesale Market Wholesale Market Wholesale Market Wholesale Market X-ray and CT findings, cases Bilateral ground-glass opacity, 40 (98%) Multiple mottling and ground-glass opacity, 14 (14%) Radiographic evidence of pneumonia Ground glass opacity, 39 (77%) NA Signs and symptoms Fever, 40 (98%) Fever, 82 (83%) Fever, with or without recorded temperature Fever, 49 (96%) Fever, 28 (97%) Cough, 31 (76%) Cough, 81 (82%) Cough, 24 (47%) Cough or expectoration, (21 -72%) Myalgia or fatigue 18 (44%) Shortness of breath, 31 (31%) Phlegm, 10 (20%) Dyspnea, 17 (59%) Sputum production, 11/39 (28%) Muscle ache, 11 (11%) Myalgia or fatigue, 16 (31%) Myalgia or fatigue, 12 (41%) Headache, 3/38 (8%) Confusion, 9 (9%) Headache and dizziness, 8 (16%) Headache, 2 (7%) Hemoptysis, 2/39 (5%) Headache, 8 (8%) Dyspnea or chest pain, 7 (14%) Diarrhea, 4 (14%) Diarrhea, 1/38 (3%) Sore throat, 5 (5%) Loss of appetite, 9 (18%) Dyspnea, 22/40 (55%) Rhinorrhea, 4 (4%) Diarrhea, 5 (10%) Chest pain, 2 (2%) Stuffy and runny nose, 2 (4%) Diarrhea, 2 (2%) Sore throat, 3 (6%) Nausea and vomiting, 1 (1%) Nausea and vomiting, 3 (6%) Complications ARDS, 12 (29%) ARDS, 17 (17%) NA NA NA RNAemia, 6 (15%) Acute renal injury, 3 (3%) Acute cardiac injury, 5 Acute respiratory injury, 8 (8%) -12% Septic shock, 4 (4%) Acute kidney injury, 3 (7%) Ventilator-associated pneumonia, 1 (1%) Treatments Antiviral, 38 (93%) Oxygen therapy, 75 (76%) NA NA NA Antibiotic, 41 (100%) Noninvasive ventilation, 13 (13%) Corticosteroid, 9 (22%) Invasive ventilation, 4 (4%) CRRT, 3 (7%) CRRT, 9 (9%) Nasal cannula, 27 (66%) ECMO, 3 (3%) Noninvasive ventilation or high-flow nasal cannula, 10 (24%) Antibiotic, 70 (71%) Invasive mechanical ventilation, 2 (5%) Antifungal, 15 (15%) Antiviral, 75 (76%) Glucocorticoids, 19 (19%) ------------------------------------- -------------------------------------------------------------- ------------------------------------------------------ --------------------------------------------- ---------------------------------------- ----------------------------------- Discussion This review included the latest studies from December 2019 to March 2019 to analyze the clinical characteristics of the novel coronavirus. New evidence supports the possibility of gastrointestinal infection with the SARS-CoV-2, as well as a possible fecal-oral route of transmission, according to the results of studies published in Gastroenterology. Currently, a SARS-CoV-2 infection causes primarily a respiratory-based constellation of symptoms. Due to the respiratory nature of symptoms, it has been hypothesized that the virus infects respiratory epithelial cells and is transmitted via respiratory droplets from human to human. However, viral target cells and organs have not been explicitly identified, creating a significant gap in the current understanding of the pathogenesis of SARS-CoV-2 and its transmission routes \[[@REF24]-[@REF26]\]. Moreover, a recent case report detected SARS-CoV-2 viral RNA in stool samples, thus challenging these hypotheses. Therefore, researchers examined stool samples obtained between February 1 and February 14, 2020, from 73 patients hospitalized as a result of SARS-CoV-2 infection for viral RNA. Results demonstrated that 53.42% of these patients had positive samples for SARS-CoV-2 RNA. The samples remained positive for a range of one to 12 days; of note, 23.29% of patients had positive stool samples after respiratory samples were negative for viral RNA. A study described one patient, a man, aged 78 years, who was admitted to a hospital in Guangdong Province, China, on January 17, 2020. rRT-PCR was positive for SARS-CoV-2, and the patient experienced 10 days of respiratory illness with the hallmarks of COVID-19. On day 10, however, he presented with coffee-ground gastric contents in his gastric drainage tube. Fecal occult blood testing indicated an upper gastrointestinal bleed; subsequent endoscopy found mucosal damage to the esophagus, and esophageal, gastric, duodenal, and colonic samples were taken for testing. Hematoxylin and eosin staining showed no significant damage, but numerous infiltrating plasma cells and lymphocytes, as well as interstitial edema, were seen in the lamina propria of the stomach, duodenum, and rectum. Most significantly, results demonstrated positive ACE2 protein staining, mainly in the cytoplasm of gastrointestinal epithelial cells. It has been previously shown that SARS-CoV-2 uses this protein as a viral receptor for its entry process. Immunofluorescence testing showed that ACE2 was abundantly expressed in the glandular cells of the gastric, duodenal and rectal epithelia of the above-mentioned patient, as well as other patients. These data demonstrated that infectious virions are being secreted from the gastrointestinal cells of people with SARS-CoV-2 infection, and thus support the potential for fecal-oral transmission of the virus. The discharge guideline depending on the respiratory tract test also meets the challenge. Interestingly, while the virus test of the nasopharyngeal swab switched from positive to negative after the treatment, the rectal swab specimens still tested positive \[[@REF29]\]. These cases remind the clinicians that the rectal swab may be equally important to the pharyngeal swab even if the patient is asymptomatic, which challenges the latest published guideline provided by National Health Commission of China that two successive, negative respiratory tract tests are regarded as the standard for discharge and termination of compulsory isolation for COVID-19 patients \[[@REF30]\]. Famous well-described clusters of infection of SARS in Amoy Gardens, Hong Kong, drew the attention of health officials on fomite transmission because two-thirds of the confirmed SARS patients in Amoy Gardens had diarrhea \[[@REF31]\]. As findings showed that patients with SARS could discharge SARS-CoV in their stool up to 73 days after symptom onset, the stools with the virus became the resource of contamination of airdrops and a variety of environmental surfaces, which may contribute to the clusters of infection \[[@REF32]\]. Similarly, evidence showed that SARS-CoV-2 were identified in four stool specimens (four out of 62), so fomite transmission should not be ignored in the transmission of SARS-CoV-2 since the virus may move from the respiratory tract into the gastrointestinal tract the recovered patients may discharge the stool with the virus for a long time \[[@REF33]\]. According to a recently published report by the Centers for Disease Control and Prevention (CDC) of China, community-acquired infections are becoming the predominant route in transmission \[[@REF34]\]. Based on these cases and the lessons from SARS, rRT-PCR testing for SARS-CoV-2 from feces should be performed in SARS-CoV-2 patients before discharge. Conclusions =========== Clinicians should recognize that digestive symptoms, such as diarrhea, may be a presenting feature of COVID-19, and that the index of suspicion may need to be raised earlier in at-risk patients presenting with digestive symptoms rather than waiting for respiratory symptoms to emerge. It should also be considered before discharge that viral gastrointestinal infection and potential fecal-oral transmission can last even after viral clearance from the respiratory tract. The authors have declared that no competing interests exist. What this study adds \- A significant portion of coronavirus patients experiences diarrhea, nausea, vomiting, and/or abdominal discomfort before the onset of respiratory symptoms. \- Viral RNA is detectable in fecal samples from suspected cases, indicating that the virus sheds into the stool. \- Viral gastrointestinal infection and potential fecal-oral transmission can last even after viral clearance from the respiratory tract.	{ "pile_set_name": "PubMed Central" }
All data are within the paper and its Supporting Information files. Introduction {#sec001} ============ Lettuce (Lactuca sativa L.) downy mildew, caused by the oomycete Bremia lactucae Regel, is a major threat to lettuce production around the world \[[@pone.0144573.ref001]--[@pone.0144573.ref002]\]. The life cycle of B. lactucae, an obligate biotrophic parasite, involves primary (sexual) and secondary (asexual) infection cycles. The sexual cycle results in soilborne oospores that are the likely source of primary inoculum \[[@pone.0144573.ref003]\]. Oospores can potentially overwinter and become a source of primary inoculum during the spring. However, there are no published data showing that B. lactucae oospores survive the harsh Canadian winter. The asexual cycle generates conidia that are adapted to aerial dispersal \[[@pone.0144573.ref004]--[@pone.0144573.ref005]\]. The conidia are produced when humidity is high and wind speed is low, and they are released in response to concomitant decreasing humidity and increasing temperature \[[@pone.0144573.ref006]\]. Viable conidia that land on leaves of susceptible lettuce plants germinate and colonize the leaves, resulting in symptoms that are visible 7 to 14 d after the initiation of infection \[[@pone.0144573.ref007]\]. First infection is then followed by successive asexual cycles occurring throughout the lettuce production season. Environmental factors such as temperature, relative humidity (RH), wind speed, solar radiation, and leaf wetness duration have been identified as factors that determine the extent of conidia production, dispersal, and survival as well as the infection processes \[[@pone.0144573.ref004], [@pone.0144573.ref006], [@pone.0144573.ref008], [@pone.0144573.ref009], [@pone.0144573.ref010]\]. Hence, conidia survival is greater at 23°C (12h) than at 31°C (2 to 5 h), regardless of RH (33% and 76%), and conidia survival increases substantially at an RH of 90% or greater \[[@pone.0144573.ref005], [@pone.0144573.ref009]\]. Wind speed plays a major role in the conidia dispersal process, and solar radiation determines the survival of the airborne conidia \[[@pone.0144573.ref005], [@pone.0144573.ref007]\]. However, the most important factor for successful infection is the duration of morning and evening leaf wetness \[[@pone.0144573.ref011]--[@pone.0144573.ref015]\]. The control strategy for lettuce downy mildew is based mostly on chemical protection with fungicide applied at fixed or weather-based intervals. A difficulty that most growers face is identifying the best timing for fungicide applications to get the optimum level of control against downy mildew. Therefore, in the province of Quebec, Canada, fungicides are applied routinely to control downy mildew even though some of these applications may be unnecessary. In the pathosystem formed by lettuce and B. lactucae, two decision-support systems (DSSs) were developed to guide decision-making concerning the timing of fungicide applications. The first DSS, which was developed in California, USA, and was later modified, is based on leaf wetness ending late in the morning (10:00 hours) to predict when infection by B. lactucae occurs \[[@pone.0144573.ref013]\]. The second DSS, BREMCAST, was developed in Quebec, Canada \[[@pone.0144573.ref012]\], use a leaf wetness duration of 3 to 5 h after dawn (continuing until 10:00 hours) as an action threshold for fungicide application. Therefore, in both of these DSSs, leaf wetness duration is used as an indicator for occurrence of an infection event \[[@pone.0144573.ref012]--[@pone.0144573.ref013], [@pone.0144573.ref016]\]. These systems assume that sporulation is nocturnal, that conidia are released at dawn, and that infections occur in the morning \[[@pone.0144573.ref012]--[@pone.0144573.ref013]\]. However, the major limitation of these DSSs is their inability to assess the presence and amount of conidia above lettuce fields, and consequently these systems can overestimate or underestimated the risk of downy mildew infection \[[@pone.0144573.ref017]\]. In fact, to evaluate the availability of B. lactucae airborne inoculum, these DSSs rely on signs of downy mildew in the field during routine scouting \[[@pone.0144573.ref012]\]. The potential risk of lettuce downy mildew development and consequent yield losses are related to the quantity of B. lactucae airborne conidia \[[@pone.0144573.ref015]\]. Fall et al. \[[@pone.0144573.ref015]\] observed a quantitative relationship between the airborne conidia concentration (ACC) of B. lactucae and the number of lesions per leaf. Hence, an ACC of 14 conidia/m^3^ can cause one lesion per leaf \[[@pone.0144573.ref015]\]. Therefore, incorporating B. lactucae airborne conidia concentration into DSSs may help to develop more effective strategies for controlling lettuce downy mildew with better timing of fungicide applications \[[@pone.0144573.ref015]\]. The amount of airborne conidia above lettuce fields is dictated by several factors, including the sporulation intensity of sporulating lesions (source), the proportion of conidia that is released, the proportion of conidia that escapes the canopy layer and becomes airborne, and the proportion of surviving conidia that are deposited on susceptible lettuce leaves. Because each of these factors has been mathematically described for B. lactucae or closely related species \[[@pone.0144573.ref012]--[@pone.0144573.ref013], [@pone.0144573.ref018]--[@pone.0144573.ref019]\], these mathematical relationships could be used to simulate the quantity of airborne conidia. Despite the amount of scientific information available on the aerobiology of plant pathogens, few published studies have actually focused on the simulation of airborne inoculum, even though it has been documented that inoculum is a key factor for the development of epidemics of several plant diseases \[[@pone.0144573.ref020]--[@pone.0144573.ref022]\]. In this study, we developed a dynamic simulation model of B. lactucae airborne conidia in an effort to improve the decision-making process for lettuce downy mildew management. The specific objectives of this study were (1) to develop a structurally dynamic model to simulate B. lactucae airborne conidia and (2) to assess the accuracy and sensitivity of the model against measured airborne conidia data. Materials and methods {#sec002} ===================== Ethics statement {#sec003} ---------------- The study did not involve endangered or protected species. All data are available in Tables, Figures and Supporting Information files. Asexual disease cycle processes {#sec004} ------------------------------- Because of the absence of information and evidence about the overwintering of oospores under Canadian weather conditions, the asexual life cycle of B. lactucae was used to build the model framework. This cycle can be divided in four major stages: the sporulation and release stage, the escape and dispersal stage, the survival and deposition stage, and the infection and germination stage. First, conidia that land on the leaves of susceptible lettuce plants germinate and colonize the leaf cells. The resulting infected leaves produce conidia under conditions of high humidity and low wind speed. These conidia are released in response to decreasing humidity and increasing temperature (sporulation and release stage). Second, increases in wind speed promote the escape of conidia from the canopy layer and the dissemination of these conidia (escape and dispersal stage). Third, weather variables, including solar radiation, temperature, and air RH, determine conidia survival, and as the wind speed decreases, the conidia are deposited gradually on lettuce leaves (survival and deposition stage). Finally, new infections by conidia require free water on leaf surfaces for germination (infection and germination stage) \[[@pone.0144573.ref008], [@pone.0144573.ref015]\]. [Fig 1](#pone.0144573.g001){ref-type="fig"} shows the different stages of the asexual life cycle of B. lactucae. ![Asexual stages of the life cycle of Bremia lactucae.](pone.0144573.g001){#pone.0144573.g001} Modeling approach {#sec005} ----------------- System dynamics is a methodology for studying and managing complex systems that change over time \[[@pone.0144573.ref023]\]. A model was built using the system dynamics methodology to simulate the asexual life cycle of B. lactucae. The model was developed according to the principles of "flow charts" in a dynamic simulation system (STELLA, v. 10.6.0). STELLA is a flexible computer modeling package with an easy, intuitive interface that allows users to construct dynamic models that can simulate biological systems. The main components of STELLA are stocks, flows, connectors, and converters. Stocks are accumulations within the system and can be different types, including reservoirs and conveyors. Flows are the movement of the stocks throughout the system and allow resources to be transported around the model. Connectors provide information links within the system, and converters contain the algebraic relationships within a model \[[@pone.0144573.ref023]\]. The change in any stock at a given time is expressed as follows: $$\text{Stocks}\left( \text{t} \right)\ = \ \text{Stocks}\left( {t\ - \ dt} \right)\ + \ \left( {n\text{Inflow}\ - \ n\text{outflow}} \right)dt$$ where t is time, dt is differential of t, and n is number of inflows or outflows The simulation time unit is hour, and the variable inputs are estimated on an hourly basis. To reduce potential integration error, a built-in simulation algorithm of the Euler integration method and a simulation time step of 0.25 are used \[[@pone.0144573.ref023]\]. As a result, for each hour of simulation, the STELLA program runs four times for the integration process. The time horizon of the model is 24 h (1 d), and the model simulates the airborne conidia of B. lactucae from July to September of each year, which is the critical period for lettuce downy mildew in Quebec, Canada. Model description {#sec006} ----------------- The model can be divided into six state variables (stocks) that are linked by flow charts and connectors ([Fig 2](#pone.0144573.g002){ref-type="fig"}, [Table 1](#pone.0144573.t001){ref-type="table"}). The first state variable is the potential for the sporulation of one lesion (PSL). It is assumed that at least one lesion per m^2^ of crop is sporulating when the temperature is between 5 and 25°C and the leaf wetness duration is greater than or equal to 2 h \[[@pone.0144573.ref015], [@pone.0144573.ref024]\]. It is also assumed that, for the lesion lifetime and for the time frame of July to September, a sporulating lesion can produce a maximum of 20,000 conidia (unpublished data) and that the mean lesion size is 0.004 m^2^ \[[@pone.0144573.ref025]\]. ![Diagram of the model simulating Bremia lactucae airborne conidia.](pone.0144573.g002){#pone.0144573.g002} 10.1371/journal.pone.0144573.t001 ###### Description of variables and parameters used in the model. ![](pone.0144573.t001){#pone.0144573.t001g} Abbreviation Description Unit --------------- ------------------------------------------------------ ----------------- PSL Potential for the sporulation of one lesion per m^2^ Number (= 1) NRC Number of released conidia per m^2^ Number (0 to ∞) NEC Number of escaped conidia per m^2^ Number (0 to ∞) NSC Number of survived conidia per m^3^ Number (0 to ∞) NDC Number of deposited conidia per m^2^ Number (0 to ∞) NGC Number of germinated conidia per m^2^ Number (0 to ∞) Sp Sporulation Number (0 to ∞) Escp rate Escape rate Number (0 to ∞) Surv rate Survival rate Number (0 to ∞) Germ Germination Number (0 to ∞) UEC Unescaped conidia Number (0 to ∞) UDC Undeposited conidia Number (0 to ∞) Asymp1 Asymptote for sporulation equation Number (0 to ∞) Asymp2 Asymptote for germination equation Number (0 to ∞) r1 Rate for sporulation equation 0.75 r2 Rate for germination equation Number (0 to ∞) Ws Wind speed m/s SR Solar radiation MJ/m^2^ RH Relative humidity \% LWD Leaf wetness duration H Tp Average temperature over leaf wetness duration °C The second state variable is the number of released conidia (NRC). The change in NRC is dictated by the conidia inflow (released conidia), the conidia outflow (escaped conidia), and the conidia death rate, as follows: $$\text{NRC}\left( \text{t} \right)\ = \text{NRC}\left( {t\ - \ dt} \right)\ + \ \left( {\text{Released} - \ \text{Escaped} - \ \text{Unescaped\ conidia}} \right)dt$$ The algebraic equations describing conidia inflow and conidia outflow for [Eq 2](#pone.0144573.e002){ref-type="disp-formula"} are defined in [Table 2](#pone.0144573.t002){ref-type="table"}. 10.1371/journal.pone.0144573.t002 ###### Algebraic equations describing the movement of the stocks throughout the model. ![](pone.0144573.t002){#pone.0144573.t002g} Variable Equation Data source ---------------- ----------------------------------------------------- -------------------------------------------------------------------------------------- Sp Asymp1(1 + 25118.86 × exp(−r1 × LWD))^−0.909^ Tchervenivanova, 1995 \[[@pone.0144573.ref018]\] Asymp1 0.996 -- 0.000051 × Tp^3^ -- 15.575/(Tp)^2^ Tchervenivanova, 1995 \[[@pone.0144573.ref018]\] Germ Asymp2 × exp(−exp(r2 × (LWD − 2))) Scherm and van Bruggen, 1993 \[[@pone.0144573.ref016]\] Asymp2 0.385 + 0.054 × Tp − 0.0024 × Tp^2^ Scherm and van Bruggen, 1993\[[@pone.0144573.ref016]\] r2 −1.154 + 0.327 × Tp − 0.011 × Tp^2^ Scherm and van Bruggen, 1993 \[[@pone.0144573.ref016]\] Escp rate 0.073 × Ws − 0.0087 Fall, unpublished data, 2013 Surv rate IF SR\<1 THEN 0.8 ELSE 0.6 Bhaskara Reddy et al., 1996 \[[@pone.0144573.ref026]\] Released PSL × Sp Escaped NRC × Esca. rate Surviving NEC × Surv. rate Deposited NSC × 1/3 Kranz, 1974 \[[@pone.0144573.ref027]\] Germinated NDC × Germ Incubation 240 h Scherm and van Bruggen,1993, 1994 \[[@pone.0144573.ref008], [@pone.0144573.ref016]\] The third state variable is the number of escaped conidia (NEC). The change in NEC is dictated by the conidia inflow (escaped conidia), the conidia outflow (surviving conidia), and the conidia death rate, as follows: $$\text{NEC}\left( \text{t} \right)\ = \text{NEC}\left( {t\ - \ dt} \right)\ + \ \left( {\text{Escaped}\ - \ \text{Surviving}\ - \ \text{Conidia\ deaths}} \right)dt$$ The algebraic equations describing conidia inflow and outflow for [Eq 3](#pone.0144573.e003){ref-type="disp-formula"} are defined in [Table 2](#pone.0144573.t002){ref-type="table"}. The fourth state variable is the number of surviving conidia (NSC). The change in NSC is a function of conidia survival and conidia deposition, as follows: $$\text{NSC}\left( \text{t} \right)\ = \ \text{NSC}\left( {t\ - \ dt} \right)\ + \ \left( {\text{Surviving}\ - \ \text{Deposited} - \ \text{Undeposited\ conidia}} \right)dt$$ The fifth state variable is the number of deposited conidia (NDC), described as follows: $$\text{NDC}\left( \text{t} \right)\ = \ \text{NDC}\left( {t\ - \ dt} \right)\ + \ \left( \text{Deposited} \right)dt$$ where the variables are defined in [Table 2](#pone.0144573.t002){ref-type="table"}. The sixth state variable is the number of germinated conidia (NGC), described as follows: $$\text{NGC}\left( \text{t} \right)\ = \ \text{NGC}\left( {t\ - \ dt} \right)\ + \ \left( {\text{Germinated}\ - \ \text{Incubation}} \right)dt$$ The germination process and the duration of incubation dictate the changes in the NGC. The duration of incubation is defined as the time from spore deposition to lesion production and is described in [Table 2](#pone.0144573.t002){ref-type="table"}. Running the model requires weather variables, including leaf wetness duration, RH, temperature, wind speed, and solar radiation ([Table 1](#pone.0144573.t001){ref-type="table"}). Each of the six state variables can be chosen as model outputs. Measurement of environmental variables {#sec007} -------------------------------------- Leaf wetness duration was assessed every 15 minutes with electrical-impedance leaf-wetness sensors (Model 237; Campbell Scientific, Edmonton, AB, Canada) placed at the height of the lettuce leaves. In 1997 and 1998, air temperature (°C) was recorded with a data logger (Model 21X; Campbell Scientific) placed near the spore sampler, and RH (%) was monitored with a probe (Model HMP35C; Campbell Scientific). In 2003 and 2004, air temperature and RH were monitored using WatchDog data loggers (Spectrum Technologies, Aurora, IL, USA) placed near the spore sampler. Weather variables were monitored every 30 min, and hourly averages were used in the analyses. The temperature and RH probes were placed in a white shelter 1.5 m above the ground. For all years, wind speed (10 m above the ground) and solar radiation data were obtained from an Environment Canada weather station located approximately 200 m from the plots \[[@pone.0144573.ref007], [@pone.0144573.ref015]\]. Model evaluation {#sec008} ---------------- The model was evaluated using four years (1997, 1998, 2003, and 2004) of hourly data on B. lactucae ACC \[7, 15\]. The data were collected between 1 July and 20 September in each of 1998, 1997, 2003, and 2004. Each year, a plot of the lettuce cultivar Ithaca was established in an organic soil at the Agriculture and Agri-Food Canada experimental farm in Ste-Clotilde, QC, Canada (latitude 45°10′ N, longitude 73°40′ W). Lettuce plants produced in a greenhouse by Les Serres Lefort (Ste-Clotilde, QC, Canada) were transplanted 0.3 m apart in the rows, with 0.35 m between rows. A 7-d volumetric spore sampler (0.94 height, Standard orifice 0.002 m x 0.014 m, Burkard Manufacturing Co., Rickmansworth, Hertfordshire, UK) placed in the center of the plot was used to monitor ACC two weeks per month (alternating sampling weeks, one on, one off) between July and September. The sampler was adjusted to sample air at 0.01 m^3^/min. Impaction tapes were coated with a thin layer of silicone grease before they were placed in the sampler. Tapes were removed at 7-days intervals, cut into 0.048 m long segments corresponding to 24 h periods. Conidia were counted on whole transects perpendicular to the tape length. These transects were fixed at 0.002 m intervals to obtain hourly counts. Conidia counts were performed with a microscope at 250× magnification (0.00075 m wide transect) and converted to conidia per cubic meter of air for each hour of the day \[[@pone.0144573.ref007]\]. For each year, the hourly-simulated conidia concentrations were compared with the hourly-observed conidia concentrations for a period of 24 h. The cumulative conidia concentration within a 24-h period was considered the daily conidia concentration, and hence the daily-simulated conidia concentrations were compared with the daily-observed conidia concentrations. In order for accuracy to be compared, regression of observed data (on the y-axis) versus simulated data (on the x-axis) was used instead of regression of simulated data (on the y-axis) versus observed data (on the x-axis) \[[@pone.0144573.ref028]--[@pone.0144573.ref029]\]. Also, when the coefficient of determination was greater than 0.7, Theil's decomposition of error was performed \[[@pone.0144573.ref030]\]. Theil's U statistic is decomposed into three coefficients of inequality ([Eq 7](#pone.0144573.e007){ref-type="disp-formula"}): mean differences between the observed and simulated airborne conidia, U~bias~; deviations from the 1:1 line, U~slope~; and the unexplained variance, U~error~. The coefficients are calculated as follows: $$U_{bias} = \ \frac{n\left( {\overline{O} - \overline{S}} \right)^{2}}{\sum{(O - S)}^{2}};U_{slope} = \ \frac{\left( {b - 1} \right)^{2}\sum\left( {S - \overline{S}} \right)^{2}}{\sum\left( {O - S} \right)^{2}};U_{error} = \ \frac{\sum{(\widetilde{O} - O)}^{2}}{\sum{(O - S)}^{2}}$$ where n is the number of simulations, O and S are the observed and simulated values of airborne conidia, respectively, $\overline{O}$ and $\overline{S}$ are the observed and simulated means, respectively, and b is the slope in the equation $\widetilde{O} = a + bS$, where $\widetilde{O}$ is calculated observed values based on simulated values. The sum of the three coefficients is 1. A numerical sensitivity analysis was carried to describe how much the model output values (simulated number of airborne conidia concentrations) are affected by the changes in the model input values (leaf wetness duration, sporulation rate \[which is defined as a function of temperature; see [Table 2](#pone.0144573.t002){ref-type="table"}\], wind speed, solar radiation, and RH). To evaluate how increasing one unit at time, of each input, will affect the increase in the simulated conidia number, the inputs variables were gradually increased and the corresponding airborne conidia concentrations were measured in each simulation run. Results {#sec009} ======= In the time window from July to September in 1997, 1998, 2003, and 2004, a total of 132 simulation runs were performed. Simulation runs (one run correspond to one day) were done 32, 36, 33, and 31 times in 1997, 1998, 2003, and 2004 respectively. There was a linear relationship between the hourly-observed number of conidia and the hourly-simulated number of conidia. In 1997, 1998, 2003, and 2004, the proportion of the linear variation in the hourly-observed values explained by the variation in the hourly-simulated values was greater than 0.7 (R^2^) in 32.5%, 94.7%, 96.8%, and 100% of the total number of simulated days, respectively ([Table 3](#pone.0144573.t003){ref-type="table"}, Figs [3](#pone.0144573.g003){ref-type="fig"} and [4](#pone.0144573.g004){ref-type="fig"}). The hourly-simulated and observed conidia concentrations had a similar trend and shape (Figs [3](#pone.0144573.g003){ref-type="fig"} and [4](#pone.0144573.g004){ref-type="fig"}). ![Examples of scatter plots of hourly-simulated versus hourly-observed airborne conidia concentrations of Bremia lactucae for the dynamic model developed to monitor airborne inoculum of the lettuce downy mildew pathogen.\ (A) Example of a case when 0.90 \< R^2^ \< 0.96, and (B) example of case when 0.60 \< R^2^ \< 0.70. The dashed line indicates 1:1 agreement between simulated and observed airborne conidia. The solid line indicates the fitted values from the regression of simulated versus observed airborne conidia. R^2^, coefficient of determination. The inset graph in each panel represents the hourly-simulated and hourly-observed airborne conidia as a function of hour of day.](pone.0144573.g003){#pone.0144573.g003} ![Examples of scatter plots of hourly-simulated versus hourly-observed airborne conidia concentrations of Bremia lactucae for the dynamic model developed to monitor airborne inoculum of the lettuce downy mildew pathogen.\ (A) Example of a case when 0.80 \< R^2^ \< 0.90, and (B) example of a case when 0.70 \< R^2^ \< 0.80. The dashed line indicates 1:1 agreement between simulated and observed airborne conidia. The solid line indicates the fitted values from the regression of simulated versus observed airborne conidia. R^2^, coefficient of determination. The inset graph in each panel represents the hourly-simulated and hourly-observed airborne conidia as a function of hour of day.](pone.0144573.g004){#pone.0144573.g004} 10.1371/journal.pone.0144573.t003 ###### Coefficients of determination of the regression model of hourly-observed versus hourly-simulated airborne conidia concentrations of Bremia lactucae in a lettuce field. n, number of simulations; R2, coefficient of determination. ![](pone.0144573.t003){#pone.0144573.t003g} Year n Coefficients of determination ---------- ----- ------------------------------- ------- ------- ------- ------- 1997 32 10.4% 15.3% 6.8% 36.8% 30.6% 1998 36 7.5% 10.6% 76.6% 5.3% 0.0% 2003 33 15.2% 16.3% 65.3% 3.0% 0.0% 2004 31 5.0% 30.9% 64.1% 0.0% 0.0% There was a linear relationship between the daily-observed number of conidia and the daily-simulated number of conidia. In 1997, 1998, 2003, and 2004, the proportions of the linear variation in the daily-observed values explained by the variation in the daily-simulated values were 0.12, 0.88, 0.74, and 0.75, respectively ([Table 4](#pone.0144573.t004){ref-type="table"}, [Fig 5](#pone.0144573.g005){ref-type="fig"}). The trends and the conidia peaks in the daily-simulated and daily-observed conidia concentrations were similar in all years except 1997, when some simulated peaks in conidia were not observed ([Fig 6](#pone.0144573.g006){ref-type="fig"}). ![Regression analysis of daily-simulated versus daily-observed airborne conidia concentrations of Bremia lactucae in 1998, 2003, and 2004 for the dynamic model developed to monitor airborne inoculum of the lettuce downy mildew pathogen.\ The solid lines indicate the fitted values from the regressions of simulated versus observed airborne conidia. R^2^, coefficient of determination.](pone.0144573.g005){#pone.0144573.g005} ![Daily-observed and daily-simulated airborne conidia of Bremia lactucae in 1997, 1998, 2003, and 2004 as function of day of the year (where day 1 is 1 January) for the dynamic model developed to monitor airborne inoculum of the lettuce downy mildew pathogen.](pone.0144573.g006){#pone.0144573.g006} 10.1371/journal.pone.0144573.t004 ###### Evaluation of the regression model of daily-observed versus daily-simulated airborne conidia concentrations of Bremia lactucae in a lettuce field. n, number of simulations; R2, coefficient of determination; Ubias, differences between the observed and simulated airborne conidia; Uslope, deviations from the 1:1 line; Uerror, the unexplained variance. ![](pone.0144573.t004){#pone.0144573.t004g} Year n R^2^ U~bias~ U~slope~ U~error~ ---------- ----- -------- --------- ---------- ---------- 1997 32 0.12 \- \- \- 1998 36 0.88 0.28 0.71 0.00 2003 33 0.74 0.14 0.85 0.00 2004 31 0.75 0.22 0.78 0.00 In 1997, as the coefficient of determination was not greater than 0.7, Theil's decomposition of error was not performed. Most of the errors in the daily-simulated conidia came from the deviations from the 1:1 line. In 1998, 71% of the error was due to the deviation from the 1:1 line, whereas the proportion of error associated with the bias was 3%. In 2003, 85% of the error was due to the deviation from the 1:1 line, whereas the proportion of error associated with the bias was 14%. In 2004, 78% of the error was due to the deviation from the 1:1 line, whereas the proportion of error associated with the bias was 22% ([Table 4](#pone.0144573.t004){ref-type="table"}). In 96%, 98%, 99%, and 99% of cases when no airborne conidia were measured, the daily-simulated airborne conidia concentrations were 0 in 1997, 1998, 2003, and 2004, respectively. Sensitivity analysis of the model showed that the model was highly sensitive to the rate of sporulation (r1 in the Sp equation in [Table 2](#pone.0144573.t002){ref-type="table"}), which is the multiplication factor of spore production. Increasing the sporulation rate to one unit increased the simulated conidia number to 1.6 units ([Fig 7](#pone.0144573.g007){ref-type="fig"}). Also, the model was not significantly sensitive to other inputs variables that were tested (leaf wetness duration, wind speed, solar radiation, and RH) in this study. ![Sensitivity analysis of the model to the sporulation rate.\ The blue (1), red (2), and purple (3) lines represent the simulated airborne conidia for sporulation rates of 0.5, 0.6, and 0.7, respectively.](pone.0144573.g007){#pone.0144573.g007} Discussion {#sec010} ========== Commercially acceptable control of lettuce downy mildew is achieved when only a few external leaves are infected. Regardless of the type of lettuce (head, leaf, or romaine), the commercialized parts must be free of disease. Because the development of lettuce downy mildew is strongly related to the environmental conditions \[[@pone.0144573.ref015]\], the decision support systems (DSS) that have been developed in the last 20 years rely on weather conditions to predict the best time for fungicide applications. However, epidemics are also driven by the presence and quantity of inoculum in the lettuce field \[[@pone.0144573.ref026]\]. Consequently, to improve the effectiveness of DSSs, it is crucial to find a way to simulate or measure airborne inoculum above the lettuce canopy \[[@pone.0144573.ref012], [@pone.0144573.ref031]\]. The results of recent studies \[[@pone.0144573.ref017], [@pone.0144573.ref031]\] suggested that inoculum could be monitored with a spore-sampling network. A high disease risk as estimated by a DSS in combination with a significant airborne spore concentration would trigger fungicide applications. However, there are advantages and limitations associated with both monitoring and simulating airborne inoculum. Spore sampling is generally representative of the real airborne inoculum concentration. However, because several samplers may be necessary to achieve an acceptable level of representativeness \[[@pone.0144573.ref017]\] the cost and time required to obtain the information are increased, especially when the spore count is done by microscopy. In contrast, using a simulation model to estimate airborne inoculum is less costly and easier to implement once the model has been developed. However, the reliability of simulations is influenced considerably by the quality of the weather data used. The ideal situation is probably to combine the monitoring and simulation of airborne inoculum. In this study, a weather-based simulation model of B. lactucae ACC was developed and validated with independent data (data not used to develop the model). For this model, the asexual life cycle of B. lactucae was divided into six state variables, and the changes from one state to the next were described using mathematical equations derived from the scientific literature or developed by assembling knowledge of the interactions between pathogen and host. The model was evaluated with four years of data by comparing model simulations with field observations of hourly and daily airborne conidia concentrations. The model followed the trend and shape of hourly and daily-observed conidia concentrations almost perfectly. In over 94% of the simulation runs, the proportion of the linear variation in the hourly-observed values explained by the variation in the hourly-simulated values was greater than 0.7 in all years except in 1997. Over the four-year simulation period, the proportion of the linear variation in the daily-observed values explained by the variation in the daily-simulated values was greater than 0.70 in all years except in 1997. Also, most of the errors in the daily-simulated conidia came from the deviations from the 1:1 line, and the proportion of error associated with the model bias was low. Overall, the model was accurate in simulating the trend and the peaks in B. lactucae ACC, even though the model overestimated the daily number of conidia on some days. This situation is due probably to the model's sensitivity to the sporulation rate. Indeed, sensitivity analysis showed that the model was highly sensitive to variation in the sporulation rate, which was defined as a function of temperature. Further field validations to calibrate the sporulation model developed by Tchervenivanova (1995) seem to be necessary. These results can be useful for decision making to improve lettuce downy mildew management. Indeed, Fall et al. \[[@pone.0144573.ref015]\] found an exponential relationship between downy mildew intensity and the airborne conidia concentration (ACC). An ACC of 14 conidia/m^3^ was required to cause one lesion per leaf in the field \[[@pone.0144573.ref015]\]. Over 95% of the time, when no airborne conidia were measured, the daily-simulated ACC was 0. Therefore, when the simulated number of airborne conidia is 0, fungicides should not be applied, whereas simulated peaks in airborne conidia should lead to fungicide applications. Indeed, the disease will develop only if the pathogen is present in the area \[[@pone.0144573.ref020], [@pone.0144573.ref022]\]. In the pathosystem formed by potato and Phytophthora infestans, an oomycete like B. lactucae, the first signs of disease were detected between 6 and 7 d after the peak in airborne spores \[[@pone.0144573.ref017], [@pone.0144573.ref032]\]. Thus, instead of waiting for signs of downy mildew in the field before running the BREMCAST DSS \[[@pone.0144573.ref012]\], the model developed in this study could be used. It may be risky to wait until downy mildew is noticed in the field before running a BREMCAST DSS. Thus, DSSs can be modified to incorporate simulated or measured airborne inoculum above the lettuce canopy. However, it will be a complex process to modify these DSSs without rebuilding the entire model because, these latter were not built using a dynamic system methodology. Also, one challenge involved in implementing a spore sampling network is to know whether the number of samplers is sufficient to obtain a representative airborne conidia concentration for the targeted area. In this context, it is probably better to combine monitoring and simulation of airborne inoculum instead of increasing the number of samplers, which comes at a cost. Nevertheless, the model missed some observed conidia on certain days. Also, some simulated peaks in conidia were not observed in 1997, and the simulations during that year were not accurate on most of the days. However. the ACC is strongly related to the number of sporulating lesions across the field, and hence it is difficult to estimate the exact ACC. The uncertainties stemming from the highly variable potential number of lesions make it difficult to quantify the airborne inoculum at a large scale. At the moment, there are no methods for quantifying the number of sporulating lesions in a given set of weather conditions. Once such methods have been developed, it will be possible to circumscribe the model at a specific scale (small or large) and improve the accuracy of quantitative simulations of airborne conidia. Conclusion {#sec011} ---------- To the authors' knowledge, this is the first time in the published literature that a quantitative dynamic simulation model of airborne conidia was developed. The model offers advantages for accurately simulating the trend and the temporal progression of B. lactucae airborne conidia. Indeed, from a strictly epidemiological point of view, the focus is on the rate of change rather than the direct stage of the process. Decision support system related to polycyclic disease such as lettuce downy mildew can be significantly improved by taking into account quantitative aspects of the asexual cycle of the pathogen. The existing DSSs in the literature did not take into consideration airborne conidia derived from the asexual life cycle of B. lactucae. The results obtained with this model generally compare favorably with field-observed data for airborne conidia. To the authors' knowledge, only the PLANT-Plus model developed by Dacom for the management of Phytophthora infestans integrates a submodel for airborne spores. Nonetheless, this submodel does not simulate the absolute number of spores; it just confirms their presence \[[@pone.0144573.ref033]\]. Moreover, the model developed in the present study may be used as a research tool for investigating the impact of agricultural practices (e.g., irrigation systems) or weather conditions (e.g., temperature, wind speed, and RH) on airborne inoculum. The next step in the development of the model will be incorporating it into DSSs in order to more efficiently predict episodes of lettuce downy mildew. Supporting Information {#sec012} ====================== ###### Observed and simulated data of airborne conidia of Bremia lactucae. (XLS) ###### Click here for additional data file. ###### All STELLA equations used in the model. (DOCX) ###### Click here for additional data file. The authors gratefully acknowledge, Mathieu Tremblay, Audrey Lavasseur, Laurianne Gauthier and Annie Lefebvre for their support and advice. [^1]: Competing Interests:This study was partly supported by Compagnie de recherche Phytodata inc., the employer of H. Van der Heyden. There are no patents, products in development or marketed products to declare. This does not alter the authors\' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors. [^2]: Conceived and designed the experiments: MFL OC. Performed the experiments: MFL. Analyzed the data: MLF. Contributed reagents/materials/analysis tools: MLF OC HvdH. Wrote the paper: MLF OC HvdH.	{ "pile_set_name": "PubMed Central" }
Introduction {#section1-1076029615601494} ============ It is over 50 years since the first license for a fibrinogen concentrate (FIBRINOGENIO HUMANO LIO.) was granted in Brazil in 1963. Historically, fibrinogen came into focus because of reports suggesting it was a cardiovascular risk factor, and there is still no final answer as to whether it is a mediator,^[@bibr1-1076029615601494]^ or just a marker^[@bibr2-1076029615601494]^ or predictor,^[@bibr3-1076029615601494]^ of cardiovascular disease. It is only recently that research started focusing on the positive role of fibrinogen as a hemostatic agent, particularly in the management of perioperative bleeding. The History of the Term "Fibrinogen" {#section2-1076029615601494} ==================================== Before we discuss the development of fibrinogen concentrate, we take a historical look at where the term "fibrinogen" first came from. As far back as the fifth- to fourth-century BC, Hippocratic doctors noted the presence of fibers in circulating blood. However, it was not until the end of the 17th century that Malpighi, using a single lens microscope, observed that whole blood clots have a fibrous and corpuscular component.^[@bibr4-1076029615601494]^ The term "fibrin" was introduced by Fourcroy in 1801. He demonstrated that plasma contained soluble substances and that the precursor of fibrin was present in plasma but not serum. A few decades later, in 1847, the term "fibrinogen" was invented by Virchow where he stated "should one want to give it a name, it could be called fibrinogen" (translated from German: "vielmehr, wollte man sie benennen, so könnte man sie höchstens Fibrinogen taufen.").^[@bibr5-1076029615601494](p.581)^ Some consider that he created the term out of the words fibrin and oxygen, since he believed that "fibrin was a homogeneous mass formed from soluble fibrin after exposure of blood or exudates to oxygen,"^[@bibr6-1076029615601494](p.301)^ and others suggest that he composed the term by adding the syllable "ogen" for a precursor substance instead of the prefix "pro." At the time, Virchow was experimenting with exudates, rather than with whole blood, and therefore mistakenly applied the term "fibrinogen" to decaying exudates rather than to the precursor of fibrin in whole blood as it was understood later. In fact, Virchow had not discovered fibrinogen, instead Denis (in his "Mémoire sur le sang" in 1859) was the first to recognize that plasma contained a clottable substance different from fibrin and went on to attempt to purify and characterize this protein. Denis proposed the name fibrinogen (independent of Virchow): "I believe that I should repeat once more that I do not mean to define a liquid fibrin, plasma fibrin, but rather a substance which is not at all fibrous and which is the origin of fibrin, clottable lymph, a substance which I would like to call 'sérofibrine' and which one might also designate by the name 'fibrinogène', if such a word would be admissible."^[@bibr7-1076029615601494]^ Nevertheless, it took a further 20 years before fibrinogen was first isolated by Hammarsten (1879) who precipitated and purified it from horse plasma by salting-out with sodium chloride in what has become the classical procedure for fibrinogen preparation.^[@bibr6-1076029615601494]^ The History of Bleeding Management {#section3-1076029615601494} ================================== Nowadays, bleeding management encompasses, among other things, the assessment of bleeding risk prior to surgery, and monitoring and managing blood loss during surgery, including appropriate management of bleeding. Historically, the development of therapeutic blood management can be traced back to the 17th century where the first animal-to-human blood transfusion was conducted in 1667.^[@bibr8-1076029615601494]^ However, it was not until 1818 that James Blundell, a British obstetrician, performed the first human-to-human transfusion---to save a patient from exsanguination.^[@bibr9-1076029615601494]^ The introduction of blood transfusion into the medical armamentarium was further advanced with the groundbreaking discovery of the ABO blood group by the Austrian--American immunologist and pathologist Karl Landsteiner in 1901, a feat for which he was later awarded the Nobel Prize.^[@bibr10-1076029615601494]^ Over the last 100 years, further developments have served to transform the practice of transfusion medicine.^[@bibr11-1076029615601494]^ One of the most relevant developments is the shift from using whole blood to using specific blood components (e.g., red blood cells, platelets, and fresh frozen plasma \[FFP\]) and further to the purified, virus-inactivated, plasma-derived products (such as factor concentrates). This latter development was made possible by the work of Edwin Joseph Cohn who discovered that the complex mixture of components in whole blood could be separated through a process of fractionation, commonly referred to as "Cohn fractionation."^[@bibr12-1076029615601494]^ In 1947, he described 6 major plasma fractions, with fraction I containing most of the fibrinogen. Cohn emphasized the importance of separating different plasma components, as it was more efficient and less wasteful to use a particular component for a specific need rather than whole plasma. This approach proved of great value during World War II with the development of products such as fibrin foam helping to solve major problems connected with hemostasis. This change from the use of specific blood components to purified plasma-derived products for bleeding management represents a continuing paradigm shift. Fibrinogen Concentrate: The Past and The Present {#section4-1076029615601494} ================================================ During major bleeding and replacement with red blood cell concentrates, fibrinogen is the first clotting factor to reach critically low levels.^[@bibr13-1076029615601494]^ Furthermore, low perioperative fibrinogen levels have been shown to be associated with increased bleeding.^[@bibr14-1076029615601494]^ Therefore, increasing the levels through fibrinogen supplementation appears essential to restore normal clotting function.^[@bibr15-1076029615601494][@bibr16-1076029615601494]--[@bibr17-1076029615601494]^ The standard practice of replacing fibrinogen has focused on infusions of therapeutic plasma (eg, FFP) or cryoprecipitate; however, there are differences concerning the volume of each product.^[@bibr18-1076029615601494]^ Furthermore, there are various well-known disadvantages of using these products.^[@bibr11-1076029615601494],[@bibr19-1076029615601494]^ Some of the issues are related to the safety and purity of these products. Therefore, with the progression from using whole blood to allogeneic blood products (eg, plasma, red blood cells, platelet concentrates, and cryoprecipitate), the next consideration should be moving toward more purified, specific products such as factor concentrates (eg, fibrinogen concentrate). This represents the move from the use of a "liquid organ" to a pharmaceutical product. Fibrinogen concentrate (FIBRINOGENIO HUMANO LIO.) was first approved for use in Brazil on March 4, 1963. This day is considered the international birth date of the product. Notably, fibrinogen concentrate and freeze-dried fraction I (prepared by Cohn fractionation of plasma) were already available before licensure of this product became obligatory.^[@bibr20-1076029615601494][@bibr21-1076029615601494]--[@bibr22-1076029615601494]^ The European Union birth date of the product is January 4, 1966,^[@bibr23-1076029615601494]^ when fibrinogen concentrate (Human-Fibrinogen Behringwerke Konzentrat) was approved in Europe for the first time by the German Federal Ministry of Health. However, Behringwerke (a predecessor company of CSL Behring GmbH) had already commenced production of human fibrinogen concentrate in 1956, before registration of a product became mandatory with the Medicinal Products Act coming in to force in 1961 in Germany. In 1985, the production process was modified to include pasteurization as a virus inactivation step, and the product was renamed Haemocomplettan P (CSL Behring GmbH). Currently, all formulations of fibrinogen concentrate manufactured by CSL Behring are pasteurized, and over 3 million grams have been used since 1985.^[@bibr24-1076029615601494]^ At the time of writing, CSL Behring's fibrinogen concentrate is approved for treatment and prophylaxis of acquired and congenital fibrinogen deficiency under the trade name Haemocomplettan P in several countries ([Table 1](#table1-1076029615601494){ref-type="table"}).^[@bibr25-1076029615601494],[@bibr26-1076029615601494]^ In Israel, Haemocomplettan P is indicated for the treatment of acute bleeding episodes in patients with congenital fibrinogen deficiency, including afibrinogenemia and hypofibrinogenemia. The same product, licensed under the trade name RiaSTAP, is also available for the treatment of acute bleeding episodes in patients with congenital fibrinogen deficiency, including afibrinogenemia and hypofibrinogenemia, in Australia, Canada, Mexico, New Zealand, Puerto Rico, United States, and several European countries ([Table 1](#table1-1076029615601494){ref-type="table"}).^[@bibr25-1076029615601494],[@bibr26-1076029615601494]^ ###### Current Licensing Status of Fibrinogen Concentrate Manufactured by CSL Behring. ![](10.1177_1076029615601494-table1) Product Name Availability^a^ Date of First Approval Indication ------------------------- ------------------------------ ------------------------------------------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------------ Haemocomplettan^®^ P^b^ Argentina August 2012 Treatment and prophylaxis of acquired and congenital fibrinogen deficiency Austria June 1994 Brazil March 1963 (March 2008^c^) Bulgaria January 2009 Czech Republic March 1993 Germany March 1966 (March 2005^c^) Hungary June 1998 Iran January 2010 Kuwait October 2003 Lebanon December 2013 Netherlands March 1997 Portugal January 1978 Romania July 1999 Switzerland November 1992 Taiwan March 1970 (January 1987^c^) Tunisia April 2013 Turkey October 1997 Uruguay October 2013 Israel August 2009 Treatment of acute bleeding episodes in patients with congenital fibrinogen deficiency, including afibrinogenemia and hypofibrinogenemia RiaSTAP^®^ Australia August 2010 Treatment of acute bleeding episodes in patients with congenital fibrinogen deficiency, including afibrinogenemia and hypofibrinogenemia Belgium October 2010 Canada September 2012 Cyprus January 2012 Denmark August 2010 Finland September 2010 France October 2010 Germany December 2009 Greece May 2011 Iceland August 2010 Ireland September 2010 Italy April 2012 Luxembourg February 2011 Malta February 2012 Mexico January 2013 New Zealand May 2011 Norway November 2010 Poland June 2011 Puerto Rico December 2009 Slovakia October 2010 Slovenia June 2011 Spain March 2011 Sweden October 2010 United Kingdom August 2010 United States January 2009 ^a^Additional countries received licenses to use the product that are no longer active, for example, pre-1985 (Colombia, Jamaica, Pakistan) and post-1985 (Croatia, India). ^b^Other trade names have been used for this product during the licensing history and in other countries, for example, FIBRINOGENIO HUMANO LIOF, Fibrinogenio Humano, Fibrinogenio, Haemocomplettan, Haemocomplettan HS, Human-Fibrinogen Behringwerke Konzentrat. ^c^Approval date of current license. Current evidence suggests that fibrinogen concentrate has a good safety profile and delivers a standardized dose of fibrinogen.^[@bibr27-1076029615601494]^ Haemocomplettan P/RiaSTAP has been shown to be effective and well tolerated in a number of clinical studies of patients with congenital or acquired fibrinogen deficiency.^[@bibr28-1076029615601494][@bibr29-1076029615601494][@bibr30-1076029615601494][@bibr31-1076029615601494][@bibr32-1076029615601494][@bibr33-1076029615601494][@bibr34-1076029615601494]--[@bibr35-1076029615601494]^ A postmarketing surveillance study of Haemocomplettan P/RiaSTAP covering a 27-year period (1986-2013) found 28 spontaneous reports of possibly related thrombotic events for 2 611 294 g of distributed product.^[@bibr27-1076029615601494]^ This represents an incidence rate of 1 event per 23 300 doses. The use of Haemocomplettan P/RiaSTAP to control bleeding and to decrease the use of allogeneic blood products during cases of acquired bleeding has been reported in a variety of clinical settings, including surgery, trauma, liver transplantation, and obstetrics.^[@bibr28-1076029615601494][@bibr29-1076029615601494]--[@bibr30-1076029615601494],[@bibr33-1076029615601494],[@bibr36-1076029615601494][@bibr37-1076029615601494][@bibr38-1076029615601494][@bibr39-1076029615601494][@bibr40-1076029615601494][@bibr41-1076029615601494][@bibr42-1076029615601494]--[@bibr43-1076029615601494]^ Furthermore, recent systematic reviews confirm that the use of fibrinogen concentrate may reduce allogeneic blood product transfusion^[@bibr44-1076029615601494][@bibr45-1076029615601494]--[@bibr46-1076029615601494]^; however, very often the trials are of low quality with a high risk of bias. Moreover, the measured outcomes, such as transfusion requirements, may be prone to performance bias.^[@bibr47-1076029615601494]^ Therefore, more randomized controlled trials are warranted before it can be confirmed whether fibrinogen concentrate has a definitive role in the management of bleeding. Additionally, some controversy still exists around the use of fibrinogen concentrate with regard to the appropriate level of fibrinogen to trigger treatment.^[@bibr48-1076029615601494],[@bibr49-1076029615601494]^ Ranucci and Solomon are right to note in their Editorial that there is currently a discrepancy among licences, guidelines, and clinical use and that there are open issues and gaps in knowledge that need to be addressed.^[@bibr50-1076029615601494]^ What Does the Future Hold for Fibrinogen Concentrate? {#section5-1076029615601494} ===================================================== Four fibrinogen concentrate products are currently available in various regions^[@bibr25-1076029615601494]^ and according to public sources more are in development.^[@bibr51-1076029615601494]^ Other developments include the production of recombinant fibrinogen, first engineered by Lord et al in 1993.^[@bibr52-1076029615601494]^ The functional properties of a recombinant fibrinogen have been tested ex vivo in whole blood samples from a group of patients undergoing cardiac surgery.^[@bibr53-1076029615601494]^ According to the clinical trials registries ([clinicaltrials.gov](http://clinicaltrials.gov) and [clinicaltrialsregister.eu](http://clinicaltrialsregister.eu)), there are currently 11 ongoing clinical studies investigating the use of fibrinogen concentrate in acquired bleeding. The results of these studies will potentially provide further evidence on the role of fibrinogen concentrate in the management of bleeding. These ongoing investigations will provide additional evidence of fibrinogen concentrate in acquired bleeding, but fibrinogen may have a number of different applications beyond bleeding management. Thrombocytopenia is a common complication in hospitalized patients which affects blood clotting, and studies have shown that fibrinogen concentrate can increase the clot strength, independent of platelet count.^[@bibr54-1076029615601494],[@bibr55-1076029615601494]^ Further clinical investigations are necessary to confirm whether bleeding complications associated with thrombocytopenia could be managed through supplementation with fibrinogen concentrate. The role of fibrinogen and interference with the immune system is also worthy of further scrutiny, as a significant decrease in sepsis-induced mortality was observed in animals treated with fibrinogen concentrate in a rat model of sepsis-induced disseminated intravascular coagulation.^[@bibr56-1076029615601494]^ Advances in science and medicine have led to improvements in blood and blood products and in the way patients are treated. Fibrinogen concentrate seems to offer a specific option in the management of bleeding; however, further investigations are required to clearly define its role. Fibrinogen concentrate was described in 2009 as a "potential universal hemostatic agent"^[@bibr57-1076029615601494]^; the next 50 years will demonstrate whether this is the case. Editorial assistance with manuscript preparation was provided by Meridian HealthComms, funded by CSL Behring. Authors' Note: Michael Wendt provided company and product information. Gerald Hochleitner developed the outline and researched the historical aspect of fibrinogen. All authors were involved in writing and reviewing the manuscript and gave final approval of the version to be published. Declaration of Conflicting Interests: The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: R.C-F. has no conflict of interest to declare. G.H. and M.W. are employees of CSL Behring. A.T. received lecture fees and travel support from CSL Behring. D.R.S. was the chairman of the ABC Faculty and is the co-chairman of the ABC-Trauma Faculty, which both are managed by Physicians World Europe GmbH, Mannheim, Germany and sponsored by unrestricted educational grants from Novo Nordisk Health Care AG, Zurich, Switzerland, CSL Behring GmbH, Marburg, Germany and LFB Biomédicaments, Courtaboeuf Cedex, France. In the past 5 years, D.R.S. has received honoraria or travel support for consulting or lecturing from the following companies: Abbott AG, Baar, Switzerland; AMGEN GmbH, Munich, Germany; AstraZeneca AG, Zug, Switzerland; Bayer (Schweiz) AG, Zürich, Switzerland; Baxter AG, Volketswil, Switzerland; Baxter S.p.A., Roma, Italy; B. Braun Melsungen AG, Melsungen, Germany; Boehringer Ingelheim (Schweiz) GmbH, Basel, Switzerland; Bristol-Myers-Squibb, Rueil-Malmaison Cedex, France and Baar, Switzerland; CSL Behring GmbH, Hattersheim am Main, Germany and Berne, Switzerland; Curacyte AG, Munich, Germany; Daiichi Sankyo (Schweiz) AG, Thalwil, Switzerland; Ethicon Biosurgery, Sommerville, New Jersey, USA; Fresenius SE, Bad Homburg v.d.H., Germany; Galenica AG, Bern, Switzerland (including Vifor SA, Villars-sur-Glâne, Switzerland); GlaxoSmithKline GmbH & Co. KG, Hamburg, Germany; Janssen-Cilag AG, Baar, Switzerland; Janssen-Cilag EMEA, Beerse, Belgium; Merck Sharp & Dohme AG, Luzern, Switzerland; Novo Nordisk A/S, Bagsvärd, Denmark; Octapharma AG, Lachen, Switzerland; Organon AG, Pfäffikon/SZ, Switzerland; Oxygen Biotherapeutics, Costa Mesa, CA, USA; PAION Deutschland GmbH, Aachen, Germany; Photonics Healthcare B.V., Utrecht, Netherlands; ratiopharm Arzneimittel Vertriebs-GmbH, Vienna, Austria; Roche Diagnostics International Ltd, Reinach, Switzerland; Roche Pharma (Schweiz) AG, Reinach, Switzerland; Schering-Plough International, Inc., Kenilworth, New Jersey, USA; Tem International GmbH, Munich, Germany; Verum Diagnostica GmbH, Munich, Germany; Vifor Pharma Deutschland GmbH, Munich, Germany; Vifor Pharma Österreich GmbH, Vienna, Austria; Vifor (International) AG, St. Gallen, Switzerland. Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: D.R.S's academic department is receiving grant support from the Swiss National Science Foundation, Berne, Switzerland (grant numbers: 100 014_138545/1, 33CM30_124117, 33CM30_140339 and 406440-131268), the Ministry of Health (Gesundheitsdirektion) of the Canton of Zurich, Switzerland for Highly Specialized Medicine, the Swiss Society of Anesthesiology and Reanimation (SGAR), Berne, Switzerland (no grant numbers attributed), the Swiss Foundation for Anesthesia Research, Zurich, Switzerland (no grant numbers attributed), Bundesprogramm Chancengleichheit, Berne, Switzerland (no grant numbers attributed), CSL Behring, Berne, Switzerland (no grant numbers attributed), Vifor SA, Villars-sur-Glâne, Switzerland (no grant numbers attributed).	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#SEC1} ============ Mutations can confer selective growth advantages on cancer cells, leading to their clonal selection and expansion. Analysis of tumor genome sequences led to the identification of several different patterns of mutations or mutational signatures ([@B1]). Many such mutational signatures are linked to underlying external causes such as exposures to sunlight or cigarette smoke. Additionally, specific defects in DNA repair mechanisms also generate unique mutational signatures, linking DNA repair abnormalities to mutation accumulation and genomic instability in cancer cells ([@B1],[@B2]). The discovery that hereditary forms of breast cancer are caused by a defect in homologous recombination (HR), a repair pathway that utilizes the undamaged sister chromatid as a template to repair DNA double-strand breaks (DSBs), led to the development of PARP inhibitors as therapeutics to selectively target cancers with HR defects ([@B3],[@B4]). To date, key HR factors, including BRCA1, BRCA2, PALB2 and RAD51, have been identified as cancer susceptibility genes. Mutations in these genes are associated with a specific mutational signature, namely signature 3 ([@B5]). Furthermore, PARP inhibitors are effective clinically for breast and ovarian cancers with this signature ([@B6]). Importantly, in HR-deficient tumors the repair of DSBs that arise during the S/G2 phases of the cell cycle occurs via less understood alternative end joining pathways, which are often error-prone ([@B7]). These pathways, initially discovered as backup pathways for non-homologous end joining (NHEJ), frequently utilize microhomologies to bridge the broken ends. They typically involve limited resection at DSBs by MRE11 and CtIP, gap-filling synthesis by DNA polymerase θ (POLQ), scaffolding by XRCC1 and ligation by DNA ligase 1 or 3 (LIG1/LIG3) ([@B12]). Although microhomology-mediated end joining (MMEJ) has a minor role in NHEJ- and HR-proficient cells, it contributes to chromosomal rearrangements and genomic instability in aneuploid cancer cells ([@B13]), and is upregulated in HR-deficient tumors, where POLQ is critical for genome maintenance and tumor survival ([@B7],[@B16]). Additional MMEJ factors, PARP1 and XRCC1, were also identified as being synthetically lethal with BRCA2 in a recent CRISPR screen ([@B17]). Error-prone DSB repair (DSBR) in sister chromatids by MMEJ contributes to deletions and translocations in HR-deficient cancers ([@B5],[@B11]). In addition, there is emerging evidence that degradation of stalled replication forks followed by replication restart also contributes to genome instability in these tumors ([@B18]). Interestingly, while BRCA1, BRCA2 and RAD51 have HR-independent roles in fork protection ([@B21],[@B23],[@B24]), MRE11 and PARP1 are involved in fork degradation ([@B25],[@B26]). Since PARP1 and MRE11 are also MMEJ factors ([@B27]), we reasoned that MMEJ likely contributes to fork degradation and subsequent replication restart at the degraded forks. This hypothesis is consistent with recent studies showing that POLQ is critical for replication-associated DSBR ([@B28],[@B29]). While XRCC1 engages with MRE11 and CtIP in response to ionizing radiation to form a MMEJ-competent complex ([@B27],[@B30]) in addition to its canonical role as a scaffolding protein that coordinates base excision repair (BER) and single-strand break repair (SSBR) ([@B31]), the role of XRCC1 at stalled replication forks is less well defined. Here, we examined whether XRCC1-dependent MMEJ contributes to the repair of single-ended DSBs (seDSBs) that arise at collapsed replication forks. We find that replication fork collapse induces formation of a MMEJ-competent XRCC1 complex containing end resection factors and DNA polymerases in BRCA2-deficient cells. Furthermore, XRCC1 promotes stalled fork degradation and replication restart in BRCA2-deficient cells. Thus, in the absence of fork protection by BRCA2, XRCC1 enables cells to complete DNA replication at the expense of increased genome instability by promoting fork resection followed by replication restart. MATERIALS AND METHODS {#SEC2} ===================== Cell culture {#SEC2-1} ------------ All in-cell and in vitro repair assays were performed with U2OS cells. Stable inducible shRNA-expressing cell lines (scrambled control, scr; and BRCA2, B2) were a generous gift from Ryan Jensen (Yale University). U2OS-EJ2 cells were a kind gift from Jeremy Stark (City of Hope) ([@B32]). All cell lines were cultured in Dulbecco's modified Eagle medium (DMEM, high-glucose, Gibco-BRL) supplemented with 10% fetal calf serum (Sigma) and 100 U/ml penicillin and 100 μg/ml streptomycin (Gibco-BRL). Inducible cell lines were cultured in 2 μg/ml puromycin (InvivoGen), and shRNA expression was induced with 10 μg/ml doxycycline for 72 h. Cells were pretreated with 100 μM mirin (Sigma) for 1 h to inhibit MRE11 exonuclease activity or 10 μM rucaparib (Selleck) to inhibit PARP activity for EJ2-U2OS assays. Ataxia telangiectasia and Rad3-related kinase inhibitor (ATRi) VE-821 (Sigma) and hydroxyurea (HU; Sigma) were used to induce replication stress as indicated in the figure legends. Plasmid and siRNA transfection {#SEC2-2} ------------------------------ The XRCC1^WT^ cDNA sequences were subcloned from respective 6X-His-tag-containing pCDE2 vectors into the p3XFLAG-CMV14 vector. myc-hPolQ-Flag vector was purchased from Addgene (Plasmid \#73132) (deposited by Agnel Sfeir) ([@B8]). Treatment with 100 nM siRNA oligonucleotides (Sigma, TX) for XRCC1, CtIP, FEN1, LIG3, POLQ, BRCA1, BRCA2 and LIG1 ([Supplementary Table S1](#sup1){ref-type="supplementary-material"}) depleted the respective proteins ([Supplementary Figure S1A](#sup1){ref-type="supplementary-material"}). Exponentially growing cells were transfected with plasmids or siRNA at the indicated concentrations with Lipofectamine 2000 in OptiMEM media (Gibco-BRL) following the manufacturer's protocol. The media were changed after 4 h, and the cells were incubated for 48 h for XRCC1 and POLQ expression and 72 h for siRNA treatment. Clonogenic cell survival assay {#SEC2-3} ------------------------------ U2OS cells were transfected with 100 nM control or XRCC1 siRNA, and after 72 h incubation the cells were treated with HU and/or the ATRi, VE-821, for the indicated times. The cells were then detached with trypsin and 500 cells from each sample were plated in triplicate in six-well dishes. After 10 days, the cells were fixed and stained with 0.5% crystal violet solution in 50% methanol and colonies were counted. MTT assay {#SEC2-4} --------- U2OS cells pretreated with siRNA and doxycycline were seeded in triplicate at a density of 3--5 × 10^3^ cells/well in a 96-well plate, incubated overnight and exposed to rucaparib (Selleck) at indicated concentrations for 120 h. The cells were then exposed to a tetrazolium compound (TACS MTT Reagent, Trevigen) for 4 h, followed by solubilization for 2 h. Absorbance at 570 nm was measured using an Infinite M1000 microplate reader (TECAN). Comet assay {#SEC2-5} ----------- Neutral comet assay was performed using the Trevigen Comet Assay Kit (4250-050-K) according to manufacturer's protocol. At least 50 random comets for each sample were analyzed using CaspLab ([@B33]). Antibodies {#SEC2-6} ---------- Primary antibodies used were mouse monoclonal ANTI-FLAG^®^ M2-peroxidase (HRP) antibody (A8592, Sigma), mouse monoclonal ANTI-FLAG^®^ M2 antibody (F1804, Sigma), rabbit polyclonal anti-DYKDDDDK tag antibody (\#2368, Cell Signaling Technology), mouse monoclonal anti-XRCC1 antibody (\#MS-434-P0, Thermo Scientific), rabbit monoclonal anti-XRCC1 antibody (ab134056, Abcam), rabbit polyclonal anti-PARP-1 antibody (H-300) (sc-25780, Santa Cruz Biotechnology), mouse monoclonal anti-DNA ligase 3 antibody (E-7) (sc-390922, Santa Cruz Biotechnology), mouse monoclonal anti-DNA ligase 1 antibody (ab615, Abcam), rabbit polyclonal anti-DNA polymerase beta antibody (18003-1-AP, Proteintech), rabbit polyclonal anti-MRE11 antibody (\#4895, Cell Signaling Technology), rabbit monoclonal anti-phospho-histone H2A.X antibody (Ser139) (20E3) (\#9718, Cell Signaling Technology), mouse monoclonal anti-phospho-histone H2A.X (Ser139) antibody (\#05-636, EMD Millipore), mouse monoclonal anti-BRCA1 antibody (D-9) (sc-6954, Santa Cruz Biotechnology), mouse monoclonal anti-BRCA2 antibody (2B) (OP95, EMD Millipore), mouse monoclonal anti-FEN1 antibody (B4) (sc-28355, Santa Cruz Biotechnology), mouse monoclonal anti-BrdU antibody (IIB5) (ab8152, Abcam) and mouse monoclonal anti-β-Actin antibody (A5316, Sigma). Secondary antibodies for western blotting were from GE Healthcare (anti-mouse, NA9310V; anti-rabbit, NA934V). Secondary antibodies for immunofluorescence were from Invitrogen (Alexa Fluor, anti-rabbit 594, A11037; anti-mouse 594, A11005; anti-rabbit 488, A11008; anti-mouse 488, A11001). γH2AX and XRCC1 foci analysis {#SEC2-7} ----------------------------- U2OS cells were transfected with 100 nM scrambled control or XRCC1 siRNA; after 48 h, the cells were plated in eight-chamber slides, incubated overnight and then treated with HU and/or ATRi as indicated. The cells were then fixed with 4% paraformaldehyde for 15 min, followed by permeabilization with 0.5% Triton X solution in phosphate-buffered saline (PBS) for 30 min. Subsequently, after blocking with 3% bovine serum albumin (BSA) solution in PBS for 1 h, the cells were incubated with anti-phosphoserine H2A.X antibody (Cell Signaling Technology, \#9718) or anti-XRCC1 antibody (\#MS-434-P0, Thermo Scientific) diluted 1:500 in PBS for 2 h at room temperature. After washing with PBS, the cells were incubated with Alexa Fluor secondary antibody (1:500). After the final wash, the slides were dried for 5--10 min at 37°C and mounted with mounting media with DAPI (Invitrogen, P36941) and coverslips. Samples were observed under 60× oil immersion lens, and images were captured from at least 10 random fields for each sample. Cells were marked positive if they contained \>10 foci. Proximity ligation assay and immunostaining {#SEC2-8} ------------------------------------------- For proximity ligation assay (PLA) and immunostaining, U2OS cells grown in an eight-chamber slide were treated as described in the figure legends and then fixed with 4% paraformaldehyde in PBS followed by permeabilization with 0.5% Triton X in PBS. For each PLA experiment, the respective primary antibodies raised in different species were used. Assays were performed with the Duolink kit (Olink Bioscience, Uppsala, Sweden) according to the manufacturer's instructions. PLA foci/immunostaining were visualized using a 60× oil immersion lens in an inverted Zeiss bright-field/fluorescent microscope. For XRCC1-BrdU PLA, asynchronous U2OS cells were pulsed with 10 μM BrdU for 15 min before treatment with replication stalling agents for various times. Genomic DNA was denatured after fixation and permeabilization. Images were merged and analyzed with ImageJ software. For the PLA to detect co-localization of XRCC1 and POLQ, U2OS cells were transfected with the construct expressing POLQ-FLAG 48 h before treatment. Chromosomal MMEJ assay {#SEC2-9} ---------------------- Incubation of U2OS-EJ2 cells with HU and/or ATRi, depletion with siRNA, DSB induction and cell harvesting were carried out as described ([@B32]). Flow cytometry was performed using a BD FACS LSRII and data were analyzed using Flowing Software (Perttu Terho, Turku Centre for Biotechnology). In-cell plasmid circularization assay {#SEC2-10} ------------------------------------- U2OS cells grown in 60 mm plates (50% confluent) were treated with HU and/or ATRi as indicated, and then transfected with 100 ng pNS (the linearized DSBR substrate, Figure [3B](#F3){ref-type="fig"}) ([@B30]) using Lipofectamine 2000 and incubated overnight (15 h). Plasmids (5 μl) were isolated using the Qiagen plasmid miniprep kit prior to transformation of XL10-gold ultracompetent Escherichia coli cells (Agilent) following the manufacturer's protocol. Forty colonies were randomly selected for plasmid sequencing using the CMV-F primer by Genewiz, Inc. Sequences with insertions/deletions of 1--4 nt at the DSB site were scored as NHEJ products, whereas the plasmids with deletion of one of the 5 nt microhomology sequences were scored as products of MMEJ/alt-EJ products, as described ([@B30]). Nonspecific extended deletions (\>10 nt) at the 3′ or 5′ end of the DSB were not considered when plotting MMEJ versus NHEJ. Preparation of whole cell and nuclear extracts {#SEC2-11} ---------------------------------------------- U2OS cells were harvested, washed with Dulbecco's phosphate-buffered saline solution (Cellgro, Corning) and then pelleted by centrifugation at 800 rpm for 5 min. For preparing whole cell lysates, the cell pellet was resuspended in whole cell lysis buffer (20 mM Tris--HCl, pH 7.5, 150 mM NaCl and 1% Triton X) with protease inhibitor cocktail (Thermo Fisher Scientific), vortexed at 4°C for 15 min and then clarified by centrifugation at 14 000 rpm. For preparation of nuclear extracts, the cell pellet was resuspended in cytoplasmic extraction buffer (10 mM Tris--HCl, pH 7.9, 0.34 M sucrose, 3 mM CaCl~2~, 2 mM MgCl~2~, 0.1 mM EDTA, 1 mM DTT and 0.1% Nonidet P-40) with protease inhibitor cocktail. The cell suspension was vortexed briefly, centrifuged at 3500 × g for 15 min at 4°C. The pellet of nuclei was resuspended in whole cell lysis buffer and a soluble extract prepared as described above. For co-immunoprecipitation assays, the nuclear extracts were incubated with 0.15 U/μl benzonase (EMD Millipore) to avoid DNA-mediated immunoprecipitation of protein complexes. In vitro MMEJ assay {#SEC2-12} --------------------- Exponentially growing U2OS cells transiently expressing XRCC1-FLAG were treated with HU and/or ATRi as indicated. After treatment, the irradiated and control cells were harvested for preparation of nuclear extracts. XRCC1-FLAG IP was isolated by incubating the nuclear extract with FLAG-M2 agarose beads for 2 h at 4°C. The beads were directly incubated for 30 min with 5 ng pNS in a reaction buffer containing 2 mM MgCl~2~, 60 mM NaCl, 50 mM HEPES, 2 mM DTT, 1 mM ATP, 1 mM dNTPs and 50 μg/ml BSA, with gentle shaking at 30°C. This was followed by addition of 14 ng XRCC1/LIG3 recombinant protein complex (as the ligation step is rate limiting) to the reaction mix for further incubation for 15 h at 16°C. After removal of the beads by low-speed centrifugation, 5 μl supernatant was used for transformation of XL10-gold ultracompetent E. coli cells (Agilent Technologies). The separated beads were also eluted with 4× LDS loading buffer for western blot analysis. The colonies in each plate were counted and submitted for sequence analysis using the CMV-F primer (Genewiz Inc.) similar to that used in the in-cell assays. Details on pNS synthesis are given in ([@B30]). SSBR assay {#SEC2-13} ---------- The in vitro ligation activity assay was performed as previously described ([@B34]). Briefly, annealed oligomers labeled with Cy3 fluorescent dye were mixed with IP complexes in 1× T4 ligation buffer, and the mixture was incubated in a water bath at 30°C for 20 min, followed by incubating with 2× TBE sample buffer at 100°C for 3 min and on ice for another 3 min. Oligomers were separated by denaturing urea polyacrylamide gel electrophoresis, and Cy3 fluorescence was detected by a Typhoon FLA 7000 system. DNA fiber analysis {#SEC2-14} ------------------ Cells were labeled with 50 μM CldU (Sigma), exposed to HU (4 mM) and labeled with 50 μM IdU (Sigma) as indicated in the figures. DNA fibers were spread as previously described ([@B35]), and fiber tracts were detected using anti-IdU (BD Biosciences, 347580) and anti-CldU (Novus Biologicals, NB500-169) primary antibodies and Alexa Fluor 488 and 555 secondary antibodies (Invitrogen). Fibers were imaged at 60× magnification with oil immersion using a Zeiss microscope and analyzed with ImageJ. Statistical and bioinformatic analyses {#SEC2-15} -------------------------------------- Most statistics were performed using Prism software. Fiber assay distributions were analyzed using the Mann--Whitney U test. PLA foci distributions, comet assay distributions, MMEJ in vitro assay, EJ2 repair events and clonogenic survival assay were analyzed using Student's t-test. Foci formation and in-cell MMEJ assay results were analyzed using Fisher's exact test. Mutational signature data were from COSMIC (<https://cancer.sanger.ac.uk/cosmic/signatures>); heat plots were generated in R using the 'gplots' package. TCGA data for gene expression were obtained through the TCGA Assembler utility and analyzed using custom scripts. P-values were obtained from a Welch's t-test with unequal variance. Survival curves and hazard ratios were obtained from the 'survminer', 'survival' and 'dplyr' R packages. ns: P \> 0.05; \*P \< 0.05; \\P \< 0.01; \\\*P \< 0.001 in all figures. RESULTS {#SEC3} ======= XRCC1 has minor impact on cell survival and DSBR after replication stress in HR-proficient cells {#SEC3-1} ------------------------------------------------------------------------------------------------ Because XRCC1-depleted cells are sensitive to inhibition of ATR, the key proximal kinase activated by replication stress ([@B36],[@B37]), we sought to examine XRCC1's role in the replication stress response. XRCC1-depleted U2OS cells (Figure [1A](#F1){ref-type="fig"}) were modestly sensitive to the ATRi VE-821 (Figure [1B](#F1){ref-type="fig"}). HU, a ribonucleotide reductase inhibitor, induces replication stress by depleting cell deoxyribonucleotides, causing replication fork stalling. Similar to the ATRi results, XRCC1-depleted cells were modestly sensitive to HU (Figure [1C](#F1){ref-type="fig"}). HU induced a larger but not significant increase in DSBs in XRCC1-depleted cells compared to control cells as measured by the comet assay (Figure [1D](#F1){ref-type="fig"}). There was also a modest but not significant reduction in the rate of repair of HU-induced DSBs in XRCC1-depleted cells (Figure [1E](#F1){ref-type="fig"}). These results demonstrate a less robust effect of ATR inhibition on XRCC1-deficient cell survival than previously reported ([@B36]), likely in part due to the usage of U2OS cells rather than CHO cells in our experiments. Taken together, these results indicate that XRCC1 has a minor role in the response to replication stress in cells that are NHEJ and HR proficient. ![Effect of XRCC1 depletion and replication stress in WT cells. (A) Western blot of XRCC1 depletion by siRNA. (B) Clonogenic survival assay of XRCC1-depleted cells to ATRi. (C) Clonogenic survival assay of XRCC1-depleted cells to HU. (D) Neutral comet assay in ctrl and XRCC1 siRNA-treated cells after HU. Cells were treated with 3 mM HU for 8 h. At least 50 cells were analyzed for each experiment. (E) Relative fraction of U2OS cells positive for the DSB marker γH2AX by immunofluorescence. Cells were treated with 3 mM HU for 3 h and then allowed to recover for the indicated times. Cells were marked positive if they contained \>10 foci. The number of cells analyzed in each experiment varied from 55 to 72. P-values obtained from z-tests at each time point varied from 0.88 (time 0) to 0.35 (2 h). The fractions of positive cells at time 0 were 0.49 (control siRNA, 55 cells) and 0.47 (XRCC1 siRNA, 65 cells).](zcaa013fig1){#F1} XRCC1 is recruited to replication stress sites and co-localizes with DNA damage response factors {#SEC3-2} ------------------------------------------------------------------------------------------------ Since most forks stalled in response to HU treatment are stable without prolonged incubation, we examined the effect of treating cells with HU and simultaneously inhibiting ATR to induce the collapse of stalled forks ([@B36]). Cortez and colleagues have demonstrated the formation of DSBs with dissociation of replisome components both after prolonged treatment of cells with HU and more quickly after combined treatment of HU and ATRi ([@B37]). As expected, distinct XRCC1 foci, which co-localized with phosphorylated histone H2AX (γH2AX) in the proximity of DSBs, formed in response to exposure to HU and to a greater extent after combined exposure to HU and ATRi (Figure [2A](#F2){ref-type="fig"} and [B](#F2){ref-type="fig"}). To determine whether XRCC1 is recruited to DSBs at stalled and collapsed replication forks, we used the PLA, which produces a fluorescent focus when two distinct antibodies are in close proximity (∼40 nm; Duolink). Asynchronous U2OS cells were pulse-labeled with BrdU to mark sites of DNA replication and then incubated with either HU or HU together with ATRi. The co-localization of XRCC1 with BrdU increased in a time-dependent manner both in the HU-treated cells and, to a greater extent, in cells co-incubated with HU and ATRi (Figure [2C](#F2){ref-type="fig"} and [E](#F2){ref-type="fig"}). ![XRCC1 is recruited to sites of replication stress-induced DNA damage. (A) XRCC1 foci formation after treatment of U2OS cells with 3 mM HU and/or 3 μM ATRi for 8 h where indicated. Cells were marked positive if they contained \>10 foci. At least 50 cells were analyzed for each experiment. (B) Co-localization of XRCC1 with γH2AX after ATRi + HU. (C) Representative images of the PLA data related to (E). (D) Representative images of the PLA data related to (F). (E) XRCC1 localization to sites of replication stress, as measured by BrdU-XRCC1 PLA. Asynchronous cells were pulsed with 10 μM BrdU for 15 min before treatment with 3 mM HU and/or 3 μM ATRi for the indicated times. At least 50 cells were analyzed for each experiment. (F) PLA between XRCC1 and the indicated factors in control cells and after treatment for 8 h with 3 mM HU and 3 μM ATRi. At least 50 cells were analyzed for each experiment.](zcaa013fig2){#F2} We then utilized PLA to quantify changes in association between XRCC1 and other DNA damage response proteins (Figure [2D](#F2){ref-type="fig"} and [F](#F2){ref-type="fig"}). Significant increases in the co-localization of XRCC1 with the MMEJ factors PARP1, MRE11 and POLQ, as well as γH2AX were observed, supporting a role for XRCC1 at DSBs formed in response to replication stress. In contrast, the association between XRCC1 and POLβ, the canonical gap-filling polymerase in BER and SSBR, did not change after replication stress. Replication stress stimulates microhomology-mediated end joining by forming an XRCC1 repair complex {#SEC3-3} --------------------------------------------------------------------------------------------------- Based on the increased association between XRCC1 and other MMEJ factors in response to replication stress and collapsed replication forks, we measured MMEJ activity utilizing a chromosomally integrated MMEJ reporter system (EJ2) (Figure [3A](#F3){ref-type="fig"}, upper panel) ([@B32]). This measures MMEJ by restoration of GFP after DSB induction by I-SceI. In parallel, we used a linearized plasmid reporter system (Figure [3B](#F3){ref-type="fig"}, upper panel) that measures both MMEJ and NHEJ based on sequencing DSB joints in plasmids recovered from transfected cells ([@B30]). While neither of these assays directly measures events at stalled/collapsed replication forks, they do detect increased MMEJ activity that includes resection, annealing and subsequent ligation. We suggest that these MMEJ assays serve as surrogates for the activity of MMEJ factors at stalled/collapsed replication forks. Treatment of U2OS-EJ2 cells with HU or a combination of HU and ATRi led to an increase in the total number of MMEJ events, although the increase with HU alone did not reach statistical significance (Figure [3A](#F3){ref-type="fig"}, lower panel). In the linearized plasmid reporter system, pretreatment of U2OS cells with HU or a combination of HU and ATRi led to an increase in the number of MMEJ events relative to NHEJ events (Figure [3B](#F3){ref-type="fig"}, lower panel). Repair sequences are listed in [Supplementary Figure S2](#sup1){ref-type="supplementary-material"}. Thus, replication stress promotes cellular MMEJ repair activity, supporting and extending previous observations ([@B38]). As expected, this enhancement was dependent on XRCC1 and CtIP levels, in addition to PARP1 and MRE11 activity ([Supplementary Figure S1B](#sup1){ref-type="supplementary-material"}). ![Replication stress activates MMEJ. (A) Upper panel: scheme for repair of I-SceI-induced DSBs via MMEJ in the EJ2-U2OS cell line; lower panel: repair of I-SceI-induced DSBs after replication stress. EJ2-U2OS cells were treated with the indicated doses of HU and VE-821 (ATRi) for 8 h before DSB induction. (B) Upper panel: scheme for repair of pNS plasmid in mammalian cells after transfection; lower panel: repair of linearized plasmid substrate pNS after replication stress. U2OS cells were treated with the indicated doses of HU and VE-821 for 8 h before transfection of pNS; repaired sequences were analyzed according to ([@B30]). (C) Upper panel: in vitro MMEJ repair activity assay scheme; lower panel: MMEJ repair activity of XRCC1-FLAG IP. U2OS cells were transfected with XRCC1-FLAG and immunoprecipitated after treatment. Immunoprecipitated XRCC1-FLAG complexes were incubated with linearized plasmid substrate and competent E. coli cells were transformed with repaired product according to ([@B30]). Total colony number reflects individual repair events. Western blot of IP bead eluate after incubation is shown.](zcaa013fig3){#F3} To further test whether replication stress and fork collapse enhance formation of active MMEJ complexes, we measured the ability of XRCC1 immunocomplexes to repair linearized pNS plasmid via MMEJ ([@B30]) (Figure [3C](#F3){ref-type="fig"}, upper panel). Pre-incubation of WT U2OS cells expressing FLAG-tagged XRCC1 with ATRi and HU significantly increased the repair of plasmid DNA by the XRCC1 immunocomplex (Figure [3C](#F3){ref-type="fig"}, lower panel). The sequences of the repaired plasmids confirmed that, as expected ([@B30]), all the repair events occurred by MMEJ rather than increased ligation activity ([Supplementary Figure S3](#sup1){ref-type="supplementary-material"}). BRCA2- and XRCC1-depleted cells are sensitive to replication stress and repair DSBs inefficiently {#SEC3-4} ------------------------------------------------------------------------------------------------- Given the synthetic lethal relationship between BRCA2 and XRCC1 ([@B17],[@B39]), we examined the impact of depleting either BRCA2 or BRCA1 on replication stress-induced MMEJ. While BRCA2 depletion led to a marked increase in chromosomal MMEJ events, BRCA1 depletion led to a decrease (Figure [4A](#F4){ref-type="fig"}). These results are consistent with published studies on the effect of BRCA1 and BRCA2 deficiency on MMEJ frequency ([@B10],[@B40]) although there is a conflicting report ([@B41]). To further test the relationship between XRCC1 and BRCA2, we utilized a set of isogenic U2OS cells that allow reversible knockdown of BRCA2 in the presence of doxycycline (Figure [4B](#F4){ref-type="fig"}). Depletion of XRCC1 in BRCA2-deficient cells significantly increased the cytotoxicity of the combination of ATRi and HU (Figure [4C](#F4){ref-type="fig"}). These cells both accumulated more DSBs (Figure [4D](#F4){ref-type="fig"}) and repaired DSBs less efficiently (Figure [4E](#F4){ref-type="fig"}) following HU treatment. We therefore investigated whether XRCC1 depletion affects the sensitivity of BRCA2-deficient cells to PARP inhibition using the MTT assay, a colorimetric method of assessing metabolic activity. As expected, XRCC1-deficient cells were moderately sensitive to rucaparib, and BRCA2-deficient cells were highly sensitive (Figure [4F](#F4){ref-type="fig"}). Surprisingly, co-depleted cells were less sensitive than BRCA2-depleted cells, raising the possibly that XRCC1 promotes formation of toxic, DNA-trapped PARP1. Collectively, these data indicate that XRCC1 plays a more prominent role in the repair of DSBs arising from replication stress in BRCA2-deficient cells. ![XRCC1 plays prominent roles in replication stress responses and DSBR in BRCA2-deficient cells. (A) Dependence of EJ2-U2OS MMEJ events on BRCA1 and BRCA2. Cells were treated with siRNA according to ([@B32]). Data are plotted relative to ctrl siRNA-treated samples and scaled according to their fold depletion as measured via western blot. (B) Western blot of inducible knockdown of BRCA2 in scr-U2OS (scr) and BRCA2-U2OS (BRCA2) cells. scr-U2OS and BRCA2-U2OS (also referred to as scr and BRCA2 or B2) cells can be induced with doxycycline to express scrambled shRNA and BRCA2 shRNA, respectively. (C) Clonogenic survival assay of scr-U2OS and BRCA2-U2OS cells treated with ctrl or XRCC1 siRNA and ATRi + HU. (D) Neutral comet assay in scr-U2OS and BRCA2-U2OS cells treated with ctrl and XRCC1 siRNA and/or HU. Cells were treated with 3 mM HU for 8 h where indicated. At least 50 cells were analyzed for each experiment. (E) Relative fraction of scr-U2OS and BRCA2-U2OS cells positive for the DSB marker γH2AX by immunofluorescence. XRCC1 or ctrl siRNA-treated cells were treated with 3 mM HU for 3 h and then allowed to recover for the indicated times. Cells were marked positive if they contained \>10 foci. The number of cells analyzed in each experiment ranged from 87 to 158. P-values were obtained from pairwise z-tests between control and each of the other conditions. \*P-value 0.034, P(α)~0.05~ 0.565; \\P-value 0.006, P(α)~0.05~ 0.793. (F) MTT assay of scr-U2OS and BRCA2-U2OS treated with either ctrl siRNA or XRCC1 siRNA and exposed to varying concentrations of the PARP inhibitor rucaparib for 120 h.](zcaa013fig4){#F4} BRCA2 suppresses MMEJ by preventing XRCC1 recruitment and repair complex formation {#SEC3-5} ---------------------------------------------------------------------------------- To better define the relationship between XRCC1 and BRCA2, we examined XRCC1 localization and complex formation in BRCA2-deficient cells. More XRCC1 foci formed in response to replication stress in BRCA2-deficient cells (Figure [5A](#F5){ref-type="fig"}), although the result was not statistically significant. This correlated with increased co-localization of XRCC1 with BrdU (Figure [5B](#F5){ref-type="fig"} and [E](#F5){ref-type="fig"}) and with γH2AX, PARP1, POLQ and MRE11 (Figure [5C](#F5){ref-type="fig"} and [F](#F5){ref-type="fig"}). To further test whether replication stress causes increased association of MMEJ factors in BRCA2-deficient cells, we analyzed co-immunoprecipitation of MRE11 and XRCC1 with FLAG-tagged POLQ expressed in BRCA2-proficient and BRCA2-deficient U2OS cell lines. In accord with the increased co-localization (Figure [5F](#F5){ref-type="fig"}), higher levels of XRCC1 (∼2.7-fold) and MRE11 (∼4-fold) were found in POLQ-FLAG IP from BRCA2-deficient cells, showing that BRCA2 suppressed the association of POLQ with both XRCC1 and MRE11 (Figure [5D](#F5){ref-type="fig"}, upper panel). Similarly, there was less POLQ-FLAG and MRE11 in endogenous XRCC1 immunoprecipitates from BRCA2-proficent cells compared with BRCA2-deficient cells (∼1.5-fold, POLQ-FLAG; ∼2-fold, MRE11) after replication stress (Figure [5D](#F5){ref-type="fig"}, lower panel). ![BRCA2 suppresses XRCC1--POLQ--MRE11 complex formation and MMEJ. (A) XRCC1 foci formation after treatment of scr-U2OS or BRCA2-U2OS cells with 3 mM HU and/or 3 μM ATRi for 8 h where indicated. Cells were marked positive if they contained \>10 foci. At least 50 cells were analyzed for each experiment. (B) Representative images of the PLA data related to (E). (C) Representative images of the PLA data related to (F). (D) Upper panel: western blot of POLQ-FLAG IP from scr-U2OS and BRCA2-U2OS cells with and without ATRi + HU treatment. Lower panel: western blot of endogenous XRCC1 IP from scr-U2OS and BRCA2-U2OS cells expressing POLQ-FLAG with and without ATRi + HU treatment. (E) XRCC1 localization to sites of replication stress in scr-U2OS or BRCA2-U2OS cells, as measured by BrdU-XRCC1 PLA. Asynchronous cells were pulsed with 10 μM BrdU for 15 min before treatment with 3 mM HU and/or 3 μM ATRi for the indicated times. At least 50 cells were analyzed for each experiment. (F) PLA between XRCC1 and the indicated factors in scr-U2OS and BRCA2-U2OS cells treated for 8 h with 3 mM HU and 3 μM ATRi. At least 50 cells were analyzed for each experiment. (G) MMEJ repair activity of XRCC1-FLAG IP. scr-U2OS and BRCA2-U2OS cells were transfected with XRCC1-FLAG and treated with 3 mM HU and 3 μM ATRi where indicated. Immunoprecipitated XRCC1-FLAG complexes were incubated with linearized plasmid substrate and competent bacterial cells were transformed with repaired product according to ([@B30]). Western blot of IP bead eluate after incubation is shown.](zcaa013fig5){#F5} To test whether replication stress causes increased activity of the MMEJ complex in BRCA2-deficient cells, we examined the effect of BRCA2 depletion on the MMEJ activity of XRCC1-FLAG immunocomplexes. The XRCC1-FLAG IP from BRCA2-depleted cells performed MMEJ at a significantly higher level than control cells (Figure [5G](#F5){ref-type="fig"}), whereas SSBR was carried out at a moderately higher level in BRCA2-depleted cells relative to control cells ([Supplementary Figure S3](#sup1){ref-type="supplementary-material"}). Together these data indicate that the presence of BRCA2 suppresses the localization of XRCC1 to sites of replication stress, formation of the XRCC1--MMEJ repair complex and the repair activity of this complex. BRCA2 deficiency changes the effect of XRCC1 depletion on replication fork dynamics and chromosome aberrations {#SEC3-6} -------------------------------------------------------------------------------------------------------------- Since XRCC1 is recruited to sites of replication stress, we considered the possibility that it has functions at the stalled replication fork in addition to MMEJ. Using DNA fiber analysis ([@B35]), we found that XRCC1 depletion did not affect DNA replication progression or fork protection in WT U2OS cells ([Supplementary Figure S4](#sup1){ref-type="supplementary-material"}), whereas it did significantly limit the extent of replication fork restart (Figure [6A](#F6){ref-type="fig"}), extending previous observations ([@B42]). ![XRCC1 depletion limits fork degradation, fork restart and chromosome aberration accumulation in BRCA2-deficient cells. (A) Effect of XRCC1 depletion on replication restart in U2OS cells. (B) Effect of XRCC1 depletion on replication restart in scr-U2OS and BRCA2-U2OS cells. (C) Effect of XRCC1 depletion and MRE11 inhibition (100 μM Mirin) on replication fork protection in scr-U2OS and BRCA2-U2OS cells. At least 200 individual fibers were analyzed for each experiment. (D) Quantification of chromosome aberrations from 20 separate metaphases in scr- and BRCA2-U2OS cells with or without XRCC1 siRNA treatment. (E) Representative metaphase chromosome spread of BRCA2-U2OS with HU treatment. (F) Western blot of BRCA2 and XRCC1 co-depletion.](zcaa013fig6){#F6} Based on the degradation of nascent DNA at stalled forks by nucleases, including MRE11, that leads to chromosomal aberrations and genomic instability in BRCA2-deficient cells ([@B21]), together with our data showing increased association of MRE11 and XRCC1 at replication forks in response to replication stress, we examined the effect of depleting XRCC1 in BRCA2-depleted and control U2OS cells (Figure [6F](#F6){ref-type="fig"}) on replication fork dynamics. Depletion of BRCA2 further inhibited replication restart caused by XRCC1 knockdown (Figure [6B](#F6){ref-type="fig"}), suggesting that BRCA2 and XRCC1 participate in distinct pathways that promote replication restart. Interestingly, XRCC1 depletion alleviated the degradation of stalled forks in BRCA2-deficient cells (Figure [6C](#F6){ref-type="fig"}) and the increase in chromosome aberrations induced by replication stress (Figure [6D](#F6){ref-type="fig"} and [E](#F6){ref-type="fig"}). XRCC1 knockdown in BRCA2-depleted cells has a similar effect to inhibiting MRE11 nuclease activity with mirin on fork degradation (Figure [6C](#F6){ref-type="fig"}), as well as a similar effect to mirin treatment in BRCA2-deficient cells on chromosome aberration number (Figure [6D](#F6){ref-type="fig"}) ([@B21]). Since MRE11 inhibition fully restores replication fork protection in BRCA2-deficient cells, MRE11 is likely the primary nuclease responsible for fork degradation in these cells. The observed effect of XRCC1 in fork protection, combined with observations indicating co-localization (Figure [5D](#F5){ref-type="fig"}) and complex formation (Figure [5F](#F5){ref-type="fig"}), leads us to reason that MRE11 and XRCC1 function in the same pathway responsible for resecting stalled forks in BRCA2-deficient cells. Collectively, our results reveal specific roles for XRCC1 in BRCA2-deficient cells that balance fork degradation and fork restart to promote chromosome aberrations and cell survival. XRCC1 and MRE11 deficiency are more strongly correlated with survival in BRCA-deficient breast cancers than BRCA-proficient breast cancers {#SEC3-7} ------------------------------------------------------------------------------------------------------------------------------------------ Based on our observation that XRCC1 and MRE11 are critical not only for DSBR by MMEJ but also for replication fork restart in BRCA2-deficient cancer cells, we analyzed XRCC1 and MRE11 expression in breast cancers compiled in The Cancer Genome Atlas (TCGA). We first partitioned all TCGA breast cancers into BRCA-proficient and BRCA-deficient groups using their signature 3 score (a pattern of single-base substitutions strongly associated with large insertions and deletions with overlapping microhomology at chromosome breakpoint junctions; signature 3+ cancers are strongly associated with homologous recombination deficiency) from COSMIC ([@B1]) (sig3+: sig3score \> 0; sig3−: sig3score = 0) (Figure [7A](#F7){ref-type="fig"}). Then, we partitioned breast cancers within these groups into high and low gene expression groups (top and bottom ∼20% of tumors) (Figure [7D](#F7){ref-type="fig"} and [G](#F7){ref-type="fig"}), and calculated the survival curves for those groups. We found that in all breast cancers (regardless of HR status), low XRCC1 expression is associated with poor survival (Figure [7B](#F7){ref-type="fig"}) (P-value = 0.031). This association is stronger in HR-proficient cells (Figure [7E](#F7){ref-type="fig"}) (P-value = 0.0008), and is absent in HR-deficient cells (Figure [7F](#F7){ref-type="fig"}) (P-value = 0.52), consistent with a distinct role of XRCC1 in the promotion of cancer fitness in HR-deficient breast cancers. Similarly, MRE11 expression is significantly associated with poor survival only in the HR-deficient subset of breast cancers (Figure [7C](#F7){ref-type="fig"}, [F](#F7){ref-type="fig"} and [I](#F7){ref-type="fig"}) (P-value = 0.016), consistent with promotion of cancer fitness by MRE11 specifically in this subset of breast cancers. In contrast, levels of the SSBR/BER polymerase POLβ and the NHEJ factor Ku70 (XRCC6) did not show association with survival in either BRCA-proficient or BRCA-deficient breast cancers ([Supplementary Figure S5](#sup1){ref-type="supplementary-material"}). These data collectively suggest that in HR-deficient breast cancers, high expression of XRCC1 and MRE11 leads to poor survival as a result of either increased genome instability or therapy resistance due to increased replication fork restart and DNA repair (Figure [8](#F8){ref-type="fig"}). ![XRCC1 and MRE11 gene expression is correlated with poor survival in HR-deficient breast cancer patients. (A) Distribution of signature 3 score across all breast cancers in TCGA. (B) Kaplan--Meier survival curve and hazard ratio for all TCGA breast cancer patients with high (top 20%) and low (bottom 20%) gene expression levels (panel D) of XRCC1. (C) Kaplan--Meier survival curve and hazard ratio for all TCGA breast cancer patients with high (top 20%) and low (bottom 20%) gene expression levels (panel G) of MRE11. Kaplan--Meier survival curves and hazard ratios for Sig3-negative (HR-positive, E) and Sig3-positive (HR-negative, F) breast cancer patients with high and low expression of XRCC1. Kaplan--Meier survival curves and hazard ratios for Sig3-negative (HR-positive, H) and Sig3-positive (HR-negative, I) breast cancer patients with high and low expression of MRE11.](zcaa013fig7){#F7} ![Graphical scheme summarizing results. In BRCA2-proficient cells, BRCA2 loads RAD51 to protect reversed replication forks, leading to replication restart by homologous recombination and genome stability. In BRCA2-deficient cells, MRE11 and XRCC1 degrade reversed forks, leading to replication restart using microhomology-based mechanisms, the accumulation of signature 3 mutations and poor patient survival.](zcaa013fig8){#F8} DISCUSSION {#SEC4} ========== XRCC1 has key scaffolding roles in coordinating repair of base damage and single-strand breaks by housekeeping DNA repair pathways ([@B31]). Also, it was proposed to have an analogous role in backup error-prone DSBR pathways, including MMEJ, that can become critical for the viability of HR-deficient tumor cells ([@B12]). Previously, we found that ionizing radiation exposure enhances formation of functional XRCC1-containing MMEJ complexes and increases MMEJ activity ([@B30]). Here, we support and extend those results to show that XRCC1-mediated MMEJ is also increased in response to replication stress and fork collapse. Furthermore, we find that XRCC1 and other MMEJ proteins are recruited to stalled and/or damaged replication forks. While this study is limited by the usage of only U2OS cells for isogenic BRCA2 depletion, we suggest that the reported results on replication, repair substrates and mutational signature are likely to be generalizable to other cell types in light of the fundamentally conserved roles of XRCC1 in DNA damage responses. Analyses of XRCC1-deficient rodent and human cells have revealed similarities but also some differences in the contribution of XRCC1 to the cellular response to DNA damage and replication stress, such as the modest sensitivity of XRCC1-depleted human U20S cells to HU compared with CHO EM9 cells ([@B43]). Additional studies in other cancer and non-malignant human cell lines will explore the role of XRCC1 in dealing with replicative stress, such as that induced by oncogenes, in the absence or presence of functional recombinational repair. A role of XRCC1 in PARP-dependent replication-associated DNA repair was identified previously ([@B44]), but was attributed to the canonical role of XRCC1 SSBR/BER without examination of possible roles of XRCC1 in replication and microhomology-mediated repair processes. We have extended that work here to demonstrate a clear role for XRCC1 in DNA replication processes beyond its SSBR/BER function. Recently, low levels of poly(ADP-ribose) (pAR) synthesis were detected at the replication fork during unperturbed DNA replication ([@B45]). As there are higher levels of pAR synthesis in DNA Ligase 1 (Lig1)-deficient cells but Lig1-null cells are viable because of the presence of the XRCC1 partner protein, Lig3, it is likely that Lig3 and XRCC1 participate in a PARP-dependent backup pathway that joins Okazaki fragments at the replication fork ([@B45]). Since depletion of XRCC1 had only a small effect on cellular sensitivity to replication stress and the level of replication stress-induced DSBs, it appears that, despite enhanced recruitment to replication forks and increased MMEJ activity, XRCC1 also has only a minor role at stalled or collapsed replication forks in HR-proficient cells. In contrast, in HR-deficient cells, it is evident that MMEJ plays a critical role in the mutagenic repair of DSBs in sister chromatids ([@B5],[@B11]). There is, however, emerging evidence that HR factors contribute to genome stability not only through error-free repair of DSBs but also by protecting stalled replication forks. In BRCA2-mutant cancer cells, stalled and collapsed forks, which are not substrates for the NHEJ machinery, are degraded by the MRE11 nuclease ([@B21]). Since this increased fork degradation was dependent upon XRCC1, we suggest that complex formation with XRCC1 at the replication fork is likely to be required for fork degradation. Given the increased association of XRCC1, MRE11 and other MMEJ factors in response to replicative stress, it appears likely that this complex carries out resection at stalled forks that is limited by the exposure and annealing of microhomologies. This type of resection is compatible with a RAD51-independent pathway of break-induced replication (BIR) termed microhomology-mediated BIR (MMBIR), which utilizes microhomology to facilitate multiple template switching events and is a significant source of rearrangements observed in human cancers and other genetic diseases ([@B46]). Notably, replication-associated seDSBs are repaired by MMBIR in HR-deficient cells that are unable to utilize RAD51-dependent, homology-dependent BIR ([@B49],[@B50]). While we are currently working to establish the cell cycle dependence of these processes, the clear effect on replication fork dynamics and its suppression by BRCA2 demonstrated here provide strong evidence that these repair processes are highly active in S/G2 phases of the cell cycle. As depletion of both BRCA2 and XRCC1 had an additive effect on replication restart, these proteins likely act in two distinct pathways for replication restart. Thus, it appears that BRCA2 may protect stalled forks from degradation and promotes replication restart, possibly via homology-dependent BIR ([@B21]). The increased formation of MMEJ complexes and the increased co-localization of MMEJ factors at stalled forks in HR-proficient tumors may reflect replication stress-induced signaling and/or recruitment to a subset of stalled forks that are not protected by BRCA2. Our assay results imply that XRCC1 promotes microhomology-mediated repair at stalled or collapsed forks. Moreover, the increased formation of MMEJ complexes at sites of stalled replication, their increased resection and annealing activities, plus promotion of fork degradation, restart and chromosome aberrations argue that the MMEJ repair complex acts at replication sites in HR-deficient cancer cells. We posit that this replication-associated activity may contribute to observed genome rearrangements in HR-deficient cancers and thus merits further investigation due to its implications for chemotherapeutic strategies targeting POLQ and PARP activities. Our collective results suggest that, in the absence of BRCA2 and homology-dependent pathways for fork restart, MMEJ directs fork degradation in order to expose microhomologies that mediate template switching and replication restart via a pathway involving XRCC1 as well as the MMEJ factors POLQ, MRE11 and PARP1 ([@B7],[@B16],[@B26]). While this pathway is mutagenic and presumably contributes to genomic instability by generating deletions with microhomologies at the breakpoints to enable replication restart at microhomologies, it would enable HR-deficient cancer cells to complete replication ([@B22],[@B51]). The enrichment of such breakpoints and the biological importance of MMEJ factors in HR-deficient cancers indicate a connection between replication restart, fork protection and MMBIR/MMEJ that warrants further exploration. For example, defining the mechanism of POLQ and PARP activity more precisely during fork degradation and replication restart will inform the applications of PARP and POLQ inhibitors to BRCA-deficient cancer therapy ([@B6]). Furthermore, specific inhibitors of the poly(ADP-ribose) glycohydrolase (PARG) cause replication fork stalling and cancer cell death ([@B52]). In concert with the results reported here, including the finding that the sensitivity of BRCA2-deficient cells to PARP inhibitors is partially dependent on XRCC1, it appears that the XRCC1--MRE11 axis connects PARP, PARG and POLQ at stalled forks as well as at DSBs. These findings are significant in the context of PARP inhibition in HR-deficient tumors, which extend beyond BRCA deficiency itself, particularly in platinum-sensitive ovarian cancer patients who demonstrate improved progression-free survival following maintenance therapy with rucaparib, olaparib, veriparib or niraparib ([@B53]). Since increased activity of this XRCC1--MRE11-dependent pathway of fork degradation and restart will lead to genome instability, our finding that expression levels of XRCC1 and MRE11 correlate with survival outcomes for patients with HR-deficient breast cancers, in contrast to NHEJ and SSBR/BER factors, supports our model in which XRCC1 and MRE11 have fundamentally different roles in HR-deficient and HR-proficient breast cancers. Notably, high expression of XRCC1 and MRE11 correlates with decreased overall survival (relative to HR-proficient survival curves) specifically in HR-deficient breast cancers. While there are multiple possible explanations for this observation, we suggest that, given the high burden of MH-containing genome rearrangements in HR-deficient cancers, the ability of HR-deficient cancers to complete replication via XRCC1 and MRE11 correlates with tumor fitness, and thus decreased overall survival. In sum, we herein have identified a novel role for an MMEJ-competent protein complex that is enhanced by replication stress at stalled replication forks in BRCA2-deficient cells. Notably, XRCC1 and MRE11 promote degradation of forks in BRCA2-deficient cells, presumably exposing microhomologies that are utilized to prime replication restart. Thus, in HR-deficient tumors, resection and annealing mediated by the MMEJ complex may be important for replication fork restart and cell survival. Overall, the identification here of an XRCC1--MRE11 axis connecting PARP, PARG and POLQ at stalled forks as well as at DSBs appears relevant to therapeutic strategies targeting POLQ and PARP activities. Supplementary Material ====================== ###### Click here for additional data file. We thank Katharina Schlacher for discussions. This research used computing resources at the Texas Advanced Computing Center, which is supported by National Science Foundation. SUPPLEMENTARY DATA {#SEC5} ================== [Supplementary Data](https://academic.oup.com/narcancer/article-lookup/doi/10.1093/narcan/zcaa013#supplementary-data) are available at NAR Cancer Online. FUNDING {#SEC6} ======= National Institutes of Health \[CA158910 and GM105090 to S.M., ES12512 to A.E.T., NS088645 to M.L.H., CA092548 to J.A.T., A.E.T. and S.M., CA220430 to J.A.T.\]; National Space Biomedical Research Institute \[NCC 9-58 to B.J.E.\]; Cancer Prevention and Research Institute of Texas (CPRIT) \[RP180813 and RP130397 to J.A.T.\]; Robert A. Welch Chemistry Chair \[to J.A.T.\]; National Science Foundation \[ACI-1134872\]. Conflict of interest statement. None declared.	{ "pile_set_name": "PubMed Central" }
There has been little progress in understanding pancreatic ASC since these aggressive tumors were first described over a century ago^[@R1],[@R2]^. One problem is that no mutations unique to this class of pancreatic tumors have been identified. Thus, while oncogenes and tumor suppressor genes have been shown to be altered in ASC---including K-RAS and p53---these same genes are also abnormal in other forms of pancreatic cancer^[@R3]--[@R5]^. In this communication, we report that ASC pancreatic tumors have somatic mutations in Up-frameshift 1 (UPF1), which encodes an RNA helicase essential for a highly conserved RNA degradation pathway called nonsense-mediated RNA decay (NMD)^[@R6]^. We were initially led to investigate the possibility that NMD had a role in ASC because reverse transcription-polymerase chain reaction (RT-PCR) analysis showed that an ASC tumor (TU) with wild-type p53 alleles expressed an abnormally large band that was not detectable in the adjacent normal pancreas (NP) ([Supplementary Fig. 1a](#SD1){ref-type="supplementary-material"}). Sequence analysis of this large band demonstrated that it was generated by alternatively splicing using a non-canonical splice donor in intron 6 and a non-canonical splice acceptor in intron 10, which both share 14 nt of sequence identity with the splice donor ([Supplementary Fig. 1b](#SD1){ref-type="supplementary-material"}). Quantitative PCR (qPCR) analysis demonstrated that the alternatively spliced p53 mRNA was only detectably expressed in tumor tissue, not adjacent normal tissue, while the normally spliced p53 mRNA was expressed at similar levels in the tumor and adjacent normal tissue ([Supplementary Fig. 1c](#SD1){ref-type="supplementary-material"}). The alternative splicing event creating this p53 isoform generates an in-frame premature termination codon (PTC) in intron 6 ([Supplementary Fig. 1b](#SD1){ref-type="supplementary-material"}). Because NMD degrades mRNAs with PTCs^[@R6],[@R7]^, the selective accumulation of this "alt-PTC-IVS6-p53" mRNA in the tumor tissue but not the adjacent normal tissue, raised the possibility that the tumor tissue had deficient NMD. To test this, we screened ASC tumor samples for mutations in core NMD genes. We found genomic mutations in the UPF1 gene in ASC tumors from 18 of 23 patients (see TU in [Fig. 1a](#F1){ref-type="fig"}), whereas the 3 other NMD genes we tested---UPF2, UPF3A, and UPF3B---did not have detectable mutations ([Supplementary Table 1](#SD1){ref-type="supplementary-material"}). The UPF1 mutations were somatic in origin, since they were not present in matched normal pancreatic tissues from these 18 patients (see NP in [Fig. 1a](#F1){ref-type="fig"}). UPF1 mutations were also not present in 50 non-ASC pancreatic and SCC lung tumors that we tested ([Supplementary Table 2](#SD1){ref-type="supplementary-material"}). Together, these results suggest that UPF1 mutations are a unique signature of ASC pancreatic tumors. The point mutations in the ASC tumors clustered in two regions of the UPF1 gene ([Fig. 1b](#F1){ref-type="fig"}). Surprisingly, these mutations were nearly equally distributed in the exons and introns in these two regions. This raised the possibility that they alter UPF1 splicing by disrupting intronic splicing enhancers (ISEs) and exonic splicing enhancers (ESEs), which are essential for the inclusion of a subset of exons during splicing^[@R8],[@R9]^. In support of this, a subset of the mutations in the ASC tumors disrupted predicted ESEs/ISEs ([Fig. 1b](#F1){ref-type="fig"} and [Supplementary Fig. 2](#SD1){ref-type="supplementary-material"}). To directly test whether the ASC mutations disrupt UPF1 splicing, we generated UPF1 mini-gene constructs corresponding to the two regions mutated in ASC tumors ([Fig. 2a,b](#F2){ref-type="fig"}). We then introduced, by site-specific mutagenesis, the mutations found in 15 of the patients. Transfection analysis showed that the wild-type versions of the two mini-gene constructs expressed normally spliced mRNA that included all exons, as determined by direct sequencing of the bands generated by RT-PCR ([Fig. 2c,d](#F2){ref-type="fig"}). No other bands were detected, indicating that most of the mature mRNA generated by these mini-gene constructs was normally spliced. In contrast, all 15 mutant mini-genes expressed alternatively spliced transcripts ([Fig. 2,d](#F2){ref-type="fig"}). While none of the patient's mutations completely eliminated normal splicing, the ratio of alternatively spliced-to-normally spliced mRNA was 10 to 1 or higher in many cases. As confirmation, analysis of endogenous UPF1 splicing in 2 of 2 frozen tumor samples revealed an alternatively spliced UPF1 mRNA that was not present in the adjacent normal tissue ([Fig. 2e](#F2){ref-type="fig"}). We conclude that ASC-specific mutations in the UPF1 gene trigger alternative splicing of UPF1 pre-mRNA. The UPF1 alternative splicing events caused by mutations in the exon 10/11 region eliminate a portion of UPF1's RNA helicase domain essential for UPF1 function^[@R6]^, while mutations in the exon 21/23 region truncate the carboxy-terminal region of UPF1, which contains \[S/T\]Q motifs phosphorylated by SMG1, a serine/threonine kinase required for NMD^[@R10]^ ([Figs. 1b](#F1){ref-type="fig"} and [2c](#F2){ref-type="fig"}). Given that essential UPF1 domains are lost, these mutant forms of UPF1 are likely to have lost their function or have dominant-negative activity. In the case of the latter, inhibited NMD would be predicted to occur when only one UPF1 allele is mutated. In support of this possibility, mutations in various regions of UPF1 have previously been shown to confer dominant-negative activity^[@R11]--[@R13]^. With regard to the former, we found that at least one ASC tumor (from patient 22) had mutations in both UPF1 alleles. Several other ASC tumors have multiple UPF1 mutations ([Supplementary Table 1](#SD1){ref-type="supplementary-material"}), consistent with some of them also having mutations in both UPF1 alleles. In further support of this notion, we observed little or no UPF1 expression in many ASC tumors compared to adjacent normal tissue, as demonstrated by immunohistochemical (IHC) analysis ([Fig. 2f](#F2){ref-type="fig"} and [Supplementary Table 1](#SD1){ref-type="supplementary-material"}). As direct evidence that NMD is inhibited in ASC tumors, we found that the NMD substrates, ATF3 and MAP3K14 mRNA^[@R14]^, were dramatically upregulated in tumor tissue relative to adjacent normal tissue ([Supplementary Fig. 3](#SD1){ref-type="supplementary-material"}). Since perturbation of NMD is known to dysregulate \~3 to 10% of mRNAs in a variety of cell lines and organisms^[@R6],[@R14],[@R15]^, it is likely this will also be the case of ASC tumors, which could shift the balance towards a more malignant phenotype^[@R16],[@R17]^. One candidate NMD substrate that could serve in this capacity is the alternatively spliced p53 transcript---alt-PTC-IVS6-p53---that we identified above ([Supplementary Fig. 1](#SD1){ref-type="supplementary-material"}), as we found it encoded a protein with dominant-negative activity ([Supplementary Fig. 4](#SD1){ref-type="supplementary-material"}). The discovery of mutations in the UPF1 gene in ASC tumors represents the first known example of genetic alterations in a NMD gene in human tumors. To our knowledge, UPF1 is also the first gene known to be selectively mutated in pancreatic ASC tumors. Other genes mutated in these tumors, including p53 and K-RAS (see [Supplementary Fig. 5](#SD1){ref-type="supplementary-material"} and [Supplementary Table 1](#SD1){ref-type="supplementary-material"}) are also mutated in other pancreatic tumor types^[@R4],[@R5]^. We note that it is possible that UPF1 mutations in ASC tumors have broad disruptive effects that that extend beyond NMD, including Staufen-mediated RNA decay^[@R18]^. Our finding that UPF1 is selectively mutated in pancreatic ASC suggests that this feature will be useful for diagnosis of this type of tumor and also raises the possibility that ASC patients can benefit from therapy designed to target NMD substrates^[@R19],[@R20]^. METHODS {#S1} ======= Methods are available in the [Supplementary Information](#SD1){ref-type="supplementary-material"} section. Supplementary Material {#S2} ====================== This work was supported by The National Key Basic Research Program of China grants 2013CB967500 and 2011DFB30010 (to GX), 2011CB965102 (to LL), the National Natural Science Foundation of China grant 81071740 (to YJ), and NIH grant GM58595 (to MFW). AUTHOR CONTRIBUTIONS C.L., R.K., Y.Z., F.S., C.W., M.S., Y.W., G.X., L.L., J.Z., and Y.M. performed the experiments. Y.Z., Y.J., G.L., W.F., M.Z., M.A.V. and M.J. provided clinical samples and expertise on pancreatic cancer. R.K. and C.L. prepared the figures and tables and wrote the manuscript. Y.L. and M.F.W. equally contributed to this study by designing and supervised all experiments and assisting in writing the manuscript. [SUPPLEMENTARY INFORMATION](#SD1){ref-type="supplementary-material"} Includes [Supplementary Figures 1 to 5](#SD1){ref-type="supplementary-material"}, [Supplementary Tables 1 to 7](#SD1){ref-type="supplementary-material"}, Methods, and [Supplementary References](#SD1){ref-type="supplementary-material"}. COMPETING FINANACIAL INTERESTS The authors declare no competing financial interests. ![Somatic UPF1 mutations in ASC tumors. (a) Chromatograms of UPF1 DNA sequences from tumor (TU) and normal adjacent pancreatic tissue (NP). A total of 36 single-base substitutions in genomic UPF1 DNA were found in the tumors; each patient had between one to five point mutations. All NP samples had wild-type UPF1 sequences, indicating mutations were somatic in origin. (b) Top: schematic of UPF1 protein domains. Bottom: the type and location of mutations found in the two regions of UPF1 that harbored mutations. ESEs and ISEs are exonic and intronic splicing enhancers, respectively (sequences are provided on [Supplementary Figure 2](#SD1){ref-type="supplementary-material"}).](nihms580422f1){#F1} ![ASC-specific UPF1 mutations trigger alternative UPF1 RNA splicing. (a) The indicated region of human UPF1 (nt 22,779-23,570, RefSeq accession number NC_000019.9) was cloned into the NdeI site of the pTBNde mini-gene construct. All mutations were generated by site-directed mutagenesis to match those in the indicated patient's tumors. (b) The indicated region of human UPF1 (nt 33,138-34,490) was cloned and mutated as described in panel a. (c) RT-PCR analysis of HEK293 cells transfected with the constructs shown in panel a (primer locations are indicated by the arrows). Direct sequencing of the large (792 nt) and small (239 nt) bands indicated that they correspond to normally spliced and exon-skipped transcripts, respectively. The numbers below the gel are the average values from five independent transfections. (d) RT-PCR analysis performed as in panel c. Direct sequencing of the bands in the gel indicated they were derived from mRNA spliced in the manner shown in the schematic. The bands corresponding to normally spliced, alt 1, and alt 2 mRNA had lengths of 742, 1117, and 619 nt, respectively. The numbers below the gel are the average values from five independent transfections. (e) RT-PCR analysis of normal pancreas (NP) and ASC frozen samples (TU) from patients 15 and 16. RT-PCR sequencing results are indicated as schematics next to the gels. (f) Immunohistochemical analysis of UPF1 staining in ASC tumor samples from patient 15. Tumor (T) and normal tissue (NT) are indicated. Bottom panel shows H & E staining, with arrows pointing to the adenocarinoma component in this ASC tumor. The scale bar represents 200 micrometers on each picture.](nihms580422f2){#F2} [^1]: These authors contributed equally to this work.	{ "pile_set_name": "PubMed Central" }
![](psycholclin69649-0001){#sp1 .97} ![](psycholclin69649-0002){#sp2 .98} ![](psycholclin69649-0003){#sp3 .99} ![](psycholclin69649-0004){#sp4 .100} ![](psycholclin69649-0005){#sp5 .101} ![](psycholclin69649-0006){#sp6 .102} ![](psycholclin69649-0007){#sp7 .103}	{ "pile_set_name": "PubMed Central" }
The importance of fruit and vegetable consumption to human health has been well established. Fruit and vegetable intake is associated with reductions in chronic diseases including heart disease, stroke and some cancers^(^ [@ref1] ^,^ [@ref2] ^)^. However, in the USA, less than 10 % of the population consumes adequate fruits and vegetables consistent with dietary recommendations^(^ [@ref3] ^,^ [@ref4] ^)^. Socio-economic disparities in fruit and vegetable consumption, while not new, have increased; for example, a 2014 study found the gap in Alternate Healthy Eating Index (AHEI) scores between adults of low and high socio-economic status had grown from 3·9 points in 1999--2000 to 7·8 points in 2009--2010 (maximum AHEI score=110; mean 2009--2010 adult score=46·8)^(^ [@ref5] ^)^. A number of possible mechanisms have been suggested as contributing factors to these disparities. One area that has received widespread attention in recent decades is the association of access to fresh fruits and vegetables in the neighbourhood food retail environment with health behaviours and outcomes^(^ [@ref6] ^--^ [@ref8] ^)^. Studies have found that children living near food stores with lower prices of fruits and vegetables have lower increases in BMI^(^ [@ref9] ^)^, and that quality, selection and convenience in the food retail environment in low-income neighbourhoods are predictors of fruit and vegetable consumption^(^ [@ref10] ^)^. Further, studies of low-income consumers suggest that both food quality and price are key factors in shopping decisions^(^ [@ref11] ^,^ [@ref12] ^)^. While many studies have examined whether residents of low-income communities have poorer access to healthy foods, fewer have examined the quality and/or price of healthy fruits and vegetables in these neighbourhoods. Most recent studies have relied upon relatively small samples of stores and the findings have been somewhat inconsistent. A 2012 systematic review of the evidence reported that six of ten studies conducted in the USA found less availability of healthy foods in low-income or high-minority communities; only two studies reported lower-quality food items in low-income neighbourhoods^(^ [@ref13] ^)^. Food price results in that systematic review also were inconsistent: six studies found that healthy foods were cheaper at larger compared with smaller grocery stores; three found healthy foods to be more expensive at convenience stores compared with grocery stores and supermarkets; and two studies found higher prices of healthy foods in low-income neighbourhoods across store types, while two others found no significant price differences when comparing supermarkets across low- and high-income communities^(^ [@ref14] ^)^. One California study conducted over a decade ago evaluated the cost of a fruit and vegetable market basket in twenty-five supermarkets and found that prices were lower in low-income neighbourhoods than in middle- and high-income neighbourhoods^(^ [@ref15] ^)^. While the latter study included bulk, independent and chain supermarkets, the assessment of prices by neighbourhood was not adjusted for store type. Currently, a good deal of community intervention work focuses on increasing access to fruits and vegetables in smaller neighbourhood stores^(^ [@ref16] ^,^ [@ref17] ^)^; however, few small food store interventions address price^(^ [@ref18] ^,^ [@ref19] ^)^. Of particular interest are opportunities for the Supplemental Nutrition Assistance Program (SNAP) and the Supplemental Nutrition Program for Women, Infants, and Children (WIC) to influence the food retail environments of income-constrained shoppers. Studies have shown that many small and mid-sized food stores participating in SNAP carry a limited selection of healthy foods^(^ [@ref20] ^)^. One recent study in New Jersey found that participation in WIC, but not SNAP, was associated with better availability of healthy foods in corner stores^(^ [@ref21] ^)^. Studies describing differences in food availability in WIC- and/or SNAP-accepting stores are relatively limited. In response to concerns about disparities in fruit and vegetable access and consumption, the Supplemental Nutrition Program Education (SNAP-Ed) efforts of the California Department of Public Health prioritized improving fruit and vegetable access in the food retail setting. In establishing this intervention focus area, the California Department of Public Health developed and reliability tested a community-driven neighbourhood food environmental assessment protocol called Communities of Excellence in Nutrition, Physical Activity, and Obesity Prevention (CX^3^) to help local health departments assess low-income neighbourhood food environments and identify retailers with whom to work to improve healthy food access^(^ [@ref22] ^)^. The present paper describes the large sample of grocery stores, small markets and convenience stores in low-income neighbourhoods that participated in CX^3^ baseline assessments from 2011 to 2015. The purpose of the current cross-sectional study was to determine the extent to which fruits and vegetables are available in stores in low-income neighbourhoods across the state and to assess the degree to which availability, quality and price vary by store type and federal food programme participation. Recognizing the price sensitivity of income-constrained consumers, we also examined the store characteristics most associated with lower produce prices. Further, we sought to assess whether produce items in our sampled stores were priced higher or lower than average chain supermarket store prices in the same counties during the same month. Methods {#sec1} ======= Sample {#sec1-1} ------ The data were collected as part of a retail food intervention project of the California Department of Public Health, Nutrition Education and Obesity Prevention Branch with SNAP-Ed funding from the US Department of Agriculture. The project aimed to prepare local community leaders to identify issues in the food retail environment that they could work to improve and focused on partnership development and stakeholder engagement^(^ [@ref22] ^)^. Local health department leaders selected low-income neighbourhoods for inclusion in the data collection efforts if the neighbourhoods were being considered for potential interventions. The local community was responsible for defining the neighbourhood boundaries based upon local conditions and residents' perceptions. Eligible stores were required to be located in a SNAP-Ed eligible census tract (defined as a tract in which at least 50 % of the population is at or below 185 % of the federal poverty level^(^ [@ref23] ^)^). In some cases, the identified neighbourhoods included a limited number of census tracts in a large city, while in other cases they included an entire town. Because census tracts and neighbourhood boundaries did not always align, some neighbourhoods selected by local leaders included stores in eligible and ineligible census tracts. Stores in ineligible census tracts were excluded. Local leaders were instructed to collect data in all eligible food retail stores in which the primary product sold was food. Liquor stores, bakeries, cafés, restaurants and other similar establishments were excluded. Data collection {#sec1-2} --------------- Data collection occurred on a rolling basis over four years, until all health departments in California participating in the Nutrition Education and Obesity Prevention Branch SNAP-Ed programme had an opportunity to participate. The CX^3^ Food Availability and Marketing Survey was used to assess community food store environments. In-person trainings were offered annually and included store visits in which data collectors had an opportunity to complete the survey form and compare their responses against those of the trainer. In some cases, local staff who attended the State training trained additional data collectors, including community partners, community residents and students. Prior to completing the on-site assessments, data collectors were trained to gain permission from the store manager. Store assessments were completed by single or pairs of data collectors and took approximately 30 min per store to complete. With a few exceptions, stores in a single neighbourhood were surveyed within the same week. Data were double entered by professional data-entry staff. Store assessment measures {#sec1-3} ------------------------- ### Store characteristics {#sec1-3-1} Local leaders obtained initial descriptions of store characteristics from Geographic Information Systems store lists using annually updated data from state databases as well as data purchased from Dunn & Bradstreet. All store information was verified during on-site visits. Store type was identified as supermarket chain stores, large grocery stores (≥4 cash registers but not large chains), small markets (\<4 registers but selling a range of foods, including fresh meat), convenience stores (selling food items and snacks but not a complete range of foods and no fresh meat; may sell gas) and other (included WIC-only stores, dollar stores and drug stores/pharmacies); this store type assessment was verified during on-site visits. Store participation in the US Department of Agriculture's WIC and/or SNAP programme(s) was obtained from the Geographic Information Systems and verified during on-site visits or by telephone as needed. ### Fruit and vegetable availability and quality {#sec1-3-2} The survey instrument included an item that assessed whether produce was sold or not. In stores where produce was available, data collectors recorded the variety of fresh fruits available (none; 1--3 types; 4--6 types; ≥7 types) and repeated the assessment for fresh vegetables. Similarly, the quality of fruits was assessed on a scale of 1--4 (1=most of the options are brown, bruised, overripe and/or wilted; 4=all or most of the options appear fresh, without soft spots, excellent colour) and was repeated for fresh vegetables. These were coded as binary variables, with 'high-quality' fruits or vegetables indicating the top two categories. ### Cost of fruits and vegetables {#sec1-3-3} A single price for a small selection of fruits and vegetables commonly found in the USA (apples, bananas, oranges, carrots, tomatoes, broccoli and cabbage) was recorded. The recorded prices were the cheapest price per pound (0·454 kg) for each type of fruit or vegetable; thus, if a store carried many varieties of apples, the price recorded was the least expensive apple in the store. When prices per pound were not available, the price per package or bunch was recorded along with the weight of the bunch. In cases where only price per piece was available, that was recorded, and a US Department of Agriculture database of standard weights was used to convert to price per pound. Mean prices were not reported for broccoli and cabbage in convenience stores, as fewer than thirty stores in this category had recorded prices for these items. To assess the store's produce price against a price point benchmarked to geography and season, retail scanner data from chain supermarkets in the county were purchased from FreshLook Marketing Group. FreshLook data provided actual retail prices of perishable items sold at twenty-seven supermarket chains (including Safeway, Target, Nob Hill Foods, Food4Less, Raley's, Vons, Walmart and others) in the sampled counties and were obtained to match the data collection month in each county. When same-month observations were not available, average in-county lowest prices across all months were used as the comparison group. Analyses {#sec1-4} -------- The 1474 stores identified as large groceries (includes supermarket chain stores and large grocery stores), small markets and convenience stores were included in the analyses. The fifty-three (4 %) excluded stores (including e.g. dollar stores, WIC-only stores and butcher's shops) did not fit any of the defined store types. Series of paired tests were performed between each of the store types to determine whether key variables differed by type. Three tests were performed for each variable, comparing grocery stores with small markets, grocery stores with convenience stores, and small markets with convenience stores. Comparisons of binary store characteristics between store types were done using paired two-sample tests of equality of proportions and continuous measures using ANOVA to estimate the studentized sample range differences. For all analyses, tests were deemed significant if they rejected the null hypothesis of coming from the same population, after controlling for test-specific family-wide error rate, with a P value of 0·05. To calculate the reference average lowest price for produce items, lowest prices were drawn from the stores in the purchased data set matched to the month and county in which sampled store observations were completed. Relative price difference was calculated as the difference between the lowest sampled store price and the average in-county, in-month chain supermarket lowest price, expressed as a percentage of the average county chain supermarket lowest price. The average relative price difference across produce types was taken as the average of the relative prices of any of the five most common produce items present in the store (apples, bananas, oranges, carrots and tomatoes). We fit ordinary least-squares regression models, performing the regression of this average relative price v. a vector of store characteristics based upon variables in the data set that were selected a priori. These characteristics included: whether the store participated in the SNAP or WIC programme, whether the store was located within half a mile (0·8 km) of a school, whether the fruits and vegetables were rated high quality, and whether the store carried four or more types of fruits or four or more types of vegetables. A fixed effect was included to control for the county in which the stores were located. Regressions were run using the 'lfe' package and all calculations were performed using R version 3.2.2. Results {#sec2} ======= The final sample included 1474 stores in 470 unique census tracts within 225 low-income neighbourhoods in cities and towns across forty-four California counties ([Fig. 1](#fig1){ref-type="fig"}). All large grocery stores sold fresh produce and nearly all of those had a wide variety of high-quality offerings ([Table 1](#tab1){ref-type="table"}). Most (80 %) small markets also sold fresh fruits and vegetables, but fewer had a wide variety and high quality compared with grocery stores. Fewer than half (41 %) of convenience stores sold produce; of those that did, only a small minority had a selection of at least four or more fruits (16 %) or vegetables (12 %), and few stores' selections were rated high quality (25 % for fruits, 13 % for vegetables). Nearly all large grocery stores (95 %) and most small markets (79 %) and convenience stores (67 %) accepted SNAP benefits. Most large grocery stores (83 %) accepted WIC, but few small markets (34 %) or convenience stores (11 %) did. We observed significant differences between each of the store types for all variables examined, except the proximity to a school.Fig. 1Locations in which stores included in the CX^3^ (Communities of Excellence in Nutrition, Physical Activity, and Obesity Prevention) sample are situated; cross-sectional statewide survey in low-income neighborhoods in California, 2011--2015 Table 1Number and proportion of stores in the sample with selected characteristics; cross-sectional statewide survey in low-income neighbourhoods in California, 2011--2015Large groceries (n 231)Small markets (n 621)Convenience stores (n 622)n%n%n%Significance[\](#tab1fn1){ref-type="fn"}Sells produce2301004978025741A, B, C≥4 Fruits227983415510016A, B, C≥4 Vegetables22999391637312A, B, CHigh-quality fruits223973605815725A, B, CHigh-quality vegetables22698368598414A, B, CAccepts SNAP219954917941767A, B, CAccepts WIC19183211346911A, B, CHalf a mile (0·8 km) from a school161704857847777A[^1][^2] WIC and/or SNAP participation was associated with better quality and variety of produce ([Table 2](#tab2){ref-type="table"}). Small markets and convenience stores participating in both WIC and SNAP were significantly more likely to sell any fresh produce and to have a wider variety and higher-quality produce items than stores not participating in either programme. SNAP-only small markets had significantly better availability, variety and quality of fresh produce than stores not participating in SNAP or WIC, and SNAP-only convenience stores were significantly more likely to sell fresh produce (43 % v.* 29 %), carry four or more fruits (15 % v. 8 %) and carry fruits rated high quality (28 % v. 16 %) than convenience stores not participating in SNAP or WIC. Similar analyses could not be completed for large grocery stores, since nearly all of them participated in SNAP and WIC.Table 2Proportion of small markets and convenience stores in the sample meeting selected availability and quality criteria, as well as relative price differences[\](#tab2fn1){ref-type="fn"}, by federal food programme participation status; cross-sectional statewide survey in low-income neighbourhoods in California, 2011--2015Small marketsConvenience storesNeither WIC or SNAP (n* 108)SNAP only (n 302)Both (n 211)Neither WIC or SNAP (n 196)SNAP only (n 357)Both (n 69)%%%Significance[†](#tab2fn2){ref-type="fn"}%%%Significance[†](#tab2fn2){ref-type="fn"}Sells fresh produce687697A, B, C294371A, B, C≥4 Fruits687697A, B, C81546A, B, C≥4 Vegetables445682A, B, C5848A, B, CHigh-quality fruits465075A, B, C162838A, B, CHigh-quality vegetables435475A, B, C81332A, B, CRelative price differencesApples715548122105111Bananas43443313212762B, COranges18292610812768CCarrots39394711010897Tomatoes−5141A, C22430Broccoli087--------Cabbage−69−8--------Five-item average40393312110674B, C[^3][^4][^5] In the sampled stores, the average recorded lowest prices for fruits and vegetables were lowest in large grocery stores for four of the items recorded (apples, bananas, carrots, broccoli) and lowest in small markets for three (oranges, tomatoes, cabbage; [Table 3](#tab3){ref-type="table"}). Convenience store prices were higher than those in both large grocery stores and small markets for apples, oranges, bananas and carrots, but only higher than those in small markets for tomatoes (too few stores sold broccoli and cabbage to allow for meaningful inclusion).Table 3Average lowest price per pound (0·454 kg) and relative price difference[\](#tab3fn1){ref-type="fn"} of seven produce items in the sample by store type; cross-sectional statewide survey in low-income neighbourhoods in California, 2011--2015Large groceriesSmall marketsConvenience storesAverage lowest price (\$US)Relative price differenceAverage lowest price (\$US)Relative price differenceAverage lowest price (\$US)Relative price differenceRelative price testItemnMeanRange[sd]{.smallcaps}(%)nMeanRange[sd]{.smallcaps}(%)nMeanRange[sd]{.smallcaps}(%)significance [†](#tab3fn2){ref-type="fn"}Apples2181·170·66--2·990·39443121·180·63--2·590·37531301·580·69--2·800·50110B, CBananas2180·700·49--1·490·16153360·850·49--2·190·31391501·300·49--2·190·46115A, B, COranges2000·950·48--1·490·40352490·880·47--2·200·4127881·360·50--2·200·59107B, CCarrots1790·840·50--2·140·33332420·910·50--2·200·3743241·220·50--2·000·58102B, CTomatoes1971·210·59--2·660·45133111·120·59--2·790·416571·210·69--2·190·3823CBroccoli1991·300·68--3·190·4911421·330·68--3·180·566----------Cabbage1720·740·38--1·590·2702610·660·34--1·500·26−1----------[^6][^7] To compare prices across store type we used 'relative price difference', the difference between the average lowest prices in our sample and the county average lowest prices in chain supermarkets, expressed as a percentage of the average county lowest price in chain supermarkets. Overall, all store types in our sample had higher lowest prices for the observed produce items than the average lowest prices available in the reference group of chain supermarkets ([Table 3](#tab3){ref-type="table"}). The relative price difference for the produce items studied ranged from 0 % to 44 % higher for large groceries, from −1 % to 53 % higher for small markets and from 23 % to 115 % higher for convenience stores. Mean relative convenience store produce prices were substantially higher than the mean relative prices in large groceries and small markets. While the relative price differences for large grocery stores in our sample were smaller than the other store types, even the large grocery stores in our sample on average sold produce at higher prices than the reference county average in chain supermarkets (27 % higher relative price difference). A few stores of each type (large groceries, small markets, convenience stores) had average lowest prices for produce that were lower than the same-month county average (see stores to the left of the diagonal line in [Fig. 2](#fig2){ref-type="fig"}) but far more stores had higher average lowest prices.Fig. 2(colour online) Average lowest price of seven common fruits and vegetables in the sample stores, by store type (![](S1368980018000058_inline1.jpg), supermarkets/large groceries; ![](S1368980018000058_inline2.jpg), small markets; ![](S1368980018000058_inline3.jpg), convenience stores), compared with average lowest price of the items in chain supermarket stores in the same county and the same month; cross-sectional statewide survey in low-income neighborhoods in California, 2011--2015. Diagonal line indicates where points would lie if average price in sampled stores and average price in same-county, same-month grocery stores were equal. Few significant relationships were observed between store produce prices and SNAP or WIC participation. In small markets, the only significant relationship observed was that relative tomato prices were significantly higher in SNAP-only small markets than in both non-participating and WIC+SNAP stores (14 % of county average v. −5 % and 1 %, respectively). In convenience stores, the combined average relative price of the five most commonly measured produce items was significantly lower in WIC+SNAP convenience stores compared with both other groups (74 % higher v.* 106 % in SNAP-only stores and 121 % in stores not participating in SNAP or WIC). Additionally, in convenience stores, relative banana prices were significantly lower in WIC+SNAP stores than in other stores (62 % higher v. 127 % in SNAP-only stores and 132 % in stores not participating in SNAP or WIC); and oranges were significantly cheaper in WIC+SNAP stores than in in SNAP-only stores (68 % higher v. 127 % higher). Regression models estimated that across all store types, the most significant predictor of price was whether a store carried four or more fruits or vegetables or not ([Table 4](#tab4){ref-type="table"}). In the combined model including all store types, carrying more than four types of fruits and vegetables were both significantly associated with lower prices, a 17 % relative reduction (95 % CI −27, −7 %) for fruit and an identical 17 % relative reduction (95 % CI −29, −6 %) for vegetables. In small markets carrying four or more types of fruit, relative prices were 25 % cheaper compared with stores carrying fewer than four types of fruit (95 % CI −37, −14%). This same relationship was observed in convenience stores, where prices were 11 % cheaper in stores carrying four or more types of fruit, although the relationship was not statistically significant (95 % CI −43, 20 %). Average prices were 1·7 % cheaper in small markets (95 % CI −16, 13 %) and 38 % cheaper in convenience stores carrying more than four types of vegetables (95 % CI −69, −7 %).Table 4Parameter estimates and 95 % CI from regression analyses predicting the average relative price of five produce items (apple, bananas, oranges, carrots and tomatoes) as a function of store characteristics among three groups of sampled stores (all stores, small markets and convenience stores); cross-sectional statewide survey in low-income neighbourhoods in California, 2011--2015All storesSmall marketsConvenience storesβ95 % CIβ95 % CIβ95 % CIAccepts SNAP−3·2−12, 5·5−1·8−11·5, 7·94·8−33·3, 42·9Accepts WIC−0·3−7·0, 6·42·1−6·0, 10·220·1−7·7, 47·9Half a mile (0·8 km) from a school−8·5−16·1, −0·9−10·8−20·7, −0·8−6·0−43·1, 31·1High-quality fruits2·5−7·7, 12·7−3·7−14·9, 7·44·7−39·9, 49·3High-quality vegetables−0·7−11·2, 9·80·7−11·1, 12·4−1·2−42·1, 39·7≥4 Fruits−17·3−27·4, −7·1−25·4−37·2, −13·6−11·4−43·2, 20·4≥4 Vegetables−17·3−28·7, −5·9−1·7−16·1, 12·7−38·1−69, −7·2Small market3·9−3·5, 11·3--------Convenience store28·016·8, 39·2--------[^8] Prices for the five produce items were lower in the full sample of stores and in small markets located within half a mile of a school, but no relationship between distance from school and price was observed in convenience stores. No clear relationship was observed between fruit or vegetable quality and price. No significant quality-related price differences were observed in the full sample of stores or small markets. Overall, in the adjusted model, store type was significantly associated with produce price only for convenience stores, where the five-item average was 28 % higher (95 % CI 17, 39 %). Discussion {#sec3} ========== Within our large sample of stores in low-income neighbourhoods in California, we find that produce availability, variety and quality are best in large grocery stores, followed by small markets, and poorest in convenience stores. While most large grocery stores and small markets in our sample sold produce that was rated high quality, the selection of fruit was rated high quality in only 25 % of convenience stores and vegetables in only 14 % of convenience stores. At the same time, convenience stores that sold produce had substantially higher prices than large groceries or small markets. Importantly, we find higher average store lowest prices in the low-income communities we assessed for the produce items examined (apples, bananas, oranges, carrots and tomatoes) compared with the county average chain supermarket prices in the same county in the same month across all store types, even in large grocery stores. While it might be expected that prices in chain supermarkets would be lower than in smaller stores, this large sample of stores in low-income neighbourhoods across California provides evidence to suggest that low-income residents who shop for food in their neighbourhoods may pay more, on average, for produce regardless of the type of store in which they shop. This is the largest sample of stores in recent decades to assess the relationship between neighbourhood sociodemographic characteristics and fruit and vegetable prices adjusted for store type. These findings are consistent with some earlier studies^(^ [@ref13] ^)^ but contradict the findings of a 2007 study of California food stores that looked at a broader basket of foods^(^ [@ref15] ^)^. Additionally, we find that a wider variety of produce items is associated with lower prices within store type. Other studies have suggested that efforts to promote healthy eating should support retailers to expand their selection of healthy foods^(^ [@ref24] ^)^. While the literature on retail interventions to improve diet has not been robust, some evidence does suggest that a wider variety of fruits and vegetables in corner stores is significantly associated with the odds of a customer purchasing those items, and the likelihood of purchasing was even stronger for customers participating in SNAP than non-participants^(^ [@ref25] ^)^. Future studies should assess whether increasing store produce variety helps stores to offer more competitive prices and whether this leads to increased purchases, particularly among SNAP participants. In our sample, SNAP and WIC programme participation among small markets and convenience stores was associated with better availability of produce overall as well as better variety and quality of fruits and vegetables. Consistent with earlier studies^(^ [@ref20] ^)^, our analysis found that many convenience stores participating in SNAP fail to provide a wide variety of fruits and vegetables; but contrary to the previous study, our results find that SNAP stores perform better than stores not participating in SNAP or WIC. Consistent with a 2016 study of corner stores^(^ [@ref21] ^)^, our data find that WIC participation (along with SNAP) of small groceries and convenience stores is more highly associated with produce availability and quality than SNAP participation alone or non-participation in either programme. Recently, efforts to revise SNAP retailer standards were undertaken at the federal level^(^ [@ref26] ^)^. Our results suggest the programme (prior to the revisions) may already have played a role in encouraging stores, or attracting stores more willing, to offer fruits and vegetables. Given the poor variety of produce offered in SNAP-participating convenience stores in the present study, SNAP could provide a mechanism for increasing the variety and quality of produce in these stores, which could potentially reduce price. Our study has a number of limitations. The cross-sectional design limits our ability to understand cause and effect, and the community-based data collection likely produces more error than researcher-controlled efforts. However, given our large sample size and the fact that these errors would not cause a clear bias, we feel these data can substantively contribute to the literature. Our data set includes price information about a limited variety of items and we expect that any given store's prices fluctuate throughout the year. Our price comparison data set contains information about supermarket chain store prices, which other studies have found to be lower, generally, than other stores. It would be stronger to have comparison price data from a wider variety of store types. The store data we used for comparison, although not an exact match for any of our sample store types, provide a common reference for comparison purposes. Further, we were able to match the comparison store prices to our sampled stores based on county location and month of the year, limiting the potential bias in the relative price analyses related to geographic or temporal differences. However, there is value in presenting findings showing that across all the store types in our study, average produce lowest prices in the study's low-income neighbourhoods are consistently higher than average supermarket lowest prices matched by county. While efforts to increase access to fresh fruits and vegetables to all shoppers are important for a variety of reasons, evidence is mounting to suggest that physical access alone is not the answer to the complex problem of dietary and health disparities. Studies repeatedly have demonstrated that consumers, particularly low-income consumers, are price sensitive^(^ [@ref27] ^--^ [@ref29] ^)^. Our study reveals fruit and vegetable price disparities between low- and higher-income neighbourhoods. Additionally, we find differences in availability, variety, quality and price of produce across store types within low-income neighbourhoods. Our finding that convenience stores on average offer higher-priced, poorer-quality produce than other stores, coupled with the finding that variety is most highly associated with produce price, highlights both the need and the challenge of working in these stores. The vast majority of groceries are not purchased in convenience stores even by low-income shoppers^(^ [@ref30] ^,^ [@ref31] ^)^. How convenience stores can contribute to improving access to affordable, high-quality produce should be further investigated. Finally, the current study finds that SNAP and WIC participation are associated with better availability, variety and quality of produce, suggesting that these programmes provide an opportunity to continue to improve healthy food access for low-income consumers. Our study examines several possible factors that may influence the food choices made by low-income consumers and suggests that efforts to eliminate dietary disparities and support all residents of the USA to eat a health-promoting diet require a deeper investigation into the underlying issues leading to suboptimal outcomes. In particular, studies are needed to identify what interventions would be most effective for ensuring that quality, affordable, healthy foods are available to all, especially to those with the most limited financial resources. Acknowledgements: The authors would like to thank the staff at the California Department of Public Health, Nutrition Education and Obesity Prevention Branch, for their work in designing and implementing the project and providing the data used in this analysis. Additionally, they would like to thank the numerous staff and volunteers throughout the state of California who participated in the data collection process. Financial support: This work was supported by the US Department of Agriculture, Supplemental Nutrition Education Program Education (SNAP-Ed), through a contract between the Nutrition Policy Institute and the California Department of Public Health. The US Department of Agriculture had no role in the design, analysis, or writing of this article. Conflict of interest: None. Authorship: W.G. directed the analysis project, developed the research questions and drafted the manuscript. D.M.B. contributed to the development of the research questions, conducted the analyses and contributed to the manuscript development. B.C.S. provided training and technical assistance to the local health departments throughout the data collection process, managed the data files, contributed to the development of research questions and supported the manuscript development. G.W.-L. contributed to manuscript development. P.B.C. served as Principal Investigator for the project, advised on the analysis and reviewed and edited the manuscript. Ethics of human subject participation: This study did not involve human subjects. [^1]: SNAP, Supplemental Nutrition Assistance Program; WIC, Supplemental Nutrition Program for Women, Infants, and Children. [^2]: The significance column indicates the significance (at P\<0·05) of two-way tests of the hypothesis that the proportions of interest differ between store type categories. The letter A indicates that 'large groceries' and 'small markets' have different proportions, B that 'large groceries' and 'convenience stores' differ, and C that 'small markets' and 'convenience stores' differ. [^3]: WIC, Supplemental Nutrition Program for Women, Infants, and Children; SNAP, Supplemental Nutrition Assistance Program. [^4]: Relative price differences are the difference between the observed lowest store price and the average lowest price for chain supermarkets in the same county that month, expressed as a percentage of the average county chain supermarket lowest price. [^5]: The significance column indicates the significance (at P\<0·05) of two-way tests of the hypothesis that the proportions differ. The letter A indicates that the 'neither' and 'SNAP only' groups differ, B that the 'neither' and 'both' groups differ, and C that the 'both' and 'SNAP only' groups differ. [^6]: Relative price differences are the difference between the observed lowest store price and the average lowest price for chain supermarkets in the same county that month, expressed as a percentage of the average county chain supermarket lowest price. [^7]: The test significance column indicates the significance (at P\<0·05) of two-way tests of the hypothesis that the relative price differences differ between store type categories. The letter A indicates that 'large groceries' and 'small markets' stores have different relative price differences, B that 'large groceries' and 'convenience stores' differ, and C that 'small markets' and 'convenience stores' differ. [^8]: SNAP, Supplemental Nutrition Assistance Program; WIC, Supplemental Nutrition Program for Women, Infants, and Children.	{ "pile_set_name": "PubMed Central" }
Introduction {#section1-1177271918756648} ============ Prion protein (PrP) is a naturally occurring glycoprotein. Prion protein binds to the cell membrane via a glycophosphatidylinositol (GPI) anchor^[@bibr1-1177271918756648]^ and is abundantly present on the cell surfaces of neurons.^[@bibr2-1177271918756648],[@bibr3-1177271918756648]^ In humans, PrP has many proposed functions, including the maintenance of myelinization homeostasis,^[@bibr4-1177271918756648]^ mitochondrial function,^[@bibr5-1177271918756648]^ and intercellular signaling.^[@bibr6-1177271918756648][@bibr7-1177271918756648]--[@bibr8-1177271918756648]^ For unknown reasons, the α-helix-rich 3-dimensional structure of the cellular form of PrP (PrP^C^) can transform into a β-sheet-rich molecule, which is the pathological form of PrP (PrP^Sc^). PrP^Sc^ is unable to perform the same functions as PrP^C^ and is prone to autocatalytic conversion and aggregation into insoluble aggregates. The accumulation of PrP^Sc^ aggregates over time results in disease. Transmissible spongiform encephalopathies (TSEs), also known as prion diseases, are rare fatal neurodegenerative disorders that occur in humans and animals. Prion diseases can be divided into 3 groups---sporadic, genetic, and iatrogenic. Sporadic diseases are the most common type, and their origins are unknown. Genetic prion diseases are the second most common type and are characterized by mutations in the PrP-encoding gene, PRNP. Iatrogenic prion diseases develop due to the consumption of PrP^Sc^-infected tissue, after transplantation, after surgical interventions, and due to the use of infected growth hormone of human origin. All 3 forms have a different clinical picture but show similar symptoms of disease, such as dementia, the loss of brain function, and spongiform deformation of the brain. Prion diseases also have important clinical and neuropathological features that are similar to those of Alzheimer disease.^[@bibr6-1177271918756648]^ Although prion disease occurs due to accumulation of toxic PrP^Sc^ aggregates in the brain, the mechanism that underlies the conversion of PrP^C^ to PrP^Sc^ and the development of prion disease remains unknown. PrP^C^ is linked to the cell membrane via a GPI anchor. Similar to other GPI-anchored proteins, PrP^C^ enters the endocytic recycling pathway.^[@bibr9-1177271918756648]^ As the protein revolves between the cell surface and the endosome, the protein encounters both neutral and acidic pH.^[@bibr10-1177271918756648],[@bibr11-1177271918756648]^ Many studies have shown that these environmental changes may cause the misfolding of PrP, leading to its transformation into the pathological form and to its accumulation.^[@bibr9-1177271918756648],[@bibr12-1177271918756648][@bibr13-1177271918756648][@bibr14-1177271918756648][@bibr15-1177271918756648]--[@bibr16-1177271918756648]^ During normal metabolism, PrP^C^ can undergo 5 posttranslational cleavages. Cleavages α and β occur within the highly conserved hydrophobic region of PrP in the endosomes and result in the release of a 10-kDa N-terminal fragment into the extracellular matrix.^[@bibr6-1177271918756648]^ Cleavage γ presumably occurs during the transport of PrP through the secretary pathway before glycosylation.^[@bibr17-1177271918756648]^ The cleavage near the GPI anchor produces a 6- to 7-kDa unglycosylated C-terminal fragment. Prion protein is also the target of shedding. Phospholipase C cleaves PrP within the GPI anchor, which leads to the release of the whole protein into the extracellular medium.^[@bibr18-1177271918756648]^ Shed PrP is also produced by cleavage of the PrP^C^ polypeptide chain at the C-terminus by ADAM10 protease.^[@bibr19-1177271918756648]^ In hamsters, approximately 15% of PrP^Sc^ molecules, isolated from the prion-infected hamster brain, terminate at Gly228.^[@bibr20-1177271918756648]^ In addition, in a recombinant hamster PrP model, ADAM10 protease cleaves PrP between amino acid residues Gly228 and Arg229.^[@bibr21-1177271918756648]^ The cleavage site of ADAM10 protease in human PrP^C^ is not unambiguously defined. A cleavage site profile of ADAM10 was recently explored using peptide libraries and cleavage liquid chromatography-mass spectrometry analysis.^[@bibr22-1177271918756648]^ The authors have shown that no unique sequence exists that induces cleavage. Therefore, ADAM10 protease can produce more than one variant of shed PrP. In the proximity of the cleavage site, the sequence and 3-dimensional structure of human PrP differ from those of hamster PrP. Because the protein sequences are not identical, we can expect a different cleavage pattern for the ADAM10 protease. Consequently, we hypothesize that the proteolytic site in the human sequence is not located between amino acid residues 228 and 229 but is instead shifted toward the N-terminus, namely, between Tyr226 and Gln227 ([Figure 1](#fig1-1177271918756648){ref-type="fig"}). ![Part of the human and hamster PrP sequence, indicating the C-terminus and potential ADAM10 protease cleavage sites. The blue arrow indicates the ADAM10 protease cleavage site in the hamster PrP, and the green arrow indicates the potential cleavage site in the human PrP, which produces PrP226\. PrP indicates prion protein.](10.1177_1177271918756648-fig1){#fig1-1177271918756648} In addition to the brain, PrP without the GPI anchor is also found in the urine, cerebrospinal fluid, and blood of patients with different prion diseases and in controls.^[@bibr23-1177271918756648][@bibr24-1177271918756648][@bibr25-1177271918756648]--[@bibr26-1177271918756648]^ Most analyses concerning PrP have focused on finding PrP^Sc^ and determining infectivity; however, studies on PrP^C^ and its variants in bodily fluids are few. Reports of PrP variants that lack the GPI anchor show that these molecules are involved in neurodegenerative diseases, where the molecules have different roles. These variants may thus be a useful tool for determining the mechanism of neurodegenerative diseases as well as for their diagnosis and therapy. PrPs That Lack the GPI Anchor {#section2-1177271918756648} ============================= Anchorless PrP {#section3-1177271918756648} -------------- Analysis of aggregates in patients with sporadic Creutzfeldt-Jakob disease (sCJD)^[@bibr27-1177271918756648],[@bibr28-1177271918756648]^ and Gerstmann-Straussler-Scheinker (GSS) syndrome^[@bibr29-1177271918756648][@bibr30-1177271918756648][@bibr31-1177271918756648]--[@bibr32-1177271918756648]^ has shown that apart from PrP^Sc^, aggregates are built from anchorless PrP and other variants of PrP without the GPI anchor. Initially, the conversion from PrP^C^ to PrP^Sc^ was thought to occur only when PrP^C^ was bound to the cell membrane by the GPI anchor, but a study by Chesebro and coworkers showed the opposite.^[@bibr33-1177271918756648]^ Experiments on transgenic mice expressing only anchorless PrP showed that infection with PrP^Sc^ encouraged the accumulation of PrP in various tissues, such as the heart, kidney, and body fat.^[@bibr33-1177271918756648],[@bibr34-1177271918756648]^ This accumulation led to the development of prion disease without clinical signs.^[@bibr33-1177271918756648]^ Later, experiments on these mice confirmed that anchorless PrP could cause prion disease, even though the protein was not covalently bound to the cell surface.^[@bibr35-1177271918756648][@bibr36-1177271918756648][@bibr37-1177271918756648][@bibr38-1177271918756648]--[@bibr39-1177271918756648]^ Anchorless PrP aggregates were found to spread and seed through the interstitial fluid and thus enable the development of cerebral amyloid angiopathy (CAA) in the brain.^[@bibr38-1177271918756648]^ In addition, the recent study on transgenic mice expressing anchorless PrP showed that PrP without GPI anchor could propagate the infectivity of prion diseases within the same and among different species.^[@bibr40-1177271918756648]^ PrP shed by ADAM10 {#section4-1177271918756648} ------------------ ADAM proteases cleave transmembrane proteins and affect cell interactions with their surroundings, cell migration, and cell signaling.^[@bibr41-1177271918756648],[@bibr42-1177271918756648]^ Several research groups have shown that ADAM10 protease cleaves PrP^C^ from the cell membrane, producing shed PrP.^[@bibr18-1177271918756648],[@bibr19-1177271918756648],[@bibr21-1177271918756648]^ The effect of the protease was shown in 2 systems: PrP^C^-expressing human embryonic kidney (HEK) cells that were transfected with complementary DNA encoding for ADAM10^21^ and neuron-specific ADAM10 knockout mice.^[@bibr43-1177271918756648]^ The experiments in HEK cells showed an increase in the amount of soluble PrP^C^ in the cell supernatant, whereas the experiments in the knockout mice showed an increase in the amount of membrane-linked PrP^C^ and an increase in PrP^Sc^ formation, compared with the wild-type mice. Shedding is presumed to have an antagonistic role in the development of prion diseases. The primary product of shedding, shed PrP, is a molecule that freely floats in bodily fluids, binds PrP^Sc^ aggregates, and prevents their replication. However, when PrP^Sc^ aggregates bind to PrP^C^, PrP^C^ is refolded into PrP^Sc^ and released from the cell membrane by ADAM10, generating new sources of infection ([Figure 2](#fig2-1177271918756648){ref-type="fig"}).^[@bibr44-1177271918756648]^ ![Binding of toxic oligomers to shed PrP and PrP^C^. (A) The toxic amyloid β (AβO) and PrP (PrP^Sc^) oligomers are bound by the shed PrP N-terminal region and are neutralized, whereas (B) the binding of toxic oligomers to the N-terminal region of PrP^C^ on the neuronal cell membrane evokes a toxic signaling pathway (thunderbolt). PrP indicates prion protein.](10.1177_1177271918756648-fig2){#fig2-1177271918756648} Prion protein is also involved in Alzheimer disease. Similar to PrP^Sc^ in prion diseases, toxic amyloid-β peptide oligomers (AβOs) bind to the N-terminus of GPI-anchored PrP^C^.^[@bibr45-1177271918756648]^ When the expression of membrane-bound PrP^C^ increases at the neuronal surface, the level of toxic AβO binding to membrane-bound PrP^C^ also increases, causing cell damage. However, in the extracellular space, shed PrP binds to AβO and neutralizes it.^[@bibr46-1177271918756648],[@bibr47-1177271918756648]^ ADAM10 protease can shed both PrP^C^ and PrP^Sc^, and therefore, the regulation of this mechanism may affect the development and occurrence of Alzheimer disease and prion diseases ([Figure 2](#fig2-1177271918756648){ref-type="fig"}).^[@bibr44-1177271918756648]^ Truncated PrP due to nonsense mutations {#section5-1177271918756648} --------------------------------------- Nonsense mutations in the PRNP* gene are another source of PrPs that lack GPI anchor. Due to nonsense mutations, a codon that encodes for a specific amino acid changes into a stop codon, which leads to premature termination of the protein. The product is a truncated PrP that lacks the GPI anchor at the time of synthesis. Several cases of prion diseases that were caused by the expression and accumulation of truncated PrP have been described. The nonsense mutations that caused the diseased were Y145X,^[@bibr48-1177271918756648]^ Q160X,^[@bibr49-1177271918756648],[@bibr50-1177271918756648]^ Y163X,^[@bibr51-1177271918756648],[@bibr52-1177271918756648]^ V203X,^[@bibr53-1177271918756648]^ Y226X,^[@bibr30-1177271918756648]^ and Q227X.^[@bibr30-1177271918756648]^ In the cases of Y145X, Y163X, and Y226X, the patients developed PrP-CAA. The PrP-CAA is characterized by amyloid deposits of PrP in cerebral vessels and by Alzheimer disease--like neurofibrillary tangles in the brain, predominantly in the hippocampus.^[@bibr54-1177271918756648]^ The distribution of PrP^Sc^ aggregates is similar to the distribution of PrP in transgenic mice expressing only anchorless PrP. The deletion of 2 base pairs in codon 178 causes a change in the protein sequence from codon 178 onward and results in a stop codon at codon 203.^[@bibr53-1177271918756648]^ This truncated PrP is deposited in almost all examined organs, namely, peripheral nerves, smooth muscles, and blood vessels in non-central nervous system tissues. The Q160X mutation leads to an illness that is similar to Alzheimer disease, whereas patients with the Q227X mutation develop a disease that is similar to the GSS syndrome. Interestingly, the insertion of stop mutations in the codon that codes for tyrosine results in PrP-CAA, whereas a premature stop in the codon sequence that codes for glutamine results in a different pathology. This difference may be due to the amino acid that terminates the variant's protein sequence. V5B2 and PrP226\* {#section6-1177271918756648} ----------------- The methods used to determine the presence of PrP^C^ and PrP^Sc^ in biological samples are based on antigen-antibody interactions. To find an antibody that discriminates between Creutzfeldt-Jakob disease (CJD) and non-CJD samples, we prepared monoclonal antibodies against PrP. BALB/c mice were immunized with a KLH-bound peptide fragment of the human PrP sequence (amino acid residues between 214 and 226). Using hybridoma technology, we prepared a panel of monoclonal antibodies against PrP, one of which was V5B2.^[@bibr55-1177271918756648]^ V5B2 differed from the other anti-PrP monoclonal antibodies because V5B2 recognized only a specific form of PrP. We determined and confirmed the epitope of monoclonal antibody V5B2 using an alanine scan and phage display approach examining the C-terminal region between amino acid residues 214 and 228.^[@bibr56-1177271918756648]^ We found that monoclonal antibody V5B2 recognized the anchorless truncated form of PrP that ended with amino acid residue Tyr226, and we named the variant PrP226\. Using monoclonal antibody V5B2, we identified PrP226\ in pathological samples and found that PrP226\* was a remarkable biomarker for diagnosing prion diseases. Our immunohistochemistry, dot blot, and Western blot analyses of samples from a group of patients with sCJD showed that monoclonal antibody V5B2 specifically discriminated between the CJD and non-CJD brain.^[@bibr29-1177271918756648],[@bibr55-1177271918756648]^ We postulated that PrP226\* incorporated into PrP^Sc^ aggregates in such a manner that the C-terminus was exposed on the aggregate surface or was hidden inside the molecule. The C-terminus of PrP226\* represents the epitope of V5B2. To undoubtedly determine the presence of PrP226\* in the PrP^Sc^ aggregates using this monoclonal antibody, we denatured the sample before the analysis to release all V5B2 epitopes from the aggregates.^[@bibr27-1177271918756648],[@bibr29-1177271918756648]^ Our monoclonal antibody also enables a sensitive determination of PrP226\* in the brains of TSE patients using dissociation-enhanced lanthanide fluorescence immunoassay^[@bibr29-1177271918756648]^ and enzyme-linked immunosorbent assay.^[@bibr27-1177271918756648]^ In addition to PrP-infected samples, we have shown that PrP226\* is also present in the healthy brain, albeit in small amounts.^[@bibr27-1177271918756648],[@bibr29-1177271918756648]^ We are not the only research group to describe this PrP form. As indicated above, PrP226\* was concurrently described by Jansen and coworkers.^[@bibr30-1177271918756648]^ The authors characterized a patient who carried a stop mutation at position Q227X and developed a disease similar to GSS syndrome. PrP226\* causes disease, and minor quantities of PrP226\* are also present in the brains of healthy individuals. We suspect that this protein is produced through natural processes, such as shedding or nonsense mutations, and that under yet undefined conditions, PrP226\* can act as a propagator of disease. Based on the nature of PrP226\, namely, its presence in human brain and body fluids (unpublished data), lack of GPI anchor, slight truncation at the C-terminus, and neutralization and pathological capacity, we speculate that PrP226\ is produced by shedding. To determine whether PrP226\* could be the result of cleavage by ADAM10, we analyzed the amount of PrP226\* in the human brain. The PrP226\* amount, which we determined in the brains of subjects with sCJD, is approximately 12% (unpublished data), which is similar to the amount of truncated PrP determined in the hamster.^[@bibr20-1177271918756648]^ Based on the properties of PrP226\* in vivo, we have indirect evidence supporting PrP226\* in humans as one of the products, if not the only product, of ADAM10 shedding ([Figure 1](#fig1-1177271918756648){ref-type="fig"}). Structure and conversion properties of PrP226\* {#section7-1177271918756648} ----------------------------------------------- Recently, we examined the structural and biochemical properties of PrP226\* and resolved the high-resolution nuclear magnetic resonance structure of the protein under acidic conditions (PDB ID 5L6R).^[@bibr57-1177271918756648]^ The structure of the protein reveals a disordered region between residues 90 and 125 and a structured region between residues 126 and 226. The structured region consists of 2 β-sheets and, compared with wild-type PrP, has 4 α-helices instead of 3 α-helices. As the C-terminus is truncated, new interactions within the structure are needed to stabilize the protein. To do that, the amino acid residues at the C-terminus are driven into the proximity of the amino acid residues of the flexible loop. This causes a minor change in the structure, electrostatic potential and solvent accessibility of the C-terminus, and flexible loop compared with the wild-type molecule. Similar observations have also been made in structural studies of various pathogenic mutants that cause TSE, namely, Q212P^[@bibr58-1177271918756648]^ and V210I.^[@bibr59-1177271918756648]^ We also determined the thermodynamic properties and in vitro conversion propensity of PrP226\* at acidic and physiological pH.^[@bibr60-1177271918756648]^ We found that in both conditions, PrP226\* is more thermodynamically destabilized than the wild-type protein. Furthermore, the fibrillization propensity of PrP226\* was similar to that of the wild-type PrP in physiological conditions, and the fibril formation onset time at acidic pH for PrP226\* was increased compared with that of the wild-type protein. Alongside PrP226\, we analyzed variants that were similar to PrP226\. One variant, PrP225\, had similar thermodynamic properties to those of PrP226\, but the fibrillization propensity was shorter than that of PrP226\. A similar trend to that observed in vitro* was observed in vivo. Jansen and coworkers^[@bibr30-1177271918756648]^ reported that the time of onset of prion disease, which resulted from expression of only PrP225\, was 27 months, whereas the time of onset of prion disease, which resulted from PrP226\ expression alone, was longer, namely, 72 months. In addition to the time of disease onset, the expression of the 2 PrP variants differentially affected the development of prion diseases, resulting in different pathologies. In conclusion, the in vivo and in vitro data suggest that a change in the protein length, even by 1 amino acid residue, can have a profound effect on in vitro and in vivo conversion. Conclusions {#section8-1177271918756648} =========== Neurotoxicity occurs at the cell surface, where toxic oligomers bind to membrane-anchored PrP^C^. Binding of PrP^Sc^ oligomers to membrane-bound PrP^C^ transforms the native molecule into PrP^Sc^, which can be shed from the cell surface by ADAM10 protease, generating a new and infective PrP^Sc^. However, the shedding of PrP^C^ monomers generates the shed PrP, which can act as a neutralizer of toxic oligomers, which is important in the prevention of neurodegenerative diseases such as Alzheimer disease. ADAM10 protease does not have a unique proteolytic site and can, under specific conditions, generate various shed PrPs, one of which is PrP226\. Due to the dual role of the shed PrP, particularly PrP226\, in the mechanism behind neurodegenerative diseases, these molecules can be potent biomarkers for the diagnosis of neurodegenerative diseases, and thus, these molecules can be powerful tools for the development of therapy. Funding:The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors gratefully acknowledge the financial support of the Slovenian Research Agency. Declaration of conflicting interests:The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Author Contributions: VK and VČŠ contributed to the writing of the manuscript, confirmed the manuscript results and conclusions, jointly developed the structure and arguments of the paper, made critical revisions and approved the final version, and have reviewed and approved the final manuscript.	{ "pile_set_name": "PubMed Central" }
All relevant data are within the paper and its Supporting Information files. Introduction {#sec001} ============ GPR91 and GPR99 are G-protein-coupled receptors (GPCRs) activated by Krebs cycle intermediates, part of the larger class of carbohydrate metabolites---an observation that renewed interest in a biochemical pathway discovered decades ago \[[@pbio.2003619.ref001],[@pbio.2003619.ref002]\]. GPR91, through its activation by succinate outside the tricarboxylic acid (TCA) cycle, has a wide range of functions in diverse diseases, such as hypertension and diabetes. Its study allowed greater understanding of the molecular links between the TCA cycle and metabolic diseases \[[@pbio.2003619.ref002],[@pbio.2003619.ref003]\]. The development of the visual system requires high levels of energy to propel mitochondrial-enriched axons properly through the nervous system, as retinal ganglion cells (RGCs) are essential for transmitting information from the retina to the brain. The growth and survival of neurons depend on mitochondria as they perform aerobic ATP synthesis and play a significant role in apoptotic and necrotic cell death \[[@pbio.2003619.ref004]\]. Thus, failures of mitochondrial function appear to be involved in degenerative diseases of the nervous system \[[@pbio.2003619.ref005]\]. One of the most mitochondria-enriched regions of the axon is the active growth cone (GC) at the tip of the axon \[[@pbio.2003619.ref006]\]. The GC contains multiple receptors that interact with guidance molecules, allowing the front end of a developing axon to navigate through the complex landscape of the early nervous system toward its appropriate targets \[[@pbio.2003619.ref007]\]. However, the role of intermediates from carbohydrate metabolism during the development of the visual system has not been well characterized. In the past decade, increasing evidence has highlighted GPCRs as mediators of both repulsive and attractive axon guidance, as their ligands may serve as guidance cues for axon pathfinding; however, GPCRs involved in axon growth still remain to be found \[[@pbio.2003619.ref008]--[@pbio.2003619.ref011]\]. In a groundbreaking study in 2004, GPR91 (succinate receptor 1 \[Sucnr1\]) and GPR99 (2-oxoglutarate receptor 1 \[Oxgr1\]) were both identified as receptors of the Krebs cycle intermediates succinate and α-ketoglutarate (α-KG), respectively \[[@pbio.2003619.ref002]\]. GPR91 and the closely related GPR99 are expressed in multiple tissues, such as the kidney \[[@pbio.2003619.ref002],[@pbio.2003619.ref012]\] and cardiac muscle \[[@pbio.2003619.ref013]--[@pbio.2003619.ref015]\]. Previous reports have shown that succinate and GPR91 regulate normal retinal vascularization, proliferative ischemic retinopathy \[[@pbio.2003619.ref016]\], and cortical revascularization post-ischemia \[[@pbio.2003619.ref017]\]. Moreover, through the activation of GPR91, succinate has been shown to have an effect on motility, migration, and growth, as it directly promotes chemotaxis and potentiates activation initiated by Toll-like receptor agonists in dendritic cells \[[@pbio.2003619.ref018],[@pbio.2003619.ref019]\]. However, to date, scarce literature exists on GPR99 functions. Human neuronal mapping and vascular innervation are closely related, as similar molecules and signaling mechanisms are shared between axon guidance, neuronal migration, and blood vessel guidance and growth. For example, the Slit/Robo pathway plays a critical role in both angiogenesis and the guidance of neuronal migration of the olfactory system \[[@pbio.2003619.ref020],[@pbio.2003619.ref021]\]. Moreover, semaphorins and their receptors play a pivotal role as axon guidance cues \[[@pbio.2003619.ref022],[@pbio.2003619.ref023]\] while also acting as a vasorepulsive force that misdirects new retinal vessels toward the vitreous in a murine model of oxygen-induced retinopathy \[[@pbio.2003619.ref024]\]. Therefore, we investigated the growth-promoting actions and guidance effects of the carbohydrate metabolites succinate and α-KG, through their respective receptor GPR91 and GPR99, during the establishment of the retino-thalamic pathway in an embryonic mouse model. Elucidating carbohydrate metabolite functions during visual development may provide crucial insights regarding their potential roles in the plasticity and regeneration of the nervous system and allow the development of further pharmacological tools, expanding and improving central and peripheral nervous system repair strategies. Results {#sec002} ======= GPR91 and GPR99 are expressed in RGCs in the developing retina {#sec003} -------------------------------------------------------------- We utilized murine retinas obtained from embryos (embryonic day 14/15 \[E14/15\]) to characterize the presence of GPR91 and GPR99 and their possible involvement during retinal projection navigation. At E14/15, GPR91 and GPR99 proteins were mainly present in the ganglion cell layer but were also detected in the ganglion cell fiber and neuroblast layers ([Fig 1A--1F](#pbio.2003619.g001){ref-type="fig"}). The retinas from adult and E14/15 knockout (KO) mice (gpr91KO or gpr99KO) showed no expression of GPR91 or GPR99, confirming the antibodies' specificity ([S1A--S1H Fig](#pbio.2003619.s002){ref-type="supplementary-material"}). In E14/15 wild-type (WT) murine retinal explants, GPR91 and GPR99 were present in neurites, GCs, and filopodia, in dendrites and axons ([Fig 1G--1R](#pbio.2003619.g001){ref-type="fig"} and [S1I--S1L Fig](#pbio.2003619.s002){ref-type="supplementary-material"}). Retinal explants obtained from gpr91KO and gpr99KO E14/15 embryos did not express GPR91 or GPR99, respectively ([S1M--S1P Fig](#pbio.2003619.s002){ref-type="supplementary-material"}), which also confirms the specificity of the antibodies used in this study. Moreover, we observed the presence of GPR91 and GPR99 at the RGC layer of P1 Syrian golden hamsters ([S1Q--S1R Fig](#pbio.2003619.s002){ref-type="supplementary-material"}). ![GPR91 and GPR99 are expressed in RCGs and in axons and dendrites of retinal explants.\ (A-F) Expression of GPR91 and GPR99 proteins in retinal sections of E14/15 WT mouse embryos. Scale bars: 10 μm. (G-R): Expression of GPR91 and GPR99 in retinal explants, GCs, and neurites of E14/15 WT mouse embryos. Double-labeling immunofluorescence illustrates the colocalization of GPR91 and GPR99 with dendrites or with axons of RGCs. Scale bars: 5 μm. E14/15, embryonic day 14/15; GC, growth cone; GCFL, ganglion cell fiber layer; GCL, ganglion cell layer; MAP2, microtubule-associated protein 2; NBL, neuroblast layer; NFM, neurofilament; RGC, retinal ganglion cell; WT, wild-type.](pbio.2003619.g001){#pbio.2003619.g001} The Krebs cycle intermediates succinate and α-KG reorganize GC morphology and increase axon growth {#sec004} -------------------------------------------------------------------------------------------------- As previous studies have shown that GPCRs are involved in axon guidance, we evaluated the roles of GPR91 and GPR99 on GC actions using retinal explants isolated from E14/15 mouse embryos after 2 days in vitro (DIVs) in culture. Explants treated for 60 min with the specific agonists succinate (100 μM) or α-KG (200 μM) showed a significant increase in the GC surface area and the number of filopodia, compared to controls ([Fig 2A--2C](#pbio.2003619.g002){ref-type="fig"} and [S2A--S2D Fig](#pbio.2003619.s003){ref-type="supplementary-material"}). As expected, the effect of succinate on GC size and filopodia number was completely abolished in gpr91KO but not in gpr99KO mouse retinal explants, demonstrating a specific action of succinate on GPR91 ([Fig 2A--2C](#pbio.2003619.g002){ref-type="fig"} and [S2A--S2D Fig](#pbio.2003619.s003){ref-type="supplementary-material"}). Similarly, α-KG effects were maintained in gpr91KO and decreased in gpr99KO. The effects of both agonists were abolished in the retinal explants from double-KO (gpr91KO/gpr99KO) mice ([Fig 2A--2C](#pbio.2003619.g002){ref-type="fig"} and [S2A--S2D Fig](#pbio.2003619.s003){ref-type="supplementary-material"}). Moreover, following 60-min succinate (100 μM) or α-KG (200 μM) treatment, similar effects were observed on GC surface area and filopodia number of cortical neurons (2 DIVs); these effects were also abolished in neurons lacking the expression of GPR91 and/or GPR99 ([S2E--S2G Fig](#pbio.2003619.s003){ref-type="supplementary-material"}). ![Succinate and α-KG modulate GC morphology and increase neurite outgrowth via GPR91 and GPR99.\ (A) Photomicrographs of E14/15 GCs of retinal projections from WT, gpr91KO, gpr99KO, and double-KO mouse embryos, untreated and after treatment with GPR91 agonist (100 μM succinate) or GPR99 agonist (200 μM α-KG). (B) Analysis of GC surface area (N = 104--271 per condition) and (C) filopodia number (N = 105--271 per condition) following a 1 h treatment with GPR91 or GPR99 ligands in WT, gpr91KO, gpr99KO, and double-KO mice. (D) Photomicrographs of retinal explants from each mouse strain cultured for 1 DIV and treated for 15 h with succinate (100 μM) or α-KG (200 μM). (E) Quantification of neurite growth after treatment with GPR91 or GPR99 agonist (N = 84--181 per condition). Scale bars: 5 μm (A); 100 μm (D). Values are presented as the means ± SEM. \# indicates significant changes compared to WT in B, C, and E; p \< 0.001. Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. α-KG, α-ketoglutarate; Ctrl, control; DIV, day in vitro; E14/15, embryonic day 14/15; GC, growth cone; KO, knockout; WT, wild-type.](pbio.2003619.g002){#pbio.2003619.g002} To further evaluate the effects of GPR91 and GPR99 ligand treatment on axon growth, retinal explants from WT mouse embryos were treated for 15 h with succinate (100 μM) or α-KG (200 μM). Both agonists induced an increase in total neurite growth ([Fig 2D and 2E](#pbio.2003619.g002){ref-type="fig"}). Moreover, stimulation of gpr91KO murine retinal explants with α-KG and the stimulation of gpr99KO murine retinal explants with succinate also induced neurite growth ([Fig 2D and 2E](#pbio.2003619.g002){ref-type="fig"}). Again, the effects of succinate were essentially abolished in gpr91KO murine retinal explants, whereas the increased outgrowth produced by α-KG was markedly reduced in gpr99KO murine retinal explants. In double-KO murine retinal explants, the effect produced by either succinate or α-KG was abolished ([Fig 2D and 2E](#pbio.2003619.g002){ref-type="fig"}). To investigate whether the effects of intermediates of carbohydrate metabolism on GC morphology and neurite outgrowth could also affect cell viability, we treated murine embryonic retinal explants or cortical neurons with succinate or α-KG and then used a LIVE/DEAD assay to evaluate cell death. Following a 15-h treatment with succinate (100 μM) or α-KG (200 μM), retinal explants or cortical neurons showed no differences in cell viability compared to control explants ([S3 Fig](#pbio.2003619.s004){ref-type="supplementary-material"}). However, we observed a high induction of cell death in the positive control condition of staurosporine-treated explants or neurons ([S3 Fig](#pbio.2003619.s004){ref-type="supplementary-material"}). Taken together, these results indicate that the Krebs cycle intermediates succinate and α-KG, via GPR91 and GPR99, increase axon growth in retinal explants and modulate GC morphology in retinal explants and primary neurons. GPR91 and GPR99 agonists modulate GC morphology and increase axon growth via the extracellular signal--regulated kinases 1 and 2 (ERK~1/2~) pathway {#sec005} --------------------------------------------------------------------------------------------------------------------------------------------------- GPR91 is coupled to at least two signaling pathways, G~i~/G~o~ and G~q11~, whereas the activation of GPR99 by α-KG triggers a G~q~-mediated pathway \[[@pbio.2003619.ref002]\]. Moreover, previous reports have demonstrated that succinate activates the mitogen-activated protein kinase (MAPK) signaling pathways via GPR91 \[[@pbio.2003619.ref002],[@pbio.2003619.ref012],[@pbio.2003619.ref013],[@pbio.2003619.ref018],[@pbio.2003619.ref019],[@pbio.2003619.ref025]\]. Since MAPKs mediate axon outgrowth, migration, and guidance \[[@pbio.2003619.ref026]\], we determined whether the effects observed with succinate/GPR91 and α-KG/GPR99 were mediated via the ERK~1/2~ pathway. ERK~1/2~ phosphorylation was significantly increased, both in vitro in neurons and ex vivo in retinal explants, following succinate and α-KG stimulation, while these effects were abrogated by CI-1040, a selective ERK~1/2~ inhibitor ([Fig 3A, 3B and 3H](#pbio.2003619.g003){ref-type="fig"}). CI-1040 treatment also abolished succinate- and α-KG-induced increases in GC surface area and filopodia number ([Fig 3C--3E](#pbio.2003619.g003){ref-type="fig"}); no significant differences were observed between the untreated control and a control pretreated with CI-1040. Inhibition of ERK~1/2~ interfered with succinate- and α-KG-induced projection length ([Fig 3F and 3G](#pbio.2003619.g003){ref-type="fig"}), whereas CI-1040 treatment alone had no significant effect on the total projection length, as observed in control conditions. Moreover, CI-1040 treatment did not affect the viability of embryonic retinal explants and cortical neurons, as no significant neuronal cell death was observed compared to controls with the LIVE/DEAD assay ([S3 Fig](#pbio.2003619.s004){ref-type="supplementary-material"}). These data implicate the ERK~1/2~ pathway in the GPR91- and GPR99-induced modulation of GC morphology and axon outgrowth via their respective TCA cycle metabolite ligand. ![GPR91 and GPR99 agonists modulate GC morphology and increase axon growth via the ERK~1/2~ pathway.\ (A) Protein expression levels of P-ERK~1/2~, ERK~1/2~, and β-actin in primary neuron cultures incubated with one of the following: 100 μM succinate or 200 μM α-KG at 37 °C for 2, 5, and 15 min. The antibody β-actin was used to verify equal loading in all lanes. (B) ERK phosphorylation state following 15 min pretreatment with CI-1040, an ERK~1/2~ inhibitor, before incubation with or without GPR91 or GPR99 agonist. (C) Photomicrographs of retinal explant GCs treated with GPR91 or GPR99 agonist in the presence or absence of CI-1040. (D) Quantification of the GC surface areas (N = 178--189 per condition) and (E) filopodia numbers (N = 178--237 per condition). (F) Photomicrographs of retinal explants treated with succinate (100 μM) or α-KG (200 μM) in the presence or absence of CI-1040. (G) Analysis of retinal projection growth of retinal explants (N = 31--86 per condition). (H) Immunohistochemical photomicrographs of the ERK~1/2~ phosphorylation state in retinal explant GCs treated with 100 μM succinate or 200 μM α-KG in the presence or absence of CI-1040. Scale bars: 5 μm (C, H); 100 μm (F). Values are presented as the means ± SEM. \* indicates a significant change compared to the Ctrl group in D, E and G; p \< 0.0001. \# indicates a significant change induced by ERK inhibitor treatment in D, E and G; p \< 0.002. Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. α-KG, α-ketoglutarate; Ctrl, control; ERK~1/2~, extracellular signal--regulated kinases 1 and 2; GC, growth cone.](pbio.2003619.g003){#pbio.2003619.g003} Role of GPR91 and GPR99 during development {#sec006} ------------------------------------------ To determine the contribution of GPR91 and GPR99 to the development of retinal projections in vivo, E14/15 murine embryos received an intraocular injection of DiI (DiIC18\[3\] \[1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate\]), a lipophilic tracer. After 7 d of tracer diffusion, surgery was performed to visualize the optic nerve, chiasm, and tract. The photomicrographs obtained revealed that genetic deletion of either GPR91 or GPR99 had no detrimental effects on RGC axon guidance, as axon steering at the optic chiasm, after a single genetic deletion of gpr91 or gpr99, was similar to the WT group ([S4A and S4B Fig](#pbio.2003619.s005){ref-type="supplementary-material"}). Moreover, succinate and α-KG treatment also failed to modulate axon steering in time-lapse microscopy experiments performed on GCs from E14/15 WT murine retinal explants at 1 DIV ([S5A--S5E Fig](#pbio.2003619.s006){ref-type="supplementary-material"}). Microgradient application of succinate or α-KG did not induce any significant directional GC turning compared to the vehicle control ([S5A--S5E Fig](#pbio.2003619.s006){ref-type="supplementary-material"}). Interestingly, short-term exposure to succinate induced an increase in the growth of retinal axons, while α-KG exposure had no significant effects ([S5E Fig](#pbio.2003619.s006){ref-type="supplementary-material"}). However, in double-KO mice, few retinal axon fibers projected to the ipsilateral side of the brain although, some extended into the contralateral optic nerve. The concomitant absence of GPR91 and GPR99 appeared to induce some abnormal projections in the ipsilateral and contralateral sides of the optic chiasm, suggesting a potential compensatory role played by each receptor in the absence of the other ([S4A and S4B Fig](#pbio.2003619.s005){ref-type="supplementary-material"}). Moreover, to assess the involvement of the citric acid cycle intermediate receptors in retino-geniculate development, we examined the projections to the dorsal lateral geniculate nucleus (dLGN) of adult mice. Contralateral and ipsilateral projections in the dLGN from all genetically modified mouse strains occupied the same area as those of WT mice ([S4C Fig](#pbio.2003619.s005){ref-type="supplementary-material"}). These data indicate a similar overlap between contralateral and ipsilateral RGC projections in the dLGN for all mouse genotypes ([S4D Fig](#pbio.2003619.s005){ref-type="supplementary-material"}). Taken together, these observations demonstrate that GPR91 and GPR99 do not appear to be implicated in guidance and target selection during the development of the retinogeniculate pathway in vivo. To investigate the in vivo effects of intermediates of carbohydrate metabolism during the development of the visual system, the mouse model presents limitations. Because the mouse visual system is completed at birth \[[@pbio.2003619.ref027]\], we further utilized a different rodent model. The Syrian golden hamster has a shorter gestation period (15 d versus 18.5 d), and pups are born with a relatively premature visual system \[[@pbio.2003619.ref027]\]. As the axons of RGCs reach their thalamic and midbrain targets at P3 in the hamster, this model allows examination of the induction of axon growth by different agonists \[[@pbio.2003619.ref010],[@pbio.2003619.ref028]\]. Taking advantage of this observation, hamsters were injected intravitreally 24 h after birth (P1) with a mixed solution of cholera toxin subunit B (CTb) with either 0.9% saline solution, 100 mM succinate, or 200 mM α-KG, and immunohistological analyses were performed at P5. Intraocular injections of CTb produced intense labeling of thalamic and midbrain targets such as the dLGN and superior colliculus, making the evaluation of the collateral growth of RGC axons difficult. Thus, we evaluated the RGC branch growth at the dorsal terminal nucleus (DTN), one of the nuclei composing the accessory visual pathway and involved in mediating visuomotor reflexes underlying the generation of optokinetic nystagmus \[[@pbio.2003619.ref029]\]. Compared with the control group, unilateral intraocular injections of succinate or α-KG induced significant increases in RGC collateral axon projection length and branch number in the DTN ([Fig 4A--4C](#pbio.2003619.g004){ref-type="fig"}). ![Succinate/GPR91 and α-KG/GPR99 are involved in RGC axon growth in vivo.\ (A) Photomicrographs of P5 hamster retinal projections in the DTN for the control, succinate (100 mM), and α-KG (200 mM) treatment groups. (B, C) Quantification of retinal projection development in the DTN: collateral axon projection length (N = 168--240 per condition) and collateral axon branch number (N = 34--57 per condition) of treated groups compared with the Ctrl group. Succinate and α-KG treatment increased axon growth and collateral branch number compared with the Ctrl. (D) Photomicrographs of P5 retinal projections in the DTN in WT, gpr91KO, gpr99KO, and double-KO mice. (E) Quantification of retinal projections in the DTN during the development of WT, gpr91KO, gpr99KO, and double-KO mice (N = 627--963 per condition). Collateral projection length is expressed as the mean ± SEM. (F) The number of collateral axon branches decreased in gpr91KO, gpr99KO, and double-KO mice compared to WT (N = 82--160 per condition). Data are presented as the means ± SEM. Scale bars: 100 μm (A, D). \* indicates a significant change compared to the Ctrl group in B and C (p \< 0.05) and in E and F (p \< 0.0001). Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. α-KG, α-ketoglutarate; Ctrl, control; DTN, dorsal terminal nucleus; KO, knockout; RGC, retinal ganglion cell; WT, wild-type.](pbio.2003619.g004){#pbio.2003619.g004} We next proceeded to investigate the impact of genetic deletions of gpr91 and gpr99 on axon growth during development in vivo. Within 24 h of birth, pups from all 4 murine genotypes received a unilateral intraocular injection of CTb to label their retinal projections. At P5, immunohistological experiments revealed the effects of GPR91 and GPR99 on RGC axon development. Investigating RGC branch growth at the DTN, we showed a significant decrease in the collateral projection lengths of the KO animals compared to the control group ([Fig 4D and 4E](#pbio.2003619.g004){ref-type="fig"}). In addition, axon collateral density was significantly decreased in gpr91KO, gpr99KO, and, to a greater extent, in double-KO mice, compared to WT controls ([Fig 4F](#pbio.2003619.g004){ref-type="fig"}). These findings demonstrate---for the first time, to our knowledge---the essential role of GPR91 and GPR99 in the growth of RGC projections. Discussion {#sec007} ========== Most functional studies of GPR91 and GPR99, receptors of intermediates of carbohydrate metabolism, have been performed outside the central nervous system, primarily in the kidney and heart \[[@pbio.2003619.ref002],[@pbio.2003619.ref014],[@pbio.2003619.ref015]\]. In the present study, we showed that GPR91 and GPR99 are expressed on axonal and dendritic projections, GCs and filopodia of murine embryonic retinal explants, and on retinal projections and cell body of RGCs during the development of the retinothalamic pathway. We demonstrated that succinate and α-KG increase ERK~1/2~ phosphorylation, corroborating a large number of studies on signaling pathways triggered by GPR91 \[[@pbio.2003619.ref002],[@pbio.2003619.ref012],[@pbio.2003619.ref018],[@pbio.2003619.ref025]\]. Moreover, stimulation of both GPR91 and GPR99 resulted in the modulation of GC morphology and an increase in RGC axon growth in an ERK~1/2~-dependent manner. The increased GC size, number of filopodia, and growth of RGC axons following stimulation of GPR91 and GPR99 by succinate and α-KG, respectively, is the first report, to our knowledge, implicating these ligands and receptors in axon growth. Interestingly, the deletion of GPR91 completely blocked the effects of succinate but also partially abolished the effects observed with α-KG. Nevertheless, in double-KO animals, the effects of both succinate and α-KG were abrogated. These results tend to demonstrate that succinate's effects on RGC axon growth were mediated only through GPR91, while α-KG could, through an as-yet-unknown mechanism, activate both GPR91 and GPR99. A possible mechanism could be the conversion of α-KG into succinate, since α-KG is a precursor of succinate in the Krebs cycle. Moreover, our findings showed that GPR91 and GPR99, while having no effect on axon guidance, have complementary roles in RGC axon growth during development. These data are consistent with previous observations in which succinate, via GPR91, has shown highly proliferative and stimulating vascular effects in different tissues \[[@pbio.2003619.ref016],[@pbio.2003619.ref017]\], to promote chemotaxis \[[@pbio.2003619.ref019],[@pbio.2003619.ref030]\] and to potentiate the activation and aggregation of platelets \[[@pbio.2003619.ref018],[@pbio.2003619.ref031]\]. Axon guidance and angiogenesis share several fundamental challenges during the formation of their extensive networks. Tip cells---specialized endothelial cells at the end of each vessel sprout---are motile and dynamically extend long filopodia protrusions reminiscent of axonal GCs \[[@pbio.2003619.ref032]\]. In light of the spatiotemporal link between axon growth and angiogenesis, as well as the morphological similarities between endothelial tip cells and axonal GCs, the observed increase in the morphology of GC and neurite growth could be explained by a similar mechanism in the presence of succinate. As the only type of neuron that sends axons out of the retina, RGCs ensure the visual and cognitive processing of information from the outside world to the brain. A combination of intrinsic and extrinsic signals also plays an important role in driving the axons through the visual pathway via responsive GCs, which detect and effectively translate a multitude of external chemotactic cues. In the mouse, the axon decussation occurs at the level of the optic chiasm at around E14--16 \[[@pbio.2003619.ref033]\]. We observed that in WT, gpr91KO, or gpr99KO mice, the optic chiasm appeared relatively normal, as the majority of the axons at the midline crossed to project contralaterally. Our results suggest that in the mouse visual system, the absence of either GPR91 or GPR99 is insufficient to affect decussation. Moreover, neither GPR91 nor GPR99 activity at the GC modulated axon turning in an ex vivo experiment of retinal explants, since GCs are not attracted nor repelled in the presence of a succinate or α-KG microgradient, whereas succinate induced significant axon extension. Based on these results, succinate plays an essential role in axon growth by increasing axon motility, but succinate and α-KG do not affect GC and axon guidance. However, the visual projections of double-KO mice showed some mild abnormalities in axon guidance that could be explained by a compensatory effect between the two receptors, which would allow a rescue of this mild phenotype in gpr91KO or gpr99KO mice. Nevertheless, further experiments are needed to study this subtle defect in a more quantitative fashion in order to draw significant conclusions. In addition, our data show that deletion of either GPR91 or GPR99 in vivo did not affect target selection of retinal projections. Indeed, during perinatal development, RGC axons connect with multiple targets in the dLGN, sharing common terminal space, while RGC axons occupy distinct eye-dependent nonoverlapping regions of the dLGN in the adult rodent. Eye-specific segregation only occurs during postnatal development \[[@pbio.2003619.ref034]\]. Accordingly, a similar relative eye-specific segregation of retinal projections was observed in the adults of all 4 mouse genotypes. Thus, our in vivo results support previous ex vivo findings that GPR91 and GPR99 do not modulate RGC axon guidance and target selection during the establishment of the visual pathway. However, we demonstrated that TCA cycle intermediates induce axon growth in vivo during the development of the visual system, as intraocular injection of succinate and α-KG induced significant increases in RGC collateral axon projection length and branch number in the DTN. Moreover, accordingly, genetic interference with GPR91 or GPR99 activity profoundly affects retinal projection growth in the DTN. We showed a significant difference between WT, gpr91KO, and gpr99KO mice in axon projection length and branching at the DTN. Furthermore, the relative lack of growth of retinal projections in double-KO mice demonstrates the fundamental role played by GPR91 and GPR99 during RGC axon growth. Nonetheless, these in vivo experiments do not conclude that the receptors involved in the growth-promoting actions of intermediates of carbohydrate metabolism are only those expressed at the GCs but could also be, to some extent, those expressed throughout the projections or on the cell body of RGCs as well. The levels of intermediates of carbohydrate metabolism adapt depending on tissue needs and the conditions in the surrounding regions. Investigating RGC projections and GC actions in the developing visual system faces technical limitations regarding intermediates of carbohydrate metabolism dosing. The amount of tissue needed (and its isolation) from mouse embryos or hamster newborn pups does not allow detection of metabolites due to the technique sensitivity and the rapid turnover of the metabolites. Nevertheless, based on previous published data and our own findings, we sought to avoid nonspecific responses by determining the lowest responsive doses for succinate and α-KG in our system, even if the physiological levels could not be measured \[[@pbio.2003619.ref002],[@pbio.2003619.ref003],[@pbio.2003619.ref016]--[@pbio.2003619.ref018]\]. In summary, this study demonstrates---for the first time, to our knowledge---a role for the intermediates of carbohydrate metabolism succinate and α-KG and their respective receptor GPR91 and GPR99 in axon growth during development in vivo. These receptors mediate axon growth in an ERK~1/2~-dependent manner, although succinate and α-KG have no effect on axon guidance. Moreover, these findings suggest a potential link between mitochondria and axon growth in development, outside the strict production of energy. This study not only demonstrates a new role for TCA cycle intermediates in the visual system development but also provides a foundation for the investigation of metabolite receptors in the visual, central, and peripheral nervous system development. This novel concept also provides new avenues for the elaboration of effective therapies aimed at the development and regeneration of the nervous system. Materials and methods {#sec008} ===================== Ethics statement {#sec009} ---------------- All experimental procedures were approved by the Animal Care Committee of Sainte-Justine's Hospital Research Center or the relevant University of Montreal animal care committee's regulations and were conducted in accordance with the Association for Research in Vision and Ophthalmology statement regarding the use of animals in ophthalmic and vision research and the guidelines established by the Canadian Council on Animal Care. Animals {#sec010} ------- The C57BL/6 WT control mice were purchased from Jackson Laboratory. Syrian golden hamsters (Charles River Laboratories, Saint-Constant, Canada) were used in this study. Generation of gpr91KO mice {#sec011} ---------------------------- Sucnr1KO mice, generated by Deltagen through partial replacement of exon 2 (5'-GGCTACCTCTTCTGCAT-3') with a lacZ-neomycin cassette, were generously provided by Dr. José M. Carbadillo at Norvartis Institutes for Biomedical Research, Vienna, Austria \[[@pbio.2003619.ref019]\]. As described by Rubic and colleagues in 2008, correctly targeted 129/OlaHsd embryonic stem cells were used for the generation of chimeric mice, which were crossed with C57BL/6 (called "WT" here). F1 mice with germline transmission of the mutated gene were further backcrossed with WT mice for 10 generations (in specific pathogen-free conditions at the Novartis Institutes for Biomedical Research, Vienna) before being intercrossed to produce homozygous gpr91KO mice. gpr91KO mice were healthy and bred normally when maintained in specific pathogen-free conditions. All experiments in the production of the gpr91KO mice were conducted in accordance with Austrian Law on Animal Experimentation and the Novartis Animal Welfare Policy. All procedures were approved by the local government and the animal care and user committee of the Novartis Institutes for Biomedical Research, Vienna. Generation of gpr99KO mice {#sec012} ---------------------------- Heterozygous (GPR99+/−) mice with mixed genetic background (C57BL/6J− Tyrc-Brd x 129 Sv/EvBrd) were developed and generously provided by Lexicon Pharmaceuticals Incorporated (The Woodlands, TX). The full-length gpr99 gene was removed by homologous recombination as the PCR-generated selection cassette was introduced in a murine genomic clone by yeast recombination, followed by the electroporation of the linearized targeting vector in 129 Sv/EvBrd embryonic stem cells. In selected clones, gpr99 deletion was confirmed by Southern hybridization followed by their injection into C57BL/6J-Tyrc-Brd blastocysts. To generate F1 heterozygous offspring, the resulting chimeras were backcrossed to C57BL/6J-Tyrc-Brd. Heterozygous mice were intercrossed to generate WT control (gpr99WT), homozygous-null (gpr99KO), and heterozygous littermates, consistent with Mendelian ratios. The resulting homozygous-null gpr99KO mice were backcrossed onto the C57BL/6 background with C57BL/6 obtained from Jackson Laboratory (Connecticut, USA) for 10 generations in CHU Sainte-Justine's Research Center animal facility before using them in experiments. The gpr99KO mice were viable, healthy, and bred normally when maintained in specific pathogen-free conditions. Generation of gpr91KO / gpr99KO double-KO mice {#sec013} -------------------------------------------------- gpr99KO / gpr91KO mice (double KO) were generated by crossing gpr99KO and gpr91KO mice to produce gpr99+/− / gpr91+/− (double-heterozygous) parents. The double-heterozygous parents were then crossed together until we obtained double-KO gpr99KO / gpr91KO (1:16 pups according to Punnett Square) male and female mice that were then crossed together to obtain a stable double-KO mouse lineage. The double-KO mice were viable, healthy, and bred normally when maintained in specific pathogen-free conditions. Genotypic screening {#sec014} ------------------- Mice were genotyped by PCR reactions of tail genomic DNA using specific primers for either the WT or mutant allele. For GPR91 mice, the primer pair WT-F: 5′-GTTCATTTTTGGACTGCTTGGG-3′ and WT-R: 5′-AATGGCAAATTCCTTCTTTTGTAGA-3′ generated a GPR91-specific fragment only present in the WT allele, while the primer pair KO-F: 5′- GGCACATATCGGTTGCTTATACAGA-3′ and KO-R: 5′- GGGTGGGATTAGATAAATGCCTGCTCT-3′ amplified a fragment specific to the selection cassette of the gpr91KO mutant allele. For GPR99 mice, a GPR99-specific fragment present in the WT but absent in the mutant allele was generated using the specific primer pair UTT069-21 (5′-GAGCCATGATTGAGCCACTG-3′) and UTT069-25 (5′-CACCACTGGCATAGTAATGG-3′). Another primer pair amplified a fragment specific to the selection cassette of the gpr99KO mutant allele: UTT069-3 (5′-CAGAGCCATGCCTACGAG-3′) and GT (5′-CCCTAGGAATGCTCGTCAAGA-3′). For double-KO mice, all pairs of primers were used (4 reactions) to determine whether both genetic modifications were present. Reagents {#sec015} -------- BSA, ciliary neurotrophic factor, DNase, forskolin, Hoechst 33258, insulin, laminin, poly-D-lysine, progesterone, selenium, putrescine, succinate, α-KG, trypsin, and triiodothyronine were purchased from Sigma Aldrich (Oakville, ON, Canada). B27, N2, Dulbecco's phosphate-buffered saline, FBS, glutamine, Neurobasal medium, penicillin-streptomycin, Minimum Essential Medium Eagle Spinner Modification (S-MEM), and sodium pyruvate were purchased from Life Technologies (Burlington, ON, Canada). The standard donkey and goat sera were from Jackson ImmunoResearch (West Grove, PA, USA). ERK~1/2~ inhibitor (CI-1040) was obtained from Selleck Chemicals (Houston, TX, USA). LNAC was acquired from EMD (La Jolla, CA, USA). The CTb was from List Biological Laboratories (Campbell, CA, USA). Triton X-100 was purchased from US Biological Life Sciences (Salem, MA, USA). DiI stain was obtained from Molecular probes (Eugene, OR, USA). Tissue preparation for immunohistochemistry {#sec016} ------------------------------------------- Adult mice and P1 hamsters were euthanized by an overdose of isoflurane. Transcardiac perfusion was conducted with phosphate-buffered 0.9% saline (PBS; 0.1 M, pH 7.4), followed by 4% formaldehyde in PBS, until the head was fixed. The nasal part of the eyes of murine embryos and adult mice was marked with a suture and removed. Two small holes were made in the cornea before a first postfixation step in formaldehyde for a period of 30 min. The cornea and lens were removed, and the eyecups were postfixed for 30 min in formaldehyde. The eyecups were then washed in PBS, cryoprotected in 30% sucrose overnight, embedded in NEG 50 tissue Embedding Media (Thermo Fisher Scientific Burlington, ON, Canada), flash-frozen, and kept at −80 °C. Sections (14-μm thick) were cut with a cryostat (Leica Microsystems, Concord, ON, Canada) and placed on gelatin/chromium-coated slides. Immunohistochemistry {#sec017} -------------------- Retinal sections were washed in 0.1 M PBS, postfixed for 5 min in a 70% solution of ethanol, rinsed in 0.03% Triton X-100 in PBS, and blocked in 10% normal donkey serum and 0.5% Triton X-100 in PBS for 1 h. The sections were then incubated overnight with antibodies against GPR91 or GPR99. The antibody Brn-3a was also used as a specific marker for RGCs. After incubation with the primary antibodies, the sections were washed in PBS, blocked for 30 min, and incubated for 1 h with the secondary antibodies Alexa Fluor 647 donkey anti-rabbit and Alexa Fluor 488 donkey anti-mouse. After washing, the sections were mounted using a homemade PVA-Dabco medium. Antibody characterization {#sec018} ------------------------- The specifications of all the antibodies used in this study are detailed in [S1 Table](#pbio.2003619.s008){ref-type="supplementary-material"}. Confocal microscopy {#sec019} ------------------- Images of the central retina (within 200 μm of the optic nerve head) were taken using a laser scanning confocal microscope (TCS SP2, Leica Microsystems) with a 40X (NA: 1.25) oil immersion objective and 488 and 633 nm lasers. Image stacks (1,024 × 1,024 pixels × 0.5 μm per stack) were captured with a frame average of 3 using the LCS software (version 2.6.1; Leica Microsystems). The stacks were taken sequentially and in distant wavelengths to ensure no "bleed through" between channels and were collapsed into projection images. All images in which labeling intensities were compared were obtained under identical conditions of gain intensity. Because gray-scale photographs provide better contrast and more detail, individual channels are presented in gray scale, and the merged images are presented in color. Retinal explant culture {#sec020} ----------------------- The retinas were isolated from E14/15 mouse embryos, dissected into small segments in ice-cold Dulbecco's phosphate-buffered saline, and plated on 12-mm glass coverslips previously coated with poly-D-Lysine (20 μg/ml) and laminin (5 μg/ml) in 24-well plates. The explants were cultured in Neurobasal supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, 5 μg/ml LNAC, 1% B27, 40 ng/ml selenium, 16 μg/ml putrescine, 0.04 ng/ml triiodo-thyronine, 100 μg/ml transferrin, 60 ng/ml progesterone, 100 μg/ml BSA, 1 mM sodium pyruvate, 2 mM glutamine, 10 ng/ml ciliary neurotrophic factor, 5 μg/ml insulin, and 10 μM forskolin at 37 °C and 5% CO~2~. At 0 DIV, 1 h following plating, the explants were treated for 15 h for projection analysis or for 1 h at 1 DIV for GC analysis. The photomicrographs were taken using an Olympus IX71 microscope (Olympus, Markham, ON, Canada) and analyzed with Image-Pro Plus 5.1 software (Media Cybernetics, Bethesda, MD, USA). The total length of axon bundles was quantified and expressed as the mean ± SEM. Statistical significance of differences between means was evaluated by analysis of variance (ANOVA) with Bonferroni's post-hoc test (Systat Software Inc, Chicago, IL, USA). Primary neuron culture {#sec021} ---------------------- Primary cortical neurons were used in this study because of the large number of neurons that can easily be cultured and harvested for biochemical assays, which is hardly possible with RGCs. C57BL/6 WT, gpr91KO, gpr99KO, and double-KO pregnant mice were used. Brains from E14/15 embryos were dissected, and the superior layer of each cortex was isolated and transferred in 2 ml S-MEM containing 2.5% trypsin and 2 mg/ml DNase and incubated at 37 °C for 15 min. The pellet was transferred into 10 ml S-MEM with 10% FBS and stored at 4 °C. After centrifugation, the pellet was again transferred in 2 ml S-MEM supplemented with 10% FBS and triturated 3 to 4 times. The supernatant was transferred in 10 ml Neurobasal medium. Dissociated neurons were counted and plated at 50,000 cells per well on 12 mm glass coverslips previously coated with poly-D-lysine (20 μg/ml) for immunocytochemistry or at 250,000 cells per 35 mm petri dish for western blot. Neurons were cultured for 2 d in Neurobasal medium supplemented with 1% B-27, 100 U/ml penicillin, 100 μg/ml streptomycin, 0.25% N2, and 0.5 mM glutamine. They were then treated with either a GPR91 agonist (100 μM succinate), GPR99 agonist (200 μM αKG), or ERK~1/2~ inhibitor (20 μM CI-1040) for 1 h to study GC morphology or 2, 5, and 15 min for ERK~1/2~ quantification using western blot analysis. Determination of cell viability {#sec022} ------------------------------- LIVE/DEAD cell viability assay: Cell viability was assessed with the LIVE/DEAD assay using an ethidium homodimer/calcein acetoxy methyl ester (L-3224, Molecular Probes, Eugene, OR, USA) combination of vital dyes, as previously described \[[@pbio.2003619.ref035],[@pbio.2003619.ref036]\]. Staurosporine (5 μM), an inducer of apoptotic cell death, was used as a positive control \[[@pbio.2003619.ref037]\]. Immunocytochemistry {#sec023} ------------------- After treatment, retinal explants and primary cortical neuron cultures were washed with PBS (pH 7.4), fixed in 4% formaldehyde (pH 7.4), and blocked with 2% normal goat serum (NGS) and 2% BSA in PBS containing 0.1% Tween 20 (pH 7.4) for 30 min at room temperature. The samples were then incubated overnight at 4 °C in a blocking solution containing anti-GAP-43, anti-GPR91, anti-GPR99, anti-MAP2, or anti-NFM. The following day, the samples were washed and labeled with Alexa Fluor 488 and 555 secondary antibodies and Hoechst 33258 (1:10,000), and the coverslips were mounted with a homemade PVA-Dabco medium \[[@pbio.2003619.ref038]\]. Western blot analysis {#sec024} --------------------- Primary cortical neurons were cultured for 2 DIVs at a density of approximately 250,000 cells/dish in 35 mm poly-D-lysine-coated petri dishes. Following treatment, neurons were washed once with ice-cold PBS (pH 7.4) and then lysed with Laemmli sample buffer. Thirty micrograms of protein/sample of the homogenate were resolved with 12% SDS-polyacrylamide gel electrophoresis, transferred onto a nitrocellulose membrane, blocked with 5% BSA, and incubated overnight with antibodies directed against ERK~1/2~, p-ERK~1/2~, and β-actin, the latter serving as a loading control. The blots were exposed to the appropriate HRP-coupled secondary antibodies (Jackson Immunoresearch Laboratories, West Grove, PA, USA). Detection was performed using homemade enhanced chemiluminescence western blotting detection reagent (final concentrations: 2.5 mM luminol, 0.4 mM p-coumaric acid, 0.1 M Tris-HCl \[pH 8.5\], 0.018% H~2~O~2~). GC behavior assay {#sec025} ----------------- Embryonic retinal explants were cultured on a coverglass in a borosilicate chamber (Lab-Tek; Rochester, NY, USA) for 2 DIVs and placed in an incubator mounted on an inverted microscope (Olympus IX71). They were maintained at 37 °C and 5% CO~2~ with a live cell chamber (Neve Bioscience, Camp Hill, PA, USA) throughout the whole experiment. A microgradient was created using a Picoplus micro-injector (Harvard Apparatus, St-Laurent, QC, Canada). Glass micropipettes with a tip of 2--3 μm diameter were positioned at 45° and at 100 μm away from the GC of interest, as described previously \[[@pbio.2003619.ref008],[@pbio.2003619.ref010],[@pbio.2003619.ref011]\]. Intraocular injections {#sec026} ---------------------- Syrian golden hamsters (Charles River) were used for investigating the in vivo implication of succinate/GPR91 and α-KG/GPR99 in RGC projection growth during postnatal development. At P1, 24 h after birth, anesthetized hamsters received a unilateral injection of 2 μl solution of CTb with either 0.9% saline solution, succinate (100 mM), or α-KG (200 mM). Briefly, under an operating microscope, a small incision was made in the eyelids to access the right eye. The injections were administered using a glass micropipette attached to a 10 μl Hamilton syringe. The micropipette was carefully inserted into the vitreous at an angle to avoid damage to the lens. Following the injection, the eyelids were closed with surgical glue (Vetbond; 3M). At P5, 4 d after the injection, hamsters were anesthetized and perfused transcardially with 0.1 M PBS, pH 7.4, followed by 4% PFA in PBS. The brains were removed, postfixed overnight at 4 °C and cryoprotected with sucrose. Then, brains were frozen and kept at −80 °C until processing by immunohistochemistry according to a protocol previously described by Argaw and colleagues in 2011 \[[@pbio.2003619.ref008]\]. Briefly, 40 μm---thick coronal sections of tissue were incubated in 90% methanol and 0.3% H~2~O~2~ in 0.1 [m]{.smallcaps} PBS, pH 7.4, for 20 min. They were then rinsed and incubated in 0.1 M glycine/PBS for 30 min, followed by an overnight incubation (4 °C) in PBS containing 4% NDS, 2.5% BSA, and 1% Triton X-100. The sections were subsequently rinsed and immersed for 48 h at room temperature in a solution containing goat anti-CTb diluted 1:4,000 in PBS with 2% NDS, 2.5% BSA, and 2% Triton X-100. Afterward, the sections were rinsed and incubated in 2% NDS and 2.5% BSA/PBS for 10 min. This was followed by a 1 h incubation in donkey anti-goat biotinylated secondary antibody diluted 1:200 in PBS with 2% NDS, 2.5% BSA, and 1% Triton X-100. Tissues were rinsed, incubated in 2% NDS and 2.5% BSA in PBS for 10 min, and subsequently processed with an avidin-biotin-peroxidase complex ABC Kit (diluted 1:100 in PBS) for 1 h in the dark at room temperature. The sections were then rinsed and preincubated in 3, 3′-diaminobenzidine tetrahydrochloride (DAB) in PBS for 5 min. The peroxidase reaction product was visualized by adding 0.004% H~2~O~2~ to the DAB solution for 2--4 min. Sections were finally washed 5 times (1 min each) with PBS, mounted on gelatin-chromium alum-subbed slides, air-dried, dehydrated in ethanol, cleared in xylenes, and mounted on coverslips with Depex (EMS). Lipophilic dye tracing and tissue clearing {#sec027} ------------------------------------------ After 14--15 d of gestation, pregnant mice (WT, gpr91KO, gpr99KO, and double KO) were euthanized, and the embryos were removed. The lambdoid sutures of the embryos were incised, and the occipital bones were removed to expose the brain to the fixative (4% formaldehyde), where they were placed for 1 wk at 4 °C until tracing with DiI. For complete optic nerve labeling, 1 eye of each embryo was enucleated and crystals of DiI implanted unilaterally into the optic disk. Embryos were incubated at 37 °C in 4% formaldehyde for 7 d. Tissue clearing was performed according to Hama and colleagues (2011) \[[@pbio.2003619.ref039]\]. Briefly, embryos were incubated for 2 d in Scale A2 solution (4 M urea, 10% glycerol, 0.1% Triton X-100, in water) followed by 2 d in Scale B4 solution (8 M urea, 0.1% Triton X-100, in water) and then to a fresh Scale A2 solution for 1 wk to complete the clearing \[[@pbio.2003619.ref039]\]. The brains were then carefully removed with their optic nerves, and the proximal visual system was imaged with a fluorescence microscope to allow the observation of subtle guidance defects at the optic chiasm. Tracing of RGC axons {#sec028} -------------------- For eye-specific segregation studies in the dLGN, C57BL/6 WT, gpr91KO, gpr99KO, and double-KO adult mice received an intraocular injection of CTb conjugated to Alexa Fluor 555 into the left eye and CTb coupled to Alexa Fluor 488 into the right eye (2 μl; 0.5% in sterile saline). Four days after the injection, the animals were anesthetized and perfused transcardially with 0.1 M PBS (pH 7.4) followed by 4% formaldehyde. The brains were removed, postfixed overnight at 4 °C, cryoprotected, frozen, and kept at −80 °C. Retinal projections marked with the CTb were visualized on brain sections washed 5 times (1 min each) with PBS, mounted on gelatin-chromium alum-subbed slides, air-dried, and mounted on coverslips with DEPEX (EMS, Hatfield, PA, USA). In vivo quantification method {#sec029} ----------------------------- The photomicrographs of the optic chiasm were taken with an IX71 microscope (Olympus, Richmond Hill, ON, Canada), an Evolution VF camera (Media Cybernetics, Warrendale, PA, USA) and Image-Pro Plus 5.1 image analysis software. Universal gains and exposures were established for each labeling. Raw images of the dLGN were imported to MATLAB (Natick, MA, USA), and an area of interest comprising the dLGN was cropped, excluding the ventral lateral geniculate nucleus and the intergeniculate leaflet. Then, the degree of left and right eye projection overlap was quantified using an established multithreshold method of analysis \[[@pbio.2003619.ref040]--[@pbio.2003619.ref042]\]. This approach allows for a better analysis of overlapping regions independent of the threshold. For these experiments, an observer "blind" to the experimental conditions to avoid any bias performed the quantification. Values are expressed as the means ± SEM. The significance of differences between means was evaluated by Student t test analysis (Systat). To assess axon growth in vivo, photomicrographs of the DTN of mice and P5 hamsters were taken with a microscope (Leica Microsystems, Concord, ON, Canada) coupled to an Evolution VF camera (Media Cybernetics). The images were quantified using Image-Pro Plus 5.1 software. The growth of axon branches was quantified on consecutive photomicrographs of coronal slices of brain tissue comprising the DTN. On each photomicrograph, the distance between the lateral border of the DTN and the tips of the longest axon branches was measured. To take into account brain size differences, axon branch lengths were normalized with the interthalamic distance (distance between the right and left lateral borders of the thalamus; see [S6A Fig](#pbio.2003619.s007){ref-type="supplementary-material"} for a schematic representation of such quantification). Axon collateral number was quantified on consecutive photomicrographs comprising the DTN using an adaptation of the Sholl technique \[[@pbio.2003619.ref043]\], as described by Duff and colleagues in 2013 \[[@pbio.2003619.ref011]\] and illustrated in [S6B Fig](#pbio.2003619.s007){ref-type="supplementary-material"}. Values are expressed as the means ± SEM. Statistical analysis {#sec030} -------------------- The significance of differences between means was evaluated by ANOVA with Bonferroni's post-hoc test (Systat). Supporting information {#sec031} ====================== ###### Underlying data for Figs [2](#pbio.2003619.g002){ref-type="fig"}, [3](#pbio.2003619.g003){ref-type="fig"} and [4](#pbio.2003619.g004){ref-type="fig"}, [S2](#pbio.2003619.s003){ref-type="supplementary-material"}, [S4](#pbio.2003619.s005){ref-type="supplementary-material"} and [S5](#pbio.2003619.s006){ref-type="supplementary-material"} Figs. (XLSX) ###### Click here for additional data file. ###### Specificity of the antibodies against GPR91 and GPR99 in the retina. (A) Expression of GPR91 protein in the WT adult mouse retina with lack of fluorescence in the gpr91KO mouse (C). (B & D) GPR99 immunoreactivity in the retina of adult WT and gpr99KO mice. (E & G) Expression of GPR91 and (F & H) GPR99 in retinal sections of E14/15 WT, gpr91KO, and gpr99KO murine embryos. Scale bars: 75 μm (A-H). Expression of GPR91 and GPR99 in retinal explants, GCs, and neurites of E14/15 WT (I-L), gpr91KO, and gpr99KO murine embryos (M-P). Scale bars: 100 μm (I & K); 10 μm (J & L). Expression of GPR91 and GPR99 proteins in retinal sections of P1 hamster pups (Q, R). Scale bars: 10 μm. E14/15, embryonic day 14/15; GC, growth cone; GCFL, ganglion cell fiber layer; GCL, ganglion cell layer; KO, knockout; NBL, neuroblast layer; WT, wild-type. (TIF) ###### Click here for additional data file. ###### Succinate and α-KG modulate GC morphology and filopodia number of retinal explants and cortical neurons via GPR91 and GPR99. (A-D) Additional representative examples of E14/15 GCs of RGC projections from WT, gpr91KO, gpr99KO, and double-KO mouse embryos (E) Photomicrographs of E14/15 GCs of cortical neurons from WT, gpr91KO, gpr99KO, and double-KO mouse embryos after a 1 h treatment with succinate (100 μM) or α-KG (200 μM). (F) Analysis of the GC surface area (N = 99--271 per condition) and (G) filopodia number (N = 177--187 per condition) of cortical neurons from WT, gpr91KO, gpr99KO, and double-KO mice, following a 1 h treatment with succinate (100 μM) or α-KG (200 μM). Scale bars: 5 μm. Values are presented as the means ± SEM. \# indicates significant changes compared to WT in F and G; p \< 0.001. Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. α-KG, α-ketoglutarate; E14/15, embryonic day 14/15; GC, growth cone; KO, knockout; RGC, retinal ganglion cell; WT, wild-type. (TIF) ###### Click here for additional data file. ###### GPR91 and GPR99 ligands have no effect on RGC viability in vitro. Photomicrographs of 1 DIV embryonic mouse retinal explants (A) and 1 DIV embryonic cortical neurons (B) taken at t = 15 h after LIVE/DEAD assay experiments, in the presence or absence of succinate (100 μM), α-KG (200 μM), CI-1040 (1 μM), or Staurosporine (5 μM; positive control). CI-1040, succinate, and α-KG produced no effect on viability compared to the control. Staurosporine induced RGC cell death. Green: living cells, Red: dead cells. Scale bars: 100 μm. DIV, day in vitro; RGC, retinal ganglion cell. (TIF) ###### Click here for additional data file. ###### GPR91 and GPR99 are not involved in guidance or target selection in vivo. (A, B) Lipophilic dye (DiI) tracings of the proximal visual pathway in E14/15 mouse embryos imaged with a dissecting (A) or a fluorescence microscope (B) (the red asterisk marks the location of the optic chiasm). (A, B) The visual pathway labeled with DiI in WT, gpr91KO, gpr99KO, and double-KO mice. The arrowhead shows the aberrant projections prior to crossing the optic chiasm, and the thin arrow highlights the fibers that have extended out of the contralateral optic tract toward the contralateral eye. (C) Retinogeniculate projection patterns visualized following different fluorescent CTb injections into both eyes of WT, gpr91KO, gpr99KO, and double-KO adult mice. (D) Quantification of the dLGN area receiving overlapping inputs (N = 21--42 per condition). Data are presented as the means ± SD. Scale bars: 100 μm. Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. Contra, contralateral pathway; CTb, cholera toxin subunit B; dLGN, dorsal lateral geniculate nucleus; E14/15, embryonic day 14/15; Ipsi, ipsilateral pathway; KO, knockout; ON, optic nerve; OT, optic tract; WT, wild-type. (TIF) ###### Click here for additional data file. ###### GPR91 and GPR99 ligands have no effect on the turning of GCs from RGCs ex vivo. (A) Photomicrographs of time-lapse microscopy from 1 DIV mouse retinal explant GCs taken at t = 0 min and t = 60 min during GC turning assay experiments, in the presence or absence of GPR91 or GPR99 agonists. Black arrows indicate the direction of the microgradient, while blue arrowheads indicate initial GC position. Green arrowheads show the GC position following neurite growth. (B-D) Superimposed RGC axon trajectories over the 60-min observation period. Succinate (100 μM) increased axon growth but had no effect on the turning. α-KG (200 μM) produced no significant change on GC behavior. (E) Quantification of neurite elongation and GC turning responses following drug stimulation (N = 13--16 per condition). Scale bars: 40 μm (A). Values are presented as the means ± SEM; \* indicates a significant change compared to the control vehicle in (E); P \<0.0001. Underlying data can be found in [S1 Data](#pbio.2003619.s001){ref-type="supplementary-material"}. α-KG, α-ketoglutarate; DIV, day in vitro; GC, growth cone; RGC, retinal ganglion cell. (TIF) ###### Click here for additional data file. ###### Quantification method of retinal projection branch length and number in the DTN. Schematic representation of the method used to measure retinal projection branch length (A) and the number of retinal axon branches (B) in the DTN. Arrowed dotted lines indicate the distance between the border of the thalamus and the end of the farthest projections (A). DTN, dorsal terminal nucleus. (TIF) ###### Click here for additional data file. ###### (TIF) ###### Click here for additional data file. ANOVA : analysis of variance CTb : cholera toxin subunit B DAB : 3, 3′-diaminobenzidine tetrahydrochloride DiI : (DiIC18\[3\] \[1,1'-dioctadecyl-3,3,3',3'-tetramethylindocarbocyanine perchlorate\]) DIV : day in vitro dLGN : dorsal lateral geniculate nucleus DTN : dorsal terminal nucleus E14/15 : embryonic day 14/15 ERK~1/2~ : extracellular signal--regulated kinases 1 and 2 GC : growth cone GCFL : ganglion cell fiber layer GPCR : G-protein-coupled receptor KO : knockout MAPK : mitogen-activated protein kinase NGS : normal goat serum Oxgr1 : 2-oxoglutarate receptor 1 RGC : retinal ganglion cell S-MEM : Minimum Essential Medium Eagle Spinner Modification Sucnr1 : succinate receptor 1 TCA : tricarboxylic acid WT : wild-type α-KG : α-ketoglutarate [^1]: The authors have declared that no competing interests exist.	{ "pile_set_name": "PubMed Central" }
Related literature {#sec1} ==================== For literature on the UV spectroscopy of proteins, see: Demchenko (1986[@bb7]). For the different polymorphic forms of fumaric acid, see: Reis & Schneider (1928[@bb15]); Yardley (1925[@bb18]); Bednowitz & Post (1966[@bb3]). For the nonlinear optical properties of organic molecules, see: Chemla & Zyss (1987[@bb6]); Zyss & Ledoux (1994[@bb19]); Zyss & Nicoud (1996[@bb20]). For the common conformations of [l]{.smallcaps}-tryptophan, see: Bye et al. (1973[@bb5]); Bakke & Mostad (1980[@bb2]). The bond lengths and angles in [l]{.smallcaps}-trypophan, see, for example: Gorbitz (2006[@bb9]); Gorbitz et al. (2012[@bb10]), and for fumaric acid, see: Goswami et al. (1999[@bb11]). For the crystal structure of [l]{.smallcaps}-tryptophan formic acid solvate, see: Hubschle et al. (2002[@bb13]). For details of the Cambridge Structural Database, see: Groom & Allen (2014[@bb12]). Experimental {#sec2} ============== Crystal data {#sec2.1} -------------- C~11~H~12~N~2~O~2~·C~4~H~4~O~4~·H~2~OM ~r~ = 338.31Monoclinic,a = 11.3928 (8) Åb = 6.6476 (4) Åc = 21.4219 (13) Åβ = 95.801 (3)°V = 1614.07 (18) Å^3^Z = 4Mo Kα radiationμ = 0.11 mm^−1^T = 296 K0.30 × 0.20 × 0.20 mm Data collection {#sec2.2} ----------------- Bruker Kappa APEXII CCD diffractometerAbsorption correction: multi-scan (SADABS; Bruker, 2004[@bb4]) T ~min~ = 0.898, T ~max~ = 0.97810737 measured reflections3157 independent reflections2731 reflections with I \> 2σ(I)R ~int~ = 0.025 Refinement {#sec2.3} ------------ R\[F ^2^ \> 2σ(F ^2^)\] = 0.034wR(F ^2^) = 0.084S = 1.043157 reflections242 parameters4 restraintsH atoms treated by a mixture of independent and constrained refinementΔρ~max~ = 0.18 e Å^−3^Δρ~min~ = −0.15 e Å^−3^ {#d5e709} Data collection: APEX2 (Bruker, 2004[@bb4]); cell refinement: APEX2 and SAINT (Bruker, 2004[@bb4]); data reduction: SAINT and XPREP (Bruker, 2004[@bb4]); program(s) used to solve structure: SIR92 (Altomare et al., 1994[@bb1]); program(s) used to refine structure: SHELXL2014 (Sheldrick, 2015[@bb16]); molecular graphics: ORTEP-3 for Windows (Farrugia, 2012[@bb8]) and Mercury (Macrae et al., 2008[@bb14]); software used to prepare material for publication: SHELXL2014 and PLATON (Spek, 2009[@bb17]). Supplementary Material ====================== Crystal structure: contains datablock(s) I, global. DOI: [10.1107/S205698901501484X/su5176sup1.cif](http://dx.doi.org/10.1107/S205698901501484X/su5176sup1.cif) Structure factors: contains datablock(s) I. DOI: [10.1107/S205698901501484X/su5176Isup2.hkl](http://dx.doi.org/10.1107/S205698901501484X/su5176Isup2.hkl) ###### Click here for additional data file. Supporting information file. DOI: [10.1107/S205698901501484X/su5176Isup3.cml](http://dx.doi.org/10.1107/S205698901501484X/su5176Isup3.cml) ###### Click here for additional data file. . DOI: [10.1107/S205698901501484X/su5176fig1.tif](http://dx.doi.org/10.1107/S205698901501484X/su5176fig1.tif) The molecular structure of the title compound, with atom labelling. Displacement ellipsoids are drawn at the 40% probability level. ###### Click here for additional data file. a . DOI: [10.1107/S205698901501484X/su5176fig2.tif](http://dx.doi.org/10.1107/S205698901501484X/su5176fig2.tif) The crystal packing of the title compound, viewed along the a axis. The hydrogen bonds are shown as dashed lines (see Table 1 for details). ###### Click here for additional data file. b . DOI: [10.1107/S205698901501484X/su5176fig3.tif](http://dx.doi.org/10.1107/S205698901501484X/su5176fig3.tif) The crystal packing of the title compound, viewed along the b axis. The hydrogen bonds are shown as dashed lines (see Table 1 for details). CCDC reference: [1417535](http://scripts.iucr.org/cgi-bin/cr.cgi?rm=csd&csdid=1417535) Additional supporting information: [crystallographic information](http://scripts.iucr.org/cgi-bin/sendsupfiles?su5176&file=su5176sup0.html&mime=text/html); [3D view](http://scripts.iucr.org/cgi-bin/sendcif?su5176sup1&Qmime=cif); [checkCIF report](http://scripts.iucr.org/cgi-bin/paper?su5176&checkcif=yes) Supporting information for this paper is available from the IUCr electronic archives (Reference: [SU5176](http://scripts.iucr.org/cgi-bin/sendsup?su5176)). The scientific support extended by the Sophisticated Analytical Instruments Facility, Indian Institute of Technology IITM, Chennai, in solving the crystal structure is greatly appreciated. The authors personally thank Professor E. M. Subramanian, retired Professor of Chemistry, Pachayappas College, Kanchipuram, Tamilnadu, for valuable suggestions. S1. Comment {#d1e22} =========== Natural aromatic amino acids, particularly tryptophan, have near UV absorption and emission properties which are utilized extensively in solution phase investigations of structure-function relationships (Demchenko, 1986). Fumaric acid is known to exist in two different polymorphic forms viz. cis and trans ( Reis & Schneider, 1928; Yardley, 1925; Bednowitz & Post, 1966). In conjunction with our ongoing work on non-linear optical organic crystals among the 20 naturally occurring amino acids, we have directed our interest to tryptophan (Trp), one of the essential amino acids for humans. The non-linear optical properties of organic molecules and crystals have been reviewed by Zyss (Chemla & Zyss, 1987; Zyss & Ledoux, 1994; Zyss & Nicoud, 1996). For similar properties and most the common confirmations of L-tryptophan have been reported (Bye et al., 1973; Bakke & Mostad, 1980). Compared with other amino acids, there are less than 30 tryptophan structures listed in the Cambridge Structural Database (Groom & Allen, 2014), due to the difficulty of obtaining good optical quality crystals; as noted by (Hubschle et al., 2002) who studied the crystal structure of L-tryptophan formic acid solvate. We successfully obtained good quality hard golden-yellow single crystals of L-tryptophan fumaric acid monohydrate, and we report herein on its synthesis and crystal structure. In the title compound, Fig. 1, L-tryptophan is zwitterionic, as are most amino acids in the solid state, and fumaric acid is neutral. The bond lengths and angles in L-trypophan and fumaric acid are similar to those reported previously (Gorbitz, 2006; Gorbitz et al., 2012; Goswami et al., 1999). In the crystal, the three components are linked by a series of O---H···O, N---H···O and C---H···O hydrogen bonds forming slabs lying parallel to (001); Table 1 and Fig. 2. The slabs are connected by O---H···O hydrogen bonds, involving inversion related fumaric acid groups, leading to the formation of a three-dimensional structure; Table 1 and Fig. 3. S2. Synthesis and crystallization {#d1e72} ================================= An aqueous solution of L-tryptophan and fumaric acid in a 1:1 stoichiometric ratio was stirred at room temperature for 6 h. The resulting yellow solution was filtered and kept in a Petri dish. Yellow prismatic-shaped hard crystals suitable for X-ray analysis were obtained over a period of 5 days. S3. Refinement {#d1e89} ============== Crystal data, data collection and structure refinement details are summarized in Table 2. The N-bound, acid and water H atoms were located in a difference Fourier map. The NH and NH~3~ H atoms were freely refined. The water H atoms were refined with distance restraints: O---H = 0.86 (2) Å, H···H = 1.388 (20) Å with U~iso~(H) = 1.5U~eq~(O). The acid (OH) H atoms and the C-bound H atoms were included in calculated positions and treated as riding atoms: U~iso~(H) = 1.5U~eq~(O,C) for OH and methyl H atoms and 1.2U~eq~(C) for other H atoms. Figures ======= ![The molecular structure of the title compound, with atom labelling. Displacement ellipsoids are drawn at the 40% probability level.](e-71-0o661-fig1){#Fap1} ![The crystal packing of the title compound, viewed along the a axis. The hydrogen bonds are shown as dashed lines (see Table 1 for details).](e-71-0o661-fig2){#Fap2} ![The crystal packing of the title compound, viewed along the b axis. The hydrogen bonds are shown as dashed lines (see Table 1 for details).](e-71-0o661-fig3){#Fap3} Crystal data {#tablewrapcrystaldatalong} ============ --------------------------------------- --------------------------------------- C~11~H~12~N~2~O~2~·C~4~H~4~O~4~·H~2~O F(000) = 712 M~r~ = 338.31 D~x~ = 1.392 Mg m^−3^ Monoclinic, C2 Mo Kα radiation, λ = 0.71073 Å a = 11.3928 (8) Å Cell parameters from 5033 reflections b = 6.6476 (4) Å θ = 2.8--27.9° c = 21.4219 (13) Å µ = 0.11 mm^−1^ β = 95.801 (3)° T = 296 K V = 1614.07 (18) Å^3^ Block, colourless Z = 4 0.30 × 0.20 × 0.20 mm --------------------------------------- --------------------------------------- Data collection {#tablewrapdatacollectionlong} =============== ------------------------------------------------------------ -------------------------------------- Bruker Kappa APEXII CCD diffractometer 2731 reflections with I \> 2σ(I) Radiation source: Sealed X-ray tube R~int~ = 0.025 ω and φ scan θ~max~ = 26.0°, θ~min~ = 2.9° Absorption correction: multi-scan (SADABS; Bruker, 2004) h = −14→14 T~min~ = 0.898, T~max~ = 0.978 k = −8→8 10737 measured reflections l = −26→26 3157 independent reflections ------------------------------------------------------------ -------------------------------------- Refinement {#tablewraprefinementdatalong} ========== ---------------------------------------------------------------- ------------------------------------------------------------------------------------------------------ Refinement on F^2^ Secondary atom site location: difference Fourier map Least-squares matrix: full Hydrogen site location: mixed R\[F^2^ \> 2σ(F^2^)\] = 0.034 H atoms treated by a mixture of independent and constrained refinement wR(F^2^) = 0.084 w = 1/\[σ^2^(F~o~^2^) + (0.0458P)^2^ + 0.2154P\] where P = (F~o~^2^ + 2F~c~^2^)/3 S = 1.04 (Δ/σ)~max~ \< 0.001 3157 reflections Δρ~max~ = 0.18 e Å^−3^ 242 parameters Δρ~min~ = −0.15 e Å^−3^ 4 restraints Extinction correction: SHELXL2014 (Sheldrick, 2015), Fc^\^=kFc\[1+0.001xFc^2^λ^3^/sin(2θ)\]^-1/4^ Primary atom site location: structure-invariant direct methods Extinction coefficient: 0.0069 (11) ---------------------------------------------------------------- ------------------------------------------------------------------------------------------------------ Special details {#specialdetails} =============== ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Geometry. All esds (except the esd in the dihedral angle between two l.s. planes) are estimated using the full covariance matrix. The cell esds are taken into account individually in the estimation of esds in distances, angles and torsion angles; correlations between esds in cell parameters are only used when they are defined by crystal symmetry. An approximate (isotropic) treatment of cell esds is used for estimating esds involving l.s. planes. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Fractional atomic coordinates and isotropic or equivalent isotropic displacement parameters (Å^2^) {#tablewrapcoords} ================================================================================================== ----- -------------- ------------ -------------- -------------------- -- x* y z U~iso~\/U~eq~ N1 0.75082 (19) 0.1813 (4) 0.77433 (10) 0.0380 (5) H1 0.825 (3) 0.191 (5) 0.7735 (13) 0.047 (8)\ N2 0.5095 (2) 0.4264 (3) 0.91895 (10) 0.0276 (5) H2A 0.472 (3) 0.534 (5) 0.9350 (15) 0.056 (10)\* H2B 0.563 (3) 0.476 (5) 0.8921 (14) 0.049 (8)\* H2C 0.554 (3) 0.365 (4) 0.9546 (15) 0.047 (9)\* O1 0.24153 (16) 0.1425 (3) 0.89730 (8) 0.0483 (5) O2 0.35274 (17) 0.2275 (3) 0.98224 (8) 0.0485 (6) O3 0.50562 (13) 0.6533 (3) 0.80307 (7) 0.0334 (4) O4 0.69014 (13) 0.6700 (3) 0.77851 (7) 0.0381 (4) H4O 0.7050 0.6615 0.8167 0.057\* O5 0.63962 (18) 0.6937 (5) 0.54724 (8) 0.0706 (7) H5O 0.6080 0.6855 0.5112 0.106\* O6 0.45352 (18) 0.6884 (4) 0.57018 (8) 0.0595 (6) O7 0.37013 (19) 0.7265 (3) 0.97085 (11) 0.0532 (6) H7A 0.304 (2) 0.705 (5) 0.9880 (16) 0.080\* H7B 0.382 (3) 0.852 (4) 0.9713 (18) 0.080\* C1 0.3327 (2) 0.2159 (3) 0.92463 (10) 0.0263 (5) C2 0.4263 (2) 0.2860 (3) 0.88365 (10) 0.0244 (5) H2 0.3876 0.3556 0.8468 0.029\* C3 0.4928 (2) 0.1034 (4) 0.86179 (12) 0.0328 (6) H3A 0.5345 0.0401 0.8983 0.039\* H3B 0.4354 0.0073 0.8432 0.039\* C4 0.5792 (2) 0.1458 (4) 0.81563 (11) 0.0302 (5) C5 0.6979 (2) 0.1602 (4) 0.82805 (11) 0.0357 (6) H5 0.7379 0.1562 0.8681 0.043\* C6 0.6657 (2) 0.1816 (4) 0.72436 (11) 0.0333 (6) C7 0.6742 (2) 0.1930 (5) 0.66020 (12) 0.0445 (7) H7 0.7471 0.2041 0.6445 0.053\* C8 0.5720 (3) 0.1874 (5) 0.62093 (12) 0.0508 (7) H8 0.5756 0.1951 0.5778 0.061\* C9 0.4630 (2) 0.1705 (5) 0.64405 (13) 0.0507 (7) H9 0.3951 0.1677 0.6160 0.061\* C10 0.4529 (2) 0.1579 (5) 0.70688 (11) 0.0417 (6) H10 0.3792 0.1472 0.7216 0.050\* C11 0.5557 (2) 0.1615 (4) 0.74880 (11) 0.0312 (5) C13 0.57628 (19) 0.6648 (4) 0.76415 (10) 0.0286 (5) C14 0.5377 (2) 0.6733 (4) 0.69680 (10) 0.0365 (6) H14 0.4570 0.6672 0.6848 0.044\* C15 0.6070 (2) 0.6888 (5) 0.65258 (11) 0.0418 (7) H15 0.6880 0.6994 0.6632 0.050\* C16 0.5603 (2) 0.6899 (5) 0.58582 (11) 0.0447 (7) ----- -------------- ------------ -------------- -------------------- -- Atomic displacement parameters (Å^2^) {#tablewrapadps} ===================================== ----- ------------- ------------- ------------- -------------- -------------- -------------- U^11^ U^22^ U^33^ U^12^ U^13^ U^23^ N1 0.0245 (11) 0.0520 (14) 0.0381 (12) 0.0037 (12) 0.0067 (9) −0.0028 (12) N2 0.0274 (12) 0.0345 (12) 0.0219 (11) −0.0020 (9) 0.0072 (9) 0.0000 (9) O1 0.0374 (11) 0.0818 (15) 0.0254 (9) −0.0268 (11) 0.0019 (7) −0.0006 (10) O2 0.0441 (11) 0.0834 (16) 0.0183 (9) −0.0273 (11) 0.0045 (7) −0.0019 (9) O3 0.0300 (9) 0.0449 (10) 0.0260 (8) −0.0024 (9) 0.0056 (7) 0.0031 (8) O4 0.0278 (9) 0.0651 (11) 0.0216 (8) −0.0034 (10) 0.0026 (6) 0.0003 (10) O5 0.0548 (13) 0.133 (2) 0.0237 (9) −0.0004 (17) 0.0058 (9) −0.0012 (15) O6 0.0464 (12) 0.1044 (17) 0.0265 (9) 0.0027 (13) −0.0027 (8) 0.0012 (12) O7 0.0511 (13) 0.0543 (13) 0.0578 (13) 0.0096 (11) 0.0227 (10) 0.0010 (11) C1 0.0249 (12) 0.0323 (13) 0.0220 (11) −0.0021 (10) 0.0034 (10) −0.0003 (10) C2 0.0250 (13) 0.0331 (12) 0.0149 (11) 0.0015 (10) 0.0011 (9) 0.0015 (9) C3 0.0399 (15) 0.0318 (13) 0.0275 (13) 0.0034 (11) 0.0073 (12) −0.0007 (10) C4 0.0304 (12) 0.0315 (12) 0.0295 (12) 0.0041 (12) 0.0074 (9) −0.0047 (11) C5 0.0371 (14) 0.0387 (13) 0.0312 (13) 0.0068 (13) 0.0029 (10) −0.0028 (12) C6 0.0327 (13) 0.0365 (13) 0.0317 (12) 0.0040 (12) 0.0076 (10) −0.0033 (12) C7 0.0416 (15) 0.0558 (17) 0.0384 (14) 0.0055 (15) 0.0158 (12) −0.0013 (14) C8 0.0600 (19) 0.0640 (19) 0.0291 (13) 0.0095 (18) 0.0084 (13) 0.0003 (15) C9 0.0446 (16) 0.0678 (19) 0.0374 (14) 0.0066 (17) −0.0071 (12) −0.0061 (16) C10 0.0301 (13) 0.0578 (16) 0.0373 (14) 0.0054 (15) 0.0033 (11) −0.0047 (15) C11 0.0286 (12) 0.0331 (12) 0.0323 (12) 0.0049 (12) 0.0056 (9) −0.0043 (12) C13 0.0286 (12) 0.0313 (12) 0.0257 (11) −0.0016 (12) 0.0024 (9) 0.0003 (11) C14 0.0328 (13) 0.0500 (15) 0.0263 (12) −0.0002 (14) 0.0002 (10) −0.0013 (14) C15 0.0339 (14) 0.0630 (18) 0.0278 (13) 0.0024 (15) −0.0002 (11) −0.0004 (14) C16 0.0429 (16) 0.0637 (19) 0.0279 (13) 0.0031 (16) 0.0053 (11) −0.0018 (15) ----- ------------- ------------- ------------- -------------- -------------- -------------- Geometric parameters (Å, º) {#tablewrapgeomlong} =========================== -------------------- ------------- ----------------------- ------------- N1---C5 1.359 (3) C3---H3A 0.9700 N1---C6 1.370 (3) C3---H3B 0.9700 N1---H1 0.85 (3) C4---C5 1.355 (3) N2---C2 1.482 (3) C4---C11 1.433 (3) N2---H2A 0.92 (4) C5---H5 0.9300 N2---H2B 0.94 (3) C6---C7 1.390 (3) N2---H2C 0.96 (3) C6---C11 1.413 (3) O1---C1 1.240 (3) C7---C8 1.366 (4) O2---C1 1.235 (3) C7---H7 0.9300 O3---C13 1.219 (3) C8---C9 1.387 (4) O4---C13 1.303 (3) C8---H8 0.9300 O4---H4O 0.8200 C9---C10 1.365 (4) O5---C16 1.285 (3) C9---H9 0.9300 O5---H5O 0.8200 C10---C11 1.401 (3) O6---C16 1.229 (3) C10---H10 0.9300 O7---H7A 0.88 (2) C13---C14 1.466 (3) O7---H7B 0.85 (2) C14---C15 1.297 (3) C1---C2 1.521 (3) C14---H14 0.9300 C2---C3 1.529 (3) C15---C16 1.475 (3) C2---H2 0.9800 C15---H15 0.9300 C3---C4 1.491 (3) C5---N1---C6 108.8 (2) N1---C5---H5 124.4 C5---N1---H1 123.6 (19) N1---C6---C7 131.2 (2) C6---N1---H1 127.6 (19) N1---C6---C11 107.1 (2) C2---N2---H2A 113 (2) C7---C6---C11 121.7 (2) C2---N2---H2B 109.4 (19) C8---C7---C6 117.9 (2) H2A---N2---H2B 108 (3) C8---C7---H7 121.1 C2---N2---H2C 113.4 (17) C6---C7---H7 121.1 H2A---N2---H2C 105 (2) C7---C8---C9 121.4 (2) H2B---N2---H2C 108 (2) C7---C8---H8 119.3 C13---O4---H4O 109.5 C9---C8---H8 119.3 C16---O5---H5O 109.5 C10---C9---C8 121.6 (3) H7A---O7---H7B 107 (3) C10---C9---H9 119.2 O2---C1---O1 123.9 (2) C8---C9---H9 119.2 O2---C1---C2 119.2 (2) C9---C10---C11 118.9 (2) O1---C1---C2 116.79 (19) C9---C10---H10 120.6 N2---C2---C1 110.35 (18) C11---C10---H10 120.6 N2---C2---C3 110.2 (2) C10---C11---C6 118.6 (2) C1---C2---C3 109.37 (19) C10---C11---C4 134.3 (2) N2---C2---H2 109.0 C6---C11---C4 107.1 (2) C1---C2---H2 109.0 O3---C13---O4 123.4 (2) C3---C2---H2 109.0 O3---C13---C14 121.5 (2) C4---C3---C2 115.7 (2) O4---C13---C14 115.07 (19) C4---C3---H3A 108.4 C15---C14---C13 125.3 (2) C2---C3---H3A 108.4 C15---C14---H14 117.4 C4---C3---H3B 108.4 C13---C14---H14 117.4 C2---C3---H3B 108.4 C14---C15---C16 121.5 (2) H3A---C3---H3B 107.4 C14---C15---H15 119.3 C5---C4---C11 105.8 (2) C16---C15---H15 119.3 C5---C4---C3 126.6 (2) O6---C16---O5 124.5 (2) C11---C4---C3 127.4 (2) O6---C16---C15 121.0 (2) C4---C5---N1 111.1 (2) O5---C16---C15 114.6 (2) C4---C5---H5 124.4 O2---C1---C2---N2 −21.1 (3) C8---C9---C10---C11 −0.3 (5) O1---C1---C2---N2 161.6 (2) C9---C10---C11---C6 1.1 (4) O2---C1---C2---C3 100.3 (3) C9---C10---C11---C4 −177.8 (3) O1---C1---C2---C3 −77.0 (3) N1---C6---C11---C10 179.9 (3) N2---C2---C3---C4 −64.7 (3) C7---C6---C11---C10 −1.5 (4) C1---C2---C3---C4 173.8 (2) N1---C6---C11---C4 −0.9 (3) C2---C3---C4---C5 101.2 (3) C7---C6---C11---C4 177.6 (3) C2---C3---C4---C11 −84.9 (3) C5---C4---C11---C10 180.0 (3) C11---C4---C5---N1 −0.7 (3) C3---C4---C11---C10 5.1 (5) C3---C4---C5---N1 174.2 (2) C5---C4---C11---C6 1.0 (3) C6---N1---C5---C4 0.1 (3) C3---C4---C11---C6 −173.9 (2) C5---N1---C6---C7 −177.9 (3) O3---C13---C14---C15 178.9 (3) C5---N1---C6---C11 0.5 (3) O4---C13---C14---C15 −1.3 (4) N1---C6---C7---C8 179.2 (3) C13---C14---C15---C16 178.1 (3) C11---C6---C7---C8 1.0 (4) C14---C15---C16---O6 4.4 (5) C6---C7---C8---C9 −0.1 (5) C14---C15---C16---O5 −175.9 (3) C7---C8---C9---C10 −0.3 (5) -------------------- ------------- ----------------------- ------------- Hydrogen-bond geometry (Å, º) {#tablewraphbondslong} ============================= ---------------------- ---------- ---------- ----------- --------------- D---H···A D---H H···A D···A D---H···A N1---H1···O3^i^ 0.85 (3) 2.11 (3) 2.912 (3) 158 (3) N2---H2A···O7 0.92 (4) 1.94 (4) 2.845 (3) 170 (3) N2---H2B···O1^ii^ 0.94 (3) 2.30 (3) 3.085 (3) 140 (2) N2---H2B···O3 0.94 (3) 2.28 (3) 2.901 (3) 123 (2) N2---H2C···O2^iii^ 0.96 (3) 1.87 (3) 2.832 (3) 174 (3) O4---H4O···O1^ii^ 0.82 1.74 2.559 (2) 178 O5---H5O···O6^iv^ 0.82 1.81 2.630 (3) 174 O7---H7A···O1^v^ 0.88 (2) 2.60 (3) 3.261 (3) 133 (3) O7---H7A···O2^v^ 0.88 (2) 1.97 (2) 2.824 (3) 165 (3) O7---H7B···O2^vi^ 0.85 (2) 2.53 (3) 3.347 (3) 162 (3) C3---H3B···O3^vii^ 0.97 2.66 3.255 (3) 120 C5---H5···O7^i^ 0.93 2.58 3.491 (3) 166 ---------------------- ---------- ---------- ----------- --------------- Symmetry codes: (i) x+1/2, y−1/2, z; (ii) x+1/2, y+1/2, z; (iii) −x+1, y, −z+2; (iv) −x+1, y, −z+1; (v) −x+1/2, y+1/2, −z+2; (vi) x, y+1, z; (vii) x, y−1, z. ###### Hydrogen-bond geometry (, ) DHA DH HA D A DHA ---------------- --------- --------- ---------- --------- N1H1O3^i^ 0.85(3) 2.11(3) 2.912(3) 158(3) N2H2AO7 0.92(4) 1.94(4) 2.845(3) 170(3) N2H2BO1^ii^ 0.94(3) 2.30(3) 3.085(3) 140(2) N2H2BO3 0.94(3) 2.28(3) 2.901(3) 123(2) N2H2CO2^iii^ 0.96(3) 1.87(3) 2.832(3) 174(3) O4H4OO1^ii^ 0.82 1.74 2.559(2) 178 O5H5OO6^iv^ 0.82 1.81 2.630(3) 174 O7H7AO1^v^ 0.88(2) 2.60(3) 3.261(3) 133(3) O7H7AO2^v^ 0.88(2) 1.97(2) 2.824(3) 165(3) O7H7BO2^vi^ 0.85(2) 2.53(3) 3.347(3) 162(3) C3H3BO3^vii^ 0.97 2.66 3.255(3) 120 C5H5O7^i^ 0.93 2.58 3.491(3) 166 Symmetry codes: (i) ; (ii) ; (iii) ; (iv) ; (v) ; (vi) ; (vii) .	{ "pile_set_name": "PubMed Central" }
Introduction {#sec1} ============ Diabetic retinopathy (DR) is a sight-threatening chronic complication that virtually harms patients with diabetes \[[@B1]\]. Currently, DR affects approximately 150 million people worldwide, and the number of DR patients is expected to double by 2025 according to the World Health Organization (WHO) \[[@B2]\]. Patients with DR exhibited increased oxidative stress in diabetes, higher superoxide levels and damaged antioxidant defense systems in the retina and the capillary cells \[[@B3]\]. It is characterized by progressive alterations in the retinal microvasculature, therefore causing increased vasopermeability, non-perfusion retinal areas and pathologic intraocular proliferation of retinal vessels responding to retinal non-perfusion \[[@B1]\]. Destruction of impaired retina by photocoagulation has been the fundamental treatment for almost 50 years after its introduction. However, with the increasingly pandemic diabetes, new approaches are eagerly needed to understand the pathophysiology and to enhance the prevention, detection, and treatment of DR \[[@B4]\]. An all-round understanding of the molecular and biochemical changes in DR particularly at early stage may be conducive to new and effective therapy methods for DR prevention and amelioration \[[@B5]\]. miRNAs are small non-coding RNAs with ∼22 nt base pairs included and the small nucleic acids are capable of regulating gene expression, resulting in transcript degradation or translational suppression, and in the whole genome, \~30% of the genes are subjected to miRNAs regulation \[[@B6]\]. The miR-200 family, including miR-200b, is a cluster of miRNAs, which are highly associated with epithelial--mesenchymal transition (EMT), wherein miR-200b was thought to be a key negative regulator of tumor metastasis, invasion, and chemo-sensitivity \[[@B7]\]. Dysregulation of miR-200b has been reported to play an essential role in the EMT and metastasis in cancers such as gastric, breast, and pancreatic carcinomas \[[@B8]--[@B10]\]. Vascular endothelial growth factor A (VEGFA) belongs to the cysteine knot family of growth factors, which also includes VEGFB, VEGFC, VEGFD, and placental growth factor \[[@B11]\]. VEGFA is a pro-angiogenic factor, which plays a role in promoting survival, migration and proliferation of endothelial cells, and enhancing vascular permeability \[[@B12]\]. In adults, VEGFA is indispensable for blood vessel growth, particularly in pathologies with vascular involvement and organ remodeling. For example, it involved in tumor angiogenesis, wound healing, DR, and age-related macular degeneration \[[@B13]\]. However, it remains unknown how VEGFA is regulated in DR. Therefore, in the present study, inclusion of DR patients and animal experiment were both performed to confirm the hypothesis that miR-200b may alleviate DR development by targeting VEGFA gene. Materials and methods {#sec2} ===================== Ethics statement {#sec2-1} ---------------- This research was approved by the Ethics Committee of Taizhou Hospital of Zhejiang Province and in accordance with the standards of the National Research Council. All animals were raised and treated in accordance with the Guide for the Care and Use of Laboratory Animals by National Institutes of Health of the U.S.A., and informed consent was obtained from each patient prior to study. Study subjects and blood sample collection {#sec2-2} ------------------------------------------ From October 2014 to July 2016, 255 patients diagnosed with DR and treated in Taizhou Hospital of Zhejiang Province were included in the case group, consisting of 134 males and 121 females (mean age 61.45 ± 11.90 years). The criteria for the diagnosis of diabetes were in accordance with the 2015 Diagnostic Criteria of Diabetes created by the American Diabetes Association (ADA) \[[@B14]\]. The diagnosis of patients with DR was done according to the Clinical Classification Criteria for the Diagnosis of Diabetic Retinopathy proposed in the 2002 by the Sydney International Clinical Trials Symposium \[[@B15]\]. All patients were to be given eye examinations (including visual acuity, intraocular pressure, fundus examination, ophthalmic B-scan ultrasonography, and slit lamp examination of the anterior segment) and a general physical checkup (including blood routine, urine routine, liver, and kidney function tests). Fasting blood glucose was controlled within 8.0 mmol/l, and 2-h post-prandial blood glucose was to be not more than 10.0 mmol/l. The exclusion criteria were as follows: no history of hepatitis, acute and chronic infection, and malignant tumor; no systemic diseases such as cardiovascular and cerebrovascular diseases, inflammatory diseases, tissue proliferative diseases, and autoimmune diseases; and no other eye infections and eye diseases. At the same time, the control group included 253 healthy people who undertook physical examination in Taizhou Hospital of Zhejiang Province, consisting of 140 males and 113 females (mean age 60.18 ± 7.68 years). Fundus photography and fundus fluorescein angiography were applied to people of the control group. Besides, their fasting blood glucose levels should have been 3.9--6.1 mmol/l and 2-h post-prandial blood glucose should have been not more than 7.0 mmol/l. The exclusion criteria were as follows: no retinopathy and other eye diseases such as age-related macular diseases and ischemic optic neuropathy; no family history of glaucoma, ocular trauma, and family history of other eye diseases. After fasting for 12--14 h, 2 ml of peripheral venous blood was extracted from all subjects. The blood was anti-coagulated with EDTA-Na2 and preserved at 4°C. After centrifugation for 15 min at the rate of 1500 rev/min and isolating the serum, the expressions of miR-200b and VEGFA mRNA were detected. Construction and activity detection of luciferase reporter vector {#sec2-3} ----------------------------------------------------------------- The VEGFA target gene fragments were inserted into wild-type VEGFA-3′-UTR-WT plasmid and mutant VEGFA-3'-UTR-MUT plasmid respectively to construct VEGFA dual-luciferase reporter gene plasmid. The targeting relationship between miR-200b and VEGFA was predicted by the biological prediction website microRNA.org and validated by dual-luciferase reporter gene assay. Then, 293T cells at the logarithmic growth phase were inoculated into 96-well plates. When the cell density reached 70%, Lipofectamine 2000 transfection was conducted to co-transfect the mixed VEGFA-3′-UTR-WT plasmid and miR-200b plasmid to 293T cells. The control groups (VEGFA-3′-UTR-WT+NC and VEGFA-3′-UTR-MUT + miR-200b) were established at the same time. After culturing for 6 h in an incubator (Thermo Fisher Scientific, San Jose, CA, U.S.A.), cells were transferred into a culture medium that contained 10% FBS to culture for another 48 h. Dual-luciferase activity was detected according to the method provided by Promega as follows: after removing cell culture fluid from 96-well plates, cells were washed softly with 100 μl of PBS. Luciferase assay reagent-I solution of 100 μl was added to each well before shaking for 15 min at room temperature, and then Glomax20/20 luminometer (Promega Corporation, Madison, WI, U.S.A.) was used for activity detection. After the measurement of firefly luciferase activity, 100 μl of luciferase assay reagent-II was added to detect the activity of renilla luciferase. Gene expressions were presented by the activity ratio of the firefly luciferase to the ranilla luciferase. Establishment of rat model of DR induced by streptozotocin {#sec2-4} ---------------------------------------------------------- Seventy adult male Wistar rats (weighing 200--220 g) were purchased from the Laboratory Animal Center of China Medical University and divided into model group (n=60) and normal control (NC) group (n=10). The rats in the model group were given fasting injection with 60 mg/kg streptozotocin (STZ, dissolved in 0.01 mol/l citrate buffer solution (pH 4.4), purchased from Sigma--Aldrich Chemical Company, St. Louis MO, U.S.A.); and rats of the NC group were injected with the same dose of citrate buffer solution. Then, rats had free access to water and food under natural sunlight. After the model establishment, rats were weighed once a week and their tail vein blood were taken once every 3 days to test blood glucose. One Touch glucometer and Standard Blood Glucose Test Paper (purchased from Johnson Lifescan Inc. of U.S.A.) were used to detect the blood glucose. Rats whose glucose concentration was continually equal to or more than 16.7 mmol/l were classified into diabetic models \[[@B16]\] and later blood glucose measurement was conducted once a month. After 3 months of regular feeding, three randomly selected rats were given a left ventricular injection of 50 mg/ml FITC-labeled dextran (FITC--dextran, purchased from Sigma--Aldrich Chemical Company, St. Louis MO, U.S.A.). At the same time, their eyes were removed and fixed in 4% paraformaldehyde. The retinal tissues spread on glass slides were observed and photographed under a fluorescent microscope (PRIMOSTAR-FL2, Carl Zeiss MicroImaging, Inc., Thornwood, NY, U.S.A.) to identify the experimental animal model. Construction of miR-200b and VEGFA plasmid vectors {#sec2-5} ------------------------------------------------------ The mature miR-200b sequence (No. MI0000342) was obtained from MiRBase database to synthesize miR-200b mimics. miR-200b inhibitor was the reverse complementary sequence of mature miR-200b ([Table 1](#T1){ref-type="table"}). VEGFA silencing was in accordance with the principle of siRNA design, and homology analysis was conducted to the alternative target sequence. The NC of dsRNA was a sequence that had no homology with mammalian genome. The synthesis of oligonucleotide sequences needed to add restriction enzyme cutting sites BamHI and XhoI, which was conducted by Shanghai GenePharma Co., Ltd. Amplified fragments were recovered, linked to pM18-T vector with T4DNA ligase, and transformed into DH5α competent bacteria. Single colonies were selected for colony PCR detection. Positive colonies were selected to extract plasmid. Double enzyme digestion with EcoR and KpnI was applied to plasmid and lentiviral expression vector pMIR, which were later connected by T4 DNA ligase to form recombinant plasmids pMIR--miR-200b mimic, pMIR--miR-200b inhibitor, pMIR--siVEGFA, and pMIR-vector \[[@B17]\]. Table 1Oligonucleotide sequences of RNA and DNASequencemiR-200b mimics5′-CAUCUUACUGGGCAGCAUUGGA -3′miR-200b inhibitor5′-UCAUCAUUACCAGGCAGUAUUA-3′siVEGFA5′-AUGUGAAUGCAGACCAAAGAA -3′NC5′-AATTCTCCGAACGTGTCACGT-3′[^1] Animal grouping {#sec2-6} --------------- The experiment included six groups with ten rats in each group. In the blank group, the rats were given no treatment; in the NC group, the rats were injected with empty plasmid in the vitreous cavity; in the miR-200b mimics group, the rats were injected with miR-200b mimics plasmid in the vitreous cavity; in the miR-200b inhibitors group, the rats were injected with miR-200b inhibitors plasmid in the vitreous cavity; in the miR-200binhibitors + si-VEGFA group, the rats were injected with miR-200b inhibitor and si-VEGFA plasmid in the vitreous cavity; and in the si-VEGFA group, the rats were injected with si-VEGFA plasmid in the vitreous cavity. The concentration for plasmid injection was 20 μM with 5 μl each time \[[@B18]\]. During detection time, the rats were killed by intraperitoneal injection with 3% pentobarbital (2 ml/kg), and their eyeballs were extracted and immersed in 10% formalin for 12 h. The crystalline lens, cornea, sclera, and choroid were removed under a dissecting microscope (Carl Zeiss MicroImaging, Inc., Thornwood, NY, U.S.A.). Free retinal tissues were obtained by removing the vitreum and the pigment epithelium on the outer layer of retina, which were preserved in liquid nitrogen for later use. Quantitative real-time PCR {#sec2-7} -------------------------- The serum RNA was extracted by QIAamp MinElute Virus Spin Kit (Qiagen Company, Hilden, Germany) and retinal tissues of rats in each group were ground with saline. The total RNA extracted by RNA Extraction Kit (Omega Bio-tek Inc, Norcross, GA, U.S.A.) was detected for RNA purity and concentration under the ultraviolet spectrophotometer (UV-1800, Shimadzu Company, Kyoto, Japan), and observed for RNA integrity with agarose gel electrophoresis. The primers of miR-200b, VEGFA, transforming growth factor-β1 (TGF-β1), hepatocyte growth factor (HGF), and pigment epithelium-derived factor (PEDF) were designed with the software Primer 5.0 and synthesized by Sangon Biotech (Shanghai) Co., Ltd. ([Table 2](#T2){ref-type="table"}). Primescript™ RT reagent kit (Takara Biotechnology Ltd., Dalian, China) was used to reversely transcribe total RNA into cDNA, and the reverse-transcription system was 10 μl with reaction conditions as follows: 16°C for 30 min, 42°C for 30 min, and 85°C for 10 min. Quantitative real-time PCR (qRT-PCR) was conducted with SYBR^®^ premix Ex Taq TM real-time quantitative PCR Kit (Takara Biotechnology Ltd., Dalian, China) and the reaction conditions were a total of 40 cycles of pre-denaturation for 2 min at 95°C, denaturation for 5 s at 95°C, annealing for 4 s at 60°C, and extending for 30 s at 72°C. The relative expression of miR-200b was calculated using 2^−△△C^~t~ (C~t~, cycle threshold) with U6 sn RNA as the internal reference gene, and the mRNA expressions of VEGFA, TGF-β1, HGF, and PEDF were measured by 2^−△△C^~t~ with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as the internal reference gene. ###### Oligonucleotide sequences Gene Primer sequence -------------------- --------------------------------- miR-200b (rat) R:5′-CTCCCTAAAGCCTCCCACC-3′ R:5′-AGGGCTTTCTGCTGTTGTCC-3′ miR-200b (human) F:5′-GCGGCTAATACTGCCTGGTAA-3′ R:5′-GTGCAGGGTCCGAGGT-3′ U6 F:5′-CGCTTCGGCAGCACATATA-3′ R:5′-TTCACGAATTTGCGTGTCAT-3′ VEGFA (rat) F:5′-ACTTTCTGCTGTCTTGGGTG-3′ R:5′-CTGCATGGTGATGTTGGACT-3′ VEGFA (human) F:5′-CCTCCGAAACCATGAACTTT-3′ R:5′-CCACTTCGTGATGATTCTGC-3′ TGF-β1 (rat) F:5′-GCCTGAGTGGCTGTCTTTTGA-3′ R:5′-GAAGCGAAAGCCCTGTATTCC-3′ HGF (rat) F:5′-GACCTTGTGAGGGAGATTAT-3′ R:5′-ATGTGCCATCCCAAATCGTCC-3′ PEDF (rat) F:5′-CCAAGTCTCTGCAGGACATGAAG-3′ R:5′-GGTTTGCCAGTAATCTTGCTG-3′ GAPDH F:5′-TGGTATCGTGGAAGGACTCA-3′ R:5′-GCAGGGATGATGTTCTGGA-3′ Western blotting {#sec2-8} ---------------- The retinal tissues of rats were ground with normal saline. After 15 min of centrifugation at the speed of 12000 rev/min, the supernatant was collected to conduct SDS/PAGE electrophoresis. Proteins after electrophoretic separation were transferred to the nitrocellulose filter by electrotransfer. Then 5% skimmed milk--PBS solution was closed for 1 h at room temperature and cultured at 4°C overnight with VEGF, TGF-β1, PEDF, and HGF antibody (1:500, purchased from Beijing Bosscn Company with batch number bs-0103R, bs-2202R, bs-0731R, and bs-1025P respectively). The filter was washed with PBS buffer for three times and cultured for 1 h at room temperature with HRP-cross-linked secondary antibody. The filter was then washed with PBS buffer for another three times and developed by ECL. With GAPDH as internal reference \[[@B19]\], the gray value ratio of target band to reference band was regarded as the relative expressions of proteins. ADPase histochemical method {#sec2-9} --------------------------- The retinal tissues of rats in each group were obtained, cut into four valves with optic papilla as the center, rinsed for 12 h with purified water, and digested with 3% trypsin for 7 h at 37°C. Then ADPase histochemical staining was conducted as follows: the prepared retinal tissues were rinsed for 15 min five times with pre-cooling Tris-maleic acid buffer (50 mM), soaked for 15 min in Tris-maleic acid buffer (0.2 Mm, containing 1 mg/ml ADP) at 37°C, rinsed for 15 min five times with Tris-maleic acid buffer (50 mM); developed for 10 min with 1:10 sulfide; rinsed for 15 min three times with Tris-maleic acid buffer (50 mM); finally 50% glycerol was used to mount the section. Results were observed under an optical microscope (Olympus Optical Co., Ltd, Tokyo, Japan) with pictures taken by a digital camera. Immunohistochemistry {#sec2-10} -------------------- The retinal tissues of rats in each group were collected, embedded by paraffin, and made into sections before baking for 20 min at 68°C. After conventional xylene dewaxing and gradient alcohol dehydration, the sections were placed at room temperature for 15 min. Next, PBS was used to wash sections for 5 min two to three times before adding normal goat serum blocking solution at room temperature for 20 min. Then CD34 antibody (1:2000, purchased from Beijing Bosscn Company with batch number bs-8996R) was added for 1 h of incubation at 37°C. After PBS wash, secondary antibody was added for another hour of incubation at 37°C. Following another PBS wash, diaminobenzidine w(DAB) as used for color development and the results were observed under microscope. Then after 2 min of hematoxylin staining, sections were dehydrated, transparentized, and mounted for observation under microscope. Vascular endothelial cells were labeled with CD34 antibody \[[@B20]\]. A brown or yellow single endothelial cell or endothelial cell string was regarded as a blood vessel, which was used to determine the level of new blood vessel formation. Hematoxylin and eosin staining {#sec2-11} ------------------------------ The retinal tissues were fixed with Davidson's solution for 24 h before conventional dehydration, transparentizing, wax filing, and paraffin embedding. Ten sections that were sliced continuously were 3 μm in thickness and baked for 1 h at 50°C. Routine hematoxylin and eosin (HE) staining was performed before observing retinal angiogenesis under an optical microscope. Double blind method was used to count the endothelial nuclei that broke into internal limiting membranes (only counting nuclei in close contact with the inner limiting membrane, and excluding those having no connection with the internal limiting membrane in vitreum), and average value was calculated. Statistical methods {#sec2-12} ------------------- The statistical analysis was conducted with SPSS 21.0. Measurement data were presented by mean ± standard deviation ($\overline{x}$ ± S.D.). Differences between groups were analyzed using t-test and multiple sets of data were analyzed using single factor variance analysis. Enumeration data were presented by percentage and analyzed using Chi-Square (χ^2^) test. P\<0.05 was considered statistically significant. Results {#sec3} ======= Baseline characteristics of subjects between the case and control groups {#sec3-1} ------------------------------------------------------------------------ There were no significant differences in gender and age between the case and control groups (P\>0.05). Patients with DR in the case group were higher than healthy people in the control group in terms of body mass index (BMI), total cholesterol, triglycerides, high-density lipoprotein, glycosylated hemoglobin, blood glucose, and blood pressure (all P\<0.05) ([Table 3](#T3){ref-type="table"}). Table 3Comparisons of baseline characteristics of study subjects between the case and the control groupsVariableControl group (n=253)Case group (n=255)Gender (male/female)140/113134/121Age (years)60.18 ± 7.6861.45 ± 11.90Disease duration (years) --8.21 ± 2.43BMI (kg/m^2^)23.83 ± 3.6425.27 ± 3.67\Total cholesterol (mmol/l)4.87 ± 1.046.01 ± 1.04\Triglycerides (mmol/l)1.29 ± 0.352.36 ± 0.62\High-density lipoprotein (mmol/l)1.29 ± 0.341.50 ± 0.45\Fasting blood glucose (mmol/l)5.09 ± 1.009.96 ± 2.53\Glycosylated hemoglobin (%)5.91 ± 1.119.49 ± 2.04\Systolic pressure (mmHg)119.52 ± 14.09134.44 ± 10.40\Diastolic pressure (mmHg)77.73 ± 12.3784.14 ± 9.82\[^2] Comparisons of the expressions of miR-200b and VEGFA mRNA between the case and control groups {#sec3-2} ------------------------------------------------------------------------------------------------- Compared with healthy people in the control group, patients with DR in the case group significantly decreased in miR-200b expression and significantly increased in VEGFA mRNA expression (P\<0.05) ([Figure 1](#F1){ref-type="fig"}A). At the same time, the correlation analysis between miR-200b and VGEFA mRNA in both groups showed that miR-200b was negatively correlated with VGEFA (r = −0.4036, P\<0.05) ([Figure 1](#F1){ref-type="fig"}B). ![The expressions of miR-200b and VEGFA mRNA between the case and control groups and correlation analysis between miR-200b and VEGFA\ Note: (A) the expressions of miR-200b and VEGFA mRNA in the two groups; (B) correlation analysis between miR-200b and VEGFA mRNA; \*P\<0.05 compared with the control group.](bsr-2016-0572i001){#F1} VEGFA confirmed as a target gene of miR-200b {#sec3-3} ---------------------------------------------- According to prediction of biological prediction website microRNA.org, miR-200b is located at the 3′-UTR intron 6 and intron 7 of VEGFA gene, and miR-200b can be partially paired with Y-UTR of VEGFA gene ([Figure 2](#F2){ref-type="fig"}A). When 293T cells were co-transfected with VEGFA-3′UTR-WT plasmid and miR-200 mimics plasmid, we found that luciferase activity ratio of firefly luciferase to ranilla luciferase (Y/H) was decreased (P\<0.05) compared with the control group (VEGFA-3′-UTR-WT + NC) ([Figure 2](#F2){ref-type="fig"}B). However, 293T cells co-transfected with VEGFA-3′UTR-MUT + miR-200b mimics plasmid showed no significant difference in Y/H with the control group (P\>0.05). All the above findings predicted by microRNA.org indicated that VEGFA gene is a direct target gene of miR-200b. ![The targeting relationship between miR-200b and VEGFA gene\ Note: (A) The VEGFA 3′-UTR loci for combining miR-200b; (B) The luciferase expression was detected 48 h after 293T cells were co-transfected with VEGFA-3′-UTR-WT plasmid + miR-200 mimics plasmid, VEGFA-3′-UTR-MUT + miR-200b/NC; \\P\<0.05 was considered statistically significant.](bsr-2016-0572i002){#F2} Successful establishment of rat models of DR induced by streptozotocin {#sec3-4} ---------------------------------------------------------------------- The survival rate of both groups was 100%. Before model establishment, there were no significant differences in body weight and blood glucose between the model group and the NC group (all P\>0.05). Two weeks after STZ induction, the rats in the model group manifested typical symptoms of diabetes such as obvious polydipsia, polyphagia, and polyuria. On the 8th week, apparent symptoms included lens opacification, extreme torso and head emancipation, withered hair, and abdominal swelling. On the 4th, 8th, and 12th week, the body weight of the rats in the NC group kept rising (all P\<0.05), and their blood glucose remained stable (all P\>0.05). Compared with the rats in the NC group, the rats in the model group significantly decreased in body weight and significantly increased in blood glucose (all P\<0.05), and their blood glucose concentration remained at about 19 mmol/l from 4th week ([Table 4](#T4){ref-type="table"}). When observing tissue sections of rats after left ventricular injection with FITC--dextran, we found that model rats showed retinal capillary dilatation, interstitial edema, irregular retinal diameter, and tortuous blood vessels. However, normal rats had smooth retinal vascular branches and uniform diameter ([Figure 3](#F3){ref-type="fig"}). The results demonstrated that the DR model was successfully established. Table 4Comparisons of body weight and blood glucose of rats in the normal control group and the model group (X ± S.D.)TimeNormal control group (n=10)Model group (n=10)Body weight (g)0 week209.76 ± 4.89207.80 ± 8.704th week227.63 ± 14.09204.50 ± 9.10\8th week251.02 ± 21.52191.80 ± 11.90\12th week280.23 ± 25.64171.10 ± 16.30\Blood glucose (mmol/l)0 week5.40 ± 0.725.48 ± 0.524th week5.44 ± 0.6718.80 ± 1.71\8th week5.32 ± 0.7721.51 ± 1.94\12th week5.46 ± 0.3225.84 ± 2.31\[^3] ![Comparisons of morphological changes of retina between the model and normal control groups (× 200)\ Note: (A) model of DR; (B) NC group.](bsr-2016-0572i003){#F3} The expressions of miR-200b and expressions of VEGFA, TGF-β1, HGF, and PEDF mRNA of rats among six groups {#sec3-5} ----------------------------------------------------------------------------------------------------------- Compared with rats in the blank group and the NC group, the rats in the miR-200b mimics group were increased in miR-200b expression, while the rats in the miR-200b inhibitors group and the miR-200b inhibitors + siVEGFA group were decreased in the expression of miR-200b (all P\<0.05). Compared with rats in the NC group, the rats in the miR-200b mimics group and the siVEGFA group were elevated in PEDF mRNA expression, and reduced in expressions of VEGFA, TGF-β1, and HGF mRNA (all P\<0.05), but the rats in the miR-200b inhibitors group were decreased in PEDF mRNA expression, and increased in expressions of VEGFA, TGF-β1, and HGF mRNA (all P\<0.05). Besides, compared with rats in the NC group, the rats in the miR-200b inhibitors + siVEGFA group showed no significant difference in the mRNA expressions of VEGFA, TGF-β1, HGF, and PEDF (all P \>0.05). These results indicated that miR-200b over-expression and VEGFA low-expression can reduce the mRNA expression of VEGFA and retinopathy-related proteins such as TGF-β1 and HGF, and enhance the expression of PEDF mRNA. In addition, miR-200b was negatively correlated with VEGFA ([Figure 4](#F4){ref-type="fig"}). ![Comparisons of the expressions of miR-200band VEGFA, TGF-β1, HGF, PEDF mRNA by qRT-PCR among six groups\ Note: \, P* \< 0.05 compared with the blank group and the NC group.](bsr-2016-0572i004){#F4} The expressions of VEGFA, TGF-β1, HGF, and PEDF protein of rats among six groups {#sec3-6} -------------------------------------------------------------------------------- There was no significant difference in the expressions of VEGFA, TGF-β1, HGF, and PEDF proteins among the blank group, the NC group, and the miR-200b inhibitors + siVEGFA group (all P\>0.05). Compared with rats in the NC group, rats in the miR-200b mimics group and the siVEGFA group were decreased in the expressions of VEGFA, TGF-β1, and HGF proteins, but increased in the expression of PEDF protein (all P\<0.05). On the other hand, rats in the miR-200b inhibitors group were increased in the expressions of VEGFA, TGF-β1, and HGF proteins, but decreased in the expression of PEDF protein (all P\<0.05). These results suggested that miR-200b over-expression and VEGFA low-expression can reduce the expression of retinopathy-related proteins such as TGF-β1 and HGF, while also enhance the expression of PEDF protein. Besides, miR-200b low-expression can improve the expression of retinopathy-related proteins such as TGF-β1 and HGF, and reduce the expression of PEDF protein ([Figure 5](#F5){ref-type="fig"}). ![Comparisons of expressions of VEGFA and retinopathy-related proteins TGF-β1, HGF, and PEDF by Western blotting among six groups\ Note: \*P\<0.05 compared with the blank group and the NC group.](bsr-2016-0572i005){#F5} Morphology of retinal neovascularization of rats among six groups {#sec3-7} ----------------------------------------------------------------- In the blank group and the NC group, rats showed dense retinal blood vessels, tortuous capillaries, irregular vessel diameter with segmental enlargement, and many visible acellular capillaries. Compared with the NC group, rats in the miR-200b mimics group, the siVEGFA group and the miR-200b inhibitors + siVEGFA group had regular blood vessel running and basically uniform capillary diameter with occasionally visible segmental expansion and few acellular capillaries. In addition, rats in the miR-200binhibitors group showed tortuous blood vessel running, irregular capillary diameter and segmental enlargement but had no acellular capillaries ([Figure 6](#F6){ref-type="fig"}). ![Comparisons of morphology of retinal neovascularization detected by ADPase histochemical method among six groups (× 400)](bsr-2016-0572i006){#F6} Retinal microvessel density of rats among six groups {#sec3-8} ---------------------------------------------------- There was no significant difference in retinal microvessel density (MVD) between the blank group, the NC group, and the miR-200b inhibitors + siVEGFA group (all P\>0.05). Compared with rats in the NC group, rats in the miR-200bmimics group and the siVEGFA group had decreasing results in MVD, while the miR-200binhibitors group were increased in MVD (P\<0.05), indicating that miR-200b over-expression and VEGFA low-expression can reduce the growth level of MVD so as to reverse the occurrence of retinal lesions ([Figure 7](#F7){ref-type="fig"}). ![Comparisons of retinal MVD by immunohistochemistry among six groups ($\overline{x}$ ± S.D.) (× 200)\ Note: \*P \<0.05 compared with the blank group and the NC group.](bsr-2016-0572i007){#F7} The number of retinal vascular endothelial nuclei of rats among six groups {#sec3-9} -------------------------------------------------------------------------- In the blank group, the NC group and the miR-200b inhibitors + siVEGFA group, rats showed mild cell edema on the retinal surface, mildly disordered cell layers, a small amount of retinal neovascularization buds, and slightly irregular expanded vascular cavity. Compared with rats in the NC group, rats in the miR-200b mimics group and the siVEGFA group had no obvious cell edema on the retinal surface. Besides, their layers of cells were arranged in order, retinal neovascularization buds were occasionally visible, vascular cavity was regular in section, and vascular endothelial nuclei were significantly reduced (P\<0.05). On the contrary, rats in the miR-200b inhibitors group showed cell edema on the retinal surface, irregularly arranged cell layers, retinal neovascularization buds, and irregularly enlarged vascular cavity section, while their vascular endothelial nuclei was significantly increased (P\<0.05) ([Figure 8](#F8){ref-type="fig"}). These results suggest that miR-200b over-expression and VEGFA low-expression can reduce the number of vascular endothelial cells and inhibit the formation of retinal neovascularization buds. ![Comparisons of the number of retinal vascular endothelial nuclei by HE staining among six groups (× 200)\ Note: (A) HE staining results; (B) comparison of retinal vascular endothelial nuclei; \*P \< 0.05 compared with the blank group and the NC group.](bsr-2016-0572i008){#F8} Discussion {#sec4} ========== DR, as a common complication of diabetes, is one of the leading causes of blindness worldwide \[[@B21]\]. As the VEGFAgene has been reportedly related to the pathogenesis of DR \[[@B22]\], exploring the involvement of VEGFA in DR would provide a theoretical foundation for a new genetically therapeutic target against the pandemic eye disease. In the present study, it is found that DR patients exhibited decreased expression of miR-200b and increased VEGFA mRNA in comparison with healthy people. miRNAs are a cluster of small non-coding RNAs capable of repressing gene expression by binding mRNA target transcripts, thus causing mRNA degradation or translational repression \[[@B23]\]. Being widely involved in various biologic processes, miRNAs may have an essential modulatory role in DR \[[@B24]\]. The presence of miR-200b was found in both human and rat retinas \[[@B25]\]. Also, our in vivo and in vitro studies demonstrating miR-200b expression in humans and rats suggest evolutionary conservation and it may reflect a conserved functional role in the mammalian retina. To confirm the results of the present study, McArthur et al. \[[@B24]\] reports that miR-200b is down-regulated in retinas of diabetic mice and in endothelial cells treated with high glucose. Besides, in the study, increasedVEGFA mRNA in DR patients was also observed. The mechanisms of hyperglycemia-induced cellular damage still remain unclear. However, evidence indicates that DNA damage might be a result of high oxidative stress \[[@B26],[@B27]\]. Increased oxidative stress causes the activation of the redox-sensitive transcription factors and then changes expression of many genes, including VEGF \[[@B28]\]. Several studies have demonstrated that VEGF gene is closely related to the severity of DR \[[@B29],[@B30]\]. Whitmire et al. \[[@B31]\] reported that diabetic retina was companied by up-regulated VEGFA, which was in accordance with the present study. Furthermore, the study demonstrates that miR-200b could negatively target VEGFA. Liu et al. \[[@B32]\] reported that they selected mRNAs with a conserved seed sequence in their 3′-UTRs for miRNAs that were diversely expressed between tumor and normal kidney, and identified target mRNAs whose expression had an negative correlation with that of miR; they found that there was an obvious inverse correlation between the miR-200 family and VEGF. Thus, this result provides a support to our conclusion that VEGF is a target gene of miR-200b as revealed by dual-luciferase report assay. Additionally, in the present study, there are lower protein and mRNA expressions of VEGFA, TGF-β1, and HGF but higher mRNA expression of PEDF in miR-200b mimics and siVEGFA groups than in the NC group, and the miR-200b inhibitors group shows the opposite results. As miR-200b could negatively regulate VEGFA (its downstream target gene), miR-200bmimics group, in which rats were injected withmiR-200b mimic plasmids, reasonably presented lower protein and mRNA expression of VEGFA. TGF-β1 is an important regulator of tissue morphogenesis and a proliferation inhibitor for most cell types \[[@B33]\]. As TGF-β1 protein production and mRNA expression could be blocked by VEGF via PI3K/Akt signaling \[[@B34]\], TGF-β1 is down-regulated accordingly with inhibited protein and mRNA expression of VEGFA. VEGFA and HGF are paracrine hormones that can regulate communication between pancreatic islet β-cells and endothelial cells \[[@B35]\]. VEGFA inhibition in experimental models could result in down-regulation of HGF \[[@B36]\], which confirms the result of the present study. However, the mechanism remains unclear. PEDF is a potent anti-angiogenic factor mediated partially by the induction of endothelial cell apoptosis \[[@B37]\]. PEDF can be inhibited by VEGFA through proteolytic degradation, where the lowest PEDF levels coincides with the highest VEGF-A levels and has associations with the development of retinal neovascularization in a study concerning retinal neovascularization \[[@B38]\]. Therefore, with down-regulated VEGFA, mRNA expression of PEDF in miR-200b mimics and siVEGFA groups increased. Judging by the above mechanisms, it is reasonable that 200b inhibitors group showed higher protein and mRNA expressions of VEGFA, TGF-β1, and HGF, but lower mRNA expression of PEDF. Finally, it is observed that miR-200b mimics and siVEGFA groups show less circuitous blood vessels, MVD, and endothelial cell nucleus in comparison with the NC group. Expression of VEGFA could activate angiogenic response, which can further generate new and morphologically distinct blood vessels \[[@B39]\], while the activation of VEGFA elicits some effects on endothelial cells, such as survival, proliferation, elevated permeability, and migration \[[@B40]\], which is the reason why the miR-200b mimics and siVEGFA groups, in which VEGFA is down-regulated, exhibits decreased circuitous blood vessels, MVD, and endothelial cell nucleus. Since DR is caused by the development of abnormal new blood vessels in the retina for some extent \[[@B41]\], down-regulating VEGFA could be a promising strategy for the treatment of DR. In summary, our study provided evidence that miR-200b might alleviate DR development by down-regulating its target gene, VEGFA. Ruiz et al. \[[@B42]\] have proven that the increase of trimethylation of lysine 27 on histone H3 (H3K27me3), as a special methylation, plays a regulatory role between polycomb repressive complex 2 (PRC2) and miR-200b on chromatin modification, and histone methyltransferase complex, PRC2 could suppress miRNAs in the development of cancers. With a high glucose environment, PRC2 could regulate the growth and proliferation of REC, improve the structure and function of vascular endothelial cells, and then slow down the deteriorating progress of DR by repressing the expression of miR-200band enhancing the expression of VEGF. Consequently, it is believed that miR-200b may serve as a promising target in regulating VEGF-mediated abnormalities in DR. Understanding the novel mechanisms will contribute to a better understanding of DR pathogenesis and will eventually lead to formulation of specific treatment. We would like to acknowledge the helpful comments on the present study received from our reviewers. Competing interests {#sec5} =================== The authors declare that there are no competing interests associated with the manuscript. Funding {#sec6} ======= No funding was declared by the authors of this manuscript. Author contribution {#sec7} =================== E.H.L, Q.Z.H. and G.C.L. designed the study. E.H.L.collated the data, designed and developed the database, Q.Z.H.carried out data anaysis and produced the initial draft of the manuscript. Z.Y.X. and X.Z. contributed to drafting the manuscript. All authors have read and approved the final submitted manuscript. DR : diabetic retinopathy EMT : epithelial--mesenchymal transition GAPDH : glyceraldehyde-3-phosphate dehydrogenase HE : hematoxylin and eosin HGF : hepatocyte growth factor MVD : microvessel density NC : normal control PEDF : pigment epithelium-derived factor PRC2 : polycomb repressive complex 2 qRT-PCR : quantitative real-time PCR siVEGFA : silencing vascular endothelial growth factor A STZ : streptozotocin TGF-β1 : transforming growth factor-β1 VEGFA : vascular endothelial growth factor A [^1]: Note: siVEGFA, silencing vascular endothelial growth factor A. [^2]: Note: \*P\<0.05 compared with the control group. [^3]: Note: \*P\<0.05 compared with the normal control group.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-polymers-10-00737} =============== The electrospinning technique is a simple procedure to obtain polymeric and subsequently metallic, semiconducting, or ceramic fibers from solutions or melts \[[@B1-polymers-10-00737],[@B2-polymers-10-00737],[@B3-polymers-10-00737],[@B4-polymers-10-00737]\]. The process was proven to be a low-cost, flexible, and scalable method to produce microfibers and nanofibers. Differently arranged (randomly or aligned) fibers with various morphologies can be obtained depending on the geometry of the electrospinning set-up \[[@B5-polymers-10-00737],[@B6-polymers-10-00737],[@B7-polymers-10-00737],[@B8-polymers-10-00737]\]. A fine control of the electrospinning process parameters consequently results in a fine-tuning of the fibers' properties \[[@B9-polymers-10-00737],[@B10-polymers-10-00737]\], such as dimension, morphology (including nanostructuring), and composition (including doping with different compounds). Polymer microfibers and nanofibers produced by the electrospinning technique are in high demand because of their multiple applications, particularly in the biomedical field \[[@B11-polymers-10-00737],[@B12-polymers-10-00737],[@B13-polymers-10-00737],[@B14-polymers-10-00737],[@B15-polymers-10-00737],[@B16-polymers-10-00737]\]. Besides medical applications, the polymer electrospun fibers are used also for filters \[[@B17-polymers-10-00737],[@B18-polymers-10-00737]\], for cosmetics \[[@B18-polymers-10-00737]\], in the clothing and textile industry \[[@B19-polymers-10-00737],[@B20-polymers-10-00737]\], for electrical conductors \[[@B21-polymers-10-00737]\], and for sensor devices \[[@B22-polymers-10-00737]\]. For optical applications, such polymeric fibers are used for ultrafast photonics \[[@B23-polymers-10-00737]\], for random lasers \[[@B24-polymers-10-00737]\], for up-conversion \[[@B25-polymers-10-00737]\], for white-light-emitting diodes \[[@B26-polymers-10-00737]\], for dye-sensitized solar cells \[[@B27-polymers-10-00737]\], for their nonlinear properties \[[@B28-polymers-10-00737]\], and for optical fibers \[[@B29-polymers-10-00737]\]. Functionalization of the polymer nanofibers may be done using nanoparticle addition or further chemical or electrochemical deposition of various compounds on the surface of the electrospun nanofibers \[[@B21-polymers-10-00737],[@B30-polymers-10-00737],[@B31-polymers-10-00737]\]. Bundles of such nanofibers were proven to be able to mimic human muscle movements \[[@B32-polymers-10-00737],[@B33-polymers-10-00737]\]. Interesting optical properties of the polymer fibers may be achieved by the electrospinning of dye-doped polymeric solutions \[[@B34-polymers-10-00737],[@B35-polymers-10-00737]\]. Dyes are fluorescent chemical compounds that can re-emit light upon light excitation. Typically, they contain several combined aromatic groups or planar or cyclic molecules \[[@B36-polymers-10-00737]\]. Dyes are commercially available and have been used in many applications, particularly in spectroscopy and medicine. Fluorescent dyes are mainly used in life sciences for non-destructive tracking or analysis of biological molecules \[[@B37-polymers-10-00737],[@B38-polymers-10-00737]\], but they find a wide use in the industry for textile dyeing and as optical brighteners in laundry detergents \[[@B39-polymers-10-00737],[@B40-polymers-10-00737]\], advanced cosmetic formulations \[[@B41-polymers-10-00737]\], safety equipment and clothing, fine arts and design (posters and paintings) \[[@B42-polymers-10-00737]\], synergists for insecticides, and experimental drugs \[[@B43-polymers-10-00737]\]. Organic light-emitting diodes (OLEDs) and solar panels that collect more light (wavelengths) also have a more commercially available use nowadays. Dye-doped polymers have been studied in detail from the point of view of emission tuning with morphology and with composition \[[@B44-polymers-10-00737],[@B45-polymers-10-00737],[@B46-polymers-10-00737]\]. By combining the wavelengths of dyes' emissions that cover the entire visible spectral domain, white-light-emitting fibers can be produced. Up to 19% of worldwide electricity is consumed by lighting; therefore, increased efforts are being made to reduce the electricity dedicated to illumination, such as through the race to replace incandescent bulbs with less-energy-consuming LED lights \[[@B47-polymers-10-00737],[@B48-polymers-10-00737],[@B49-polymers-10-00737]\]. Light adjusts people's daily rhythms, and improving lighting can improve people's well-being (at home) and productivity (in the workplace) \[[@B50-polymers-10-00737]\]. In order to understand how to make this improvement, we have to understand how the human eye and light work. Because humans are trichromats (having in the middle part of the retina three types of cones---the color receptor cells: blue, green, and red), the perceived colors from the visible spectrum are made by mixing various percentages of blue, green, and red \[[@B51-polymers-10-00737]\]. Sunlight has been researched since the beginning of optical science, and numerous efforts have been made for its standardization \[[@B52-polymers-10-00737]\]. The complexity of daylight does not reside only in the mixture of the various colors, but also in the fact that there are changes in the intensities of different components depending on the season, month, place, and even time of day; that is, the morning hours' light is an intense light suited for working, containing an enhanced blue-green component, while the light at dawn is a soothing light with an increased orange-red component. Thus, although commercial LEDs emit differently from sunlight, we perceive white light (of different types) when we use artificial lighting \[[@B53-polymers-10-00737]\]. White-light-emitting low-dimensional structures have a wide range of applications, ranging from spectroscopic lab-on-a-chip sensing devices to IT and communications \[[@B54-polymers-10-00737],[@B55-polymers-10-00737]\]. Numerous types of low-dimensional structures possessing white emission have been developed using different approaches and methods of preparation \[[@B56-polymers-10-00737],[@B57-polymers-10-00737]\]. One can differentiate between these structures either by type of materials (ranging from polymers to oxides or composites) or by morphology and dimensions (ranging from zero-dimensional quantum dots to quasi one-dimensional fibers or complex-shaped structures) \[[@B58-polymers-10-00737],[@B59-polymers-10-00737]\]. By tuning the composition and shape, one can easily define an application field in which these intrinsic properties can be exploited. The aim of this study is to present a simple and straightforward method to produce white-light-emitting structures. Thus, electrospinning was employed to fabricate submicrometric dye-doped polymer fiber mats. The photoemission of these submicronic fiber-based architectures could be tuned depending on the mode in which the doping was performed. As a consequence of the intensities' values corresponding to each dyes' emission maxima in the visible spectral domain, different types of white lighting could be obtained, including sun-like distributions. Two approaches were studied and are described in the paper, mats obtained by combining three single-dye-doped types of fibers and mats of fibers obtained by electrospinning mixtures of solutions doped with several dyes simultaneously. In both cases, high luminescence tunability was obtained, and white light was measured. The luminescent structures obtained in this way possess a specific design that can be employed in a wide range of applications in which one can exploit the fibrous structure of the building blocks, ranging from biological probes to high-surface-area/low-cost lighting devices. Being based on polymer fibers, the structures have the advantages of being lightweight and highly flexible, enabling their integration into different types of devices for which these characteristics are sought. The electrospinning technique allows such fibers to be fabricated simply and inexpensively and ensures they are readily scalable. Moreover, the structures are all organic, with low environmental impact. 2. Experimental Details {#sec2-polymers-10-00737} ======================= For obtaining the dye-doped polymer submicronic fibers, we used poly(methyl methacrylate) (PMMA) with a Mw of 300,000 and chemical formula of (C~5~O~2~H~8~)n at a concentration of 12.5 wt %. Dimethylformamide (DMF; anhydrous, 99.8%) was chosen as the solvent. The polymer solution was prepared at room temperature and stirred overnight. Blue-green- and orange-red-emitting dyes were used as dopants in order for the emitted lights' wavelengths to cover a large area within the visible spectrum. Thus, the polymer solutions were doped with coumarin 7 (C7), coumarin 6 (C6), fluorescein (F), rhodamine 6G (Rh 6G), sulforhodamine B (SRh B), and sulforhodamine 101 (SRh 101). The chemical formula and optical properties of each dye used for doping are listed in [Table 1](#polymers-10-00737-t001){ref-type="table"}. The dyes had 10^−3^ M concentrations in the polymer solutions. All chemical compounds were purchased from Sigma-Aldrich (St. Louis, MO, USA) and were used without any further purification. During the fibers' production through the electrospinning process, the fluid was pumped at a rate of 0.5 mL/h through a syringe with an automatic pump (New Era Pump System). The syringe needle was positively charged using a high-voltage supply (DC power supply: 30 kV Spellman SL300, New York, NY, USA). The resulting electric field caused solution jets to be pulled from the end of the droplet formed on the needle tip (the droplet formed a Taylor cone under the influence of the electric field) reaching, as fibers, the grounded metal collector consisting of a copper frame. The electrodes were positioned at distances between 15 and 20 cm. The high voltage of 20 kV produced the electrically charged jet of the polymer solution, which dried and solidified into a polymer fiber along the trajectory towards the collector. Using high voltages resulted in the drawing of very fine fibers (submicrometric or even nanoscale) from a liquid (polymer solution or melt) with high aspect ratios and with the fiber diameter inversely proportional to the applied voltage. The set-up used to produce the dye-doped polymer fibers is illustrated in [Figure 1](#polymers-10-00737-f001){ref-type="fig"}. Imaging of the solutions and fiber mats collected on copper frames was done at a 365 nm wavelength using a UVIlite lamp LF106L (Cambridge, UK) (6 W, filter 50 × 150 mm, 700 µW/cm^2^). The characterization of the dye-doped polymer submicronic fibers from the point of view of structure and morphology was performed using scanning electron microscopy (SEM) (ZEISS EVO 50 XVP). For SEM imaging, small pieces of the mats were placed on sticky carbon discs. Because of the method used, the density of the fibers observed in the lower-magnification images differed from one sample to another. Prior to the imaging using the scanning electron microscope, the samples were sputtered with a thin layer of gold using an ANATECH Hummer VI DC sputtering system. In order to evaluate the optical properties of the electrospun nanofibers, the absorption and transmission spectroscopy measurements were made using a UV-vis-NIR CARY 5000 spectrophotometer, and the photoluminescence (PL) was evaluated using a FLS920 EDINBURGH INSTRUMENTS spectrofluorimeter (equipped with a 450 W Xe lamp and double monochromators in both excitation and emission). The chromaticity measurements were made using a Minolta CS-2000 spectroradiometer (Konica Minolta, Tokyo, Japan). 3. Results and Discussion {#sec3-polymers-10-00737} ========================= 3.1. Single-Dye-Doped Polymer Fibers {#sec3dot1-polymers-10-00737} ------------------------------------ Submicronic fiber mats were produced using copper frames as collectors during the electrospinning process. Images of the single-dye-doped polymer solutions used for the electrospinning method and their corresponding self-supporting electrospun fiber mats under either visible or UV light are presented in [Figure 2](#polymers-10-00737-f002){ref-type="fig"}. There are many parameters that have to be controlled during the electrospinning process. One has to control the ambient parameters (temperature, humidity, air currents, etc.), the solution parameters such as the polymer concentration (MW and solvent), and also the set-up parameters (the collector geometry that results in different fiber alignments, the distance between electrodes, the applied voltage, the flow rate, and the needle diameter). By varying these parameters, the tailoring of the fibers' morphological properties (diameter, texture, and pattern) becomes possible. The PMMA matrix was chosen as the carrier polymer because the hydrophobic properties of PMMA would ensure the fibers' integrity from both morphological and optical points of view. Besides its favorable optical properties, our experiments have shown that PMMA also offers the mechanical strength necessary for the collection of suspended fibers with submicronic diameters and lengths on the centimeter scale (sample size was 3 cm × 3 cm). Lower- and higher-magnification SEM images of the single-dye-doped polymer fibers produced by electrospinning are presented in [Figure 3](#polymers-10-00737-f003){ref-type="fig"} for all the dyes that were used. The lower-magnification SEM images obtained at 1000× illustrate the random placement of the nanofibers and the uniformity of the diameters' distributions, which are difficult to observe in the higher-magnification SEM images. The fibers' diameters ranged between 300 nm and 1 µm. The thicknesses of the electrospun mats are very important parameters, particularly for the intensities of absorption and luminescence. However, it is difficult to precisely measure such thicknesses because of the fact that the electrospun mats are fluffy (aired), containing lots of empty space, and because their thicknesses cannot be assimilated with the thicknesses of a certain number of overlapped fibers. Because of the way an electrospun fiber mat is formed in the optically transparent regime (i.e., at low fiber densities), it is difficult to define a meaningful geometrical thickness, as up to 90--95% of the mat can be empty space. This is further complicated by the flexibility of the fibers, which makes the mats compressible in an uncertain way. However, given the optical nature of the application at hand, a relevant characteristic is the optical transmission of the fiber mats in the visible range. This is directly linked to the true geometrical thickness but is far easier to measure. As it depends on the fiber density, this "thickness" can be controlled by varying the collection time. All the samples were produced using the same collection time, which means that they should have had approximately the same transmission values and fiber densities. In order to demonstrate the similarity of the samples regarding the measured thicknesses, we evaluated the transmission spectra in the interval 600--800 nm in which the transmittance could be compared, because in this wavelength interval, there were no absorption bands of the dyes. Information regarding the optical transmission values of the samples are presented in the [Supplementary Materials, Figure S1](#app1-polymers-10-00737){ref-type="app"}. Previous studies have shown that the presence of different dyes in polymer solutions leads to changes in the morphologies of the electrospun nanofibers as a result of the different sizes of dye molecules \[[@B45-polymers-10-00737]\]. In order to avoid fabricating fibers with different morphologies, the concentration of the polymer in the solution was increased compared to prior investigations \[[@B44-polymers-10-00737],[@B45-polymers-10-00737]\]. Thus, for a 12.5 wt % concentration of PMMA in DMF, uniform submicronic fibers were obtained by electrospinning for all single-dye-doped polymer solutions. It could be observed that all the fibers had smooth surfaces (e.g., surfaces of Rh 6G-doped fibers are presented in [Figure 4](#polymers-10-00737-f004){ref-type="fig"}). Rh 6G was chosen as an example because its molecule has an average size when compared with the other dyes used in this study. However, for the C6-doped polymer fibers, we observed the rare occurrence of a beads-on-string morphology ([Figure 5](#polymers-10-00737-f005){ref-type="fig"}), likely formed at the beginning of the electrospinning process during the period when a stabilization of the process takes place. The appearance of such a specific polymer fiber morphology for the electrospinning of the C6-dye-doped solution is consistent with the previous studies \[[@B45-polymers-10-00737]\]. The appearance of beads on fibers is strongly related to the viscosity and surface tension (among other parameters) \[[@B60-polymers-10-00737],[@B61-polymers-10-00737]\]. In the C6-doped PMMA fibers, the appearance of beads was likely related to the intrinsic properties of C6, leading to the highest probability for the occurrence of such morphology. It is well known that C6 fluorescence in solvents depends critically on concentration and on the pattern of substitution for the amine group \[[@B62-polymers-10-00737]\]. Relatively large shifts in both absorption and fluorescence were reported not only in solutions but also in polymers \[[@B63-polymers-10-00737]\]. C6 is known for its high degree of sensitivity to its local environment, which may result in alterations to the morphological and optical host's properties, including viscosity \[[@B64-polymers-10-00737],[@B65-polymers-10-00737]\]. Thus, the presence of C6 may have affected the viscosity and the surface tension of the polymer solution more so than for the other dyes, leading to the rare appearance of beads during the electrospinning process. The optical absorbance and PL spectra of the single-dye-doped fiber mats were measured and plotted, and the results are presented in [Figure 6](#polymers-10-00737-f006){ref-type="fig"}a,b. As expected, the optical absorptions of the single-dye-doped fiber mats covered the spectral domain 400--600 nm with increasing intensities from the blue region towards the red region, while the emissions covered the spectral domain 450--650 nm with higher intensities in the blue-green region and significantly lower intensities of the emission in the orange-red region of the visible spectra. The decreasing of the PL intensities was in total agreement with studies from the literature related to the efficiencies of the dyes when dissolved in different solvents \[[@B66-polymers-10-00737],[@B67-polymers-10-00737],[@B68-polymers-10-00737]\]. There is a strong relation between PL and the molecular structure of dyes, indicating different efficiencies. This further leads to different PL intensities, although the dyes had the same concentration (10^−3^ M) in the solution of PMMA (12.5 wt % in DMF). Depending on the conditions and solvent, fluorescein may dimerize. A dimer--monomer equilibrium appeared, thus leading to the appearance of different peaks in the absorption spectra \[[@B69-polymers-10-00737]\]. Moreover, different forms (cationic, neutral, anionic, and dianionic) of the fluorescein's molecule may be found in a dye-doped solution \[[@B70-polymers-10-00737]\]. Therefore, we may observe more peaks in both absorption and emission spectra of a F-doped polymer fiber mat. Absorption peaks appeared also at wavelengths of less than 350 nm, in the UV region of the spectrum. The 420 nm excitation with C7-doped fibers revealed the overlapping of two peaks appearing at 485 and 505 nm, whereas for C6-doped fibers, two peaks at 495 and 515 nm appeared under 440 nm excitation. The appearance of the shoulder in PL may have been due to the protonated form of the dye, which was previously reported for coumarins \[[@B66-polymers-10-00737]\]. The 490 nm excitation with F-doped fibers revealed a broad PL spectrum formed by an overlapping of peaks between 515 and 575 nm. The Rh 6G-doped fibers exhibited a peak at 560 nm under 510 nm excitation, while under 550 nm excitation, the SRh B-doped fibers presented a PL peak at 580 nm and the SRh 101-doped fibers presented an emission peak at 590 nm. Next, the emissive properties of the overlapping single-dye-doped fibers' layers were evaluated. By varying the thickness of the layers, we were able to tune the emission spectra of the overlapped fiber mats and thus able to enhance the blue and red contributions to the spectra or to obtain a balanced combination of the peaks. The downside of such an approach is that there are a limited number of layers that can be used and that too many layers overlap, further resulting in the obstruction of light and the screening of the first layers' emissions. PL excited at 420 nm for different overlapping combinations of three single-dye-doped fiber mats, C6, Rh 6G, and SRh 101, are presented in [Figure 7](#polymers-10-00737-f007){ref-type="fig"}. The volumes of the dye-doped solutions electrospun for the different overlapping fiber mats were varied, leading to a variation in the thicknesses of the layers, until a balanced PL spectrum was achieved. Thus, the balance in sample 2 was achieved by overlapping C6, Rh 6G, and SRh 101 mats in a 1/5/10 ratio, while sample 1 was formed by overlapping C6, Rh 6G, and SRh 101 submicronic fibers in a 10/5/5 ratio and sample 3 was formed by overlapping C6, Rh 6G, and SRh 101 submicronic fibers in an equal 10/10/10 ratio. 3.2. Dye-Doped Polymer Fibers Obtained from Mixed Solutions {#sec3dot2-polymers-10-00737} ----------------------------------------------------------- Copying the complexity of white light requires a more versatile approach; therefore, we used mixed single-dye-doped polymer solutions in different percentages in order to be able to tune the emissions of the electrospun fiber mats. [Table 2](#polymers-10-00737-t002){ref-type="table"} presents the whole range of mixed solutions used for testing the emissive properties of the polymer fiber mats. Nine mixtures were prepared using single-dye-doped polymer solutions. Mixtures a, b, and c were prepared without the F solution, and a pronounced gap appeared in the PL spectra in the range 520--540 nm (observed in [Figure 7](#polymers-10-00737-f007){ref-type="fig"} for overlapping mats). Therefore, although having a complicated overlapping of peaks in the PL spectrum due to the presence of various aggregates, as previously explained, fluorescein was found to be a good solution for leveling the spectrum in the yellow region. Thus, six additional mixtures containing six different dye-doped polymer solutions in different ratios (including fluorescein) were prepared. The spectra of mixtures d, e, and f were only slightly different from the spectrum of mixture 3. Therefore, only three mixtures were chosen to be further analyzed by optical and morphological measurements, namely, mixtures 1, 2, and 3, which illustrate the possibility of balancing the intensities between red and blue emissions. PL spectra of the fiber mats produced by the electrospinning of different combinations of single-dye-doped polymer solutions (mixtures 1, 2, and 3) are presented in [Figure 8](#polymers-10-00737-f008){ref-type="fig"}a. The spectra illustrate the possibility to balance the luminescence between blue and red emissions, depending on the ratios of the different dye-doped solutions. Commission Internationale de L'Eclairage (CIE) coordinates of the fibers obtained by electrospinning of different mixtures are presented in [Figure 8](#polymers-10-00737-f008){ref-type="fig"}b. The CIE measurements depicted the fine-tuning of the intensities of colors emitted by the fiber mats composing the white light. Thus, the emissions of the submicronic fibers obtained from the C7/C6/F/Rh6G/SRhB/SRh101 mixtures had the CIE coordinates of (0.27, 0.29) for the composition 10/20/20/20/10/20 (mixture 3), (0.38, 0.44) for the composition 15/30/30/10/7.5/7.5 (mixture 2), and (0.44, 0.44) for the composition 12.5/25/40/7.5/7.5/7.5 (mixture 1). The excitation wavelength for all the fibers obtained by the electrospinning of mixed solutions was 420 nm, corresponding to the excitation wavelength of C7, the dye with the smallest values for both the excitation and the emission wavelengths. The electrospinning of the mixture formed by C7/C6/F/Rh6G/SRhB/SRh101 in a ratio of 15/30/30/10/7.5/7.5 (mixture 2; presented in [Figure 8](#polymers-10-00737-f008){ref-type="fig"}a) resulted in a submicronic fiber mat that exhibited white-light emission. The UV image of the white-light-emitting fiber mat (mixture 2) compared with the green-emitting and red-emitting fiber mats is presented in [Figure 9](#polymers-10-00737-f009){ref-type="fig"}. It could be observed both for the overlapping fiber mats and for the fiber mats obtained by electrospinning mixtures of single-dye-doped polymer solutions that for equal amounts of different dyes, the red contribution to the spectra had a higher intensity. The increased intensity of the red contribution appeared as a result of the fact that the red component was excited by both the wavelength of excitation and by the light emitted by the green-yellow components of the mixtures. Morphology studies have showed that electrospinning mixtures of dye-doped polymer solutions in different combinations results also in submicronic fibers with smooth surfaces, uniform dimensional distribution, and diameters comparable to those of single-dye-doped electrospun polymer fibers. The diameters of the fibers obtained by electrospinning mixtures of solutions also ranged between 300 nm and 1 µm. The SEM images of the submicronic fibers are presented in [Figure 10](#polymers-10-00737-f010){ref-type="fig"}. The dye-doped polymer fibers produced by electrospinning presented similar optical properties over a long period of time. The PL of the electrospun fibers was very stable. The fibers showed identical PL signals after storage under normal atmospheric conditions. Surface enrichment of the dye is commonly observed in dye-doped polymeric systems \[[@B71-polymers-10-00737]\]. The fibers may have presented a possible segregation of the dyes on their surfaces during their formation, but this was difficult to be confirmed experimentally because of the fact that the solidification of fibers takes place over a very short period of time. 4. Conclusions {#sec4-polymers-10-00737} ============== We present a simple method to fabricate flexible, white-light-emitting nanofibers using the electrospinning methods of dye-doped polymers. Different mixtures of single-dye-doped polymer solutions were used in order to tune the emissions of electrospun nanofibers. The present study shows that electrospun nanofibers obtained from mixtures of single-dye-doped solutions present white-light emissions and may have potential applications as new light sources or sensory materials for smart textiles. White-light-emitting electrospun submicronic fibers were successfully prepared through the electrospinning of mixtures of single-dye-doped PMMA solutions in DMF using a single capillary spinneret. The emission intensities of colors composing the white light changed as the composition of the mixtures changed, increasing either the blue-green component or the orange-red component of PL. The white-light-emitting submicronic fibers obtained from the C7/C6/F/Rh6G/SRhB/SRh101 mixture with composition 15/30/30/10/7.5/7.5 had CIE coordinates of (0.38, 0.44). Our results show that white-light-emitting fibers can be produced through the use of multicomponent single-dye-doped PMMA mixtures. The authors acknowledge the financial support provided in the frame of Romania-France IFA-CEA collaboration from Project No. C5-08/2016. The authors thank Silviu Polosan for help and assistance with the chromaticity measurements. The following are available online at <http://www.mdpi.com/2073-4360/10/7/737/s1>, Figure S1: Transmission spectra for the single-dye-doped polymer fiber mats produced by electrospinning. ###### Click here for additional data file. M.E. carried out electrospun fibers' morphology and optical characterization measurements. A.E. fabricated the dye-doped self-supporting PMMA mats and optimized the conditions to obtain homogeneous electrospun fibers. M.E. and I.E. proposed and planned the research and cooperated with the analysis of results. All the authors collaborated with the writing of the reported work. This research was funded by IFA-CEA Project number C5-08/2018. The authors declare no conflict of interest. ![Schematic illustration of the experimental setup used for electrospinning of the dye-doped polymer solutions.](polymers-10-00737-g001){#polymers-10-00737-f001} ![Images of the polymer solutions and the corresponding electrospun submicronic fiber mats collected on copper frames in visible light (upper) and UV light (lower).](polymers-10-00737-g002){#polymers-10-00737-f002} ###### Scanning electron microscopy (SEM) images at two different magnifications (1000× and 10,000×) of the electrospun polymer fibers doped with (a,a') coumarin 7, (b,b') coumarin 6, (c,c') fluorescein, (d,d') rhodamine 6G, (e,e') sulforhodamine B, and (f,f') sulforhodamine 101. ![](polymers-10-00737-g003a) ![](polymers-10-00737-g003b) ![Scanning electron microscopy (SEM) image presenting the surface uniformity for rhodamine 6G (Rh 6G)-doped polymer fibers.](polymers-10-00737-g004){#polymers-10-00737-f004} ![Beads-on-string morphology observed for coumarin 6 (C6) sample: 1000× (a), and 10,000× (b).](polymers-10-00737-g005){#polymers-10-00737-f005} ![Absorption (a), and photoemission (b) spectra of single-dye-doped polymer fibers.](polymers-10-00737-g006){#polymers-10-00737-f006} ![Photoluminescence (PL) spectra of overlapping mats of coumarin 6, rhodamine 6G, and sulforhodamine 101 doped fibers with different thicknesses.](polymers-10-00737-g007){#polymers-10-00737-f007} ![(a) Photoluminescence (PL) spectra of submicronic fibers produced by electrospinning from different mixtures of dye-doped polymer solutions, and (b) their corresponding coordinates on Commission Internationale de L'Eclairage (CIE) 1931 color space chromaticity diagram.](polymers-10-00737-g008){#polymers-10-00737-f008} ![Blue-green (coumarin 7 (C7))-, white-light (mixture 2)-, and orange-red (sulforhodamine 101 (SRh 101))-emitting flexible and thin fiber mats under UV light excitation (365 nm).](polymers-10-00737-g009){#polymers-10-00737-f009} ###### Scanning electron microscopy (SEM) images of submicronic fibers obtained by electrospinning of mixtures of single-dye-doped polymer fibers: (a,a') mix 1; (b,b') mix 2; (c,c') mix 3. ![](polymers-10-00737-g010a) ![](polymers-10-00737-g010b) polymers-10-00737-t001_Table 1 ###### Chemical formulas, molecular structures, and optical properties of the dyes used for doping the polymer solutions. Dye Chemical Formula Molecular Structure Absorption Wavelengths Emission Wavelengths -------------------- -------------------------- --------------------- ------------------------ ---------------------- Coumarin 7 C~20~H~19~N~3~O~2~ 437 and 455 nm 485 and 505 nm Coumarin 6 C~20~H~18~N~2~O~2~S 448 and 465 nm 495 and 515 nm Fluorescein C~20~H~12~O~5~ 430, 457, and 487 nm 515, 550, and 575 nm Rhodamine 6G C~28~H~31~N~2~O~3~Cl 534 nm 560 nm Sulforhodamine B C~27~H~29~N~2~NaO~7~S~2~ 550 nm 580 nm Sulforhodamine 101 C~31~H~29~ClN~2~O~6~S~2~ 575 nm 590 nm polymers-10-00737-t002_Table 2 ###### Composition of dye-doped polymer solution mixtures. Solution Percentage of Single-Dye Solution ---------- ----------------------------------- ------- ------ ------- ------ ------ Mix a 20 20 \- 20 20 20 Mix b 16.66 33.33 \- 33.33 8.33 8.33 Mix c 12.5 25 \- 25 12.5 25 Mix 3 10 20 20 20 10 20 Mix d 12.5 22.5 22.5 22.5 10 10 Mix e 17 22.5 22.5 22.5 7.5 7.5 Mix f 5 25 25 25 5 5 Mix 2 15 30 30 10 7.5 7.5 Mix 1 12.5 25 40 7.5 7.5 7.5	{ "pile_set_name": "PubMed Central" }
Intensive care medicine also referred to as critical care medicine is that body of specialist knowledge and practice which is concerned with the treatment of patients who are at risk of recovering from potentially life threatening failure of one or more of the body organ systems. It includes provision of organ support system, investigations, diagnosis and treatment of acute illness, systems management and patient safety, ethics, end of life care and support to families. Cardiothoracic anaesthesiology is a sub specialty of anesthesiology devoted to pre, intra and postoperative care of patients undergoing cardiothoracic surgery and related procedures. Cardiac surgery patients are different; they are older, sicker and more frail. Also newer procedures like transcutaneous aortic valve implantation (TAVI) and transcutaneous endovascular aortic repair (TEVAR) are performed in moribund patients\[[@ref1]\] so their postoperative critical care becomes even more important. Therefore delivery of high quality critical care medicine is vital to the success of cardiac surgery.\[[@ref2]\] One needs the right operator to use right information making the right decision to use the right tool to perform the right task at the right time in the right manner! Critical care is growing at a rate of 1% of GNP in the US.\[[@ref3]\] Nearly three quarters of the care by intensivist is delivered by the 'open intensive care unit (ICU)' model.\[[@ref4]\] Intensivist makes recommendations but has no authority over patient care while the admitting physician who is neither trained in critical care nor is available 24 × 7 makes the final treatment decisions. Various other consultant who look at single organs make conflicting therapeutic decisions resulting in confusing orders, longer ICU stay and cost. Hanson et al. showed that patients cared for by the critical care team spent less time in the ICU, used fewer resources with lesser cost and all this was more obvious in sicker patients i.e. patients with higher APACHE II scores.\[[@ref5]\] Also in metaanalyses high intensity or closed ICU was associated with lower ICU mortality and length of stay than in open ICU.\[[@ref6][@ref7]\] In a Canadian study in a large university trauma centre with a mixed (medical + surgical ICU) the authors found no significant difference in mortality in patients managed by intensivist with core training backgrounds in either internal medicine or in surgery/anaesthesiology.\[[@ref8]\] So, it is recognised that closed ICU is superior in terms of outcome compared with open ICU and anaesthesiologists is as good if not superior to someone from internal medicine background! Post cardiac surgery patients require intensive monitoring, judicious use of cardiovascular drugs, effective pain control, early mobilization and intensive respiratory therapy.\[[@ref9]\] The cardiothoracic intensivist should be expert at perioprative care with timely manipulation of cardiopulmonary physiology through precise and advanced application of pharmacology, resuscitative techniques, critical care medicine and invasive procedures. In monitoring techniques pulmonary artery catheter (PAC) is still widely used in cardiac critical care unlike general ICU where it is on the decline. A multicentre study in the setting of non-emergency coronary artery bypass graft surgery (CABG) concluded that use of PAC was associated with increased hospital mortality, greater length of stay and higher total cost particularly with low volume PAC use and \<50 surgeries per annum.\[[@ref10]\] This just proves the point that with increased experience e.g. large PAC use in a busy cardiac ICU with experienced cardiac anaesthetists as intensivist this could be different. We in our group perform about 5000 cardiac surgical cases per year and most of the patients get a PAC and all of these are inserted by the anaesthesiologist. So which other specialist would be more experienced to insert, interpret and utilize PAC for hemodynamic manipulations than us in a cardiac ICU? The other modality for haemodynamic monitoring particularly for cardiac surgical patients preoperatively is Transesophageal echocardiography (TEE). This has become the 'Gold Standard' for preload estimation and assessment of new segmental wall motion abnormality (SWMA) indicating ischaemia, global systolic and diastolic ventricular function, valvular pathology and aortic dissection. Most of the publications on TEE are either by cardiac anaesthesiologists or by non invasive cardiologists (which is an extinct species now!) so who is better qualified to perform TEE than cardiac anaesthesiologists? In fact even the guidelines for TEE are jointly made by Society of Cardiac Anesthesiogy and American Society of Echocardiography.\[[@ref11]\] We have also found TEE to be the best modality for quick diagnosis of hameodynamic instability in patients in cardiac surgical ICU\[[@ref12]\] and also to recognize embolisation and subsequent stroke due to embolisation of aortic mobile atheroma.\[[@ref13]\] The only formal training in India with certification for TEE is done by Indian Association of Cardiovascular Throacic Anaestheisa (IACTA).\[[@ref14]\] This itself is enough evidence to prove that cardiac anaesthesiologist is best equipped to carry out a postoperative TEE. With the decline in use of PAC in critical care worldwide, newer methods of haemodynamic assessment are being investigated.\[[@ref15]\] TEE, pulse contour analysis (PiCCO),\[[@ref16]\] lithium dilution technique (LiDCCO),\[[@ref17]\] flotrac (Edwards Lifesciences Corp, Irvine, CA 92614),\[[@ref18]\] Transesophageal doppler (TECO),\[[@ref19]\] have been investigated in cardiac critical care and some of these are being used, in cardiac surgical ICU\'s. Most of these have been validated against PAC and most of the studies are by cardiac anaesthesiologists! Who is better equipped to use, interpret and take appropriated therapeutic actions based on these in a cardiac surgical patient than the cardiac anaesthesiologists? In a national survey in U.K. it was found that of the 39 cardiac intensive cares contacted, 94% had consultants cardiac anaesthesiologist as the senior most incharge physician.\[[@ref20]\] With the advent of off pump coronary artery bypass graft surgery (OPCAB) the morbidity associated with cardiopulmonary bypass (CPB) like systemic inflammatory response syndrome (SIRS), platelet destruction and cougulopathy with subsequent blood and blood product requirement, renal dysfunction and neurocognitive dysfunction and stroke have been significantly reduced.\[[@ref21]\] Also shorter anaesthetic and surgery with minimal access should lead to earlier extubation i.e. fast tracking.\[[@ref22]\] Who understands and can implement it better than the cardiac anaesthesiologist? Also postoperative analgesia is an extremely important part of the post operative care of these patients for early mobilization and shorter ICU length of stay and attenuation of stress response. For these various regional analgesia techniques can be used like thoracic epidural and interpleural anaesthesia,\[[@ref23]\] subarchanoid analgesia,\[[@ref24]\] paravertebral block\[[@ref25]\] and nowadays local anaesthetic infusion into the wound, Cardiac anaesthetists can manage all these better than other critical care specialists or cardiac surgeons. Ventilatory strategy is important in post cardiac surgical patients. Incidence of ventilator associated pneumonia (VAP) is directly proportional to the duration of ventilation.\[[@ref26]\] ARDS network stratetgy\[[@ref27]\] may have to be modified as permissive hypercapnia, high PEEP and traditional recruitment manoeuvres are detrimental in cardiac surgical patients. Prone ventilation has been used in these patients\[[@ref28]\] and recently it has been shown to have mortality benefit.\[[@ref29]\] Nitric oxide (NO) is another modality which has been used in refractory hypoxaemic and right heart failure with high pulmonary artery pressures.\[[@ref30]\] In refractory cases extracorporeal membrane oxygenation (ECMO) is being used\[[@ref31]\] although with a higher incidence of sepsis. Cardiac anaesthesiologist is familiar and comfortable to deal with all these technologies. Another important aspect of cardiac critical care is blood sugar control. The important paper which changed practice was by Van der Bergh\[[@ref32]\] which showed mortality benefit in mostly cardiac surgical ICU by a tight sugar control. Subsequent work although has not reproduced it and the current practice is to keep blood sugar between 120-180 mg%.\[[@ref33]\] We have recently shown how a graded scale for sugar control in cardiac surgical ICU can be successfully implemented.\[[@ref34]\] So although in my opinion cardiac anaesthesiologists is best equipped to be incharge of cardiac ICU but they need also to be trained in the other critical care management and procedures may be through a formal training by IDCCM, IFCCM or FNB. A lot has been done but a lot needs to be achieved still in this field. The woods are lovely dark and deep but I have promises to keep and miles to go before I sleep and miles to go before I sleep! (Robert Frost).	{ "pile_set_name": "PubMed Central" }
Gene transfer into experimental animals or humans resulting in generalized or tissue-specific expression may allow precise in-vivo manipulation of biological processes to cure diseases and induce immune responses to pathogens ([@B1], [@B2]). The basic challenge in gene therapy is to develop approaches to the delivery of genetic material to appropriate cells in a way that is specific, efficient, and safe. Continuous efforts have been applied to the development of gene delivery systems known as vectors, which encapsulate the gene and guide it to the target cell. Viruses are ideal vectors, being naturally suited to the highly efficient transfection of genetic material to cells. The use of viral vectors is, however, limited by safety concerns related to the immune and inflammatory responses they trigger and immune rejection phenomena arising due to repeated administration ([@B1], [@B3], [@B4]). As an alternative to viral vectors, various non-viral gene delivery systems, including cationic lipids, cationic polymers, and naked DNA, have been prepared for use in gene therapy. In liver-directed gene therapy involving radiologists, approach routes may include direct percutaneous injection, and transcatheter intra-arterial or intraportal administration. In the case of viral vectors, it has been reported that intra-arterial administration is more efficient than intravenous administration ([@B5]), though cationic lipid vectors have not been compared in this way. The purpose of this study was to compare the efficiency of intra-arterial, intraportal, and intravenous administration of cationic lipid emulsion/DNA complex, as used for gene transfer to rat liver. MATERIALS AND METHODS ===================== Preparation of Cationic Lipid Emulsion -------------------------------------- The emulsion we used contained 100 µL/mL oil (squalene) and lipid emulsifiers \[1, 2-dioleyl-sn-glycro-3-trimethylammonium-propane (DOTAP), and 1, 2-dioleyl-sn-glycro-3-phosphoethanolamine (DOPE), combined in a ratio of 5:1 by weight\], and was prepared as described previously ([@B6]). Briefly, lipid emulsifiers were weighed and dispersed in water, and the resulting mixture was sonicated in an ice/water bath using a probe type sonicator (high intensity ultrasonic processor, 600 W model; Sonic and Materials, Danbury, Conn., U.S.A.). The lipid solution was added to oil, and the mixture was sonicated further in an ice/water bath. Prior to use, the cationic lipid emulsion thus prepared was kept at 4℃, and its short-term stability was monitored by measuring the time-dependant absorbance changes occurring at 600 nm. The average size of emulsion particle was 164.5 nm. Preparation of Plasmid DNA -------------------------- As reporter genes, we used pCMV-Luc+ and pCMV-β. The latter, encoding Escherichia coli (E. coli) l acZ (β-galactosidase) gene expression plasmid driven by the human cytomegalovirus immediate-early promoter, was supplied by Clontech Laboratories (Palo Alto, Cal., U.S.A.), and the pCMV-Luc, consisting of the cytosolic form of Phontinus pyralis luciferase cDNA, was obtained from pGL3 (Promega, Madison, Wis., U.S.A.) using Xba I and Hind III restriction and was subcloned into the plasmid pcDNA3.1 (Invitrogen, Seoul, Korea). Both plasmids were amplified in the E. coli DH5-α strain and purified using a Qiagen mega-kit (Qiagen Inc., Chatsworth, Cal., U.S.A.), according to the manufacturer\'s instructions. The purity of the DNA used (OD~260~/OD~280~≥1.8) was determined by agarose gel electrophoresis and the measurement of optical density. Animal Studies -------------- To prepare DNA-carrier complexes for this experiment, 20 µg of pCMV-Luc+ and the carrier, the amount of which corresponded to the weight ratio between cationic lipid in the lipid formulation and DNA in the complex that showed the maximum transfection efficiency, were diluted with 0.5 mL of DMEM (Dulbecco\'s modified Eagle\'s medium) solution and mixed by inversion. The total volume of the mixture was 1 mL, and the time interval between mixing and infusion was minimized. Twenty-four 8-week-old Sprague-Dawley rats, each weighing 200-300 gms, were used in this study; their housing and the procedure employed were in accordance with the National Institutes of Health guidelines. The animals were divided into three groups according to whether injection was performed intra-arterially (n=9), intraportally (n=8), or intravenously (n=7). Three other rats which did not undergo treatment were included as negative controls. The animals were anesthetized in an ether jar prior to the injection of a 1:1 mixture of xylazine hydrochloride (Rompun; Bayer Korea, Seoul, Korea) and ketamine hydrochloride (Ketara; Yuhan Yanghang, Seoul, Korea) into the peritoneal cavity. Those in which administration was intra-arterial or intraportal were prepared for surgery, and a midline ventral incision was made. The liver was then retracted and connective tissues were cleared to visualize the gastroduodenal artery or mesenteric veins. For intra-arterial administration, the gastroduodenal artery was exposed and cannulated with a 28-G needle, and the DNA-carrier complex was slowly injected into the proper hepatic artery at a rate of 1 mL/min. The catheter was then removed and the gastroduodenal artery was ligated; the abdominal incision was sutured using the continuous interlocking method. For intraportal administration, a tributary of the superior mesenteric vein was selected, a 28-G needle was positioned in it and secured, and the DNA-carrier complex was delivered at the same rate as previously. The catheter was removed and the catheterized vein ligated, and the abdominal incision was closed in the same manner as before. For intravenous administration, the DNA-carrier complex was injected into the vein of the tail using a 24-G catheter, which was subsequently slowly flushed with saline and removed. Luciferase Activity Assay ------------------------- Two rats in Group II, in which administration was intraportal, expired prior to sacrifice. The remaining 22 rats survived and were sacrificed 24 hours after the procedure. The heart, lung, liver, spleen and kidneys were removed from each animal and underwent, separately, high-speed homogenization using T-25-Ultra-Turrax homogenizer (Janke & Kunkel GmbH, KG, Germany) and, for each collected organ, 5 µl/mg of pH 7.8 lysis buffer comprising 0.1 M Tris-HCl, 2 mM EDTA and 0.1% Triton X-100. After two freeze/thaw cycles, the homogenized organ lysates were centrifuged for 10 mins at 4℃ and at 12,000 rpm in an Eppendorf centrifuge. A portion of the supernatants was assayed for protein concentration using a Bio-Rad protein assay kit (Bio-Rad Laboratories, Hercules, Cal., U.S.A.). Luciferase activity in the lysates was quantified using a Promega kit (Madison, Wis., U.S.A.) with a luminometer (Turner Designs Luminometer, Model TD-20/20; Promega); peak light emission was measured for 20 seconds at room temperature, and the luciferase count was calculated from relative light units using a standard curve obtained from purified firefly luciferase (Sigma). The luciferase count of each organ in each rat was tabulated, and intra-arterial, intraportal, and intravenous administration were compared in terms of the efficiency with which they facilitated gene transfer to the liver. For all groups, mean values and standard deviations of luciferase enzyme activity were calculated, and using Wilcoxon\'s rank sum test, the statistical significance of differences in enzyme activity between the different routes of administration was determined. RESULTS ======= The amount of luciferase enzyme in each organ, which varied according to the route of administration of the cationic lipid emulsion/DNA complex, is summarized in [Table 1](#T1){ref-type="table"} and [Figure 1](#F1){ref-type="fig"}. In groups I and II, luciferase activity was highest in the liver (22.94 and 115.21 pg/mg protein, respectively), but in group III, it was highest in the spleen (5.94 pg/mg protein). Luciferase activity in each organ of untreated animals was negligible. Luciferase values in the liver were significantly higher in groups I and II than in groups III (p = 0.0080 and 0.0034, respectively), though between groups I and II there was no significant difference (p = 1.0). DISCUSSION ========== Since the liver is the site of many metabolic diseases and malignancies, it is an organ in which gene therapy may be particularly valuable. The potential indications for liver-directed gene therapy include inherited liver disorders such as familial hypercholesterolemia ([@B7]) and ornithine transcarbamylase deficiency ([@B8]); inherited systemic disorders including hemophilia ([@B9]); viral hepatitis, liver tumors, allograft or xenograft rejection, and ischemia/reperfusion injury ([@B10]), and are expanding continuously. The availability of both the intravenous approach and local delivery routes is another important reason why the liver has been a model organ in the development of gene transfer technology. Except for recombinant adenoviruses and a few vectors with targeting moiety for hepatocytes, most viral and non-viral vectors are not liver specific. To achieve efficient delivery, these vectors have been injected into the liver or peritoneal cavity; infused into the portal vein or hepatic artery; or instilled into the bile duct. The purposes of local administration of vectors are to limit their transgenic expression in the liver and to enhance the efficiency of transfection. In an experimental study using a recombinant adenovirus encoding wild-type p53 or β-galactosidase, intra-arterial delivery increased the transgenic expression in tumor tissue and decreased systemic exposure in comparison with intravenous delivery ([@B5]). In murine liver metastases, tumors showed significantly higher transgenic expression after portal venous or intraperitoneal virus administration ([@B11]). There are two kinds of cationic lipid-based gene delivery systems, namely liposome and emulsion. A cationic liposome is a closed double layer of cationic lipids filled with water. It is positively charged and interacts with negatively charged DNA molecules to form a stable complex. Cationic liposomes have been used widely in gene transfer both in vitro and in vivo ([@B12]). The oil-in-water emulsion used consists of oil dispersed in the aqueous phase with a suitable emulsifying agent such as a phospholipid or nonionic surfactant. The use of cationic lipid emulsions as gene delivery vectors has not, however, been widely investigated. Cationic lipid-mediated gene delivery is not yet as efficient as the use of a viral vector. Although intravenously-injected liposomes have an intrinsic affinity for the liver, the uptake of liposomes by this organ in the absence of a targeting mechanism is highly variable ([@B13]-[@B15]), and most liposomal carriers lose their ability to transfer DNA in the presence of ≥10% serum ([@B12], [@B16], [@B17]). Because of the physical instability of the liposome/DNA complex and the inactivation of complexes by negatively charged molecules in serum, its transfection efficiency has been found to be relatively low compared with that of viral vectors ([@B18], [@B19]). To enhance efficiency, a new formulation offering serum stability and a strategy for local administration is therefore needed. To overcome the problems associated with liposomes, castor oil emulsions have been introduced as alternative gene transfer vectors ([@B12]); new cationic lipid emulsions use squalene or soybean oil as a core lipid ([@B20]). It has been shown that in the presence of serum, the use of a cationic lipid emulsion/DNA complex has led to efficient cell transfection ([@B6]). In in-vitro transfection assay of this complex, more than 60% of transfection efficiency was retained in the presence of up to 90% serum ([@B6]). Unlike liposomal carriers, cationic lipid emulsion retained its physical integrity while forming a complex with DNA. In-vitro DNA release tests have shown that binding of the emulsion/DNA complex was strong, and it was thus unaffected by the anionic polyion, poly-L-aspartic acid ([@B16]). In-vivo, this stable emulsion system delivered genes to endothelial cells in the mouse nasal cavity more efficiently than the commercially available liposomes, Lipofectamine and Lipofectin ([@B4]). We conducted this animal experiment to determine whether local delivery of a cationic lipid emulsion/DNA complex through the hepatic artery or portal vein is advantageous, and our results clearly showed that intra-arterial or intraportal administration is superior to intravenous administration. Several important issues in cationic lipid-mediated gene delivery await solution. For example, the distribution of liposomal vectors to various organs, and the efficiency with which this is achieved, are affected by factors which include liposome/DNA complex particle size ([@B21]-[@B22]). The use of a cationic lipid emulsion for gene transfection has not been thoroughly investigated, though it has been shown that after its intravenous administration, gene expression in the lung is usually higher than in the spleen and other organs ([@B23]). In our study, on the other hand, luciferase activity in the lung was quite low, probably because of the stability of the complex used in the experiment. Further investigation of the biodistribution of lipid emulsion in each specific organ is, however, required. In hepatic gene delivery, the optimal particle size of a liposome or emulsion, for example, may differ according to whether administration is intra-arterial or intraportal, or intravenous, and research into this is also needed. As a further example, the liver contains cell types which include hepatocytes, Kupffer cells, endothelial cells and various tumor cells, and in hepatic gene therapy, target cells can vary according to the disease being treated. In metabolic diseases, the primary target is hepatocytes, while in hepatic tumors, gene expression in tumor cells should be maximized and gene delivery to normal parenchyma minimized. In clinical or experimental studies using various different vectors, a clear understanding of exactly where transgenic expression should occur is therefore important. Because the majority of liposomes administered intravenously are endocytosed by the reticuloendothelial system ([@B15], [@B17]), the reduction of Kupffer cell uptake and enhancement of hepatocyte or tumor uptake are further challenges faced by research into liposome targeting. To enhance targeted gene delivery, lipid carriers can be modified ([@B24]): polyethylene glycol conjugate or ligands for receptors, for example, can be incorporated into lipid carriers. An amphipathic polyethylene glycol such as Tween 80 can provide a steric barrier, as well as increasing the hydrophilicity of the liposome surface, both of which may reduce interaction between the liposome and plasma proteins and their recognition and uptake by macrophages of the reticuloendothelial system ([@B25]). Hepatocytes have glycosyl receptors on their surface, and the hepatic uptake of glycosylated liposome can be enhanced by receptor-mediated endocytosis. Other ligands for receptors under investigation include lactose, transferrin, and antibodies. A recent report described selective gene expression in hepatic VX2 carcinoma after intra-arterial delivery of the DNA/liposome/transferrin complex ([@B26]). The hemodynamic characteristics of this and other hepatic tumors are quite different from those of normal liver parenchyma: while normal hepatocytes, for example, receive blood primarily from the portal vein, the hepatic artery supplies nearly all the blood required by a hepatic malignancy. Because of these hemodynamic differences, the optimal conditions for lipid carriers may differ according to the target cells involved (hepatocytes or hepatic malignancies). Although local delivery of a cationic lipid emulsion/DNA complex via the hepatic artery or portal vein increases hepatic transgenic expression, it may also maximize the hepatic toxicity of the complex. Where the emulsion is used clinically, the hepatic toxicity of its therapeutic dose should therefore be investigated. In conclusion, our findings show that in hepatic gene transfer, intra-arterial or intraportal delivery of a cationic lipid emulsion/DNA complex was superior to systemic delivery. Further investigation of the effect of surface modifications, the optimal conditions for the delivery route and target cells, and the hepatic toxicity of the emulsion is warranted. ![The graph illustrates average and standard deviation of the amount of luciferase enzyme in each organ of the three groups which underwent, respectively, intra-arterial (IA), intraportal (IP), and intravenous (IV) administration.](kjr-3-194-g001){#F1} ###### Average and Standard Deviation of the Amount of Luciferase Enzyme (expressed in pg/mg protein) in Each Organ of the Three Groups of Rats \[Intra-arterial (IA), Intraportal (IP), and Intravenous (IV) Administration\] ![](kjr-3-194-i001)	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ Solid tumor tissues often contain hypoxic regions, in which the supply of oxygen and nutrition is reduced because of an immature vascular network, and in which there is rapid tumor progression [@pone.0039292-Majmundar1]. Hypoxia is a critical microenvironmental factor that contributes to tumor angiogenesis, invasion, progression and metastasis [@pone.0039292-Majmundar1], [@pone.0039292-Wilson1]. Indeed, hypoxic conditions have been shown to be associated with cancer progression and poor prognosis [@pone.0039292-Hckel1]--[@pone.0039292-Harrison1]. Furthermore, recent accumulated evidence suggests that hypoxia induces cancer progression-related characteristics such as epithelial-mesenchymal transition (EMT) [@pone.0039292-Yang1], [@pone.0039292-Yoo1] and stemness properties [@pone.0039292-Keith1]--[@pone.0039292-Chang1] of tumor cells. Acquisition of such properties by tumor cells within hypoxic areas of tumor tissues would greatly contribute to tumor progression and recurrence. Hypoxic tumor cells are known to be highly resistant to conventional chemoradiotherapy, leading to poor prognosis [@pone.0039292-Harrison1], [@pone.0039292-Tredan1]. To improve clinical outcome, novel antitumor agents that efficiently eradicate tumor cells under hypoxic conditions as well as under normoxic conditions are required. Oncolytic virotherapy has emerged as a promising novel antitumor therapy [@pone.0039292-Kirn1]. We previously generated a telomerase-specific replication-competent oncolytic adenovirus (OBP-301: Telomelysin), in which the human telomerase reverse transcriptase (hTERT) promoter element drives E1 gene expression. OBP-301 efficiently kills human cancer cells but not normal human somatic cells [@pone.0039292-Kawashima1]. hTERT is a catalytic subunit of human telomerase and is highly expressed in tumor cells, but not in normal cells. hTERT expression closely correlates with telomerase activity [@pone.0039292-Kim1]--[@pone.0039292-Nakayama1]. Tumor-specific antitumor activity of OBP-301 against various types of human cancer cells with high telomerase activity has been demonstrated in both in vitro and in vivo settings [@pone.0039292-Kawashima1], [@pone.0039292-Umeoka1], [@pone.0039292-Hashimoto1]. Furthermore, the feasibility of OBP-301 for clinical use has been demonstrated in a recently completed phase I clinical trial in the USA of OBP-301 in patients with advanced solid tumors [@pone.0039292-Nemunaitis1]. However, whether OBP-301 has an antitumor effect against hypoxic tumor cells remains unclear. Hypoxia-inducible factor 1 (HIF-1) is a master transcription factor that is activated by hypoxia [@pone.0039292-Majmundar1]. HIF-1 consists of α and β subunits and HIF-1α expression is tightly regulated by oxygen concentration. The HIF-1α protein is stabilized under hypoxic conditions, whereas it is immediately degraded under normoxic conditions. HIF-1α induces the expression of many down-stream target genes that are associated with cellular metabolism, proliferation, survival, apoptosis, neovascularization and migration [@pone.0039292-Bertout1]. The expression of many target genes is activated by HIF-1 through binding to a cis-acting hypoxia response element (HRE) located at their enhancer or promoter regions [@pone.0039292-Bertout1], [@pone.0039292-Pouyssgur1], [@pone.0039292-Ke1]. The hTERT gene is also a HIF-1-target gene. Two HREs that are present in the hTERT gene promoter are involved in hypoxia-mediated hTERT gene upregulation [@pone.0039292-Nishi1]--[@pone.0039292-Anderson1]. In contrast, it has also been shown that hypoxic conditions impair the replication of wild-type adenovirus in tumor cells [@pone.0039292-Pipiya1], [@pone.0039292-Shen1]. Based on these findings, we hypothesized that the cytopathic activity of OBP-301 that is regulated by the hTERT gene promoter would be much stronger against hypoxic tumor cells than that of wild-type adenovirus due to hypoxia-induced enhancement of OBP-301 virus replication. In the present study, we evaluated whether hypoxic conditions affect the expression levels of hTERT and the coxsackie and adenovirus receptor (CAR) in human cancer cells. We next assessed the antitumor effects of OBP-301 and Ad5 against human cancer cells under normoxic or hypoxic conditions. We further evaluated the replication of OBP-301 within hypoxic areas of human xenograft tumors. Results {#s2} ======= Maintenance of human cancer cells under hypoxic conditions {#s2a} ---------------------------------------------------------- A hypoxia chamber filled with a gas mixture of 1% O~2~, 5% CO~2~ and 94% N~2~ was used to maintain human cancer cells under hypoxic conditions. Human cancer cells were also maintained under normoxic conditions, consisting of 20% O~2~ and 5% CO~2~. To first confirm that the tumor cells were efficiently exposed to hypoxia in the chamber, the expression of HIF-1α, which is the main transcription factor induced by hypoxia [@pone.0039292-Majmundar1], was evaluated using Western blot analysis. Consistent with HIF-1α induction by exposure to cobalt chloride (CoCl~2~), HIF-1α expression was strongly induced in human cancer cells (HT29, DLD-1, H1299) maintained in the hypoxia chamber ([Fig. 1A](#pone-0039292-g001){ref-type="fig"}). However, no, or slight, HIF-1α expression was detected under normoxic conditions. Moreover, using immunocytochemistry, we further confirmed that HIF-1α was expressed and accumulated in the nuclei of human cancer cells under hypoxic conditions, but not under normoxic conditions ([Fig. 1B](#pone-0039292-g001){ref-type="fig"}). These results indicate that human cancer cells are maintained under hypoxic conditions in the hypoxia chamber. ![Increased HIF-1α expression in human cancer cells under hypoxic conditions.\ A, Western blot analysis of HIF-1α protein expression in human cancer cells (HT29, DLD-1 and H1299) under normoxic (Nx) or hypoxic (Hx) conditions. Cells were maintained under a normoxic (20% O~2~) or a hypoxic (1% O~2~) condition for 18 h. HT29 cells were also exposed to CoCl~2~ as a positive control. Cell lysates were subjected to Western blot analysis using an anti- HIF-1α antibody. β-actin was assayed as a loading control. B, Subcellular localization of HIF-1α expression in human cancer cells under normoxia or hypoxia was assessed using immunofluorescent staining. Cells cultured under a normoxic or a hypoxic condition for 18 h were stained with anti-HIF-1α antibody (red). Nuclei were counterstained with DAPI (blue). Scale bars = 50 µm.](pone.0039292.g001){#pone-0039292-g001} Expression of hTERT and the adenovirus receptor in human cancer cells under hypoxic conditions {#s2b} ---------------------------------------------------------------------------------------------- OBP-301 contains the hTERT gene promoter, which allows tumor-specific regulation of the gene expression of E1A and E1B that are required for viral replication [@pone.0039292-Kawashima1]. The activity of the hTERT gene promoter in human cancer cells has been shown to be upregulated under hypoxic conditions [@pone.0039292-Nishi1]--[@pone.0039292-Anderson1], suggesting that hypoxia would enhance OBP-301 replication through upregulation of hTERT gene promoter activity. To evaluate the effect of hypoxic conditions on the activity of the hTERT gene promoter in tumor cells, we first investigated the expression level of hTERT mRNA in human tumor cells under normoxic or hypoxic conditions by quantitative real-time RT-PCR analysis. The expression of hTERT mRNA was increased in all tumor cells under the hypoxic condition by 1.3 to 4.3-fold compared to the normoxic condition ([Fig. 2A](#pone-0039292-g002){ref-type="fig"}). Despite evident inductions of HIF-1α by hypoxia, the increases were not statistically significant in HT29 and DLD-1 cells. Because it is known that hTERT expression is regulated not only transcriptionally but also post-transcriptionally by alternative splicing [@pone.0039292-Anderson1], we further examined the effects of hypoxia on activity of exogenous hTERT gene promoter using luciferase reporter assay. Hypoxia activated the hTERT gene promoter by at least 3-fold compared to the hTERT gene promoter activity under normoxia ([Fig. 2B](#pone-0039292-g002){ref-type="fig"}). To further confirm hypoxia-induced hTERT promoter activation, we used chemical inhibitor of HIF-1α. The protein expression of HIF-1α and the activity of hTERT gene promoter were significantly decreased in HT29 and H1299 cells treated with 30 µM HIF-1α inhibitor LW6 and cultured in hypoxic condition ([Fig. S1](#pone.0039292.s001){ref-type="supplementary-material"} and [Fig. 2C](#pone-0039292-g002){ref-type="fig"}). Moreover, we confirmed that hTERT protein was expressed and accumulated in nuclei of human cancer cells under hypoxic conditions by immunofluorescence staining ([Fig. 2D](#pone-0039292-g002){ref-type="fig"}). These results suggest that the hTERT gene promoter in OBP-301 is more strongly activated under the hypoxic condition than under the normoxic condition. ![Effect of hypoxia on hTERT and CAR expression in human cancer cells.\ A, hTERT mRNA expression was assessed in human cancer cells that were maintained under normoxia (Nx) or hypoxia (Hx) for 18 h, using quantitative real-time RT- PCR analysis. The levels of hTERT mRNA were plotted as fold induction relative to the values of hTERT mRNA in HT29 cells incubated under normoxia, which was set at 1.0. Data are shown as mean values ± SD of triplicate experiments. Statistical significance (\) was determined as P\<0.05 (Student\'s t* test). B and C, hTERT gene promoter activity was assessed in human cancer cells that were transfected with the hTERT reporter vector (pGL3-hTERT) and then cultured under normoxia or hypoxia for 24 h, using luciferase reporter assay. The GFP expression vector (pCMV-EGFP) was used as a reporter for transfection efficiency, and the activities of hTERT promoter were determined as ratio of luciferase activity to GFP expression. Data are shown as mean values ± SD of triplicate experiments. Statistical significance (\) was determined as P\<0.05 (Student\'s t* test). C, HT29 and H1299 cells were treated with 30 µM HIF-1α inhibitor or DMSO solvent control in hypoxic condition. The levels of luminescence were plotted as fold induction relative to the values of luminescence in cancer cells incubated under normoxia, which were set at 1.0. D, Subcellular localization of hTERT protein expression in human cancer cells under normoxia or hypoxia was assessed using immunofluorescent staining. Cells cultured under a normoxic or hypoxic condition for 48 h were stained with anti-hTERT antibody (green). Nuclei were counterstained with DAPI (blue). Scale bars = 50 µm. E, Flow cytometric analysis of CAR expression in human cancer cells maintained under normoxia (green) or hypoxia (red) for 18 h. Cells were incubated with a mouse anti-CAR antibody followed by FITC-labeled rabbit anti-mouse IgG. An isotype-matched normal mouse IgG was used as a control (black).](pone.0039292.g002){#pone-0039292-g002} The infection efficiency of Ad5-based viral vectors depends mainly on the expression of the adenoviral receptor CAR in target cells [@pone.0039292-Bergelson1]. Therefore, to evaluate whether hypoxic conditions affect the expression of CAR in tumor cells, we examined the expression level of CAR in all tumor cells under normoxic or hypoxic conditions by flow cytometry. CAR expression was clearly detected in all tumor cells tested: the percentage of CAR-positive cells was 99.5%, 99.3% and 98.8% for HT29, DLD-1 and H1299 cells, respectively ([Fig. 2E](#pone-0039292-g002){ref-type="fig"}). All tumor cell lines showed similar expression levels of CAR under normoxic and hypoxic conditions. These results indicate that tumor cells show high CAR expression under hypoxic conditions as well as under normoxic conditions. Antitumor activities of OBP-301 and Ad5 against hypoxic tumor cells {#s2c} ------------------------------------------------------------------- To explore the potential antitumor activities of the telomerase-dependent oncolytic adenovirus OBP-301 against normoxic and hypoxic tumor cells, we investigated the cytopathic activities of OBP-301 and Ad5 against tumor cells under normoxic or hypoxic conditions. Under normoxic conditions, the cytopathic activities of OBP-301 and Ad5 against HT29 and DLD-1 cells were very similar. The cytopathic activity of OBP-301 against H1299 cells was significantly higher than that of Ad5 at a low dose of infection, whereas it was similar to that of Ad5 at a high dose of infection ([Fig. 3A](#pone-0039292-g003){ref-type="fig"}). In contrast, under hypoxic conditions, the cytopathic activity of OBP-301 against all tumor cells was significantly higher than that of Ad5, especially at a high dose of infection ([Fig. 3B](#pone-0039292-g003){ref-type="fig"}). To further evaluate the antitumor activities of OBP-301 and Ad5 against hypoxic tumor cells, the 50% inhibiting dose (ID~50~) values of OBP-301 and Ad5 under a hypoxic or a normoxic condition were calculated. The calculated ID~50~ values indicated that the cytopathic activity of OBP-301 against all tumor cells under hypoxic conditions was higher than that of Ad5, although the cytopathic activities of OBP-301 and Ad5 were very similar under normoxic conditions ([Table 1](#pone-0039292-t001){ref-type="table"}). These results suggest that the cytopathic activity of OBP-301 against hypoxic tumor cells is more efficient than that of Ad5. ![Cytopathic effect of OBP-301 and wild-type adenovirus serotype 5 (Ad5) under normoxic or hypoxic conditions.\ Cells were infected with OBP-301 (solid bars) or wild-type Ad5 (diagonal bars) at the indicated MOIs under normoxic (A) or hypoxic (B) conditions for 3 days. Cell viability was determined using an XTT assay. Cell viability was calculated relative to that of mock-treated cells, whose viability was set at 100%. Cytopathic activity was further calculated using the following formula; Cytopathic activity (%) = 100 (%) -- cell viability (%). The results shown are the mean values ± SD of quadruplicate experiments. Statistical significance (\) was determined as P\<0.05 (Student\'s t* test).](pone.0039292.g003){#pone-0039292-g003} 10.1371/journal.pone.0039292.t001 ###### Comparison of the ID~50~ values of OBP-301 and Ad5 against human cancer cells under normoxia and hypoxia. ![](pone.0039292.t001){#pone-0039292-t001-1} Cell lines Viruses ID~50~ value[a)](#nt101){ref-type="table-fn"} (MOI) Ratio[b)](#nt102){ref-type="table-fn"} (Hx/Nx) ------------ --------- ----------------------------------------------------- ------------------------------------------------ ------- HT29 OBP-301 20.2±1.5 51.7±7.3 2.5 Ad5 25.6±4.3 212.8±52.0 9.4 DLD-1 OBP-301 8.2±2.7 26.3±10.3 3.0 Ad5 6.6±1.9 2359.6±440.7 597.1 H1299 OBP-301 8.1±1.2 5.3±0.9 0.8 Ad5 15.1±0.9 18.8±7.8 1.3 The ID~50~ values of OBP-301 and Ad5 were calculated from the data of the XTT assay at day 3 after infection. Data are shown as the mean values ± SE of triplicate experiments. The ratio was calculated by division of the ID~50~ value under hypoxia (Hx) by the ID~50~ value under normoxia (Nx). Increased replication of OBP-301 compared to that of Ad5 under hypoxic conditions {#s2d} --------------------------------------------------------------------------------- We next examined the ability of OBP-301 and Ad5 to replicate in HT29, DLD-1 and H1299 cells, which showed almost similar sensitivity to OBP-301 and Ad5 under normoxic conditions but different sensitivity under hypoxic conditions ([Fig. 3](#pone-0039292-g003){ref-type="fig"}). The replication ability of OBP-301 and Ad5 was quantified by measuring viral E1A DNA in tumor cells infected with OBP-301 or Ad5 using quantitative real-time PCR analysis. Under normoxic conditions, the amount of virus production was very similar in tumor cells after infection with OBP-301 or Ad5 ([Fig. 4](#pone-0039292-g004){ref-type="fig"}). In contrast, under hypoxic conditions, viral production was significantly increased in OBP-301-infected tumor cells compared to Ad5-infected cells ([Fig. 4](#pone-0039292-g004){ref-type="fig"}). These results suggest that the replication of OBP-301 within hypoxic tumor cells is more efficient than that of Ad5. ![Quantification of viral DNA replication in human cancer cells under normoxia or hypoxia.\ The indicated human cancer cells were infected with OBP-301 or Ad5 at an MOI of 50 PFU/cell for 1 h, and were further incubated under normoxic (Nx) or hypoxic (Hx) conditions for 48 h. After incubation, cells were harvested and counted. E1A copy number in the cells at 48 h after incubation under normoxia or hypoxia was analyzed by quantitative PCR analysis. The amount of virus production was defined as the value of the E1A copy number relative to the number of cancer cells. Data are shown as the mean values ± SE of triplicate experiments. Statistical significance (\) was determined as P\<0.05 (Student\'s t* test).](pone.0039292.g004){#pone-0039292-g004} OBP-301-mediated E1A expression in the hypoxic regions of xenograft tumor tissues {#s2e} --------------------------------------------------------------------------------- To investigate whether OBP-301 actually replicates in the hypoxic regions of tumor tissues, we examined HT29 and DLD-1 xenograft tumors after intratumoral injection of OBP-301. OBP-301-mediated E1A protein expression was assessed in HT29 and DLD-1 xenograft tumors by immunohistochemistry. Hypoxic areas in tumor tissues were detected by immunohistochemical analysis of the exogenous hypoxic marker, pimonidazole hydrochloride. OBP-301-mediated E1A was expressed in the normoxic regions ([Fig. 5Aa and 5Ba](#pone-0039292-g005){ref-type="fig"}) and the regions that were confirmed to be hypoxic by detection of pimonidazole expression ([Fig. 5Ab and 5Bb](#pone-0039292-g005){ref-type="fig"}). Moreover, the quantitative image analysis of immunohistochemical stains showed that the E1A-positive areas were almost equal in pimonidazole-negative and pimonidazole-positive regions ([Fig. 5C and 5D](#pone-0039292-g005){ref-type="fig"}). These results suggest that OBP-301 replicates in hypoxic tumor cells within the hypoxic areas of tumor tissues. ![E1A expression in hypoxic areas of human xenograft tumors intratumorally injected with OBP-301.\ HT29 (A and C) and DLD-1 (B and D) tumor cells (5×10^6^cells/mouse) were injected subcutaneously into the flank of athymic nude mice. Two weeks after inoculation, OBP-301 (1×10^8^ PFU/tumor) was injected into the tumor for three cycles every 2 days. One day after final administration of OBP-301, the mice were intraperitoneally injected with the hypoxia marker pimonidazole hydrochloride (120 mg/kg). Thirty minutes after injection of pimonidazole hydrochloride, the mice were sacrificed and the tumors were harvested. Paraffin-embedded sections of HT29 and DLD-1 tumors were stained with hematoxylin and eosin (H&E). Tumor sections were also immunostained with an anti-pimonidazole antibody and an anti-adenovirus E1A antibody. A and B, Middle (a) and right (b) panels are higher magnifications of the boxed regions in the left panels. Original magnification: ×4 (left panels), ×40 (middle and right panels). Scale bars = 100 µm. C and D, Quantitative analysis of the E1A-positive areas in the normoxic and hypoxic regions of human xenografts tumor tissues. Data are shown as mean values ± SD of quadruplicate experiments.](pone.0039292.g005){#pone-0039292-g005} Discussion {#s3} ========== Hypoxic microenvironments contribute to tumor invasion, progression, metastasis and resistance to conventional antitumor therapy, such as chemotherapy and radiotherapy, leading to poor prognosis [@pone.0039292-Majmundar1]--[@pone.0039292-Harrison1]. The development of novel antitumor therapies that efficiently eliminate hypoxic tumor cells is an urgent issue for improvement of the clinical outcome of cancer patients. Although adenovirus-based oncolytic virotherapy has recently emerged as a promising antitumor therapy, a hypoxic microenvironment has been shown to reduce the replication of wild-type adenovirus in target tumor cells [@pone.0039292-Pipiya1], [@pone.0039292-Shen1]. Therefore, efficient replication of an oncolytic adenovirus under hypoxic conditions is a critical factor for the eradication of hypoxic tumor cells. In this study, our goal was to assess whether the telomerase-specific oncolytic adenovirus OBP-301 that is regulated by the hTERT gene promoter shows cytopathic activity against human tumor cells under hypoxic conditions. We demonstrated that the cytopathic activity of OBP-301 against hypoxic tumor cells was much stronger than that of wild-type adenovirus ([Fig. 3](#pone-0039292-g003){ref-type="fig"} and [Table 1](#pone-0039292-t001){ref-type="table"}). Hypoxia-mediated activation of the hTERT gene promoter was involved in the enhancement of virus replication in hypoxic tumor cells ([Fig. 2](#pone-0039292-g002){ref-type="fig"} and [4](#pone-0039292-g004){ref-type="fig"}). These results suggest that the hTERT gene promoter is useful for regulation of the replication of oncolytic adenoviruses in tumor cells in a hypoxic microenvironment. The replication of OBP-301 depends on the activity of the hTERT gene promoter, which contains two HREs and is activated by HIF-1α under hypoxic conditions [@pone.0039292-Nishi1]--[@pone.0039292-Anderson1]. Hypoxic conditions that induced nuclear accumulation of HIF-1α ([Fig. 1](#pone-0039292-g001){ref-type="fig"}) upregulated hTERT gene promoter activity in human cancer cells ([Fig. 2B and 2C](#pone-0039292-g002){ref-type="fig"}). Consistent with this hTERT gene promoter activation, OBP-301 replication was significantly higher than that of Ad5 with the endogenous E1 promoter ([Fig. 4](#pone-0039292-g004){ref-type="fig"}). These findings suggest that hypoxia enhances OBP-301 virus replication through HIF-1α-mediated activation of the hTERT gene promoter. Recently, oncolytic virotherapy has garnered interest as potential therapeutic strategy for hypoxic tumors [@pone.0039292-Guo1]. A hypoxia-responsive promoter that is upregulated by HIF-1 has been used for the tumor-specific replication of an oncolytic adenovirus [@pone.0039292-HernandezAlcoceba1]--[@pone.0039292-Kwon1]. Although an oncolytic adenovirus that is regulated by a hypoxia-responsive promoter will also be effective against hypoxic tumor cells following HIF-1 activation, non-hypoxic tumor cells in which HIF-1 is not activated may be less sensitive to these viruses. In contrast, the hTERT gene promoter-regulated oncolytic adenovirus OBP-301 would be effective against both hypoxic and normoxic tumor cells through hTERT activation. The infection efficacy of Ad5-based oncolytic adenoviruses has been suggested to depend mainly on the expression level of CAR on the target cell surface [@pone.0039292-Bergelson1]. Hypoxia has been shown to downregulate CAR expression in tumor cells in a HIF-1α dependent manner [@pone.0039292-Kster1]. However, in the present study, high CAR expression was maintained in all of the human cancer cells tested, even under hypoxic conditions ([Fig. 2E](#pone-0039292-g002){ref-type="fig"}). These results are consistent with a previous report [@pone.0039292-Shen1], which demonstrated that hypoxia has no influence on adenoviral infectivity of target cancer cells. The expression levels of integrin αvβ3 and αvβ5 are also involved in the infection efficacy of adenoviruses [@pone.0039292-Wickham1]. Previous reports have shown that hypoxia upregulates the expression levels of integrin αvβ3 and αvβ5 in tumor cells [@pone.0039292-CowdenDahl1], [@pone.0039292-Niu1]. These results suggest that hypoxic conditions would mainly suppress the replication of an adenovirus rather than the infection efficiency of the adenovirus. Tumor tissues frequently contain hypoxic areas due to an immature vascular network. Various exogenous and endogenous hypoxia-related proteins have recently been developed as markers for identification of hypoxic regions of tumor tissues. Increased HIF-1 expression is a useful endogenous marker of hypoxic areas close to blood vessels. Expression of the exogenous hypoxia marker, Pimonidazole, is as effective a marker as HIF-1 for the detection of severely hypoxic regions [@pone.0039292-KizakaKondoh1]. In this study, OBP-301-mediated E1A expression was detected in pimonidazole-positive regions as well as normoxic regions ([Fig. 5](#pone-0039292-g005){ref-type="fig"}). These results indicate that the telomerase-specific oncolytic adenovirus OBP-301 could infect and replicate in tumor cells under a hypoxic microenvironment including in tumor cells in which HIF-1 was active. Recent advances in our knowledge of tumor microenvironments have provided evidence that hypoxic tumor cells contribute to cancer progression. For example, hypoxia activates the metastatic potential of tumor cells by inducing EMT [@pone.0039292-Yang1], [@pone.0039292-Yoo1] and facilitates the maintenance of cancer stem cells [@pone.0039292-Keith1]--[@pone.0039292-Chang1]. Therefore, the complete elimination of hypoxic tumor cells with metastatic and stemness properties is important for improvement of the clinical outcome of cancer patients. Recent reports have suggested that tumor cells undergoing EMT show reduced CAR expression [@pone.0039292-Lacher1], [@pone.0039292-Lacher2], suggesting that tumor cells undergoing EMT are less sensitive to oncolytic adenovirus infection. Further study to investigate the cytopathic effect of OBP-301 in tumor cells undergoing EMT is warranted. In contrast, recent reports have shown that an oncolytic adenovirus induces oncolytic cell death in cancer stem cells [@pone.0039292-Jiang1]--[@pone.0039292-Short1]. Cancer stem cells have recently been shown to have increased hTERT expression compared to non-cancer stem cells [@pone.0039292-Hiyama1], [@pone.0039292-CasteloBranco1]. Consistent with this high hTERT expression in cancer stem cells, Hemminki et al. has suggested that an oncolytic adenovirus that is regulated by specific promoters for hTERT, cyclooxygenase-2 or multidrug resistance, shows efficient cytopathic activity against human breast cancer stem cells [@pone.0039292-Bauerschmitz1]. Thus, the hTERT promoter-regulated oncolytic adenovirus OBP-301 may have the potential to eliminate highly progressive tumor cells in a hypoxic microenvironment, thereby contributing to the improvement of its therapeutic benefit against malignant tumors. In conclusion, we have clearly demonstrated that the antitumor effect of the telomerase-specific oncolytic adenovirus OBP-301 against tumor cells in a hypoxic microenvironment is much stronger than that of a wild-type adenovirus. Regulation of virus replication by the hTERT gene promoter would be an effective antitumor strategy that would enhance the cytopathic activity of an oncolytic adenovirus against hypoxic tumor cells. Materials and Methods {#s4} ===================== Cell lines {#s4a} ---------- The human colorectal cancer (DLD-1 and HT29) and non-small cell lung cancer (H1299) cell lines were purchased from the American Type Culture Collection (ATCC) (Manassas, VA, USA). Although cell lines were not authenticated by the authors, cells were immediately expanded after receipt and stored in liquid N~2~. Cells were not cultured for more than 5 months following resuscitation. DLD-1 and H1299 cells were propagated as monolayer cultures in RPMI-1640 medium. HT29 was grown in McCoy\'s 5A medium. The transformed embryonic kidney cell line 293 obtained from the ATCC was maintained in Dulbecco\'s modified Eagle\'s medium containing high glucose (4.5 g/L). All media were supplemented with 10% heat-inactivated fetal calf serum, 100 units/ml penicillin G and 100 µg/ml streptomycin. To maintain human cancer cells under hypoxic conditions, the cells were incubated in a hypoxic chamber (Modular Incubator Chamber; Billups-Rothenberg, Del Mar, CA, USA) filled with a gas mixture of 1% O~2~, 5% CO~2~ and N~2~. The cells were also incubated under normoxic conditions at 37°C in a humidified atmosphere with 5% CO~2~ and 20% O~2~. HIF-1α inhibitor LW6 was purchased from Calbiochem (San Diego, CA, USA) and used at the concentration of 30 µM. Recombinant adenoviruses {#s4b} ------------------------ The recombinant replication-selective, tumor-specific adenovirus OBP-301 (Telomelysin), in which elements within the hTERT gene promoter drive the expression of E1A and E1B genes linked with an internal ribosome entry site, was previously constructed and characterized [@pone.0039292-Kawashima1]. The wild-type Ad5 was used as a control vector. OBP-301 and Ad5 were generated in 293 cells and purified by cesium chloride step-gradient ultracentrifugation. Their infectious titers were determined by a plaque-forming assay using 293 cells. The ratios of viral particle/plaque-forming unit of OBP-301 and Ad5 are 26 and 27, respectively. Viruses were stored at −80°C. Western blot analysis {#s4c} --------------------- Cells were maintained under a hypoxic or a normoxic condition for 18 h or 24 h. Whole cell lysates were then prepared in a lysis buffer (10 mM Tris (pH 7.5), 150 mM NaCl, 1 mM EDTA, 10% glycerol, 0.5% NP40) containing a protease inhibitor mixture (Complete Mini; Roche, Indianapolis, IN, USA). Lysates were electrophoresed on 4%--7% SDS polyacrylamide gels and proteins were transferred to polyvinylidene difluoride membranes (Hybond-P; GE Healthcare, Buckinghamshire, UK). The primary antibodies used for Western blotting were: mouse anti-HIF-1α monoclonal antibody (mAb) (BD Biosciences, San Diego, CA, USA) and mouse anti-β-actin mAb (Sigma, St. Louis, MO, USA). Horseradish peroxidase-conjugated antibody against mouse IgG (GE Healthcare) was used as the secondary antibody. Immunoreactive bands on the blots were visualized using enhanced chemiluminescence substrates (ECL Plus; GE Healthcare). Immunofluorescence staining {#s4d} --------------------------- Cells grown in chamber slides were washed twice with ice-cold PBS, and then fixed with cold 4% paraformaldehyde in PBS for 15 min on ice. The cells were permeabilized by incubation with 0.2% Triton X-100 in PBS for 5 min on ice and then blocked with 3% bovine serum albumin in PBS for 30 min at room temperature. The slides were subsequently incubated with mouse anti-HIF-1α mAb (BD Biosciences) or mouse anti-hTERT mAb (KYOWA Medex, Tokyo, JP) for 1 h at room temperature. After two washes with PBS, the slides were incubated with Alexa Fluor 488- or Alexa Fluor 568-labeled goat anti-mouse IgG antibody (Invitrogen, Carlsbad, CA, USA) for 1 h. The slides were further stained with 10 mg/ml 4′,6-diamidino-2-phenylindole (DAPI), mounted using Fluorescence Mounting Medium (Dako, Glostrup, Denmark), and then photographed using a fluorescence microscope (IX71; Olympus, Tokyo, Japan). Quantitative real-time RT-PCR analysis {#s4e} -------------------------------------- Total RNA was extracted from cancer cells maintained under hypoxic or normoxic conditions for 18 h using the RNA-Bee regent (Tel-test; Friendswood, TX, USA). The hTERT mRNA copy number was determined by quantitative real-time reverse transcription-polymerase chain reaction (RT-PCR) using a LightCycler instrument and a LightCycler TeloTAGGG hTERT Quantification Kit (Roche Diagnostics, Bagel, Switzerland). Data analysis was performed using LightCycler Software. The expression of hTERT mRNA was defined from the threshold cycle (Ct), and relative expression levels were calculated after normalization with reference to the expression of porphobilinogen deaminase (PBGD). Transfection and luciferase reporter assay {#s4f} ------------------------------------------ Cells were seeded on 6-well plates at a density of 4×10^5^ cells/well and incubated overnight. Each cell line was transfected with 3 µg of hTERT reporter plasmid (pGL3-hTERT) and 3 µg of GFP expression vector (pCMV-EGFP) as a reporter for transfection efficiency, using Lipofectamin LTX (Invitrogen) following the manufacturer\'s recommendations. Cells were then incubated under normoxic or hypoxic conditions. After 24 h incubation, luciferase activity was determined using a Bright-Glo reagent (Promega Corporation, Madison, WI, USA). Results presented are the ratios of luciferase activity to GFP fluorescent intensity and the means of three independent experiments. Flow cytometric analysis {#s4g} ------------------------ The cells that were maintained under hypoxic or normoxic conditions for 18 h were labeled with a mouse anti-CAR mAb (Upstate Biotechnology, Lake Placid, NY, USA) for 30 min at 4°C. An isotype-matched normal mouse IgG1 (Serotec, Oxfordshire, UK) was used as a negative control. The cells were then incubated with a fluorescein isothiocyanate (FITC)-conjugated rabbit anti-mouse IgG second antibody (Zymed Laboratories, San Francisco, CA, USA) and were analyzed using flow cytometry (FACSCalibur; Becton Dickinson, Mountain View, CA, USA). Cell viability assay {#s4h} -------------------- Cells were seeded on 96-well plates at a density of 1×10^4^ cells/well 20 h before viral infection. All cell lines were infected with OBP-301 or wild-type Ad5 at multiplicity of infections (MOI) of 0, 1, 5, 10, 50 or using 100 plaque-forming units (PFU)/cell. The cells were then incubated under normoxic or hypoxic conditions for 3 days. Cell viability was determined using a Cell Proliferation Kit II (Roche Diagnostics) that was based on a sodium 3′-\[1-(phenylaminocarbonyl)-3,4-tetrazolium\]-bis (4-methoxy-6-nitro) benzene sulfonic acid hydrate (XTT) assay according to the manufacturer\'s protocol. The cytotoxic activity and the ID~50~ value of each virus was calculated using cell viability data. Each experiment was performed in quadruplicate during the same day and repeated at least three times. In vitro virus replication assay {#s4i} ---------------------------------- Cells were seeded on 6-well plates at a density of 3×10^5^ cells/well 20 h before viral infection and were infected with OBP-301 or wild-type Ad5 at an MOI of 50 for 1 h. Following removal of the viral inocula, the cells were further maintained under hypoxic or normoxic conditions and were then harvested at 48 h after virus infection. After cell counting, DNA was purified using the QIAmp DNA mini Kit (Qiagen, Hilden, Germany) according to the manufacturer\'s protocol. E1A copy numbers were determined by quantitative real-time PCR using the StepOnePlus Real Time PCR System (Applied Biosystems, Carlsbad, CA, USA) and TaqMan Gene Expression Assays (Applied Biosystems). The sequences of the specific primers and probe used in this experiment were: E1A primers, 5′-CCT GAG ACG CCC GAC ATC-3′ and 5′-GGA CCG GAG TCA CAG CTA TCC-3′; E1A probe, 5′-FAM-CTG TGT CTA GAG AAT GC-MGB-3′. Data analysis was carried out using StepOne Software (Applied Biosystems). In vivo human xenograft tumor models {#s4j} -------------------------------------- Animal experimental protocols were approved by the Ethics Review Committee for Animal Experimentation of Okayama University School of Medicine (Approval ID: OKU-2009051). The HT29 and DLD-1 cells (5×10^6^ cells per site) were inoculated subcutaneously into the flank of 5- to 6-week-old female BALB/c nu/nu mice (Japan SLC, Shizuoka, Japan). When the tumor size reached approximately 10 mm in diameter, OBP-301 was injected into the tumors at a dose of 1×10^8^ PFU/tumor every 2 days for three cycles. To detect hypoxic areas within tumor tissues, pimonidazole hydrochloride (Hypoxyprobe -1; Hypoxyprobe Inc., Burlington, MA, USA) was injected intraperitoneally at a dose of 120 mg/kg body weight 24 h after the final treatment. The mice were then sacrificed and the tumors were harvested 30 min after pimonidazole injection. Four mice were used for each group. Immunohistochemistry {#s4k} -------------------- Tumors were fixed in 10% neutralized formalin and embedded in paraffin blocks. Sections (4 µm) were prepared for hematoxylin/eosin staining and also for immunohistochemical examination. After deparaffinization and rehydration, antigen retrieval was performed by microwave irradiation in 10 mM citrate buffer (pH 6.0). After quenching of endogenous tissue peroxidase, tissue sections were incubated with mouse anti-adenovirus type 5 E1A mAb (BD Biosciences) and mouse anti-Hypoxyprobe-1 mAb (Hypoxyprobe Inc.). The sections were then incubated using the Histofine Mouse Stain Kit (Nichirei Biosciences, Tokyo, Japan). Immunoreactive signals were visualized by using 3,3′-diaminobenzidine tetrahydrochloride solution, and the nuclei were counterstained with hematoxylin. Signals were viewed under a microscope (BX50; Olympus). The percentage of the positive area in each field was analyzed using Image J software (version 1.45). Statistical analysis {#s4l} -------------------- Determination of significant differences among groups was assessed by using the Student\'s t test. P\<0.05 was considered significant. Supporting Information {#s5} ====================== ###### Suppression of HIF-1α expression in human cancer cells under hypoxic conditions by HIF-1 inhibitor. A, Western blot analysis of HIF-1α protein expression in human cancer cells (HT29 and H1299) under normoxic or hypoxic conditions. Cells were treated with 30 mM HIF-1α inhibitor or DMSO solvent control under hypoxic condition for 24 h. Cell lysates were subjected to Western blot analysis using an anti- HIF-1α antibody. β-actin was assayed as a loading control. B, Subcellular localization of HIF-1α expression in human cancer cells treated with 30 mM HIF-1α inhibitor or DMSO solvent control under hypoxia was assessed using immunofluorescent staining. Cells cultured under a hypoxic condition for 24 h were stained with anti-HIF-1α antibody (red). Nuclei were counterstained with DAPI (blue). Scale bars = 50 µm. (TIF) ###### Click here for additional data file. We thank the members of our laboratories for helpful comments and discussions. We also thank Tomoko Sueishi for her excellent technical support. Competing Interests:YU is an employee of Oncolys BioPharma, Inc., the manufacturer of OBP-301. TF is a consultant of Oncolys BioPharma, Inc. This does not alter the authors\' adherence to all the PLoS ONE policies on sharing data and materials. Funding:This work was supported in part by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan (to TF and FT, <http://www.mext.go.jp/english/>) and by grants from the Ministry of Health, Labour and Welfare of Japan (to TF, <http://www.mhlw.go.jp/english/>). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: YH HT FT TF. Performed the experiments: YH FT TK YW SY. Analyzed the data: YH HT FT FU SK TF. Contributed reagents/materials/analysis tools: YU. Wrote the paper: YH HT TF.	{ "pile_set_name": "PubMed Central" }
Our Neanderthal genes {#sec1} ===================== Pandora remains fascinated with our Neanderthal ancestors and, fortunately, so are the scientists! Although contact with humans didn\'t help their survival, evidence that some of the Neanderthal DNA has survived in us is increasing. These Neanderthal genes have not only contributed to our visuospatial ability, as discussed in the last issue of Pandora\'s Box, they also have a role in many other human characteristics. Using baseline phenotypes collected for 112 000 individuals by the UK Biobank, researchers from the Max Planck Institute found that Neanderthal DNA affects our skin tone, hair colour and height. Some of the Neanderthal alleles contribute to lighter and others to darker skin and hair in modern Europeans, which suggests that the Neanderthals also had variable hair and skin tone. But their influence on us doesn\'t stop there. Neanderthal DNA can also influence our sleep patterns and our mood. All these traits are attributed to adaptation to lower and variable sun exposure over the thousands of years the Neanderthals lived in Eurasia prior to the arrival of humans from Africa 100 000 years ago. DannemannM. & KelsoJ. (2017) The contribution of Neanderthals to phenotypic variation in modern humans. American Journal of Human Genetics, 101(4), 578--589.28985494 Can we stop ourselves growing old? {#sec2} ================================== Growing old is a natural process, yet we humans try so hard to stop the clock ticking. We change our appearance with aesthetic surgery, improve our health with good nutrition and exercise (well a few of us do!), and treat with increasing success a number of illnesses that can kill us; but can we really stop or reverse the physiological process of ageing? Yes, we can say scientists! {#sec2-1} --------------------------- Researchers from New South Wales claim a discovery that could actually reverse the ageing process. Our cells' ability to repair themselves is slowed down by increasing age and radiation. Experimenting on mice, which they exposed to radiation, they discovered a critical step in the molecular process that allows cells to repair damaged DNA. This concerns the metabolite NAD^+^, which is present normally in every cell of our body, where its key role is to regulate protein interactions that control DNA repair. Treating mice with this agent, 'the cells of the old mice were indistinguishable from those of the young mice, after just one week of treatment', claim the researchers. A clinical trial is to start shortly, and apparently NASA is interested in these findings as part of their plans to send humans to Mars! In case you didn\'t know, cosmic radiation accelerates ageing in astronauts. LiJ., BonkowskiM. S., MoniotS., (2017) A conserved NAD binding pocket that regulates protein-protein interactions during aging. Science, 355(6331), 1312--1317.28336669 Oh no we cannot, say other scientists! {#sec2-2} -------------------------------------- Stopping ourselves growing old is impossible, claim scientists from Arizona. They dispute theories that attribute ageing to a failure of selection, which allow for the possibility that if senescence-causing alleles are identified we may be able to delay the process of growing old. Focusing on the role of intercellular competition within an organism, selection between somatic cells may delay ageing by getting rid of non-functioning cells. However, they claim, ageing does not depend only on the functional capacity of individual cells, but also on how well cells work together. So, while intercellular competition selects to get rid of non-functional cells, it may also select to keep cells that do not cooperate. Does that sound like our selection of politicians? Well, back to science. Researchers conclude that intercellular competition creates an 'inescapable double bind that makes ageing inevitable in multicellular organisms'. We continue to grow old and weak because our cells, like us, cannot work together to the common good! NelsonP. & MaselJ. (2017) Intercellular competition and the inevitability of multicellular aging. PNAS, 114(49), 12982--12987.29087299 If the previous chat didn\'t depress you enough, here is some more! Is it worth striving for longevity for our progeny and ourselves if our world, as we know it, is to become extinct in just over 80 years? The sixth mass world extinction due in 2100 {#sec3} =========================================== Our world has been witness to five 'mass extinction events' in the past 550 million years, each one taking millions of years to occur. Scientists believe these to be due to the 'cycling of carbon'. All of the mass extinction events happened after an upheaval of the carbon cycling between the atmosphere and the oceans. Analysing the significant changes over the millions of years, researchers were able to determine a threshold of catastrophe to the carbon cycle. Beyond this limit, enough instability can occur to lead to mass extinction. It is estimated that 310 gigatons of carbon need to be added for a mass catastrophe to occur. Should we worry? Scientists have worked out, based on the current carbon emission rates, that the year 2100 may mark the demise of our global community! The tragedy is that it will be our fault. The ecologists have been warning us for years, but do we believe the evidence? RothmanD. H. (2017) Thresholds of catastrophe in the Earth system. Science Advances, 3(9), e1700906.28948221 How dogmatic can you be? {#sec4} ======================== We have already talked in previous issues of the journal about some of the science behind why so many of us are so sure of our beliefs and cannot change our minds despite irrefutable evidence to the contrary. The previous study focused on the rigidity with which some of us hold political views, while this one examines religious and non-religious dogmatism. Examining the personality characteristics which are thought to drive religious and non-religious dogmatism, in 900 individuals, researchers found similarities as well as important differences between the two groups. Higher critical reasoning was associated with less dogmatism in both religious and non-religious groups. The more dogmatic they were, whether religious or non-religious, the less likely they were to consider the perspectives of others. However, the religious people differed in how moral concern influenced the rigidity of their thinking. The religious, say the authors 'may cling to certain beliefs, especially those at odds with analytical thinking, because those beliefs resonate with their moral sentiments and this affirms their beliefs'. In previous work, they identified two brain networks, one for empathy and one for analytic thinking, which are in tension with each other. Our thought processes move between the two, choosing the appropriate network for the issue under consideration. In the religious dogmatist\'s mind the empathetic network dominates, while in the non-religious dogmatist\'s mind the analytic network takes over. Dogmatism is not restricted to religion. In healthy people, it applies to beliefs and opinions about politics, diet (vegan, vegetarian or carnivorous), climate and other ecological matters, and more. FriedmanJ. P. & JackA. I. (2017) What makes you so sure? Dogmatism, fundamentalism, analytic thinking, perspective taking and moral concern in the religious and nonreligious. Journal of Religion and Health, DOI: . Nationalism, populism and feeling disadvantaged {#sec5} =============================================== Nationalism and populism appear to be on the increase but what does this mean? Three studies in different parts of the world (Warsaw, Poland; Kent, UK; and Maryland, USA) found that 'national collective narcissism', which is an unrealistic belief in the greatness of the nation, is linked to support for populism. In the UK study, collective narcissism predicted support for Brexit; in the US study, it predicted support for Donald Trump; and in the Polish study, it predicted support for the populist Law and Justice Party. These associations were present even when controlling for conventional national identification. Importantly, the researchers also found that collective narcissism was increased in groups that felt disadvantaged in their country. The researchers suggest that narcissistic beliefs in the nation\'s greatness are a way to compensate for feeling worse off than other groups. Populist politicians fuel the narrative of relative disadvantage, and this reinforces national narcissistic beliefs and prejudices against other groups such as immigrants. MarchlewskaM., CichockaA., PanayiotouO., (2017) Populism as identity politics: perceived in-group disadvantage, collective narcissism, and support for populism. Social Psychological and Personality Science, DOI:. If all this has upset you, talk to yourself -- but make sure you do it in the third person {#sec6} ========================================================================================== Some of you may do this already. If so, you\'ll be pleased to know that there is scientific evidence it works! Researchers asked participants in two studies to reflect on feelings, using 'I' or their name, while viewing aversive images and having their neural activity measured with ERPs (event-related potentials) in study 1, and recalling negative personal memories while undergoing brain neuroimaging (functional magnetic resonance imaging; fMRI) in study 2. The researchers demonstrated that, when talking to themselves in the third person, subjects reduced their emotional reactivity as measured by ERP on viewing aversive images, while those who reflected on negative memories showed no increase in brain activity on fMRI markers of cognitive control. Furthermore, measurements of participants' effort-related brain activity showed that talking to themselves in the third person required no more effort than talking to themselves in the first person. This is what one would call self-therapy: effective, free of charge, practical and discreet (as long as you talk to yourself in private or quietly!) MoserJ. S., DoughertyA., MattsonW. I., (2017) Third-person self-talk facilitates emotion regulation without engaging cognitive control: converging evidence from ERP and fMRI. Scientific Reports, 7(1), 4519.28674404 Are we really as politically correct as we claim? Our brains say no! {#sec7} ==================================================================== You are shown a film where one sister refuses to donate an organ to the other who has been diagnosed with cancer. How would you react? A study examined the reactions of women who were told either that the sisters were biological siblings or that the younger sister was adopted to the family as an infant. They were asked how they felt about the sister\'s refusal to be a donor, while they were undergoing neuroimaging of their brains. Ninety per cent of the participants responded that there should not be any difference in the decision to donate an organ to the sick sister, irrespective of whether the sisters were genetically related. However, their brains responded differently to this moral question. There were significantly stronger correlations in areas of the brain associated with emotional conflict regulation, decision-making and mentalising (insula, cingulate, medial and lateral prefrontal, superior temporal and superior parietal cortices) when the participants believed that the sisters were genetically related. It suggests that we do put those like us first! Bacha-TramsM., GlereanE., DunbarR., (2017) Differential inter-subject correlation of brain activity when kinship is a variable in moral dilemma. Scientific Reports, 7(1), 14244.29079809	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ Reputable scientific journals only publish manuscripts that have been subjected to peer review -- that is, critical scrutiny by scientific experts. When a manuscript is submitted, reviewers who are researching and publishing work in the same field (peers) are asked to evaluate the content of the manuscript (e.g., significance and originality of the research findings) and recommend to the editor that the manuscript be published, revised and then published, or rejected [@pone.0011344-Sense1]. On the basis of these recommendations, the journal editor makes the decision to accept or reject for publication. This means that although experts in a research area are consulted for the reviewing of the manuscript, the reviewing does not include the selection decision on the submissions: Peer review forms the basis for the selection decision made by an editor. The conscientious editor must decide on the validity of peer reviews. "Especially in controversial and highly competitive fields, the responsible editor is called upon to do more than simply average the reviewers\' opinions. Reviews may be unavoidably and understandably biased by personal involvement and convergence may occur only slightly more often than expected by chance. The editor must glean what seems objective and logical from the review, and the editor\'s ability to do so depends on the quality of the review, and on his or her familiarity with the subject of the paper, as well as on a sense of where the paper stands in light of the standards and constituency of the intended audience" [@pone.0011344-Stossel1]. The editorial decision can later prove to be successful (if manuscripts are accepted that after publication are useful for the further research in a field) or not (if useful manuscripts are not accepted for publication). The objective of this study is to investigate whether peer review contributes to valid editorial decisions or not. We have examined in three publications [@pone.0011344-Bornmann1], [@pone.0011344-Bornmann2], [@pone.0011344-Bornmann3] the predictive validity of the selection decisions at the journal Angewandte Chemie International Edition (AC-IE). AC-IE is a chemistry journal with a higher annual Journal Impact Factor (JIF) (provided by Thomson Reuters, Philadelphia, PA) than the JIFs of comparable journals (10.879 in the 2008 Journal Citation Reports, Science Edition). It is a journal of the German Chemical Society (Gesellschaft Deutscher Chemiker (GDCh), Frankfurt am Main, Germany) and is published by Wiley-VCH (Weinheim, Germany). The journal introduced peer review in 1982, primarily in conjunction with one of the document types published in the journal, "Communications," which are short reports on works in progress or recently concluded experimental or theoretical investigations. What the editors of AC-IE look for most of all is excellence in chemical research. Submissions that reviewers deem to be of high quality are selected for publication. For the investigation of the predictive validity of the AC-IE publication decisions citation counts for the accepted and rejected (but published elsewhere) manuscripts were used. In the absence of other operationalizable criteria, a conventional approach is to use citation counts as a proxy for research quality, since they measure the international impact of scientific work [@pone.0011344-vanRaan1]. These analyses showed that on average (arithmetic mean and median), accepted manuscripts have clearly higher citation counts than rejected manuscripts that are published elsewhere [@pone.0011344-Bornmann1], [@pone.0011344-Bornmann2], [@pone.0011344-Bornmann3]. A comparison of average citation counts of accepted and rejected (but published elsewhere) manuscripts with international scientific reference standards found that mean citation counts below baseline values were significantly less frequent for accepted manuscripts than for rejected (but published elsewhere) manuscripts [@pone.0011344-Bornmann1]. Both results suggest for AC-IE that the editorial decisions are on average related to the manuscripts\' future scientific impact and thus have high predictive validity. Using an approach developed by Bornmann et al. [@pone.0011344-Bornmann4], [@pone.0011344-Bornmann5], [@pone.0011344-Bornmann6] and, independently of our work, discussed by Straub [@pone.0011344-Straub1], [@pone.0011344-Straub2] and Thorngate, Dawes, and Foddy [@pone.0011344-Thorngate1], we determined for AC-IE manuscript selection decisions the extent of 'erroneous' and 'correct' decisions. For erroneous decisions we distinguished type I and type II errors: In type I errors, the editors concluded that a manuscript had the scientific potential for publication and accepted it, when it in fact did not, as reflected in a manuscript\'s low scientific impact subsequent to publication. In type II errors, the editor concluded that a manuscript did not have the scientific potential for publication and rejected it, when it actually did -- as reflected in a high scientific impact subsequent to publication. We found that the AC-IE editors\' decisions regarding 15% of the manuscripts demonstrate a type I error (accepted manuscripts that did not perform as well as or worse than the average rejected manuscript) [@pone.0011344-Bornmann3]. Moreover, the decisions regarding 15% of the manuscripts showed a type II error (rejected manuscripts that performed equal to or above the average accepted manuscript). The large part of the AC-IE editorial decisions (70%) could be classified as either correctly accepted (later highly cited) or correctly rejected (later low citation counts). In the present study, we expand the approach that we used in previous investigations of the predictive validity of the manuscript selection process at AC-IE to include the model developed by Taylor and Russell [@pone.0011344-Taylor1], in that we included in the analysis the base rate (proportion of manuscripts fundamentally suitable for publication), the selection rate (percentage of manuscripts accepted), and the ratings of the reviewers (upon which the editorial decisions were made). Using this model we are able to determine not only the validity of the editorial decisions (percentage of erroneous and correct decisions) but also (and especially) the usefulness of the peer review for the selection decisions. For example, if a high proportion of manuscripts submitted to a journal show very high quality (measured as the citation counts of these papers after publication) and if a large percentage of these manuscripts are published, then peer review is of only limited utility for the selection decision according to the model developed by Taylor and Russell [@pone.0011344-Taylor1]. The Taylor and Russell model {#s1a} ---------------------------- The importance of the base rate and selection rate in assessment of a selection process was established by Taylor and Russell [@pone.0011344-Taylor1] in a paper titled, "The relationship of validity coefficients to the practical effectiveness of tests in selection: [Discussion](#s4){ref-type="sec"} and tables." Taylor and Russell\'s model has been called "the most well-known utility model" [@pone.0011344-Cascio1]. In the paper Taylor and Russell presented a method for determining the success of a personnel selection process that is still used today; in the present study we apply their approach to the success of the selection of submissions to a scientific journal. The model consists of four parameters [@pone.0011344-Taylor1]. Applied to the manuscript selection process they are: 1. The validity coefficient r~xy~ indicates the strength of the relationship between the predictor value (x, outcome of peer review) and the criterion value (y, the scientific quality of a submission). 2. The base rate is the percentage of submissions that are 'qualified' submissions. Qualified submissions are papers that are fundamentally suitable for publication. After publication they are comparatively frequently cited. 3. The selection rate is the percentage of submissions that are accepted for publication. 4. The success rate is the percentage of submissions accepted that are qualified submissions. This percentage can either be computed from the available data, as shown in the following, or taken from the tables provided by Taylor and Russell [@pone.0011344-Taylor1], based on certain assumptions. The three graphs (A, B and C) in [Figure 1](#pone-0011344-g001){ref-type="fig"} show the relationship between the four parameters for a fictitious manuscript selection process. The gray point cloud in the center of each graph represents the validity of the process. On the x-axis is the outcome of peer review (reviewers\' ratings) and on the y-axis the scientific quality of a submission (measured as citation counts determined ex post). For every submission this yields a point in the point cloud. If both the predictor and the criterion are normally distributed, a linear relationship between them can be assumed, and if there is a minimum of validity (r~xy~\>0), the points of all submissions produce a point cloud in the form of an ellipse. However, if r~xy~ = 1 or r~xy~ = −1 the result is a line (maximal validity), and if r~xy~ = 0, the result is a circle (no validity). ![Dependency of the number of accepted and qualified submissions (AQ) on base rate, selection rate and validity coefficient (r~xy~).](pone.0011344.g001){#pone-0011344-g001} Areas AQ and RN in the three graphs of [Figure 1](#pone-0011344-g001){ref-type="fig"} represent correct decisions -- that is, if the editors used favorable peer review ratings to select submissions (above the predictor cutoff), those submissions in area AQ would be selected, and after publication they would make a substantial contribution to scientific advancement in a research field (they would be cited with above-average frequency). Those submissions in area RN would be rejected correctly, because they received unfavorable ratings by the reviewers (below the predictor cutoff) and the scientific impact of the manuscripts is low (after publication they would not contribute towards scientific advancement). Areas AN and RQ represent erroneous decisions -- those in area AN would be selected because of favorable ratings by the reviewers, but as low quality research they would not contribute towards scientific advancement in a research field (this type of error was called type I error above). Submissions in area RQ would be rejected because of unfavorable ratings by the reviewers, even though they would later turn out to be successful, high impact research (called type II error above). Based on the number of submissions found in each of the four areas, it is possible to calculate the base rate, selection rate, and success rate: If there is no correlation between the selection process and the success rate (then the point cloud is a circle) and the base and selection rates are 50% (meaning that one-half of the submissions is qualified and one-half of the submissions is selected), only one-half of the accepted submissions will be qualified (see graph A in [Figure 1](#pone-0011344-g001){ref-type="fig"}). This is equivalent to a random selection: The peer reviewing cannot contribute towards increasing the probability that qualified submissions will be accepted. If the point cloud is not a circle but an ellipse, based on peer review more qualified than non-qualified submissions will usually be accepted -- if the base and selection rates remain unchanged (see graph B in [Figure 1](#pone-0011344-g001){ref-type="fig"}). The higher the validity coefficient (the coefficient for the correlation between the predictor value and the criterion value), the narrower the ellipse is and the more qualified submissions that are accepted. An increase in the validity of a manuscript selection process hence increases the probability that qualified submissions (later highly-cited publications) are selected and unqualified submissions (later publications with low citation counts) rejected, other things being equal [@pone.0011344-Cabrera1]. However, the success of a manuscript selection process can also be increased by raising the base rate and thus the number of submissions fundamentally suitable (qualified) for publication. This can be achieved, for example, through successful submission policy (in this case, the point cloud in the graphs would shift upwards). Even if a manuscript selection process is not valid (r~xy~ = 0, see graph A), as a result there would be more qualified than non-qualified submissions among the accepted submissions. Another possible way to increase the success of a manuscript selection process is to reduce the selection rate. However, this measure is only successful if there is a correlation between the outcome of peer review and the scientific quality of a submission (r~xy~\>0; that is, in graph A this measure would not result in an improved success rate). If the vertical line (the predictor cutoff) in graph B in [Figure 1](#pone-0011344-g001){ref-type="fig"} is shifted to the right, fewer non-qualified submissions are selected. Taylor and Russell [@pone.0011344-Taylor1] provided eleven tables from which, based on the given base rate, selection rate, and validity coefficient, the percentage of correctly accepted submissions (AQ/(AQ+AN)) can be taken. For one, the Taylor-Russell tables can be used as an alternative to calculation of the success rate using equation (3). For another, using the tables we can examine how the success rate of a selection process (utility) would be changed by different (hypothetical) values for base rate, selection rate, and/or validity coefficient. Taylor and Russell developed their tables "making use of Pearson\'s Tables for Finding the Volumes of the Normal Bivariate Surface (1931)" [@pone.0011344-Cascio1]. That means the Taylor-Russell tables are appropriate only if "the assumptions of bivariate normal, linear, homoscedastic relationships between predictor and criterion" [@pone.0011344-Cascio1] are fulfilled. For instance, with base and selection rates of 50% and a validity coefficient of r~xy~ = .5 (see graph B in [Figure 1](#pone-0011344-g001){ref-type="fig"}), as the Taylor-Russell table shows, 67% of the submissions accepted for publication are correctly accepted (that is, they are qualified submissions). In contrast to graph B, in graph C in [Figure 1](#pone-0011344-g001){ref-type="fig"} the base rate has been raised to 80% (that means that more submissions are fundamentally qualified for publication) and the selection rate has been reduced to 30% (that means that fewer submissions are selected by the editors). Even though the validity coefficient in graph C is exactly the same as in graph B (r~xy~ = .5), in graph C submissions will hardly be selected that later turn out to be non-qualified (that is, publications with low impact). According to the corresponding Taylor-Russell table, with a base rate of 80%, a selection rate of 30%, and a validity coefficient of r~xy~ = .5, 94% of submissions accepted for publication are correctly accepted. However, as graph C shows, also many submissions are rejected that would have been fundamentally suitable for publication. Methods {#s2} ======= Manuscript reviewing at AC-IE {#s2a} ----------------------------- A manuscript submitted to AC-IE is usually subject to internal and external review. First, editors at the journal evaluate whether the manuscript contributes to the development of an important area of research (internal review). If the editors find this to be the case, the submitted manuscript is sent to usually three independent reviewers (external review); the reviewers are asked to review it using an evaluation form and a comment sheet. The reviewers know the authors\' identities, but their reviews are not signed (single blinding). In the year 2000 the AC-IE evaluation form for reviewers contained the following four questions (in 2008 this was changed to five questions): (1) 'How important do you consider the results?' (four response categories: very important; important; less important; unimportant); (2) 'Do the data obtained by experiment or calculation verify the hypothesis and conclusions?' (two response categories: yes; no); (3) 'Is the length of the manuscript appropriate to its contents?' (three response categories: yes; no - the manuscript is too short; no - the manuscript is too long); (4) 'Do you recommend acceptance of the Communication?' (four response categories: yes - without alterations; yes - after minor alterations; yes - but only after major alterations; no). If reviewers find a manuscript unsuitable for AC-IE, they are asked to name another journal in which the study findings might more suitably be published. Once they have received the reviewers\' reports, the editors make the decision to accept or reject a manuscript for publication. Our previous findings show that in general a manuscript is published in AC-IE only if two reviewers rate the results of the study as 'important' or 'very important' (on question 1 above) and also recommend publication in the journal (do not answer 'no' to question 4 above) [@pone.0011344-Bornmann7], [@pone.0011344-Bornmann8]. The AC-IE calls this their 'clear-cut rule,' and the editors use it when they base their publication decisions on either two reviews (one of the reviewers asked to review the manuscript did not do so or did not complete the review on time) or three reviews. If the journal editor makes the decision based on only one review (this was the case for about 100 manuscripts in the present study), a manuscript is accepted for publication only if the reviewer has given positive answers to both of the questions on the evaluation form mentioned above. The editor will make a decision based on only one review if the other reviewers have not sent in their reviews despite several reminders and if the editor is of the opinion that the decision can be made based on one review. For about one-fifth of the manuscripts that were reviewed at AC-IE in the year 2000, the editors (1) requested a review from a so-called top adviser, or (2) had a reviewer review a manuscript that had been revised by the author, or (3) had an appeal reviewed by a reviewer that an author had filed against the rejection of his/her manuscript [@pone.0011344-Bornmann8]. Decision rules like the AC-IE\'s 'clear-cut rule' have become a new research topic in recent publications on the journal peer review process [@pone.0011344-Schultz1], [@pone.0011344-Egghe1]. Database for the present study and conducting of citation analysis {#s2b} ------------------------------------------------------------------ For the investigation of the manuscript selection process at AC-IE, we used information on all 1899 manuscripts that went through internal and external review in the year 2000. Of the 1899 manuscripts, 46% (n = 878) were accepted for publication in AC-IE, and 54% (n = 1021) were rejected. A search in the literature databases Science Citation Index (SCI) (Thomson Reuters) and Chemical Abstracts (CA) (Chemical Abstracts Services, CAS, Columbus, OH) revealed that of the 1021 rejected manuscripts, 959 (94%) were later published in 136 other (different) journals. For accepted manuscripts and manuscripts that were rejected (but published elsewhere), we determined the number of citations for a fixed time window of three years after the publication year. "Fixed citation windows are a standard method in bibliometric analysis, in order to give equal time spans for citation to articles published in different years, or at different times in the same year" [@pone.0011344-Craig1]\]. The citation analyses for the present study were conducted in the year 2007 based on CA. CA is a comprehensive database of publicly disclosed research in chemistry and related sciences (see <http://www.cas.org/>). Citation counts are attractive raw data for the evaluation of research output: they are "unobtrusive measures that do not require the cooperation of a respondent and do not themselves contaminate the response (i.e., they are non-reactive)" [@pone.0011344-Smith1]. Although citations have been a controversial measure of both quality and scientific progress [@pone.0011344-Bornmann9], they are still accepted as a measure of scientific impact, and thus as a partial aspect of scientific quality [@pone.0011344-Martin1]: According to van Raan [@pone.0011344-vanRaan2], citations provide "a good to even very good quantitative impression of at least one important aspect of quality, namely international impact" (p. 404). For Lindsey, citations are "our most reliable convenient measure of quality in science -- a measure that will continue to be widely used" [@pone.0011344-Lindsey1]. In an article by Jefferson, Wager and Davidoff [@pone.0011344-Jefferson1] it was pointed out that measuring peer review requires a clear statement of the metric. They identified several, of which several (importance, usefulness and relevancy) mapped to citation counts. In their study, the authors are very explicit about using citation count as their metric. To find out whether a publication has a high or low citation impact, its performance is compared with international scientific reference standards [@pone.0011344-Bornmann10]. For this, Vinkler [@pone.0011344-Vinkler1] recommends use of a worldwide reference standard: "Relative Subfield Citedness (R~w~) (where W refers to 'world') relates the number of citations obtained by the set of papers evaluated to the number of citations received by a same number of papers published in journals dedicated to the respective discipline, field or subfield" [@pone.0011344-Vinkler2]. To calculate R~w~ for the manuscripts in this study, specific reference standards were used that refer to the subject areas of CA [@pone.0011344-vanLeeuwen1], [@pone.0011344-Neuhaus1]. CAS categorizes chemical publications into 80 different subject areas (called sections). Every publication becomes associated with a single principal entry that makes clearly apparent the most important aspect of the work [@pone.0011344-Daniel1]. In contrast to the journal sets provided by Thomson Reuters, CA sections are assigned on a paper-by-paper basis [@pone.0011344-Bornmann10]. To determine R~w~ in this study, the number of citations for accepted or rejected (but published elsewhere) manuscripts were divided by the (arithmetic) mean number of citations of all publications in a corresponding subject area [@pone.0011344-Radicchi1]. According to van Raan [@pone.0011344-vanRaan1] the R~w~ quotient allows determination of whether the citation impact of the accepted and rejected (but published elsewhere) manuscripts is far below (R~w~\<0.5), below (R~w~ = 0.5--0.8), approximately the same as (R~w~ = 0.8--1.2), above (R~w~ = 1.2--1.5), or far above (R~w~\>1.5) the international (primarily the Western world) citation impact baseline for the corresponding subject areas. With R~w~ values above 1.5, the probability of identifying excellent contributions is very high [@pone.0011344-vanRaan1]. Of all 1837 manuscripts published in the AC-IE (accepted manuscripts) or another journal (rejected manuscripts), 906 could be included in the analysis of this study. The reduced number was due to missing values for one or more of the variables included in this study. Reviewer\'s ratings were not available for all manuscripts; for example, some reviewers filled out only the comment sheet and did not fill out the evaluation form with the 2 questions (importance of a manuscript; reviewer\'s recommendation concerning publication). In addition, for some manuscripts, no citation counts and/or reference values were available [@pone.0011344-Bornmann1]. Results {#s3} ======= If the usefulness of peer reviewing for the manuscript selection process is determined using the model developed by Taylor and Russell [@pone.0011344-Taylor1], it must first be established what criterion will be used for considering a submission suitable (qualified) for publication. This can be done only based on the aims of the scholarly journal. As described above, AC-IE has one of the highest JIFs of the journals in the field of chemistry. To guarantee that AC-IE also has a high JIF in future, it is necessary that the editors accept for publication only those submissions that after publication will have far above-average citation counts. For this reason, in this study we rated a submission as qualified for publication in AC-IE, if it showed R~w~\>1.5 (criterion cutoff). Hence, all submissions with R~w~\>1.5 were categorized as qualified (that is, after publication they made a far above-average contribution to scientific advancement in their subfields) and all submission with R~w~≤1.5 as non-qualified (that is, after publication they do not make this significant contribution to scientific advancement in their subfields). In the three graphs (A, B, and C) in [Figure 2](#pone-0011344-g002){ref-type="fig"}, the R~w~ criterion cutoff is plotted as a red line starting from the y-axis. As the R~w~ values were logarithmized (log(x+1)) in order that the distribution of data more likely approximates a normal distribution [@pone.0011344-Stewart1], the criterion cutoff lies at log(1.5+1) = .92. Graph A shows those submissions for which the AC-IE editor based the publication decision on one review; for the submissions in graphs B and C the editor made the publication decisions based on two and three reviews, respectively. In all three graphs in [Figure 2](#pone-0011344-g002){ref-type="fig"} the ratings of the reviewers that formed the basis for the publication decision are on the x-axis. For these ratings we used the sum of the reviewer\'s answers to two questions: (1) How important do you consider the results? (four response categories: very important = 3, important = 2, less important = 1, unimportant = 0), and (2) Do you recommend acceptance of the Communication? (four response categories: yes--without alterations = 3; yes--after minor alterations = 2; yes -- but only after major alterations = 1; no = 0). For example, if for a submission in graph A a reviewer answered 'important' on question (1) and 'yes -- without alterations' on question (2)‚ the value 5 (2+3) was plotted on the x-axis (reviewer\'s ratings). If the editor based his/her decision on three reviews, as is the case for the submissions in graph C, we used the sum of the three reviewers\' answers on a submission to the two questions: For example, if Reviewer 1 answered 'less important' and 'yes - but after major alterations,' Reviewer 2 answered 'unimportant' and 'no,' and Reviewer 3 answered 'important' and 'yes - without alterations,' then the value 7 (1+1+0+0+2+3) was entered into graph C on the x-axis (reviewers\' ratings). ![Base rate, selection rate, success rate, and validity coefficient (r~xy~) in the reviewing of AC-IE submissions by one (Graph A), two (Graph B), or three (Graph C) reviewers.\ The green line in the graphs along with 95% confidence interval (gray lines) is the prediction for the criterion, based on a linear regression of the criterion on the predictor.](pone.0011344.g002){#pone-0011344-g002} The three graphs show accepted manuscripts as blue circles and rejected manuscripts as red circles. The distributions of the blue and red circles in the graphs make it clearly visible that for nearly all of the manuscripts, values for the reviewers\' ratings above a certain predictor cutoff led to acceptance by the AC-IE editor and values below the cutoff led to rejection. When the publication decision is based on one review, the predictor cutoff is 3; when the publication decision is based on two and three reviews, this value is 6 and 8.5, respectively. As the reviewers usually recommend that a submission should be published if it is in their opinion important, the above-mentioned "clear-cut rule" at AC-IE results in most cases in an unambiguous assignment of the reviewers\' ratings to the editorial decisions. For only a few manuscripts is this not the case. As presented in Bornmann and Daniel [@pone.0011344-Bornmann7], [@pone.0011344-Bornmann8], the editor\'s deviations from the clear-cut rule on few submission can be explained well through further information from the journal\'s archives (the manuscripts were for example withdrawn by the author during the review process despite a positive review, or the editor suggested to the author that a manuscript be submitted to a specialized journal -- the readership of AC-IE are chemists of all subfields). On the basis of the predictor (reviewers\' ratings) and criterion (R~w~) cutoffs, we determined for the submissions in the three graphs of [Figure 2](#pone-0011344-g002){ref-type="fig"} the base and selection rates. Whereas the base rates at AC-IE are 59% (with two reviews), 61% (with three reviews), and 64% (with one review), the selection rates are 44% (with one or with two reviews) and 50% (with three reviews). The resulting validity coefficients (Spearman\'s rank-order correlations) are r~xy~ = .34 (with one review), r~xy~ = .26 (with two reviews), and r~xy~ = .17 (with three reviews). Given the typical use of two or three reviewers assigned to manuscripts submitted to a journal, we would expect to see that the use of more reviewers per manuscript yields more valid recommendations and thus greater utility. But, the validity coefficients actually drop with more reviews per manuscript. This unexpected result is due to the usual decision process of the AC-IE editors: They wait for further reviews when a publication decision does not seem possible on the received reviews [@pone.0011344-Bornmann8]. To test whether more reviews actually lead to a higher coefficient we correlated predictor (reviewers\' ratings) and criterion (R~w~) for one review (first review) and two reviews (first and second review) for those 205 manuscripts that have received three reviews. This analyses show the expected order of coefficients: r~xy~ = .05 (with one review), r~xy~ = .11 (with two reviews) and r~xy~ = .17 (with three reviews). If Eq. (3) is used for calculating the success rates for AC-IE, 78% of the submissions accepted are correctly accepted for publication (with a predictor cutoff\>3) when the editor\'s decision is based on one review, 69% of the submissions are correctly accepted for publication (with a predictor cutoff\>6) when the editor\'s decision is based on two reviews, and 65% of the submissions are correctly accepted for publication when the editor\'s decision is based on three reviews (with a predictor cutoff\>8.5). As the values in [Figure 2](#pone-0011344-g002){ref-type="fig"} show, the percentages of submissions accepted that are qualified submissions are calculated using Eq. (3) differ only a little from the values in the Taylor-Russell tables [@pone.0011344-Taylor1] (even though the assumptions for use of the tables are not completely fulfilled): For the editor\'s decision based on one review, the Taylor-Russell table percentage is 73% (difference of 5 percentage points), for the editor\'s decision based on two reviews it is 69% (no difference,) and for editor\'s decision based on three reviews it is 66% (difference of 1 percentage point). Independently of the number of reviews on which the editor\'s decision is based, the percentage of submissions accepted that are qualified submissions is above the base rates; however, the differences between the base rates and success rates are not large. But with base rates of approximately 60% in the AC-IE selection process and validity coefficients of r~xy~ = .2 and r~xy~ = .3, respectively, it is difficult, to achieve considerably better success rates than base rates using the predictor (the reviewers\' ratings). Using the tables in Taylor and Russell [@pone.0011344-Taylor1], it can then be estimated through what changes in the selection rate, base rate or validity coefficient a higher success rate (utility) in the AC-IE selection process can be achieved. Here the values in the Taylor-Russell tables should be viewed as only rough approximate values, because the assumptions mentioned above for use of the tables are not completely fulfilled. If AC-IE could increase the validity to a value of r~xy~ = .55 -- for example by reviewers having precise information on what makes a manuscript a high quality manuscript (for instance, at AC-IE operationalized as 'most accessed' or 'most cited' papers, or qualified as 'Highlights' by the journal) and information on how high the selection rate at AC-IE is -- then according to the Taylor-Russell table, with a base rate of 60% and a selection rate of 40%, there would be 81% submissions accepted that are qualified submissions -- an approximately 20 percentage-point gain in utility over the base rate. About the same gain could be achieved through lowering the selection rate to about 5% -- with an unchanged validity coefficient and unchanged base rate. Discussion {#s4} ========== In this paper we presented an approach that in addition to validity points to the importance of the base rate and selection rate for determining the utility of peer review for editors\' publication decisions. High predictive validity (meaning a high correlation between the outcome of peer review and the scientific impact of the later publication) does not inevitably mean that the ratings of the reviewers are very useful for the selection of submissions. In addition to considering the base rate and selection rate when evaluating a manuscript selection process, according to Cascio [@pone.0011344-Cascio1] there is further advantage of utility analysis, as it "provides a framework for making decisions by forcing the decision maker to define goals clearly, to enumerate the expected consequences or possible outcomes of his/her decision, and to attach differing utilities or values to each. Such an approach has merit because resulting decisions are likely to rest on a foundation of sound reasoning and conscious forethought" (p. 191). The Taylor and Russell [@pone.0011344-Taylor1] model basically indicates that an effective submission selection policy should aim to provide for a high base rate, that is, a high percentage of potentially suitable submissions. A good submission policy (such as through providing on a journal\'s Web site a precise description of what the journal is looking for in a manuscript) can increase the number of suitable manuscripts submitted and decrease the number of unsuitable ones. In other words, the applied policy should lead to self-selection among submissions. Authors should not submit manuscripts that cannot meet the high quality standards of a journal or that deal with a topic that does not fit the journal. This would minimize the required effort/expense for the selection process, to which belong, for example, correspondence with the author (such as acknowledgement of receipt of a submission or rejection notice after review) and peer reviews that turn out to be 'superfluous.' This self-selection should be the first step of a manuscript selection process. The second step should be pre-selection using a pre-screening procedure, so that the later time-consuming and costly process is carried out only for those submissions that seem potentially suitable. A successful pre-selection increases the base rate for the subsequent selection process. For example, the journal Atmospheric Chemistry and Physics (ACP) conducts an access review [@pone.0011344-Pschl1], by which the designated reviewers are asked the following questions about whether a manuscript meets the ACP\'s principal evaluation criteria: (1) scientific significance ('Does the manuscript represent a substantial contribution to scientific progress within the scope of ACP (substantial new concepts, ideas, methods, or data?'), (2) scientific quality ('Are the scientific approach and applied methods valid? Are the results discussed in an appropriate and balanced way (consideration of related work, including appropriate references)?'), and (3) presentation quality ('Are the scientific results and conclusions presented in a clear, concise, and well-structured way (number and quality of figures/tables, appropriate use of English language)?'). The response categories for the three questions are: (1) excellent, (2) good, (3) fair, and (4) poor. Only those manuscripts that are rated at least 'good' on all criteria are sent to the referees that already participated in the access review, and perhaps to additional referees, for complete commenting (reviewing). Complete commenting of the manuscript is also not least designed to provide authors suggestions for improvement of the manuscript. Through this procedure, totally unsuitable manuscripts are excluded from the complete commenting stage. Following similar procedures to those used at ACP, editors and reviewers at other journals could conduct initial pre-screenings of submitted papers to check whether the manuscripts are basically suitable for publication in the journal and whether there are compelling arguments against publication in the journal. The task of the actual review process with extensive commenting would then be to mainly formulate recommendations for improvement of the submitted manuscript and if necessary to recommend a more suitable journal for publication of the manuscript. The usefulness of peer review for editors\' decisions to accept submissions for publication cannot be fully determined using Taylor and Russell\'s [@pone.0011344-Taylor1] model and tables. For one, use of the Taylor-Russell tables assumes "bivariate normal, linear, homoscedastic relationships between predictor and criterion" [@pone.0011344-Cascio1]. If these assumptions are not fulfilled (for bibliometric data it can be generally assumed that these assumptions are not fulfilled completely even after logarithmic transformation), the table values can be used only with reservations. For another, the reviewing of submitted manuscripts in many cases results in improvement of the final publication [@pone.0011344-Daniel1]; this kind of usefulness of peer review is difficult to quantify, however. Regardless of the applicability of the Taylor-Russell tables and the improvement function of peer review, we want to stimulate further studies which test our approach for determining the usefulness of peer reviews for the selection of manuscripts for publication by other journals. It would be especially interesting to base single future studies on more than one journal. This would allow comparisons between the results for different journals. There are some limitations of our study that we would like to point out: First, we used post-publication above average citation counts as the measure for deeming a submission 'suitable for publication.' It can be questioned, whether citation counts is a measure of publication suitability? Recently, Straub [@pone.0011344-Straub3] offered a number of criteria for high quality research, e.g., logical rigor or replicability of research. Do these criteria relate to, or are reflected in citation counts? Second, we used an ex-post measure (at t = 1) to determine the ex-ante suitability of a paper (at t = 0). Yet, there are many things that could have happened between t = 0 and t = 1 that could have contributed to the citation count, without the paper being 'good.' For instance, the journal could have risen in attractiveness or ranking, thereby motivating more scholars to cite papers of this journal. Or, the topic of a paper could have been on the peak of a "fashion wave", i.e., a hot topic [@pone.0011344-Baskerville1], even though the paper itself may or may not be a 'good' contribution. Third, we chose the thresholds of R~w~\> and \<1.5 to measure 'suitability.' This is an the 'experience-based' assignment [@pone.0011344-vanRaan1] to impact classes. There might be other ways of measuring 'suitability' altogether. The authors would like to thank Dr. Christophe Weymuth (formerly at the Organic Chemistry Institute of the University of Zurich and now at BIOSYNTH AG, Switzerland) for investigation of the manuscripts rejected by Angewandte Chemie International Edition and published elsewhere. We also thank Dr. Werner Marx and Dr. Hermann Schier of the Central Information Service for the Institutes of the Chemical Physical Technical (CPT) Section of the Max Planck Society (located at the Max Planck Institute for Solid State Research in Stuttgart, Germany) for conducting the citation search for citations of the accepted and rejected (but published elsewhere) manuscripts in the literature database Chemical Abstracts (CA). We thank Dr. Peter Gölitz, Editor-in-Chief of Angewandte Chemie, the Editorial Board of Angewandte Chemie, and the German Chemical Society (GDCh, Frankfurt am Main, Germany) for permission to conduct the evaluation of the selection process of the journal and the members of the editorial office for their generous support during the carrying out of the study. Competing Interests:The authors have declared that no competing interests exist. Funding:The study, which is also investigating quality assurance of open access journals, is supported by a grant from the Max Planck Society ([www.mpg.de](http://www.mpg.de)). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: LB. Performed the experiments: LB. Analyzed the data: LB. Contributed reagents/materials/analysis tools: LB. Wrote the paper: LB HDD.	{ "pile_set_name": "PubMed Central" }
Background ========== Internationally, there is substantial evidence of unfair inequalities in health between ethnic groups and, for many countries with histories of colonization, inequalities between indigenous and non-indigenous peoples within the same territory (e.g. New Zealand, Canada, and Australia). In New Zealand, there are significant and long-standing inequalities in a range of health outcomes, risk factors and healthcare between Māori (indigenous peoples) and Pākehā (European) \[[@B1]-[@B3]\]. Māori comprise approximately 15% of the New Zealand population \[[@B4]\]. Within New Zealand and internationally, there is recognition of the important role of racism as a basic underlying cause of ethnic inequalities in health \[[@B5]-[@B10]\]. Jones defines racism as, "a system (consisting of structures, policies, practices, and norms) that structures opportunity and assigns value based on ...the way people look \[racially\]" \[[@B11]\], p9. Racist practices can operate at the level of institutions and individuals \[[@B6],[@B8]-[@B13]\] with a number of pathways through which it may impact health \[[@B6],[@B8]-[@B13]\]. Racism can act directly to affect health through trauma and stress pathways, as well as acting indirectly by shaping the distribution of societal resources and individual determinants of health by ethnicity, and influencing access to health care and quality of care by ethnicity \[[@B12],[@B13]\]. In New Zealand, Māori report experiencing disproportionately higher racial discrimination at an individual level that has been linked to a range of adverse health outcomes, heightened health risk and poorer health care as well as contributing to ethnic health inequalities between Māori and Pākehā (European) \[[@B5],[@B14],[@B15]\]. At a structural level, the distribution of socioeconomic resources in New Zealand is highly racialized, with Māori more disadvantaged socioeconomically compared to European \[[@B1]\]. Racialization, or the categorization and stratification of social groups on the basis of 'race', is a fundamental element of racism \[[@B16]\]. 'Race' or 'ethnicity' is increasingly understood as socially constructed, with official measurement of ethnicity in New Zealand now based on self-identified cultural affiliation, following a shift over time from more biological based definitions \[[@B17]\]. However, in racialized societies, race/ethnicity remains a salient externally-imposed category that continues to draw on discredited notions of biology, blood and genes \[[@B17]\]. This study examines the concept of socially-assigned ethnicity and its relationship to health for Māori. Socially-assigned (or socially-ascribed) race/ethnicity refers to the way in which an individual is classified by others racially (or ethnically), based on supposed 'racial' markers, principally physical appearance \[[@B18]\]. The way people are racially or ethnically viewed by others is linked to how they are treated and the opportunities they receive in society. Jones \[[@B11]\] suggests that it is socially-assigned race that results in health impacts, rather than self-identified race per se. There has been some initial support for this hypothesis, with a study using data from the United States 2004 Behavioral Risk Factor Surveillance System (BRFSS) showing that being socially-assigned as 'white' confers a health advantage, irrespective of one's self-identified ethnicity \[[@B18]\]. Our study uses data from the 2006/07 New Zealand Health Survey to examine health among Māori participants, specifically the association of socially-assigned ethnicity with health and with individual experiences of racism amongst Māori. We hypothesize that among people who self-identify as Māori, being socially-assigned as of the dominant ethnic group ('European') will be associated with health advantage. We hypothesize that this will operate in part due to lower exposure to racial discrimination among people socially-assigned as European. Methods ======= Study design ------------ The study uses data from the 2006/07 New Zealand Health Survey (for adult respondents). This is the fourth national population based health survey conducted by the New Zealand Ministry of Health. It provides nationally representative information on measures of self-reported health, health service utilization, health risk and protective factors, and demographic variables among the usually resident population aged 15 years and over \[[@B19]\]. ### Survey design The survey uses an area based sampling frame with meshblocks (small geographical areas of approximately 100 people) as the primary sampling units. Sampling was undertaken using a multi-stage, stratified, probability-proportional-to-size design. 1385 meshblocks were initially selected, followed by dwellings within each meshblock and finally one eligible adult from each selected dwelling. Increased sampling of Māori, Pacific and Asian peoples (relative to population size) was achieved mostly through the use of a large screening sample, and to a limited extent through disproportionate sampling of areas with higher proportions of Māori at the District Health Board level. A total of 12488 adults were surveyed in face-to-face interviews between 6 October 2006 and 29 November 2007. This included 3160 people who self-identified as Māori (either alone or in combination with other ethnicities). The weighted response rate was 67.9% overall and 67.5% for Māori. Further detail on the survey design can be found elsewhere \[[@B19]\]. Ethics approval for the study reported in this paper was granted by the New Zealand Health and Disability Multi-Region Ethics Committee (MEC/10/050/EXP). Key variables ------------- ### Self-identified ethnicity This study is restricted to those people who self-identified as Māori (either alone or in combination with another ethnicity, n = 3160). In New Zealand, ethnicity is officially conceptualized as a social construct of group affiliation and cultural identity. The term 'ethnicity' is used rather than 'race' in the collection of official statistics with standard protocols in the health sector for the collection, recording and outputting of ethnicity data for statistical purposes \[[@B20]\]. The 2006/07 NZHS uses the same ethnicity question as the New Zealand Census. This asks people to identify the ethnic group or groups they feel they belong to. People can identify with more than one ethnic group. A number of ethnic groups are given as response options as well as an 'other' option that allows people to write their ethnicity. In line with indigenous rights to self-identification \[[@B21]\], we considered Māori as anyone who self-identified as Māori, either alone or in combination with another ethnicity. It is standard practice to present population health statistics and ethnic inequalities by self-identified ethnicity in New Zealand \[[@B20]\]. ### Socially-assigned ethnicity In addition to self-identified ethnicity, participants were also asked to report their socially-assigned ethnicity. Participants were asked, "Earlier you told me your ethnicity. Now I will ask you some questions about reactions to your ethnicity. How do other people usually classify you in New Zealand?" Response options were the same as for the self-identified ethnicity question and people could respond with multiple ethnicities. The socially-assigned ethnicity question was based on a similar question developed for the 'Reactions to race' module of the Behavioral Risk Factor Surveillance System (BRFSS) in the United States \[[@B22]\]. In order to test the hypothesis that being socially-assigned as dominant will be associated with health advantage and lower exposure to racial discrimination, socially-assigned ethnicity responses were grouped into dominant (European-only, n = 771) and non-dominant (any socially-assigned Māori or other non-European ethnicity, either alone or in combination with socially-assigned European ethnicity, n = 2389) groups. This approach is similar to that of other studies on the impact of socially-assigned ethnicity on health \[[@B18],[@B23]\]. ### Racial discrimination People were asked about their individual experience of racial discrimination in New Zealand in five settings. These were experience of an ethnically motivated verbal attack, physical attack, unfair treatment because of ethnicity by a health professional, unfair treatment in work or gaining a job, and unfair treatment when renting or buying housing. Prevalence of 'ever' experiencing racial discrimination for each of these domains was analysed. For the purposes of logistic regression modelling, these racial discrimination items were reduced to a three level variable when used as an explanatory variable (i.e. never, one domain ever, two domains or more ever), or a binary variable of any experience of racial discrimination ever when modelling experience of racism as an outcome (yes/no). Additional information on development, use and variable reduction of the racial discrimination questions can be found elsewhere \[[@B24]\]. ### Health outcomes Two health measures were examined, general self-rated health \[[@B3]\] and non-specific psychological distress (using the Kessler 10-item scale) \[[@B25]\]. In terms of general self-rated health, participants were asked, 'In general, would you say that your health is: excellent, very good, good, fair or poor'. In order to assess health advantage related to socially-assigned ethnicity, responses were dichotomized to excellent/very good versus good/fair/poor. This is in line with the two similar studies in this area \[[@B18],[@B23]\]. General self-rated health has been shown to be a valid indicator of health status with strong evidence it is predictive of long term illness and mortality \[[@B26],[@B27]\]. Psychological distress was measured using the Kessler 10-item scale. This scale asks 10 questions on how frequently participants' felt symptoms of psychological distress in the past four weeks. Responses range from 'None of the time' to 'all of the time' on a five-point scale (0 to 4 per item). An overall score is generated from 0 to 40 with higher scores reflecting higher levels of psychological distress \[[@B3],[@B25],[@B28]\]. K10 scores were treated as a continuous outcome variable in these analyses \[[@B29]\]. ### Other covariates Other variables adjusted for in multivariable analyses included gender (men, women (reference)) and age group (15---24 (reference), 25---34, 35---44, 45---54, 55---64, 65---74, 75+ years). Four measures of socioeconomic position were also adjusted for in regression analyses. These included education qualification, living standards, individual level deprivation and area-based deprivation. Education qualification was classified as no secondary qualification attained (reference) and attained secondary qualification or higher. Living standards were measured using the Economic Living Standards Index short form (ELSI-SF). This scale includes 25 items that cover questions on the three conventional constructs of living standards measures: economizing behaviours, ownership of durable assets, and social participation restrictions. Scores range from 0---31 and are classified into seven categories of living standards: very good, good, comfortable (reference), fairly comfortable, some hardship, significant hardship, and severe hardship \[[@B30],[@B31]\]. Individual level deprivation was measured using the New Zealand Index of Socioeconomic Deprivation for Individuals (NZiDep) \[[@B32]\]. The index comprises 8 questions that are scored to create a five-point individual-level index of socioeconomic deprivation. Questions ask individuals about experiencing particular situations in the last 12 months e.g. being forced to buy cheaper food, being out of paid work, receiving a benefit, putting up with feeling cold to save on heating costs. Scores range from 1 (no deprivation characteristics \[reference\]), 2 (one characteristic), 3 (two characteristics), 4 (three or four characteristics) and 5 (five or more deprivation characteristics). Area deprivation was measured using the New Zealand Index of Deprivation 2006 (NZDep2006). This provides a deprivation score for each meshblock (small areas of approximately 100 people) in New Zealand. It was created by combining (by principal component analysis) nine variables from the 2006 Census (benefit receipt, house-hold income, living in a single-parent family, home ownership, employment status, access to a telephone, qualifications, living space, and access to a car). Scores are categorized into deciles (1 (ref) ---10) with higher scores indicating greater deprivation \[[@B33]\]. Data analysis ------------- We analysed data in SAS 9.2 (SAS Institute, NC) using Survey analysis based procedures. All analyses account for the stratification and clustering elements of the survey design, with the exception of the descriptive statistics covering the cross-tabulation of self-identified and socially-assigned ethnicity. All data are weighted for probability of selection and non-response using survey weights to produce representative estimates for the New Zealand adult population and for appropriate calculation of confidence intervals. Unweighted frequencies and weighted prevalence of demographic variables, experience of racial discrimination and self-rated health outcomes were calculated for people socially-assigned as dominant and non-dominant ethnicity. Weighted means, medians and interquartile ranges were estimated for K10 scores. Logistic regression (with the Proc Surveylogistic procedure in SAS) was used to examine the independent association of socially-assigned ethnicity with any experience of racial discrimination ever (as a binary variable). Unadjusted and sequentially adjusted models (adding new adjustment covariates to the preceding model) are presented. Model covariates were chosen prior to analysis, and include age, gender, and socioeconomic position (qualification, ELSI, NZiDep, NZDep2006). Logistic regression (with the Proc Surveylogistic procedure in SAS) was also used to examine the independent association of socially-assigned ethnicity with optimal self-rated general health (excellent or very good self-rated health). Unadjusted and sequentially adjusted models are presented. Covariates again include age, gender, socioeconomic position (qualification, ELSI, NZiDep, NZDep2006) and individual experience of racial discrimination ever (in three levels). Modelling of K10 scores was conducted using Proc Surveyreg, using a linear model approach. Model covariates, and the sequential sets of covariates adjusted for at each stage of modelling, were the same as described for the self-rated health analysis above. K10 scores in the general population are known to be right skewed rather than normally distributed, which might generally suggest that linear regression on raw data is not appropriate. However, group comparison of means are still valid for these data, as the central limit theorem dictates that in a suitably large sample the standard error will be a valid estimate of sampling variation processes, and by extension confidence intervals and hypothesis tests of regression parameters, will also be valid \[[@B34]\]. These methods have been applied previously to analysis of K10 data from a complex survey in New Zealand \[[@B29]\]. In the models examining the association between socially-assigned ethnicity and health (general self-rated health and psychological distress), age and sex were adjusted for as potential confounders, while socioeconomic measures and racial discrimination were adjusted for as markers of institutional and individual racial discrimination and potential pathway variables by which racial discrimination based on socially-assigned ethnicity may be operating to affect health. Reductions in ORs or mean differences after adjustment for each set of variables would be consistent with this theory. ### Self-identified ethnicity combinations All regression models with health and experience of racial discrimination as outcomes were also undertaken with self-identified ethnicity (reported both European and Māori ethnicity, versus reported Māori ethnicity but not European) included in addition to socially-assigned ethnicity to test whether health and racial discrimination outcomes were associated with self-identified ethnicity above and beyond the association seen with socially-assigned ethnicity. Self-identified ethnicity was not related to outcomes after accounting for socially-assigned ethnicity and so was not included in the final models (data not shown). Examination of point estimates for the socially-assigned ethnicity variable in these models indicated that the inclusion of the different Māori self-identified categories did not affect the relationship between socially-assigned ethnicity and the outcome variables of interest. Interaction terms between the new self-identified ethnicity and the socially assigned ethnicity (coded as per the main analysis) were also examined: these interaction terms were not significant for the health outcomes (suggesting that the impact of socially assigned ethnicity does not depend on self-identified ethnicity; interaction term likelihood ratio test p-values = 0.681 and 0.512, for SF-36 and K-10 outcomes respectively). For the racial discrimination outcome, this interaction was significant (LR test p-value = 0.002,) and this relationship is described more fully in the Results section. Results ======= Of the 12488 participants in the 2006/07 New Zealand Health Survey, 3160 self-identified Māori as one of their ethnic groups. Among this Māori sample, Table [1](#T1){ref-type="table"} shows the four most common self-identified ethnicity combinations (98.7%, n = 3118) with the remaining Māori ethnic combinations grouped together as Māori/Other (1.3%, n = 42). Within each of these self-identified combinations, Table [1](#T1){ref-type="table"} also shows the most common socially-assigned ethnicity combinations. Among Māori included in non-European socially-assigned ethnic groups, most (2240 of 2389, 94%) were socially-assigned as Māori either alone or in combination with other ethnic groups while 149 (6%) were socially-assigned as another non-Māori/non-European ethnicity group. ###### Cross tabulation of self-identified and socially-assigned ethnicity combinations among people who self-identify as Māori Socially-assigned ethnicity n (% within self-identified ethnicity row) ------------------ ---------------------------------------------------------------------------- -------- ----- -------- ---- -------- ---- ------- ----- -------- ---- -------- ---- -------- M (n = 1660) 1327 (79.9) 64 (3.9) 44 (2.7) 27 (1.6) 113 (6.8) 45 (2.7) 40 (2.4) M, E (n = 1351) 438 (32.4) 205 (15.2) 13 (1.0) 16 (1.2) 635 (47.0) 14 (1.0) 30 (2.2) M, P (n = 58) 38 (65.5) 0 8 (13.8) 1 (1.7) 1 (1.7) 8 (13.8) 2 (3.4) M, P, E (n = 49) 23 (46.9) 1 (2.0) 6 (12.2) 3 (6.1) 12 (24.9) 1 (2.0) 3 (6.1) M, O (n = 42) 22 (52.4) 2 (4.8) 1 (2.4) 1 (2.4) 10 (23.8) 0 6 (14.3) M=Māori, P=Pacific, E=European, O=other ethnicity or ethnic combinations including Asian. Agreement between self-identified and socially-assigned ethnic groups was varied. The two most common self-identified ethnicity combinations among the study group were Māori-only (n = 1660) and Māori/European (n = 1351). Among participants who self-identified as Māori-only, 80% reported their socially-assigned ethnicity as Māori-only, while 7% reported being socially-assigned as European-only and 4% as both. Among those who self-identified as both Māori and European, 32% were socially-assigned as Māori-only, 47% as European-only and 15% as both Māori and European. Table [2](#T2){ref-type="table"} shows the weighted prevalences of demographic variables for Māori participants socially-assigned as European-only, and those socially-assigned to any Māori or other non-European ethnicity. These groups had similar age and sex profiles. However, Māori who were socially-assigned as being European-only tended to be more advantaged with regards to socioeconomic measures than Māori socially-assigned as being any Māori or other non-European ethnicity. Māori socially-assigned as European-only were more likely to have at least a secondary educational qualification, report higher living standards, be less socioeconomically deprived at an individual level (NZiDep) and also tended to live in less deprived areas (NZDep decile) than Māori who were socially-assigned to Māori or non-European ethnic groups (Table [2](#T2){ref-type="table"}). ###### Distribution of demographic variables by socially-assigned ethnicity combinations among people who self-identify as Māori Socially-assigned ethnicity ------------------------------------- --------------------------------- ------------- Gender Female 484 (54.5) 1471 (53.3) Male 287 (45.5) 918 (46.7) Age group 15--24 years 162 (27.7) 482 (27.0) 25--34 years 172 (21.6) 553 (20.8) 35--44 years 166 (21.3) 504 (20.5) 45--54 years 108 (13.6) 413 (16.6) 55--64 years 83 (9.2) 223 (8.7) 65--74 years 52 (4.5) 150 (4.8) 75+ years 28 (2.1) 64 (1.6) Qualification No secondary 279 (32.6) 1183 (48.5) Secondary education 492 (67.4) 1198 (51.5) ELSI Severe hardship 23 (1.9) 111 (4.0) Significant hardship 31 (4.8) 144 (5.4) Some hardship 49 (5.6) 226 (8.5) Fairly comfortable 85 (8.4) 335 (14.7) Comfortable 183 (25.6) 539 (23.7) Good 283 (39.5) 783 (33.9) Very good 102 (14.2) 206 (9.8) Area deprivation (NZDep06) Decile 1 (least deprived) 53 (7.9) 42 (2.1) Decile 2 50 (6.8) 102 (4.8) Decile 3 63 (9.0) 86 (3.6) Decile 4 80 (10.7) 118 (6.0) Decile 5 64 (8.0) 144 (6.5) Decile 6 91 (10.1) 204 (8.5) Decile 7 96 (13.5) 251 (10.9) Decile 8 100 (13.3) 273 (12.2) Decile 9 71 (10.9) 438 (17.8) Decile 10 (most deprived) 103 (9.9) 731 (27.5) Individual deprivation (NZiDep) Level 1 (least deprived) 408 (54.3) 988 (43.0) Level 2 160 (22.1) 483 (21.5) Level 3 71 (8.6) 309 (12.6) Level 4 84 (10.2) 378 (15.2) Level 5 (most deprived) 48 (4.9) 226 (7.7) \Any socially-assigned Māori or other non-European ethnicity, either alone or in combination with socially-assigned European ethnicity (1848 Māori only, 288 Māori and European, 105 Māori and non-European, 148 other socially-assigned combinations). Univariate analyses of socially-assigned ethnicity and key variables of individual experience of racial discrimination and self-rated health showed that being socially-assigned as European-only tended to be associated with lower reporting of experience of racial discrimination and better self-rated health than being socially-assigned to any non-European group, including Māori (Table [3](#T3){ref-type="table"}). People socially-assigned as European-only tended to have lower mean psychological distress scores than those who were socially-assigned as non-European. ###### Distribution of key variables by socially-assigned ethnicity combinations among people who self-identify as Māori Characteristic* Socially-assigned ethnicity ------------------------------------------------------------ --------------------------------- -------------------------- Racial discrimination (reported individual experience) Verbal assault (ever) 147, 18.2% (14.5, 21.8) 578, 23.7% (21.5, 25.9) Physical assault (ever) 27, 3.4% (1.6, 5.2) 152, 6.1% (4.9, 7.3) Discrimination in health (ever) 29, 3.4% (1.8, 4.9) 147, 5.6% (4.5, 6.6) Discrimination in work (ever) 19, 1.9% (0.8, 3.1) 169, 6.6% (5.4, 7.8) Discrimination in housing (ever) 22, 2.2% (1.0, 3.3) 235, 8.1% (6.9, 9.4) No experiences 594, 78.2% (74, 82.4) 1583, 67.7% (65.3, 70.1) One experience 123, 16.3% (12.8, 19.8) 474, 20.0% (18.1, 22.0) Two or more 50, 5.5% (3.4, 7.6) 319, 12.2% (10.7, 13.8) Health outcome Self-rated health Excellent self-rated health 113, 12.9% (9.9, 16.0) 305, 13.9% (12.0, 15.8) Very good self-rated health 310, 44.4% (39.5, 49.3) 861, 35.4% (32.9, 37.8) Good self-rated health 265, 32.6% (27.9, 37.2) 857, 35.9% (33.5, 38.3) Fair self-rated health 63, 7.4% (5.2, 9.6) 295, 12.2% (10.6, 13.8) Poor self-rated health 20, 2.7% (1.2, 4.2) 71, 2.7% (1.9, 3.4) Kessler Psychological Distress Scale (K10) Mean (95% CI) 4.12 (3.66, 4.58) 4.77 (4.47, 5.06) Median (Interquartile range) 2.1 (0.2, 5.0) 2.2 (0.2, 6.2) Note: unweighted frequencies; weighted prevalences, means, medians and 95% CIs. \Any socially-assigned Māori or other non-European ethnicity, either alone or in combination with socially-assigned European ethnicity (1848 Māori only, 288 Māori and European, 105 Māori and non-European, 148 other socially-assigned combinations). Māori who were socially-assigned as being European-only were significantly less likely to report experience of racial discrimination (Table [4](#T4){ref-type="table"}; OR = 0.59, 95% CI = 0.45, 0.76). Sequential adjustment for sociodemographic and socioeconomic factors had little impact on this estimate (fully adjusted OR = 0.58, 95% CI = 0.44, 0.78). A significant interaction suggested that this association depended upon the respondent's self-identified ethnicity. For those who self-identified as Māori and European, the impact of being socially-assigned as European was protective of experiencing racial discrimination (OR for socially-assigned ethnicity = 0.49, 95% CI = 0.35, 0.68); for those who self-identified as Māori with no European ethnicity, the impact on experience of racial discrimination of being socially assigned as European was close to null (OR = 1.07, 95% CI = 0.6, 1.88). These odds ratios are from the unadjusted model; the interaction followed a similar pattern in the fully-adjusted model. ###### Odds ratios from logistic regression models for reported individual experience of racial discrimination (ever) Socially-assigned ethnicity, OR for experience of racism (95% CI)* ---------------------------------- ----------------------------------------------------------------------- ------------------- Unadjusted model Ref 0.59 (0.45, 0.76) Adjusted models (sequential) \+ age, sex Ref 0.59 (0.45, 0.77) \+ qualification Ref 0.58 (0.44, 0.75) \+ ELSI, NZiDep, NZDep06 Ref 0.58 (0.44, 0.78) \Includes reporting any socially-assigned non-European ethnicity (largely Māori) either alone or in combination with socially-assigned European ethnicity. + indicates adding new adjustment covariates to the preceding model. In multivariable analysis, Māori who were socially-assigned as European-only had a significant health advantage (reported excellent or very good health) compared to other Māori after adjusting for age and sex (OR = 1.39, 95% CI = 1.10, 1.74; Table [5](#T5){ref-type="table"}). This appeared to be due to the socioeconomic advantage and lower reporting of racial discrimination experiences associated with being socially-assigned as European-only (Adjusting for experience of racial discrimination immediately after adjusting for age and sex slightly attenuated this association, OR = 1.33, 95% CI = 1.05, 1.68 \[not shown in table\]). The difference between Māori socially-assigned as European-only and those assigned as any non-European group (including Māori) was attenuated after the sequential adjustment of socioeconomic measures (OR = 1.15, 95% CI 0.90, 1.46) and racial discrimination (OR = 1.11, 95% CI = 0.87, 1.42). ###### Odds ratios from logistic regression models for excellent/very good health (compared to good/fair/poor health) Socially-assigned ethnicity, OR (95% CI)* ---------------------------------------------- ---------------------------------------------- ------------------- Unadjusted model Ref 1.39 (1.10, 1.74) Adjusted for age, sex Ref 1.39 (1.10, 1.74) Adjusted for potential pathway variables \+ qualification Ref 1.28 (1.02, 1.61) \+ ELSI, NZiDep, NZDep06 Ref 1.15 (0.90, 1.46) \+ Racism (individual experience) Ref 1.11 (0.87, 1.42) \Includes reporting any socially-assigned non-European ethnicity (largely Māori) either alone or in combination with socially-assigned European ethnicity. + indicates adding new adjustment covariates to the preceding model. Analysis of difference in mean K10 scores also showed that Māori who were socially-assigned as any Māori or non-European ethnic group had significantly higher psychological distress scores in both the unadjusted and the age/sex adjusted models compared to Māori who were socially-assigned as European-only. This appeared to operate via socioeconomic advantage and lower experience of racial discrimination among Māori socially-assigned as European-only, with differences in mean K10 scores attenuated after adjusting for these variables (Table [6](#T6){ref-type="table"}). Adjusting for experience of racial discrimination immediately after adjustment for age and sex had a considerable attenuating effect on the estimate of K10 group difference (mean difference = 0.35, 95% CI = −0.89, 0.20 \[not shown in table\]). ###### Linear regression estimates for difference in mean K10 score Socially-assigned ethnicity* ---------------------------------------------- --------------------------------- ---------------------- Unadjusted model Ref -0.62 (-1.19, -0.05) Adjusted for age, sex Ref -0.66 (-1.22, -0.10) Adjusted for potential pathway variables \+ education Ref -0.52 (-1.08, 0.03) \+ ELSI, NZiDep, NZDep06 Ref -0.17 (-0.65, 0.32) \+ Racism (individual experience) Ref 0.04 (-0.44, 0.51) \*Includes reporting any socially-assigned non-European ethnicity (largely Māori) either alone or in combination with socially-assigned European ethnicity. + indicates adding new adjustment covariates to the preceding model. Discussion ========== Our study builds on literature examining socially-assigned race/ethnicity and health, and suggests that ethnic appearance is an important determinant of health, for Māori at least. We found that among the self-identified Māori population, Māori who reported being socially-assigned as European-only had a health advantage compared with those who were socially-assigned as Māori and/or any other non-European group. Māori who were socially-assigned as European-only were significantly more likely to report their health as excellent or very good and have lower levels of psychological distress than other Māori. This relationship was attenuated after adjusting for socioeconomic measures and individual experience of racial discrimination (and no longer statistically significant after adjustment). The relationship between socially-assigned ethnicity and health was independent of self-identified ethnic group combinations. Our findings are consistent with those of Jones et al. who analysed data from a large United States study of participants from eight states using the Reactions to Race module of the 2004 BRFSS \[[@B18]\]. Jones et al. found that being classified by others as 'White' was associated with health advantage even among participants who did not self-identify as 'White' and included 'Hispanic', 'American Indian' and 'people who identified with more than one race' \[[@B18]\]. The authors conclude that this reflects the effects of racism on health in the United States, not only in terms of disadvantage, but also the advantages of 'whiteness' in a race conscious society \[[@B18]\]. In contrast to Jones' findings, analysis of the Michigan/Wisconsin BRFSS data found that there was not a health advantage associated with being socially-assigned to the White group for (self-identified) non-White groups, and that self-assessment of race/ethnicity predicted health status better than socially-assigned race/ethnicity \[[@B23]\]. The authors suggest that because of the systemic and societal nature of racism, and the segregated histories of the particular communities they investigated, "...there may be little opportunity for White privilege to aggregate to the community level, and thus the potential privilege from being socially-assigned as White may be partially or wholly un-realized". Our study expands on the analyses of Jones' et al. \[[@B18]\] and Ridings et al. \[[@B23]\] to also examine the association between socially-assigned ethnicity and reported individual experience of racism, with assignment to the dominant European ethnic group significantly related to reduced exposure to individual level racial discrimination, as well as exploring an additional health measure of psychological distress. It also provides evidence in a different social context and for an indigenous population. Other studies have also examined health and social differentials within the Māori population and shown that health and socioeconomic differences exist for different Māori populations based on their self-identified ethnicity. For example, people who identify solely as Māori have been shown to have more disadvantaged socioeconomic status and worse health than people who identify as Māori and European \[[@B35]-[@B38]\]. Various hypotheses have been posited to explain differences in the ways Māori self-identify their ethnicity or ethnicities that may also be linked to more or less disadvantage. These include 'cultural' affiliation or strength of identity reasons \[[@B35]\] as well as skin colour and appearance \[[@B39]\]. Kukutai \[[@B35]\] notes that "orientation towards the European mainstream confers benefits in terms of better outcomes" (p 100). Our study suggests that these types of analyses may have been confounded by socially-assigned ethnicity. The relationship between socially-assigned ethnicity and health was unchanged when Māori self-identified categories were added to the models (people self-identifying as Māori alone or in combination with another minority ethnic group compared with people identifying as Māori and European). Māori self-identified categories were not associated with health when adjusted for socially-assigned ethnicity. The relationship between socially-assigned ethnicity and health appears to operate via more advantaged socioeconomic position and reduced exposure to individual experience of racism. It is acknowledged that racism plays a fundamental role in structuring social and economic ethnic disparities \[[@B8]\]. The relationship between socially-assigned ethnicity, socioeconomic position and individual experience of racism in our study, while not definitive, is consistent with racism operating at both the institutional and individual levels. In our analysis, socioeconomic status and individual experience of racial discrimination are both viewed as markers of racism and potential pathway indicators by which racial discrimination based on socially-assigned ethnicity may be operating to affect health. This is the first time socially-assigned ethnicity has been analysed from the New Zealand Health Survey and provides further insight into intra-group health differences within the Māori population. It expands on the few international studies in the area to directly demonstrate the link between socially-assigned ethnicity and individual experience of racism and examines an additional health measure in psychological distress. A major study strength is the ability to provide nationally representative information for Māori. In addition, this study examines health advantage, rather than disadvantage, and how privilege may be afforded to people based on their appearance as European, even when they self-identify as a different ethnicity. The measure of socially-assigned ethnicity may reflect additional aspects of racial discrimination that are not fully captured by self-report of individual experiences \[[@B18]\], which can be limited by peoples' willingness and ability to report them \[[@B6]\]. In our study, adjusting for individual experience of racial discrimination attenuated the differences in health between the socially-assigned groups. While this was to a limited degree, the questions on individual experience of racism were only asked in a very specific number of situations and may well underestimate people's actual exposure \[[@B5]\]. There are also some important limitations to consider in the interpretation of our study findings. Our study is restricted to the Māori population, and so cannot be generalized to consider the effect of socially-assigned ethnicity on other groups and on ethnic inequalities between groups, although further work is underway to examine this, including how socially-assigned ethnicity impacts on the health of people who self-identify as European or other ethnic groups. The measures are self-reported with recognised limitations of such measures in health and racism literature, including the role of social desirability that has the potential to impact on reporting of ethnicity, discrimination and health measures in this study \[[@B6]\]. The New Zealand Health Survey is a cross-sectional survey, and so usual caveats on attributing causality apply, particularly temporality. For example, it is possible that experience of racial discrimination may enforce the feeling of being classified as non-European: thus self-reported socially-assigned ethnicity could potentially be influenced by experience of racial discrimination. There are some other potential issues with the measure of socially-assigned ethnicity. While it is intended to capture how other people view a person's ethnicity, it is self-reported socially-assigned ethnicity, and not observer reported \[[@B40]\]. Unlike the United States, where multiple ethnicities are not commonly reported, multiple reporting of self-identified ethnic groups in New Zealand is not uncommon. While the wording of the socially-assigned question implies a single response, "How do other people usually classify you?", multiple responses were allowed and were fairly common, although not to the same extent as the self-identified question. This complicates the analysis and also the interpretation of multiple socially-assigned ethnic groups as it is unclear whether a person is sometimes seen as one ethnicity and sometimes another, or whether they may be viewed as both ethnic groups simultaneously. In addition, the context a person is in may influence their responses to this question e.g. they may be viewed as different ethnic groups in different contexts \[[@B40]-[@B42]\]. As the question is trying to capture the ethnic group that other people recognize and react to in wider society, a more context specific question with single reporting options may be better. For example, in Australia, a similar question for use in the indigenous population asks, "Do people you meet for the first time know that you are Indigenous?" capturing both context and a single response \[[@B43]\]. While this study cannot definitely say that socially-assigned ethnicity reflects differential experiences of racism on health, it is consistent with this theory. Our study does not examine the specific mechanisms through which socially assigned ethnicity impacts on health for Māori and there are likely to be a number of factors through which this may occur including associations with possible individual risk and protective factors and interactions and experiences with health services. Finally, the use of excellent or very good health to ascertain health advantage is not the usual cut point for analysis of the general self-rated health question. This cut point was used to examine optimal health both from a theoretical perspective in order to explore health advantage rather than disadvantage and also for comparability with other studies in the field of socially-assigned ethnicity and health \[[@B18],[@B23]\]. Jones et al. \[[@B18]\] notes the importance of measuring both socially-assigned ethnicity and self-identified ethnicity in order to better understand how race and ethnicity influence health. In New Zealand for example, self-identified ethnicity remains a key measure of ethnic inequalities, with socially-assigned ethnicity a potentially stronger marker of risk for exposure to racism \[[@B44]\] rather than a measure of how an individual or group necessarily views or expresses their ethnic identity. For Māori and other indigenous peoples, socially-assigned ethnicity is unlikely to fully capture the Māori population who have particular indigenous or treaty rights, including the right to self-determination of their identity and the right to monitor Crown action and inaction in relation to Māori health and other outcomes \[[@B1],[@B17],[@B35]\]. Socially-assigned ethnicity does, however, provide insight into the determinants of health for Māori and potentially for other ethnic groups in New Zealand and more broadly. Conclusions =========== Results of this study suggest that, in a race conscious society, the way people's ethnicities are viewed by others appears to have tangible health risk or advantage, and this is consistent with an understanding of racism as a health determinant. Dismantling the structures of racism is complex yet vital in our efforts to achieve a fair society that facilitates equitable outcomes in health and other social indicators and also enables self-determination of priorities and solutions for Māori. Abbreviations ============= CI: Confidence interval; ELSI: Economic living standards index short form; Ethnic groups (M: Māori; P: Pacific; E: European; O: other ethnicity or ethnic combinations including Asian); K10: Kessler 10-item scale; NZDep06: New Zealand Index of Deprivation 2006; NZiDep: New Zealand Index of Socioeconomic Deprivation for Individuals; OR: Odds ratio; Ref: reference group. Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== RH and DC conceived the study. All authors participated in the study design and interpretation of findings. JS undertook the data analysis. RH led the initial drafting of the manuscript. DC and JS contributed to the writing of subsequent versions. All authors have read and approved the final version of the manuscript. The results presented in this paper are the work of the authors. Authors' information ==================== RH (MB ChB, MPH, FNZCPHM), senior research fellow. DC (PhD), senior research fellow. JS (PhD), senior research fellow and biostatistician. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2458/13/844/prepub> Acknowledgements ================ The Crown is the owner of the copyright of the data and the Ministry of Health is the funder of the data collection. Health and Disability Intelligence (NZ Ministry of Health) provided access to the data. The Health Research Council of New Zealand provided funding for RH, DC and JS to undertake this study. We would like to thank the participants of the New Zealand Health Survey. Thanks to other team members, Martin Tobias (who provided advice and peer review), Ruruhira Rameka (who provided administrative support) and Joanna Minster (who provided initial research assistance). Thanks also to Bridget Robson, Natalie Talamaivao, Roimata Timutimu and Li-Chia Yeh for their advice.	{ "pile_set_name": "PubMed Central" }
Published: August 11, 2016 Introduction {#sec1} ============ Invasive lobular carcinoma (ILC) is the second most common subtype of breast cancer, comprising 8%--14% of all breast cancer cases ([@bib17], [@bib27]). ILC is histologically characterized by single-cell files of tumor cells in a fibrous stroma. Due to its invasive growth behavior, this tumor type does not immediately form a defined tumor mass and, therefore, is difficult to detect by physical examination or mammography ([@bib14]). Despite being classified as a histologic low-grade tumor type with a low-to-intermediate mitotic index, the infiltrating nature of ILC complicates surgical removal. Moreover, ILCs typically respond poorly to classical regimes of chemotherapy. To improve treatment options, more insight into ILC development is required. On the molecular level, ILC is characterized by loss of the cell-adhesion protein E-cadherin ([@bib23]), an event that has been shown to drive tumor invasion. Nevertheless, somatic inactivation of E-cadherin in mouse mammary epithelium does not induce spontaneous mammary tumor development ([@bib3], [@bib8]), indicating that additional (epi)genetic lesions are required for ILC formation. Activation of the phosphatidylinositol 3-kinase (PI3K) pathway, as displayed by either activating PIK3CA mutations or homozygous deletions or inactivating mutations in PTEN, is frequently observed in ILC ([@bib4], [@bib6], [@bib21], [@bib22]). PI3K signaling controls multiple cellular processes, such as cell proliferation, survival, and migration, by the conversion of phosphatidylinositol 4,5-bisphosphate (PIP~2~) to phosphatidylinositol (3,4,5)-trisphosphate (PIP~3~), driving activation of downstream targets such as AKT. To test whether constitutive activation of PI3K signaling can, indeed, drive the formation of ILC, we generated mice with mammary-specific Cre-conditional inactivation of E-cadherin and PTEN. We found that combined loss of both E-cadherin and PTEN in mouse mammary epithelial cells induces mouse mammary tumors that closely resemble classical invasive lobular carcinoma (CLC), the main subtype of human ILC. Results {#sec2} ======= Inactivation of E-Cadherin and PTEN In Vitro Alters Growth and Invasion of Mammary Epithelium {#sec2.1} --------------------------------------------------------------------------------------------- To study the direct consequence of inactivation of E-cadherin (encoded by the Cdh1 gene) and PTEN in primary mammary epithelial cells, we generated mammary organoids from Cdh1^F/F^;Pten^F/F^ mice ([@bib8], [@bib20]). In vitro transduction of organoids using a Cre-recombinase-encoding adenovirus (AdCre) allowed us to test the direct effects of E-cadherin and PTEN loss in mammary epithelium. To confirm efficient AdCre-mediated recombination in vitro, we used organoids from mTmG reporter mice that express tandem dimer Tomato (tdT) prior to recombination and GFP after Cre-mediated excision of tdT ([@bib24]). AdCre transduction of mTmG organoids induced loss of tdT and expression of GFP in nearly all cells ([Figure S1](#mmc1){ref-type="supplementary-material"}A). Importantly, the phenotype of AdCre-transduced mTmG organoids appeared unaltered as compared to non-transduced organoids from the same mouse, indicating that organoid growth rate and phenotype are not affected by AdCre transduction ([Figure S1](#mmc1){ref-type="supplementary-material"}A). This strategy was, therefore, applied to inactivate E-cadherin and/or PTEN in Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ organoids, respectively. All non-transduced organoids displayed similar-sized cyst-like structures after 7 days of culture ([Figure S1](#mmc1){ref-type="supplementary-material"}B). The inactivation of E-cadherin altered mammary organoid phenotype by inducing discohesive cellular structures surrounding the organoid cores, while the inactivation of PTEN resulted in enlarged smooth-edged organoids ([Figure S1](#mmc1){ref-type="supplementary-material"}B). Concurrent inactivation of E-cadherin and PTEN revealed a combined phenotype of Cdh1^Δ/Δ^ and Pten^Δ/Δ^ organoids, as Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids were both enlarged and showed discohesive cellular structures ([Figure S1](#mmc1){ref-type="supplementary-material"}B). Efficient recombination was confirmed by western blotting and PCR, showing strong reduction of E-cadherin and/or PTEN protein expression and deletion of the Cdh1^F^ and Pten^F^ alleles, respectively ([Figures 1](#fig1){ref-type="fig"}A and [S1](#mmc1){ref-type="supplementary-material"}C). In addition, AdCre-transduced Pten^F/F^ and Cdh1^F/F^;Pten^F/F^ organoids showed a strong induction of AKT phosphorylation, indicating activation of PI3K signaling due to PTEN loss ([Figure 1](#fig1){ref-type="fig"}A). To study the long-term consequences of E-cadherin and/or PTEN loss in mammary epithelial cells, we monitored organoid development over a period of 21 days. Non-transduced organoids from each genotype decreased similarly in size and were hardly detectable after 21 days of culture ([Figure 1](#fig1){ref-type="fig"}B). The size of Cdh1^Δ/Δ^ organoids remained stable, while the size of Pten^Δ/Δ^ and Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids continued to increase ([Figures 1](#fig1){ref-type="fig"}B and 1C). Again, the margins of both Cdh1^Δ/Δ^ and Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids showed discohesive cellular structures ([Figure 1](#fig1){ref-type="fig"}B). To examine why growth kinetics differed between Cdh1^Δ/Δ^ and Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids, we analyzed cell proliferation and apoptosis by immunofluorescence (IF) with antibodies against Ki-67 and cleaved caspase-3 (CC3), respectively. Cdh1^Δ/Δ^ organoids showed a balanced number of cells in proliferation or apoptosis, which might explain their sustained size over time ([Figures 1](#fig1){ref-type="fig"}D and 1E). Consistent with a survival role for the PI3K signaling pathway, both Pten^Δ/Δ^ and Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids showed almost no apoptotic cells ([Figure 1](#fig1){ref-type="fig"}E), resulting in an increase in organoid size over time ([Figures 1](#fig1){ref-type="fig"}B and 1C). In contrast to Pten^Δ/Δ^ organoids, Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids displayed an invasive response upon renewal of the cell culture medium ([Figures 1](#fig1){ref-type="fig"}F and [S1](#mmc1){ref-type="supplementary-material"}D). These invasive cellular strands displayed single files of cells, indicating an increased invasive potential of Cdh1^Δ/Δ^;Pten^Δ/Δ^ mammary epithelium upon stimulation by growth factors and/or chemoattractants in the culture medium. Together, these observations suggest that PI3K pathway activation through the disruption of PTEN can rescue apoptosis induced by E-cadherin loss in mammary epithelial organoids, resulting in invasive and proliferative behavior in vitro. Inactivation of E-Cadherin and PTEN in Mammary Epithelium In Vivo Induces Rapid Formation of Hyperplastic and Infiltrative Lesions {#sec2.2} ---------------------------------------------------------------------------------------------------------------------------------- To study the consequences of E-cadherin and PTEN loss in vivo, we generated Wcre;Cdh1^F/F^;Pten^F/F^ female mice, in which mammary-epithelium-specific inactivation of E-cadherin and PTEN is induced by whey acidic protein (WAP) gene promoter-driven Cre recombinase (Wcre) ([@bib8]). We also generated Wcre;Cdh1^F/F^ and Wcre;Pten^F/F^ female mice to study the effects of E-cadherin or PTEN loss alone. Since Wcre activity is first detectable in the fourth week of mammary gland development (data not shown), we analyzed the consequences of somatic inactivation of E-cadherin and/or PTEN during mammary gland development. Histological analysis and immunohistochemistry (IHC) for cytokeratin 8 (CK8; marking luminal mammary epithelial cells) indicated inactivation of E-cadherin to cause hyperplastic and dysplastic ducts, as well as a marked accumulation of dissociated mammary epithelial cells in the lumen of mammary epithelium from 6-week-old Wcre;Cdh1^F/F^ mice ([Figure 2](#fig2){ref-type="fig"}A). Inactivation of PTEN alone did not reveal distinct alterations in mammary epithelial composition in Wcre;Pten^F/F^ ([Figure 2](#fig2){ref-type="fig"}A). In contrast, mammary epithelium from 6-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice displayed multifocal neoplastic lesions infiltrating the surrounding stroma ([Figure 2](#fig2){ref-type="fig"}A). Analysis of over 20 nulliparous Wcre;Cdh1^F/F^;Pten^F/F^ mice confirmed a robust synergy between loss of E-cadherin and PTEN in driving formation of these mammary epithelial lesions, as all mice displayed neoplastic lesions at 6 weeks of age. The infiltration pattern of these neoplastic lesions showed strong resemblance with the characteristic trabecular and discohesive growth pattern of tumor cells in human CLC. We confirmed Wcre-mediated inactivation of E-cadherin and/or PTEN in luminal epithelial cells by IF ([Figures 2](#fig2){ref-type="fig"}B and [S2](#mmc1){ref-type="supplementary-material"}A). Analysis of Wcre;Cdh1^F/F^ glands showed that E-cadherin-deficient cells in the lumen are CK8 positive, indicating that these cells indeed dissociated from the luminal epithelium ([Figure S2](#mmc1){ref-type="supplementary-material"}A). This phenomenon is likely caused by reduced junctional integrity following E-cadherin loss ([@bib30]). Even though Wcre;Pten^F/F^ glands did not display neoplastic lesions within 6 weeks, we clearly observed stochastic loss of PTEN in luminal epithelial cells ([Figure S2](#mmc1){ref-type="supplementary-material"}A). The neoplastic lesions observed in Wcre;Cdh1^F/F^;Pten^F/F^ mammary glands displayed discohesive CK8-marked luminal epithelial cells that, indeed, lost E-cadherin and PTEN, whereas surrounding stromal cells retained PTEN expression ([Figure 2](#fig2){ref-type="fig"}B). Moreover, E-cadherin-deficient cells still expressed CK8, indicating that these cells did not undergo epithelial-to-mesenchymal transition (EMT), as also previously observed ([@bib8]). Since inactivation of PTEN rescued apoptosis and promoted survival of E-cadherin-deficient mammary epithelial cells in Cdh1^Δ/Δ^;Pten^Δ/Δ^ organoids, we analyzed apoptosis by CC3 IF staining in mammary gland sections from 6-week-old Wcre;Cdh1^F/F^, Wcre;Pten^F/F^, and Wcre;Cdh1^F/F^;Pten^F/F^ mice. While no apoptosis was observed in Wcre;Pten^F/F^ and Wcre;Cdh1^F/F^;Pten^F/F^ mammary glands, CK8-positive cells in the lumen of mammary ducts in Wcre;Cdh1^F/F^ mice displayed CC3-marked apoptosis ([Figures 2](#fig2){ref-type="fig"}C and 2D). Compared to Cdh1^Δ/Δ^ organoids, Wcre;Cdh1^F/F^ mammary glands showed only limited numbers of CC3-marked cells, most likely due to rapid clearance of apoptotic cells in the lumen of mammary ducts. These data underscore the notion that the loss of PTEN promotes survival of E-cadherin-deficient mammary epithelial cells through inhibition of apoptosis. To further investigate the effects of concomitant inactivation of E-cadherin and PTEN on mammary gland development and functionality, we analyzed mammary gland outgrowth during puberty and mammary gland maturation during pregnancy in nine Wcre;Cdh1^F/F^;Pten^F/F^ mice. Whole-mount analysis of mammary glands from 8-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice showed, beside neoplastic lesions and hyperplastic ducts, a delay in mammary gland development, as compared to age-matched wild-type mammary glands ([Figure S2](#mmc1){ref-type="supplementary-material"}B). Impaired mammary gland maturation was also observed during pregnancy, as pups from five 12-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice grew poorly and died 6 days after birth, while littermates fostered onto wild-type dams showed normal weight and survival ([Figure S2](#mmc1){ref-type="supplementary-material"}C). This phenomenon is likely caused by the reduced numbers of milk-producing glands in Wcre;Cdh1^F/F^;Pten^F/F^ mice, as compared to wild-type foster dams ([Figure S2](#mmc1){ref-type="supplementary-material"}D). Thus, both the development and maturation of Cdh1^Δ/Δ^;Pten^Δ/Δ^ mammary epithelium is strongly impaired, which is likely caused by the loss of mammary epithelial integrity and formation of multi-focal neoplasia. Inactivation of E-Cadherin and PTEN Drives Formation of Mammary Tumors that Closely Resemble Human CLC {#sec2.3} ------------------------------------------------------------------------------------------------------ To examine potential synergy between E-cadherin and PTEN loss in mammary tumorigenesis, we monitored tumor formation, as detected by palpation, in Wcre;Pten^F/+^, Wcre;Pten^F/F^, Wcre;Cdh1^F/F^;Pten^F/+^, and Wcre;Cdh1^F/F^;Pten^F/F^ mice. In line with previous literature, female Wcre;Pten^F/+^ and Wcre;Pten^F/F^ mice developed heterozygous or homozygous mammary tumors, with median latencies of 615 and 250 days, respectively ([Figure 3](#fig3){ref-type="fig"}A) ([@bib16]). Additional inactivation of E-cadherin significantly decreased mammary tumor latency to 109 days of age in Wcre;Cdh1^F/F^;Pten^F/F^ mice, demonstrating strong synergy between E-cadherin and PTEN loss in mammary tumorigenesis ([Figure 3](#fig3){ref-type="fig"}A). Interestingly, Wcre;Cdh1^F/F^;Pten^F/+^ and Wcre;Pten^F/+^ female mice showed a similar tumor-free survival, indicating that complete loss of PTEN is required for malignant transformation of E-cadherin-deficient mammary epithelial cells ([Figure 3](#fig3){ref-type="fig"}A). To analyze the mammary tumors for each genotype, mice were sacrificed when tumors reached 225 mm^2^, except for Wcre;Cdh1^F/F^;Pten^F/F^ mice, which were sacrificed after a median period of 101 days after tumor detection due to cumulative burden of multifocal tumors, abnormal locomotory behavior, or skin ulceration (both associated with infiltrating tumor cells in surrounding tissues) ([Figure S3](#mmc1){ref-type="supplementary-material"}A). Mammary tumors from Wcre; Pten^F/+^mice displayed diverse histological phenotypes, including solid carcinomas, squamous metaplastic carcinomas, and EMT-like tumors, while Wcre; Pten^F/F^ mice solely developed squamous metaplastic carcinomas ([Figures 3](#fig3){ref-type="fig"}B and [S3](#mmc1){ref-type="supplementary-material"}A), suggesting that somatic PTEN loss, on its own, preferentially drives the development of this particular tumor type. Notably, additional inactivation of E-cadherin completely shifted mammary tumor morphology. About half (57%) of mammary tumors in Wcre;Cdh1^F/F^;Pten^F/+^ mice and nearly all (96%) tumors in Wcre;Cdh1^F/F^;Pten^F/F^ mice displayed infiltrative patterns of single files of tumor cells into the surrounding stroma ([Figures 3](#fig3){ref-type="fig"}B and 3C). IHC staining of tumor sections confirmed the absence of E-cadherin and PTEN protein in CK8-marked tumor cells, whereas surrounding stromal cells retained PTEN expression ([Figure 3](#fig3){ref-type="fig"}C). Moreover, IHC staining revealed elevated levels of phosphorylated (p)-AKT, p-4ebp1, and p-S6 in tumor cells, confirming the activation of PI3K signaling due to loss of PTEN ([Figure S3](#mmc1){ref-type="supplementary-material"}B). Given that this phenotype comprises key characteristics of human CLC, we designated these tumors as mouse CLC (mCLC). Interestingly, mCLCs from Wcre;Cdh1^F/F^;Pten^F/+^ mice completely lost PTEN protein expression as analyzed by IHC, indicating that ILC formation requires complete loss of PTEN expression ([Figure S3](#mmc1){ref-type="supplementary-material"}E). The remaining mammary tumors from Wcre;Cdh1^F/F^;Pten^F/+^ mice were classified as solid ILC, squamous metaplastic carcinomas, or EMT-like tumors that possibly developed due to incomplete loss of PTEN expression ([Figure 3](#fig3){ref-type="fig"}B). Since the majority of human CLCs are estrogen receptor (ER) positive, we analyzed ER expression in mCLCs from Wcre;Cdh1^F/F^;Pten^F/F^ mice, revealing a mean of 26% ER-positive tumor cells ([Figures 3](#fig3){ref-type="fig"}C, 3D, and [S3](#mmc1){ref-type="supplementary-material"}C). In contrast, a mean of only 0.1% ER-positive cells was observed in Brca1^Δ/Δ^;Trp53^Δ/Δ^ mouse mammary tumors, which are known to be ER negative ([@bib18]). Also, the expression of Esr1 mRNA, encoding ERα, was found to be increased in Cdh1^Δ/Δ^;Pten^Δ/Δ^ mCLCs, as compared to Brca1^Δ/Δ^;Trp53^Δ/Δ^ mouse mammary tumors ([Figure S3](#mmc1){ref-type="supplementary-material"}F). Expression of Esr1 mRNA was also observed in FACS (fluorescence-activated cell sorting)-sorted EpCAM (epithelial cell adhesion molecule)-positive tumor cells, confirming the ER positivity of tumor cells in mCLC ([Figures S3](#mmc1){ref-type="supplementary-material"}C and S3F). Based on their gene expression profiles, human CLCs are classified as luminal-A tumors ([@bib6], [@bib15]). To assess the molecular subtype of mCLCs from Wcre;Cdh1^F/F^;Pten^F/F^ mice, hierarchical clustering of RNA-seq data from 956 human breast cancers and 42 mouse mammary tumors was performed using the PAM50 gene set, which allows classification of breast tumors into luminal, basal-like, HER2-positive and normal-like subgroups ([@bib26]). Whereas the 22 Brca1^Δ/Δ^;Trp53^Δ/Δ^ mammary tumors clustered together with human basal-like breast cancers, all 20 Cdh1^Δ/Δ^;Pten^Δ/Δ^ mCLCs were classified as luminal-A tumors, indicating that these mouse tumors closely resemble the molecular subtype of human CLC ([Figure 3](#fig3){ref-type="fig"}E). In summary, the combined loss of E-cadherin and PTEN drives a rapid onset of mCLCs that recapitulate the morphology, ER positivity, and molecular subtype of human CLC. Mammary Gland Reconstitution Exposes Slow and Linear Growth of De Novo mCLCs {#sec2.4} ---------------------------------------------------------------------------- Human CLCs are typically slow-growing tumors with a low-to-intermediate mitotic index. Unfortunately, mCLC growth kinetics could not be measured in Wcre;Cdh1^F/F^;Pten^F/F^ mice, which display multifocal tumor formation in all mammary glands. To assess the growth kinetics of mCLCs, we performed mammary gland reconstitutions by orthotopic transplantation of small fragments of tumor-free mammary gland tissue from Wcre;Cdh1^F/F^;Pten^F/F^ mice into cleared mammary fat pads of wild-type mice. In 82 out of 104 reconstituted mammary glands, a palpable tumor was detected after a median latency period of 102 days, which is similar to the median latency of 109 days in the spontaneous mouse model ([Figure 4](#fig4){ref-type="fig"}A). Since tumor development after mammary gland reconstitution occurred at a single location, it allowed reliable assessment of tumor size. Up to 250 days after tumor detection, which is more than twice the age the spontaneous mouse model reaches, a linear and slow growth rate was observed for all 82 tumors to an average size of only 8.6 mm^2^ ([Figures 4](#fig4){ref-type="fig"}B, 4C, and [S4](#mmc1){ref-type="supplementary-material"}A). These linear growth kinetics are unique for mouse tumors, which generally exhibit exponential tumor growth, and are very comparable to the slow growth kinetics of human CLC. The fact that mCLCs display constant and linear growth kinetics from the time of detection onward suggests that no additional mutations are required for CLC formation besides loss of E-cadherin and PTEN. Interestingly, 17 out of 82 mCLCs showed a sudden transition from linear to exponential growth kinetics between 8 and 21 months after tumor detection ([Figure 4](#fig4){ref-type="fig"}D). These exponentially growing tumors displayed non-CLC phenotypes, including solid ILC and EMT-like tumors, which were also observed in a small fraction of the spontaneous tumors from Wcre;Cdh1^F/F^;Pten^F/F^ mice ([Figures 3](#fig3){ref-type="fig"}B and [S3](#mmc1){ref-type="supplementary-material"}B). The sudden and stochastic transition from linear to exponential growth kinetics suggests the acquisition of additional mutations in these tumors over time. Indeed, whole-exome sequencing of seven randomly selected exponentially growing tumors and 14 mCLCs revealed that all exponentially growing tumors harbored mutations in at least one cancer-related gene, while none of the mCLCs showed additional mutations ([Figure 4](#fig4){ref-type="fig"}E). Similarly, DNA copy-number variation (CNV) analysis showed that mCLCs displayed relatively stable genomic profiles, whereas exponentially growing tumors contained multiple DNA copy-number aberrations ([Figures 4](#fig4){ref-type="fig"}F and [S4](#mmc1){ref-type="supplementary-material"}B). In summary, we successfully developed a mammary gland reconstitution model for CLC, highlighting a linear growth rate of these tumors and indicating that combined E-cadherin and PTEN loss is sufficient and, thus, causal for CLC development. Metastatic Spectrum of mCLCs Mimics that of Human CLC {#sec2.5} ----------------------------------------------------- Human CLC is characterized by its invasive and metastatic behavior. Similarly, mCLCs displayed tumor invasion, as detected by the examination of tumor sections, revealing irregular tumor edges with files of tumor cells invading surrounding tissues such as fat, muscle, and nerves ([Figures 5](#fig5){ref-type="fig"}A and 5B). In line with previous observations ([@bib5]), we observed a mixture of CK8 and cytokeratin-14 (CK14) double-positive tumor cells and CK14 single-positive tumor cells at the invasive front ([Figure S5](#mmc1){ref-type="supplementary-material"}A). We also investigated metastatic behavior for each tumor genotype. In Wcre;Pten^F/+^ and Wcre;Pten^F/F^ mice, no microscopic metastases could be detected in any organ, possibly due to the presence of intact adherens junctions ([Figure 5](#fig5){ref-type="fig"}C). In contrast, microscopic or macroscopic metastases were observed in 34% and 5% of tumor-bearing Wcre;Cdh1^F/F^;Pten^F/+^ and Wcre;Cdh1^F/F^;Pten^F/F^ mice, respectively ([Figure 5](#fig5){ref-type="fig"}C). These metastases were detected at various distant sites that are also typical for human CLC, including lymph nodes, lungs, pleura, peritoneum, and abdominal viscera ([Figures 5](#fig5){ref-type="fig"}C and 5D). The distribution of the primary tumor phenotypes in mice that developed metastases was similar to the overall distribution, with the majority of primary tumors displaying an mCLC phenotype ([Figures 3](#fig3){ref-type="fig"}B and [S5](#mmc1){ref-type="supplementary-material"}B). A plausible explanation for the lower number of metastases in Wcre;Cdh1^F/F^;Pten^F/F^ mice lies in the fact that these mice were sacrificed at a relatively young age (median, 210 days), as compared to Wcre;Cdh1^F/F^;Pten^F/+^ mice, which were sacrificed at a median age of 634 days ([Figure S5](#mmc1){ref-type="supplementary-material"}C). Therefore, Wcre;Cdh1^F/F^;Pten^F/F^ mice might have been too young at the time of sacrifice to permit detection of (slow-growing) metastases. This notion is supported by the fact that the Wcre;Cdh1^F/F^;Pten^F/F^ mice with detectable metastases were the oldest of all mice analyzed ([Figure S5](#mmc1){ref-type="supplementary-material"}D). However, no metastatic lesions were observed in lungs from ten aged mice that had been reconstituted with mammary gland tissue from Wcre;Cdh1^F/F^;Pten^F/F^ mice as described earlier ([Figure 4](#fig4){ref-type="fig"}B). Therefore, it cannot be excluded that the late onset of tumor formation in Wcre;Cdh1^F/F^;Pten^F/+^ mice might favor the development of distinct ILCs with increased metastatic capacity. Next, we used FACS analysis to search for single mCLC cells in the circulation or in common sites of human CLC metastasis, such as bone marrow or lungs. To facilitate the identification of single tumor cells, we used Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice in which GFP expression marks Wcre-mediated loss of E-cadherin and PTEN ([@bib24]). IHC staining of mammary gland sections indeed confirmed the formation of GFP-positive mCLCs in Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice ([Figure S5](#mmc1){ref-type="supplementary-material"}E). FACS analysis of primary tumor tissue from five Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice revealed 10%--15% GFP-positive cells, implying that the majority of cells in mCLC are non-tumor cells ([Figure 5](#fig5){ref-type="fig"}E). As expected, FACS analysis of blood samples revealed low numbers of circulating tumor cells (CTCs) in all Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice analyzed, suggesting that tumor cells are able to shed from the primary tumor and intravasate into the blood circulation ([Figures 5](#fig5){ref-type="fig"}E and 5F). Furthermore, low numbers of GFP-positive cells were detected in bone marrow and lung tissue, indicating infiltration of disseminated tumor cells into these organs ([Figures 5](#fig5){ref-type="fig"}E and 5F). The low number of GFP-marked tumor cells might underestimate the actual number of disseminated tumor cells, as the presence of Tomato and GFP double-positive CTCs in the blood revealed incomplete switching of the homozygous mTmG reporter alleles. Nevertheless, these results show that mCLCs exhibit an invasive and metastatic behavior that strongly resembles that of human CLC. Early Stromal Recruitment in CLC {#sec2.6} -------------------------------- Histological and FACS analysis indicated mCLCs to contain a large proportion of non-tumor cells ([Figures 3](#fig3){ref-type="fig"}C and [5](#fig5){ref-type="fig"}E). Therefore, we performed a detailed histological characterization of the tumor microenvironment of mCLCs and compared it to that of human CLCs. Besides H&E staining, we included Masson's trichrome staining to visualize collagen fibers, a major component of the extracellular matrix, and IHC stainings to detect fibroblasts (PDGFRβ), T cells as a major component of the adaptive immune system (CD3), and macrophages as a major component of the innate immune system (F4/80 or CD68). We analyzed both early lesions and more advanced tumors in 6-week-old and 25-week-old wild-type, Wcre;Cdh1^F/F^, Wcre;Pten^F/F^, and Wcre;Cdh1^F/F^;Pten^F/F^ mice, respectively ([Figure 6](#fig6){ref-type="fig"}A). In 6-week-old wild-type and Wcre;Pten^F/F^ mice ([Figures 6](#fig6){ref-type="fig"}A, [S6](#mmc1){ref-type="supplementary-material"}A, and S6B) and 25-week-old wild-type mice (data not shown), mammary ducts were surrounded by a thin layer of blue-colored collagen fibers and PDGFRβ-positive fibroblasts, containing few CD3-positive T cells and F4/80-positive macrophages. In contrast, mammary ducts of both Wcre;Cdh1^F/F^ and Wcre;Cdh1^F/F^;Pten^F/F^ mice clearly showed an altered microenvironment. In Wcre;Cdh1^F/F^ mammary glands, a relatively normal epithelial layer was surrounded by an thickened layer of collagen fibers and fibroblasts, showing slightly increased numbers of macrophages, but no notable changes in T cell numbers ([Figure 6](#fig6){ref-type="fig"}A). Even more profound changes in the microenvironment of mammary ducts were observed in Wcre;Cdh1^F/F^;Pten^F/F^ mice. Early lesions of 6-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice showed a clear increase in collagen and numbers of fibroblasts and displayed increased numbers of infiltrating macrophages and T cells, especially at the periphery ([Figure 6](#fig6){ref-type="fig"}A). Similar phenotypes were observed in more advanced tumors of 25-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice, indicating that stromal recruitment is a continuous process with preservation of the composition ([Figure 6](#fig6){ref-type="fig"}A). To investigate whether the stromal composition of mCLC resembles that of human CLC, we analyzed breast tissue samples from five healthy individuals and tumor tissue from 12 CLC patients. Analysis of normal breast tissue yielded similar results as those from wild-type mice, showing little collagen and low numbers of fibroblasts, T cells, or macrophages surrounding mammary ducts ([Figure 6](#fig6){ref-type="fig"}B). Similar to mCLCs, human CLCs showed abundant collagen fibers and fibroblasts, infiltrating macrophages, and increased numbers of T cells that were mainly located at the tumor periphery ([Figure 6](#fig6){ref-type="fig"}B). In summary, both human CLCs and mCLCs arising in Wcre;Cdh1^F/F^;Pten^F/F^ mice are heterotypic tumors characterized by a rich stromal compartment consisting mainly of collagen and fibroblasts. Inhibition of PI3K Signaling Has Antitumor Activity in Wcre;Cdh1^F/F^;Pten^F/F^ Mice {#sec2.7} -------------------------------------------------------------------------------------------- As previously mentioned, the PI3K signaling pathway is frequently mutated in human ILC. In line with the role of PTEN as a negative regulator of PI3K signaling, CLCs from our Wcre;Cdh1^F/F^;Pten^F/F^ mouse model showed activated PI3K signaling, as evidenced by elevated levels of p-AKT and p-S6 ([Figure 7](#fig7){ref-type="fig"}A). Given that improved treatments for human CLC are highly desirable, we asked whether the PI3K signaling pathway might be a potential therapeutic target for CLC. To this end, we performed a preclinical intervention study in Wcre;Cdh1^F/F^;Pten^F/F^ female mice with the dual PI3K/mTOR (mammalian target of rapamycin) inhibitor BEZ235 ([@bib19]). Eight-week-old littermates were treated with BEZ235 or vehicle for a period of 28 days. Target inhibition of the PI3K signaling pathway by BEZ235 was confirmed by IHC and western blotting, showing decreased phosphorylation of both AKT and S6 in CLCs from mice treated with BEZ235 ([Figures 7](#fig7){ref-type="fig"}A and 7B). As expected, vehicle-treated Wcre;Cdh1^F/F^;Pten^F/F^ female mice showed, at the end of treatment, more and larger lesions than age-matched Wcre;Cdh1^F/F^;Pten^F/F^ mice did at the start of treatment (t = 0) ([Figures 7](#fig7){ref-type="fig"}C and 7D). Significantly fewer and smaller CLC lesions were observed in the mammary glands of BEZ235-treated littermates compared to t = 0 and vehicle-treated Wcre;Cdh1^F/F^;Pten^F/F^ mice, indicating that this PI3K/mTOR inhibitor strongly inhibited tumor formation in our mCLC model ([Figures 7](#fig7){ref-type="fig"}C and 7D). Together, these findings indicate that pharmacological inhibition of PI3K signaling may be a promising therapeutic strategy to target CLCs that harbor activating mutations in this pathway. Discussion {#sec3} ========== In this study, we show that concomitant inactivation of E-cadherin and PTEN leads to rapid formation of mouse mammary tumors that faithfully recapitulate the histopathology, molecular phenotype, ER status, growth kinetics, metastatic behavior, and tumor microenvironment of human CLC. Inactivating mutations of E-cadherin and PTEN co-occur in a substantial fraction (13%) of human ILCs ([@bib6]). In line with this, our data revealed potent synergism and causality between E-cadherin and PTEN inactivation in CLC formation. Previously, we reported that inactivation of E-cadherin and P53 in mammary epithelium results in the development of pleomorphic ILC, a relatively rare subtype of ILC ([@bib8], [@bib9]). In contrast, inactivation of PTEN alters the tumor outcome of E-cadherin-deficient cells toward CLC, which is the most common subtype of ILC ([@bib15]). Although E-cadherin inactivation alone is insufficient to induce tumor formation ([@bib3], [@bib8], [@bib9]), it appears to be required for ILC formation by dictating tumor outcome of P53- and PTEN-deficient mammary epithelial cells toward pleomorphic ILC and CLC, respectively. Thus, besides being recognized as a hallmark of ILC ([@bib32]), E-cadherin inactivation is also required for ILC formation. In apparent contrast to its tumor-suppressor role, E-cadherin inactivation limits mammary epithelial cell survival in vivo as E-cadherin-deficient luminal epithelial cells were found to accumulate in the lumen of mammary ducts and undergo apoptosis, which is most likely due to reduced junctional integrity caused by E-cadherin loss ([@bib30]). Previously, E-cadherin inactivation in the mammary gland was found to limit the survival of alveolar cells during the onset of lactation ([@bib3]), supporting a role for E-cadherin in maintaining the survival of luminal mammary epithelial cells. E-cadherin-deficient mammary epithelium was also unable to expand in vitro and required concomitant inactivation of PTEN for outgrowth and invasion. PTEN inactivation also enabled the survival of E-cadherin-deficient mammary epithelial cells in vivo, permitting the formation of CLC. Importantly, as tumor onset in female Wcre;Cdh1^F/F^;Pten^F/F^ mice was extremely rapid, and no additional mutations were detected in the ensuing mCLCs, the combined inactivation of E-cadherin and PTEN appears to be sufficient and, thus, causal for CLC formation. PTEN is an important negative regulator of the PI3K signaling pathway, which controls cell survival ([@bib7]). It is, therefore, likely that activation of PI3K signaling due to loss of PTEN promotes the survival of E-cadherin-deficient mammary epithelial cells and subsequent formation of CLC. Indeed, mCLCs from our Wcre;Cdh1^F/F^;Pten^F/F^ mouse model showed activated PI3K signaling, as evidenced by expression of p-AKT and p-S6. Also, human ILCs were found to show frequent activation of PI3K signaling, as evidenced by increased AKT phosphorylation, reduced PTEN expression, and a high incidence of somatic mutations leading to PI3K pathway activation ([@bib6]). Together, these data suggest that PI3K pathway inhibitors might effectively target CLC. In line with this, mCLC formation in female Wcre;Cdh1^F/F^;Pten^F/F^ mice was significantly decreased by treatment with the dual PI3K/mTOR inhibitor BEZ235, indicating that the PI3K signaling pathway is an important Achilles' heel of CLC. At present, BEZ235 is under evaluation in phase I/II clinical trials ([@bib1]), and future studies are needed to investigate its efficacy in patients with CLC. However, it should be noted that antitumoral efficacy of PI3K pathway inhibitors might be reduced, in case CLCs acquire additional activating mutations in oncoproteins, such as KRAS, or loss-of-function mutations in tumor suppressors, such as PTPN11/SHP2 and p53, which we identified to alter tumor outcome and growth kinetics of mCLCs. Other strategies to target CLC might involve targeting the tumor microenvironment, which can promote growth, metastasis, and therapy resistance of breast cancer ([@bib2], [@bib13], [@bib31]). Furthermore, the massive deposition of collagen in the tumor microenvironment of both mouse and human CLC suggests that collagen production is important for CLC development and progression. Collagen has been implicated in the induction of tumor cell invasion, as it was found to accumulate at invasive edges in breast tumors ([@bib10], [@bib28]) and to induce in vitro invasion of cultured breast tumor cells ([@bib25]). Similarly, tumor cells in mCLCs align with collagen as they invade the surrounding tissue. Both mouse and human CLC showed massive infiltration of fibroblasts, which are, in general, the main source of collagen deposition. Future studies will be required to investigate the importance of these cancer-associated fibroblasts (CAFs) for mCLC development and progression and whether targeting CAFs might provide a useful strategy for treatment of human CLC. In conclusion, in this study we presented concomitant inactivation of E-cadherin and PTEN in mammary epithelium to drive the formation of mCLC, thus developing a pertinent mouse model for human CLC. This mouse model provides a powerful in vivo tool to develop novel therapeutic strategies for CLC patients, which may extend the relative limited treatment options that are currently available in the clinic. Experimental Procedures {#sec4} ======================= Generation of Mice {#sec4.1} ------------------ We intercrossed Wcre;Cdh1^F/F^ mice ([@bib9], [@bib8]) with Pten^F/F^ mice ([@bib20]) and mTmG reporter mice ([@bib24]). Mice were bred onto an FVB/N background and genotyped as described previously ([@bib8], [@bib20]); ([@bib24]). All animal experiments were approved by the Animal Ethical Committee and were conducted in compliance with the Netherlands Cancer Institute and Dutch Animal Welfare guidelines. Isolation of Primary Mammary Epithelium for Organoid Generation {#sec4.2} --------------------------------------------------------------- Purified primary mammary epithelial fragments were obtained from isolated mammary glands by mechanical disruption, collagenase/trypsin digestion, and differential centrifugation to generate organoids as previously described ([@bib12]). For additional details, see the [Supplemental Information](#app3){ref-type="sec"}. Immunoblotting, IF, and Histopathology {#sec4.3} -------------------------------------- Immunoblotting, IF, and histopathological analysis was performed as described in the [Supplemental Information](#app3){ref-type="sec"}. Mammary Gland Reconstitution {#sec4.4} ---------------------------- Mammary gland reconstitutions were performed based on previously described mouse mammary epithelial cell transplantation ([@bib11]). In brief, the fourth mammary glands of 18- to 21-day-old wild-type or mTmG mice on FVB/N background were cleared by surgical removal of the epithelium-containing part, including the lymph node, and were engrafted with small fragments (1-2 mm in diameter) of mammary gland tissue from 5- to 6-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice. Transplanted mice were checked biweekly for the development of tumors. Genomic Analysis {#sec4.5} ---------------- RNA sequencing, whole-exome sequencing, and CNV analysis was performed as described in the [Supplemental Information](#app3){ref-type="sec"}. Flow Cytometry {#sec4.6} -------------- Blood samples were collected in tubes containing heparin. Tumors and lungs were mechanically chopped using a McIlwain tissue chopper (Mickle Laboratory Engineering), followed by digestion for 1 hr at 37°C in 3 mg/ml collagenase type A (Roche) and 25 μg/ml DNase (Sigma-Aldrich) in serum-free DMEM medium. Enzyme activity was neutralized by the addition of cold DMEM/10% fetal bovine serum (FBS), and suspensions were dispersed through a 70-μm cell strainer. Bone marrow was isolated through the flushing of two femurs using PBS. All single-cell suspensions were treated with erythrocyte lysis buffer (0.15 M NH~4~Cl, 10 mM KHCO~3~, 0.5 M EDTA) and collected in PBS + 1% BSA. DAPI (1:20) was added to exclude dead cells. GFP-positive cells were analyzed using a Beckman Coulter CyAn ADP flow cytometer and FlowJo software. BEZ235 Intervention Study {#sec4.7} ------------------------- Eight-week-old littermates were orally dosed on a daily basis with 30 mg/kg BEZ235 dissolved in 10% N-methyl-2-pyrrolidone/90% PEG 400 (polyethylene glycol 400) (Sigma-Aldrich) or vehicle for a period of 28 days. Two hours after the last dosing, mice were sacrificed, and mammary glands were harvested for further analysis. Antitumor activity was assessed by calculating the ratio between the sum of all CLC regions and the total mammary gland area using ImageJ. Statistics {#sec4.8} ---------- The following statistics were used for the indicated figures: for [Figures 1](#fig1){ref-type="fig"}C, 1E, [2](#fig2){ref-type="fig"}D, [S2](#mmc1){ref-type="supplementary-material"}C, and S2D, an unpaired t test was used; for [Figures 3](#fig3){ref-type="fig"}A and [S3](#mmc1){ref-type="supplementary-material"}A, we used a log-rank Mantel-Cox test; for [Figure 3](#fig3){ref-type="fig"}D, a Mann-Whitney test was used; for [Figure S3](#mmc1){ref-type="supplementary-material"}C, differential expression was assessed using limma ([@bib29]); and a one-tailed Mann-Whitney U test was used for [Figure 4](#fig4){ref-type="fig"}E. For [Figure 7](#fig7){ref-type="fig"}D, a paired t test was performed on these ratios between BEZ235- and vehicle-treated littermates. Author Contributions {#sec5} ==================== M.C.B., M.N., S.K., R.v.A., and J.J. conceived experiments and analyzed data. M.C.B., M.N., S.K., E.S., N.B., A.L.Z., J.R.d.R., E.W., and E.v.d.B. conducted experiments. L.W. provided expertise and feedback. M.C.B., M.N., and J.J. wrote the manuscript. Accession Numbers {#app1} ================= The accession number for the raw data of RNA sequencing and exome sequencing reported in this paper is European Nucleotide Archive (ENA): PRJEB14147. Supplemental Information {#app3} ======================== Document S1. Supplemental Experimental Procedures, Figures S1--S6, and Tables S1 and S2Document S2. Article plus Supplemental Information We are grateful to Anne Paulien Drenth, Ute Boon, and Renske de Korte-Grimmerink for help with the mammary gland reconstitutions and to the NKI Molecular Pathology & Biobanking Facility for supplying human tumor material and lab support. We thank the NKI animal facility, animal pathology facility, flow cytometry facility, and genomics core facility for technical support. This work was supported by grants from the Dutch Cancer Society (NKI 2015-7589); the Netherlands Organization for Scientific Research (\[VENI 016156012, NGI-Zenith 93512009, and VICI 91814643\], Cancer Genomics Netherlands \[CGCNL\], Cancer Systems Biology Center \[CSBC\], and a National Roadmap grant for the Mouse Clinic for Cancer and Aging research \[MCCA\]); the European Union Seventh Framework Programme (EurocanPlatform project 260791 and Infrafrontier-I3 project 312325); and the European Research Council (ERC Synergy project CombatCancer). Supplemental Information includes Supplemental Experimental Procedures, six figures, and two tables and can be found with this article online at [http://dx.doi.org/10.1016/j.celrep.2016.07.059](10.1016/j.celrep.2016.07.059){#intref0010}. ![Inactivation of E-Cadherin and PTEN Alters Mammary Epithelial Morphology and Growth In Vitro\ (A) Western blot analyses of E-cadherin, PTEN, and p-AKT(Ser473)/AKT protein expression of untransduced and AdCre-transduced Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ mammary organoids.\ (B) Representative bright-field images of untransduced and AdCre-transduced Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ mammary organoids at day 21 of organoid culture. The dashed circle depicts the surface size of AdCre-transduced Cdh1^F/F^ organoids. Scale bar, 50 μm.\ (C) Average measurements of surface size (n = 10 per genotype) of untransduced and AdCre-transduced Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ organoids at days 7, 14, and 21 of organoid culture.\ (D) Expression of Ki67 and CC3 measured by whole-mount immunostaining of AdCre-transduced Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ organoids. Scale bars, 50 μm.\ (E) Quantification of the fraction of Ki67- and CC3-positive cells over the total cell number (determined by Hoechst) in AdCre-transduced Cdh1^F/F^, Pten^F/F^, and Cdh1^F/F^;Pten^F/F^ (n = 10 per genotype).\ (F) Representative bright-field images of a Cdh1^F/F^;Pten^F/F^ organoid displaying invasive protrusions of single rows of cells (arrows) 1 day after medium change. Scale bars, 25 μm.](gr1){#fig1} ![Inactivation of E-Cadherin and PTEN in Mammary Epithelium In Vivo Induces Rapid Formation of Neoplastic Lesions\ (A) H&E staining and IHC of CK8 in mammary glands from 6-week-old wild-type, Wcre;Cdh1^F/F^, Wcre;Pten^F/F^, and Wcre;Cdh1^F/F^;Pten^F/F^ mice. Scale bars, 150 μm.\ (B) IF analysis of CK8, E-cadherin, and PTEN in mammary glands from 6-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice. Scale bar, 20 μm. Zooms of the asterisk-marked area reveal CK8-positive neoplastic cells that are negative for of E-cadherin and PTEN (indicated by arrows), flanked by PTEN-positive stromal cells.\ (C) IF analysis of CK8 (green) and CC3 (red) in mammary glands from 6-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice. Arrows mark CC3-positive apoptotic cells in the mammary duct. Scale bar, 20 μm.\ (D) Average numbers of CC3-positive shedded cells per 500-μm^2^ area in mammary gland sections from Wcre;Cdh1^F/F^, Wcre;Pten^F/F^and Wcre;Cdh1^F/F^;Pten^F/F^ mice.](gr2){#fig2} ![Somatic Inactivation of E-Cadherin and PTEN Induces Mammary Tumors that Closely Resemble Human CLC\ (A) Mammary tumor-free survival plot of Wcre;Pten^F/+^ (n = 26), Wcre;Cdh1^F/F^;Pten^F/+^ (n = 39), Wcre;Pten^F/F^ (n = 23), and Wcre;Cdh1^F/F^;Pten^F/F^ (n = 60) mice. Latency is shown in days of age. n.s., not significant.\ (B) Pie charts showing the distribution of mammary tumor phenotypes in Wcre;Pten^F/+^ (n = 7), Wcre;Cdh1^F/F^; Pten^F/+^ (n = 23), Wcre;Pten^F/F^ (n = 28), and Wcre;Cdh1^F/F^;Pten^F/F^ (n = 238) mice.\ (C) Representative H&E staining and IHC of CK8, Vimentin, E-cadherin, PTEN, and ER in mammary tumors from 12-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice. Scale bars, 25 μm. The zoom of the asterisk-marked area reveals PTEN-negative tumor cells flanked by PTEN-positive stromal cells.\ (D) Quantification of ER-positive tumor cells detected by IHC of ER in Brca1^Δ/Δ^;Trp53^Δ/Δ^ (n = 10) and Cdh1^Δ/Δ^;Pten^Δ/Δ^ (n = 14) mammary tumors.\ (E) Unsupervised hierarchical clustering of 956 human breast cancer samples, 22 Brca1^Δ/Δ^;Trp53^Δ/Δ^ mammary tumor samples, and 20 Cdh1^Δ/Δ^;Pten^Δ/Δ^ mCLC samples. Sidebars indicate tumor subtype according to PAM50 for the human breast tumors, genotypes for the mouse mammary tumors, and Cdh1 mutation status for all tumors.](gr3){#fig3} ![Mammary Gland Transplantation Reveals Linear Growth Kinetics of mCLCs\ (A) Mammary-tumor-free survival in Wcre;Cdh1^F/F^;Pten^F/F^ mice and wild-type mice transplanted with small precancerous mammary gland fragments from Wcre;Cdh1^F/F^;Pten^F/F^ mice.\ (B) Average size of tumors in mice transplanted with precancerous Wcre;Cdh1^F/F^;Pten^F/F^ mammary gland fragments (n = 82).\ (C) Pie chart showing distribution of tumor phenotypes from mice transplanted with precancerous Wcre;Cdh1^F/F^;Pten^F/F^ mammary gland fragments (n = 58).\ (D) A fraction (n = 17) of mice transplanted with precancerous Wcre;Cdh1^F/F^;Pten^F/F^ mammary gland fragments showed a sudden transition from linear to exponential growth 250--450 days after tumor onset.\ (E) Overview of oncogenic mutations identified in exponentially growing solid ILCs and EMT tumors.\ (F) Number of genomic aberrations in linear-growing mCLCs (n = 14) or exponential-growing solid ILCs and EMT tumors (n = 7).](gr4){#fig4} ![Metastatic Spectrum of mCLCs\ (A and B) Representative H&E stainings revealing tumor cell invasion into surrounding muscular tissue (A) and nerve tissue (B) in mammary glands from 12-week-old Wcre;Cdh1^F/F^;Pten^F/F^ mice (scale bars, 100 μm).\ (C) Incidence of distant metastases in Wcre;Pten^F/+^ (n = 15), Wcre;Cdh1^F/F^;Pten^F/+^ (n = 32), Wcre;Pten^F/F^ (n = 26), and Wcre;Cdh1^F/F^;Pten^F/F^ (n = 62) mice. Colors depict different metastatic sites in metastasis-bearing Wcre;Cdh1^F/F^;Pten^F/+^ (n = 11) and Wcre;Cdh1^F/F^;Pten^F/F^ (n = 3) mice.\ (D) Representative H&E stainings of tumor infiltration (marked by asterisks) in lymph node, lung, pleura, peritoneum, and spleen. Scale bars, 200 μm.\ (E) Representative FACS plots of GFP-sorted cells obtained from mammary tumors, blood, lungs, and bone marrow of tumor-bearing Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice, revealing dissemination of GFP-marked tumor cells to distant organs.\ (F) Fractions of Wcre;Cdh1^F/F^;Pten^F/F^;mTmG mice (n = 5) harboring GFP-marked tumor cells in indicated organs.](gr5){#fig5} ![Early Stromal Recruitment in Mouse and Human CLC\ (A and B) In (A), H&E and Masson's trichrome staining (revealing collagen) and IHC of PDGFRβ (fibroblasts), CD3 (T cells), and F4/80 (macrophages) in tumor-free mammary glands and mCLCs from wild-type, Wcre;Cdh1^F/F^, Wcre;Pten^F/F^, and Wcre;Cdh1^F/F^;Pten^F/F^ mice of indicated age were used to examine stromal compositions in each indicated genotype, which were subsequently compared to (B) the microenvironment of healthy human mammary glands and the tumor microenvironment of human CLCs.](gr6){#fig6} ![Inhibition of PI3K Signaling Has Antitumor Activity in Wcre;Cdh1^F/F^;Pten^F/F^ Mice\ (A) Representative IHC of p-AKT and p-S6, reflecting PI3K signaling activity in mCLCs from Wcre;Cdh1^F/F^;Pten^F/F^ mice treated with either vehicle or BEZ235.\ (B) Western blot analysis of p-AKT/AKT and p-S6/S6 expression, reflecting PI3K signaling activity in mCLCs from vehicle-treated and BEZ235-treated Wcre;Cdh1^F/F^;Pten^F/F^ mice.\ (C) H&E staining of mammary glands from Wcre;Cdh1^F/F^;Pten^F/F^ mice treated with vehicle or BEZ235. Lesions are indicated by arrows.\ (D) Quantification of lesion formation by measuring total lesion area over total mammary gland tissue area for vehicle-treated (n = 6) and BEZ235-treated (n = 6) Wcre;Cdh1^F/F^;Pten^F/F^ littermates, compared to age-matched Wcre;Cdh1^F/F^;Pten^F/F^ mice (n = 9) at start of treatment (t = 0).](gr7){#fig7} [^1]: Co-first author [^2]: Lead Contact	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec0005} =============== Severe acute respiratory syndrome-related coronavirus-2 (SARS-CoV-2) was first recognised in January 2020 and rapidly spread world-wide \[[@bib0005]\]. Tests designed to measure antibodies to SARS-CoV-2 antigens were rapidly developed and are important for diagnostics and seroprevalence studies. The latter could help inform disease burden estimates, studies of transmission dynamics and modelling of the epidemic. Antibody tests are particularly important in the context of mild or asymptomatic disease where a swab reverse transcriptase polymerase chain reaction (RT-PCR) test may be negative. For this reason, an understanding of the sensitivity and specificity of the tests being used is critical. The trimeric spike (S) protein of SARS-CoV-2 is present on the viral surface and in most cases is cleaved by host proteases into the S1 and S2 subunits, responsible for receptor recognition and membrane fusion respectively. S1 uses a region of the molecule, known as the receptor binding domain (RBD) to bind to host ACE-2 receptor and thereby gain entry to the cell \[[@bib0010]\]. The N terminal domain (NTD) of the spike protein does not interact with the receptor but contains the functional elements required for membrane fusion of the virion. The nucleocapsid (N) protein plays an important role in transcription enhancement and viral assembly \[[@bib0015]\]. Specific immunoglobulin-G (IgG) and IgM antibody responses to SARS-CoV-2 S, N and RBD of the spike protein develop between 6--15 days following disease-onset \[[@bib0020]\]. Despite a rapid increase in the number and availability of SARS-CoV-2 serologic assays, most have undergone minimal external evaluation and validation \[[@bib0025]\]. A recent large scale Spanish seroprevalence study used a point of care IgG test with a stated sensitivity of 97.2 % but on verification found it to have a sensitivity of either 82.1 %, 89.7 %, 99.6 % or 100 % depending on the sample sets used for evaluation \[[@bib0030]\]. All assays currently suffer from the absence of a defined standard serum so results are reported as positive or negative or as optical density readouts complicating the comparison between assays and studies and for many binding assays the relationship between antibody concentration and function is unclear. We have evaluated a novel assay designed to simultaneously measure IgG to four SARS-CoV-2 antigens; full-length trimeric S, RBD and NTD of spike as well as N protein. The assay, based on Meso Scale Discovery (MSD) technology, utilises a 96-well based solid-phase antigen printed plate and an electrochemiluminescent detection system. In addition this assay can measure the ability of serum to inhibit the interaction between spike protein components and soluble ACE-2, also called a pseudo-neutralisation assay \[[@bib0035]\]. To evaluate the sensitivity and specificity of the MSD assay, we were able to utilise a relatively large number of samples obtained from SARS-CoV-2 RT-PCR positive health care workers or patients as well as antibody positive health care staff enrolling in a large SARS-CoV-2 cohort study. 2. Materials and methods {#sec0010} ======================== 2.1. Serum samples {#sec0015} ------------------ Sera were obtained from Great Ormond Street Children's Hospital NHS Foundation Trust (GOSH) and came from; (i) Symptomatic RT-PCR + healthcare workers (ii) staff enrolling in a prospective longitudinal cohort study of SARS-CoV-Serology (COSTARS, IRAS 282713, ClinicalTrials.gov Identifier: NCT04380896) who tested positive for anti-Nucleocapsid IgG (Epitope Diagnostics Inc, San Diego, USA) (iii) Sera from RT-PCR + hospitalised children (n = 10). Sera for specificity pre-dated 2019 and derived from anonymised samples from healthy adults enrolled in previous studies. Pooled serum from two individuals with high convalescent antibody levels were used as an interim standard serum calibrated against research reagents NIBSC 20/130 and NIBSC 20/124 (National Institute for Standards and Biological Control, Potters Bar, UK, <https://www.nibsc.org/>) obtained from COVID-19 recovered patients. 2.2. Serological assays {#sec0020} ----------------------- Samples were screened for IgG to SARS-CoV-2 N protein using a commercially available kit (Epitope Diagnostics Inc, San Diego, USA) as previously described \[[@bib0040]\]. 2.3. Meso scale discovery coronavirus panel for COVID-19 serology {#sec0025} ----------------------------------------------------------------- A multiplexed MSD immunoassay (MSD, Rockville, MD) was used to measure the responses to SARS-CoV-2 and other respiratory pathogens. A MULTI-SPOT® 96-well, 10 Spot Plate was coated with four SARS CoV-2 antigens (S, RBD, NTD and N), SARS-CoV-1 and MERS spike trimers, spike proteins from seasonal coronaviruses OCV43S and HKU1, influenza A antigen derived from H3/HongKong and Bovine Serum Antigen. Antigens were spotted at 200−400 μg/mL in a proprietary buffer, washed, dried and packaged for further use (MSD® Coronavirus Plate 1). Proteins were expressed in a mammalian cell expression system (Expi 293 F), purified by ion exchange chromatography, affinity purification, and size exclusion chromatography; the spike proteins were produced as trimers in the pre-fusion form. These assays were developed by MSD in collaboration with the Vaccine Research Center at NIAID (A. McDermott). To measure IgG antibodies, plates were blocked with MSD Blocker A following which reference standard, controls and samples diluted 1:500 in diluent buffer were added. After incubation, detection antibody was added (MSD SULFO-TAG™ Anti-Human IgG Antibody) and then MSD GOLD™ Read Buffer B was added and plates read using a MESO® SECTOR S 600 Reader. 2.4. Meso scale discovery pseudo-neutralisation assay {#sec0030} ----------------------------------------------------- Plates were blocked and washed as above, assay calibrator (COVID-19 neutralising antibody; monoclonal antibody against S protein; 200 μg/mL), control sera and test sera samples diluted 1 in 10 in assay diluent were added to the plates. Following incubation Plates an 0.25 μg/mL solution of MSD SULFO-TAG™ conjugated ACE-2 was added after which plates were read as above. Percentage inhibition was calculated relative to the assay calibrator (maximum 100 % inhibition). 2.5. Statistical analysis {#sec0035} ------------------------- Statistical analysis was performed using MSD Discovery Workbench and GraphPad Prism version 8.0 (GraphPad, San Diego, CA). Antibody concentration in arbitrary units (AU) was interpolated from the ECL signal of the internal standard sample using a 4-parameter logistic curve fit. ROC curves showing the sensitivity and specificity (plotted as 100 %-specificity %) calculated using each value in the data as a cut-off were plotted for each antigen. A cut-off antibody concentration was chosen based on the lowest value leading to a positive likelihood ratio (LR) of \>10, in order to maximise sensitivity while providing strong evidence to rule-in infection \[[@bib0045]\]. For S antigen binding, all LR's were above 10, therefore the LLOD was used as the cut-off for this antigen. Comparisons between groups were performed by Kruskal-Wallis one-way ANOVA with Dunn's correction for multiple comparisons. Correlation analysis was performed using Spearman correlation. P values of \<0.05 were considered as significant. Latent class models with two classes were fitted with the binary antibody responses as outcome variables, using the poLCA package in the R statistical environment. The code used for the latent class analysis is available on request. 3. Results {#sec0040} ========== 3.1. Participants and samples {#sec0045} ----------------------------- SARS-CoV-2 positive samples (COVID-19 cohort) comprised 169 PCR positive and 27 anti-N IgG positive serum samples from mild symptomatic or asymptomatic cases (total n = 196, 138 females, 56 males \[2 missing\], median age 37 years). Time between symptom onset and sampling ranged from 4 to 63 days for 168 subjects with verified onset date. Control serum samples comprised 194 anonymised legacy samples obtained from healthy adults, aged predominantly over 50 years. 3.2. Standard serum assignment {#sec0050} ------------------------------ An internal standard serum (ISS) was assigned values for S, RBD and N by calibration against the NIBSC control sera. NIBSC 20/130 was used to assign arbitrary unit (AU) values for S and RBD and NIBSC 20/124 for N (Supplementary Figure S1). No endpoint titre corresponding to NTD antigen was available for ISS assignment. The interim values assigned were S 2154 AU, RBD 1837 AU and N 3549 AU and 1000 AU for all other antigens. 3.3. Evaluation of the coronavirus panel for COVID-19 serology {#sec0055} -------------------------------------------------------------- The lower limit of detection (LLOD) was assigned as 1% of the standard value in AU, and upper limit of detection (ULOD) was assigned for NTD and RBD only as the S and N antigen did not reach an upper limit ([Table 1](#tbl0005){ref-type="table"} ). For statistical purposes, ULOD was assigned the highest calculated concentration plus 20 % and LLOD as 0.5 %.Table 1The lower limit of detection (LLOD), upper limit of detection (ULOD), quality control (QC) sample range in arbitrary units (AU) and positive/negative cut-off for each SARS-CoV-2 antigen analysed.Table 1AntigenLLOD (max.) (AU)ULOD (min.) (AU)QC sample range (AU)Positive/ negative cut-offCoV-2 S21.54NA1092−147821.5CoV-2 RBD18.371254772176−2944201.7CoV-2 N35.49NA3627−4907185.4CoV-2 NTD10.00194521004−13591924 The mean coefficient of variation (CV) between duplicates was \<15 % for all except NTD (17.4 %, data not shown). The mean intra-assay CV was 6.2 % and inter-assay variation \<15 % across all SARS-CoV-2 antigens except NTD (19.0 %) on one of four samples (Supplementary [Table 1](#tbl0005){ref-type="table"}). A QC sample was run on each plate (average CV 10.3 %) and an acceptable performance range was set as within 3 SD of the mean. 3.4. Assay sensitivity and specificity {#sec0060} -------------------------------------- [Fig. 1](#fig0005){ref-type="fig"} A-D shows the concentration of IgG to each SARS-CoV-2 antigen.Fig. 1Anti-SARS-CoV-2 IgG concentration.The concentration of SARS-CoV-2 antibody against (a) spike (S), (b) receptor binding domain (RBD), (c) nucleocapsid (N) and (d) N terminal domain (NTD) was measured using the MSD coronavirus panel. Graphs show data in arbitrary units (AU) (based on the calibrated internal standard serum) in the COVID-19 cohort (n = 196) and controls (n = 194, pre-December 2019). Line shows positive/negative discrimination cut-off.Fig. 1 ROC curves were plotted to visualise the trade-off between sensitivity and specificity for each antigen ([Fig. 2](#fig0010){ref-type="fig"} A-D). The high area under the curve (AUC) values for S (0.95 %; 95 %CI 0.93 to 0.97), RBD (0.92 %, 0.89−0.95) and N (0.90 %, 0.87−0.94) indicates the high accuracy of these tests. [Table 1](#tbl0005){ref-type="table"} shows the Positive/negative cut-off values calculated from the ROC using LR \> 10. NTD data was less consistent than the other SARS-CoV-2 antigens and demonstrated lower sensitivity and specificity so this antigen was not evaluated further.Fig. 2Receiver Operating Characteristic (ROC) curves for each SARS-CoV-2 antigen.Sensitivity and specificity were calculated using each value in the data table as a cut-off value (n = 390). Graphs show the sensitivity vs 100 %-specificity of SARS-CoV-2 antigen (a) spike (S), (b) receptor binding domain (RBD), (c) nucleocapsid (N) and (d) N terminal domain (NTD). The area under curve (AUC) and 95 % CI is also shown for each antigen.Fig. 2 The specificity for S, RBD and N assays are shown in [Table 2](#tbl0010){ref-type="table"} . Assay sensitivity was initially calculated on the entire COVID-19 cohort; S antigen had the highest AUC and was the most sensitive and specific at 90.8 % and 97.4 % respectively.Table 2Assay specificity calculated for each SARS-CoV-2 antigen from the control cohort.Table 2AntigennPositiveNegativeSpecificity (95 % CI) (%)CoV-2 S194518997.4 % (94.1 to 98.9)CoV-2 RBD1941517992.3% (87.6 to 95.3)CoV-2 N1941418092.8 % (88.2 to 95.7) 3.5. Evaluation of sensitivity according to time since onset of symptoms {#sec0065} ------------------------------------------------------------------------ [Fig. 3](#fig0015){ref-type="fig"} shows the anti-S, RBD and N IgG concentration split into time since onset of symptom intervals of 0−7 days, 8--14 days, 15--21 days and over 21 days. For all three antigens, the median antibody concentration increased significantly with time since symptom onset (SARS-CoV-2 S, Spearman correlation (r) = 0.453; SARS-CoV-2 RBD, ; SARS-CoV-2 N, r = 0.392, all p=\<0.0001 (Supplementary [Fig. 2](#fig0010){ref-type="fig"}A-C)) and at all time points were higher than controls (p=\<0.0001) ([Fig. 3](#fig0015){ref-type="fig"}A-C).Fig. 3Anti-SARS-CoV-2 IgG concentration according to time since onset of symptoms.Graphs show the concentration of SARS-CoV-2 antibody against (a) spike (S), (b) receptor binding domain (RBD) and (c) nucleocapsid (N) in arbitrary units (AU) (based on the calibrated internal standard serum) of the COVID-19 cohort split in to intervals of 0−7 days, 8--14 days, 15--21 days and over 21 (\>21) days since symptom onset (to sample collection). Error bars show geometric mean with 95 % CI, line shows positive/negative discrimination cut-off, \p \< 0.05, \\* p \< 0.01 determined by Dunn's multiple comparisons test. Comparisons across interval groups had p \< 0.0001 by one-way ANOVA Kruskal-Wallis test. The assay sensitivity at each time point is shown in [Table 3](#tbl0015){ref-type="table"}.Fig. 3 Sensitivity and specificity was calculated for groups 0−7d, \>7d, \>14d and \>21d since the onset of symptom The S antigen was the most sensitive of the three, with a sensitivity of 96.2 % and 97.9 % \>14 days and \>21 days respectively ([Table 3](#tbl0015){ref-type="table"} ).Table 3Assay sensitivity by time since onset of symptoms for each SARS-CoV-2 antigen calculated using the COVID-19 cohort with verified time between onset of symptoms and blood sampling. Time was divided into 0-7 days, over 7 days, over 14 days and over 21 days since the onset of symptoms.Table 3AntigenGroupnPositiveNegativeSensitivity (95 % CI) (%)CoV-2 STotal1961781890.8 % (86.0--94.1)Time since onset of symptoms0−7 days2015575.0% (53.1--88.8)Over 7 days1481381093.2% (88.0--96.3)Over 14 days7875396.2 % (89.3--99.0)Over 21 days4746197.9 % (88.8--99.9)CoV-2 RBDTotal1961534378.1% (71.8--83.3)Time since onset of symptoms0−7 days2012860.0% (38.7--78.1)Over 7 days1481192980.4% (73.3--86.0)Over 14 days7871791.0% (82.6--95.6)Over 21 days4744393.6% (82.8--97.8)CoV-2 NTotal1961435373.0% (66.3--78.7)Time since onset of symptoms0−7 days2012860.0% (38.7--78.1)Over 7 days1481064271.6% (63.9--78.3)Over 14 days78661284.6% (75.0--91.0)Over 21 days4741687.2% (74.8--94.0) 3.6. Antibody concentration relationship between antigens {#sec0070} --------------------------------------------------------- The concentration of anti-S, RBD and N all correlated significantly with each other (p \< 0.0001; [Fig. 4](#fig0020){ref-type="fig"} A-C), the strongest association was between S and RBD (r = 0.882) ([Fig. 4](#fig0020){ref-type="fig"}A). Our two-class latent class model built using binary S, RBD and N antigen results predicted known status with 81.1 % (95 %CI 74.8--86.2) sensitivity and 99.0 % (95 %CI 95.9--99.8) specificity. It therefore had lower sensitivity and no meaningful improvement in specificity, compared to using the concentration of S antibody alone, with the 21.54 AU cut-off.Fig. 4IgG concentration relationship between antigens.Correlation between anti-SARS-CoV-2 antibody concentration of all COVID-19 group samples (n = 196) (a) S vs RBD, (b) S vs N and (c) N vs RBD. r and p value were determined by Spearman correlation. p values of \<0.05 were considered as significant.Fig. 4 3.7. Pseudo-neutralisation {#sec0075} -------------------------- 183 COVID-19 cohort samples with sufficient volume and 194 control group samples were evaluated in the pseudo-neutralisation assay. The percentage inhibition of ACE-2 receptor binding to the S and RBD antigens for the COVID-19 cohort was significantly higher than the controls (S, median 1.94 % (95 %CI 1.36--2.25) vs 0.063 % (95 %CI 0.053−0.073), p=\<0.0001; RBD, 1.50 % (95 %CI 1.064--2.11) vs 0.38 % (95 %CI 0.36−0.39); p=\<0.0001) ([Fig. 5](#fig0025){ref-type="fig"} A-B) and correlated with IgG concentration for both S and RBD antigens (Spearman correlation (r) = 0.805 and r = 0.834 respectively, p=\<0.0001) ([Fig. 5](#fig0025){ref-type="fig"}C-D).Fig. 5Percentage inhibition by anti-SARS-CoV-2 S and RBD antibody measured by MSD pseudo-neutralisation assay.Inhibition of ACE-2 binding by SARS-CoV-2 antibody against (a) spike (S) and (b) receptor binding domain (RBD) was measured using the MSD coronavirus pseudo-neutralisation assay. 183 COVID-19 cohort samples and 194 control samples were analysed. Graphs show median and 95 % CI with a line showing neutralisation assay positive/negative discrimination cut-off determined by ROC. The correlation between antibody concentration and percentage inhibition of (c) S and (d) RBD antigens in all positive group samples was assessed and r and p was determined by Spearman correlation, line shows binding assay positive/negative discrimination cut-off.Fig. 5 Cut-offs (LR \> 10) were 0.162 % for S and 0.524 % for RBD (shown by the dotted line on [Fig. 5](#fig0025){ref-type="fig"}A-B). Sensitivity and specificity for S were 97.8 % and 97.9 % respectively but lower for RBD (77.2 % and 92.8 % respectively). In the COVID-19 cohort there were some IgG positive sera that did not demonstrate neutralisation (below cut-off, n = 4 for S and 36 for RBD). These sera were predominantly those taken soon after the onset of symptoms; 22 between 0−7 days, 9 over 14 days and 5 over 21 days. 4. Discussion {#sec0080} ============= Accurate tests of SARS-CoV-2 antibodies are critical for reliably evaluating exposure to the virus causing COVID-19. Despite a large number of assays rapidly becoming available, many have not undergone rigorous evaluation. In this study we describe a novel assay that can measure antibody to several SARS-CoV-2 antigens simultaneously as well as evaluating the functional capacity of anti-Spike antibodies. The assay we used is based on existing technology developed by Meso Scale Discovery. We decided to evaluate IgG only as the kinetics of IgM responses appear to mimic those of IgG and thus add little value \[[@bib0020]\]. Unlike the majority of studies published to date, we were able to utilise a panel of COVID-19 convalescent plasma recently distributed by WHO to calibrate an internal standard made from pooled convalescent serum. This allowed us to express titres in arbitrary units that can then be compared to other assays that report values calibrated against the WHO panel. The assays performed reliably and consistently over the period of study and passed all the performance criteria expected for a solid-phase based assay. Using a carefully defined cohort of known SARS-CoV-2 exposed individuals and relevant controls we were able to show the sensitivity and specificity of the assay for the four antigens of interest. Comparing the performance of S and RBD assays in a recently published systematic review and metanalysis of the diagnostic accuracy of serological tests for COVID-19 \[[@bib0050]\] the S assay we evaluated had superior sensitivity to all of the assays included in the review while RBD performance was superior to most. The reason for this could be related to the technical aspects of the assay itself including the integrity of the antigen used and the sensitivity of the detection platform but also the use of a well-defined cohort of individuals with known exposure to SARS-CoV-2. Only the N terminal domain of the spike protein did not perform well in this assay with poor sensitivity due to the overlap in antibody titres between the COVID-19 cohort and controls. The assay format permitted the measurement of antibody against spike protein derived from SARS-1, MERS and two seasonal coronaviruses, but the results of antibody binding to these antigens could not be assessed in the same way as for the SARS-CoV-2 antigens due to the absence of defined negative and positive serum sets. An advantage of this assay is its ability to measure antibody induced inhibition of ACE-2 receptor-spike interaction thought to be the major mechanism by which SARS viruses, including SARS-CoV-2 attach to host cell surfaces \[[@bib0055],[@bib0060]\]. In the COVID-19 cohort, there was a good correlation between anti-S and anti-RBD IgG and function although a few sera bound antigen but did not neutralize. These were dominated by sera taken soon after infection and as recently described, could be non-neutralising and targeting epitopes outside the RBD \[[@bib0065]\]. Few of the control cohort sera had any pseudo-neutralisation activity despite pre-existing IgG to seasonal Coronavirus spike proteins suggesting season Coronavirus exposure is unlikely to modify interaction with SARS-CoV-2. Other cross reactive immunological mechanisms (eg T cells) cannot be ruled out and may explain the varied clinical response following exposure to SARS-CoV-2 \[[@bib0070]\]. This pseudo-neutralisation assay has been shown to correlate well with neutralisation assays using live SARS-CoV-2 (MSD, personal communication). In summary, the MSD multiplexed coronavirus panel assay evaluated in this study is highly reproducible, specific and sensitive for the detection of anti-SARS-CoV-2 antibody over 14 days since the onset of COVID-19 symptoms. The assay can be adapted to measure antibody function which corelated well with spike protein antibody concentration. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declarations of Competing Interest {#sec0085} ================================== None Appendix A. Supplementary data {#sec0095} ============================== The following are Supplementary data to this article: The study team would like to thank Meso Scale Discovery for the donation of the plates and reagents that allowed us to complete the work, the COSTARS study team at Great Ormond Street Children's Hospital, staff in the Great Ormond Street Children's Hospital Clinical Immunology Laboratory for additional support and the NIHR UCL Great Ormond Street Biomedical Research Centre for underpinning infrastructure support that facilities translation research at GOSH. Supplementary material related to this article can be found, in the online version, at doi:<https://doi.org/10.1016/j.jcv.2020.104572>.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1} =============== Since the discovery of the first microRNA in 1993, this type of small non-coding RNAs has attracted the attention of scientists in nearly all fields of researches and has proved themselves as a superstar \[[@bib1],[@bib2]\]. They have been reported to participate in various biological development processes and cell activities and are implicated in human diseases with diagnostic and therapeutic potential \[[@bib3],[@bib4]\]. According to the microRNA registry miRbase ([mirbase.org](http://mirbase.org){#intref0010}), the number of annotated human microRNAs has been expanded from 56 in miRBase release 1.0 in December 2002 to 2656 in the current release and is expected to be again growing in the future. They are scattered on all chromosomes except the Y chromosome in humans and locates in intergenic, intronic or exonic regions as either single microRNA or miR clusters \[[@bib5]\]. They are briefly small 20--24 nucleotide noncoding RNAs derived from primary microRNA transcripts and processed by a two-step cleavage involving enzymes like Drosha and Dicer. The mature microRNAs are believed to function mainly as, although not exclusively, negative regulators of target genes through RNA decay (which is major) or translation inhibition \[[@bib6]\]. Canonically, they bind to 3′-UTR of target genes through a complementary sequence of the 2--8 nucleotides in their 5′ end (so-called seed region) \[[@bib7]\]. With the development of many microRNA target prediction algorithms, it is estimated that nearly half or even more of all human genes are directly regulated by microRNAs \[[@bib8]\]. As such interesting and important players in cellular processes, microRNAs have been implicated in nearly all types of human cancers \[[@bib9], [@bib10], [@bib11]\]. Considering the relatively small size of microRNAs and especially the small size of seed regions which are critical for microRNA target recognition, even a single nucleotide variation in the pre-miRNA regions particularly those in the mature and seed regions may greatly influence either the maturation, the binding affinity/stability or even target selectivity of microRNAs. Here, I summarized the current knowledge of microRNA biogenesis, targets recognition through canonical and non-canonical binding, and microRNAs in human cancers and gave a focused review on common SNPs in pre-microRNAs, isomiRs and somatic mutations of microRNAs. I found this is still by far a not fully explored field and warrants further investigations. Considering the great potential of both microRNAs and genome variations in precision medicine \[[@bib12], [@bib13], [@bib14], [@bib15], [@bib16], [@bib17]\], I believe future studies in this field are of great importance and significance. 2. MicroRNA biogenesis and target recognition {#sec2} ============================================= MicroRNAs are small noncoding RNAs that are 19--24 nt long. They are transcribed mainly through the RNA polymerase II and following the transcription the canonical biogenesis of microRNAs is a two-step process (cropping and dicing) which includes one cleavage event in the nucleus by RNase III enzyme Drosha and the other in the cytoplasm by another RNase III enzyme Dicer \[[@bib18]\]. Drosha process the pri-microRNA into a \~60--100 nt stem-loop structured microRNA precursor (pre-microRNA). Then pre-microRNAs are exported into the cytoplasm with the help of exportins and then Dicer will cleave the pre-microRNA into a double stranded RNAs. Except this Drosha-dependent and Dicer-dependent canonical pathway, other non-canonical pathways of microRNA biogenesis have also been reported \[[@bib19],[@bib20]\]. For example, some microRNAs are Drosha independent or Dicer independent. Good example of Drosha-independent microRNAs are mirtrons which are microRNAs derived from the introns of mRNAs and produced directly through RNA splicing. m7G (7-methylguanosine)-capped pre-microRNAs are another example which does not need Drosha cleavage but could directly be exported to the cytoplasm for Dicer processing. Also, it has been found that biogenesis of miR-451 is one rare event that requires Drosha but not Dicer and was cleaved by Ago2 instead in the cytoplasm \[[@bib20]\]. After mature microRNAs are formed, they are incorporated into the RISC complex. Most microRNAs target the 3′-UTR regions of mRNAs through complementary Watson-Crick pairing of their seed region which is the 5′ 2--8 base of the mature microRNA. They either inhibits the translation or cause mRNA decapping and degradation. Many microRNA target prediction algorithms based on this theory have been developed and widely accepted including TargetScan, Pictar, RNA22, PITA etc \[[@bib8]\]. However, it was found that all these algorithms showed a high false positive prediction rates and it was reported that up to 50% of the predicted targets are false positives \[[@bib21]\]. This indicates that microRNAs may function through other direct or indirect mechanisms. Interestingly, it was found that some microRNAs could bind to the 5′-UTR, the CDS or the promoter regions to even upregulate transcription or enhance translation \[[@bib22]\]. miR10a that binds to the 5′UTR, miR373 binding to promoter and miR34a binding to both 5′-UTR and 3′-UTR are all good examples for these nonclassical binding \[[@bib23], [@bib24], [@bib25]\]. What is more, some argue that microRNAs may target transcription factors and exert their function through a two-layer miR-TF-mRNA axis \[[@bib26], [@bib27], [@bib28], [@bib29], [@bib30]\]. All in all, microRNAs in human cells play important roles and they regulate gene expression through several different types of regulatory mechanisms. 3. Common single nucleotide polymorphisms (SNPs) in pre-microRNAs {#sec3} ================================================================= Single nucleotide polymorphisms frequently called SNPs, are the most common type of genetic variations in human. It was estimated that there are more than 100 million SNPs in human genome and about 7.5 million of them are common SNPs (SNPs with minor allele frequency (MAF) no less than 5%) \[[@bib31]\]. With limitations of samples size and statistical power, past case-control studies have also revealed the biological significance and functionality of common SNPs but not rare SNPs. To begin with, I first obtained all SNP data located in the pre-microRNAs from Ensemble and ranked them with MAF and number of related publications. In brief, there are in total 322 SNPs located in the microRNA precursor regions including 37 in the seed regions, 65 in the mature regions and the rest 220 in the precursor regions ([Supplementary Table 1](#appsec1){ref-type="sec"}). Surprisingly, only 16 of these common SNPs were well studied with more than 10 related publication up to date ([Table 1](#tbl1){ref-type="table"}), with only 1 located in the seed regions and 3 in the mature regions and 12 in precursors.Table 1List of well-studied SNPs located in the pre-microRNA regions.Table 1Gene nameVariant nameMinor allele frequencySNP positions\# of related publicationsMIR196A2rs116149130.332668in mature425MIR499Ars37464440.183506in seed289MIR499Brs37464440.183506in mature289MIR149rs22928320.386581in precursor157MIR27Ars8958190.363818in precursor136MIR608rs49195100.363818in mature82MIR423rs65051620.497804in precursor77MIR1908rs1745610.279553in precursor31MIR492rs22890300.114417in precursor25MIR605rs20435560.259585in precursor24MIR149rs714284390.14397in precursor15MIR449Brs100611330.122204in mature13MIR604rs23683920.323083in precursor12MIR1307rs79114880.303714in precursor12MIR618rs26828180.242412in precursor12MIR202rs123558400.31889in precursor11 3.1. miR196A2 rs11614913 {#sec3.1} ------------------------ rs11614913 located in the mature miR-196a-3p is the most widely studied microRNA SNP so far. It was first discovered as a genetic variant impacting patients' survival in non-small cell lung cancer \[[@bib32]\] and breast cancer \[[@bib33]\]. Since then, it has been implicated in various human cancers including breast, lung, prostate, colorectal, gastric, head and neck, hepatocellular carcinomas, etc \[[@bib34], [@bib35], [@bib36], [@bib37], [@bib38], [@bib39], [@bib40], [@bib41]\]. Scientists found this SNP influenced the pre-microRNA processing and maturation. Also, this position underwent somatic mutation in breast cancer in 14% of all patients and correlates with higher expression of miR196a \[[@bib42]\]. However, conflicting results from large case-control studies have been observed for this SNP and it seems that this SNP has population and cancer type specific effect. For example, C allele is correlated with high risk of hepatocellular carcinoma \[[@bib43]\] while T allele is correlated with high risk of Glioma, prostate cancer and colorectal carcinoma \[[@bib44], [@bib45], [@bib46]\]. In a meta-analysis of 46 studies involving 20673 cases and 25143 controls, it was discovered that C allele showed significant association with cancer risk in Asians but not in Caucasians \[[@bib38]\]. In another study, it was shown that rs11614913 T allele frequency is significantly low in African population compared to Non-African population \[[@bib47]\]. 3.2. miR499 rs3746444 {#sec3.2} --------------------- The miR499 locus encodes two microRNAs miR499a (transcribing from the forward strand) and miR499b (transcribing from the reverse strand). rs3746444 is located in the seed region of miR499a-3p while in the mature region of miR499b-5p. It was first discovered together with miR196a2 SNP rs11614913 from a case-control study as risk variants for breast cancer involving 1009 cases and 1093 controls \[[@bib33]\]. It also shows a population and caner type specific effects. In Chen 2014, T allele predicts risk for esophageal cancer while C allele for all other cancers \[[@bib48]\]. Mi et al. found that C allele is associated with increased cancer risk in Asian but not in Caucasians \[[@bib46]\]. 3.3. miR149 SNPs rs2292832 and rs71428439 {#sec3.3} ----------------------------------------- These two SNPs are the only common SNPs (MAF = 0.39 for rs2292832 and 0.14 for rs71428439) located in a single microRNA locus which are both well studied up to date. They are only 2 base pair away from each other and locates in the 3' end of the pre-miR149. Rs71428439 G allele showed reduced level of maturation of miR149 compared with the A allele \[[@bib49]\] and predicts high risk for lung cancer and hepatocellular carcinomas \[[@bib50],[@bib51]\]. rs2292832 was found to be related to risk of multiple cancers including colorectal cancer, liver cancer and breast cancer \[[@bib52], [@bib53], [@bib54], [@bib55]\] and in a recent study \[[@bib56]\], it was shown that this SNP may have a population specific effect in gastric cancer patients. While T allele carriers in Caucasians displayed decreased risk of gastric cancer, they have increased risk among Asians. 3.4. miR423 rs6505162 {#sec3.4} --------------------- rs6505162 is located in the precursor region of miR423 and was discovered to influence the maturation and expression of miR423 \[[@bib57],[@bib58]\]. It ranks \#1 in minor allele frequency (49.78%, see [Table 1](#tbl1){ref-type="table"}) of all the 322 SNPs located in the pre-microRNA regions, which indicates its high prevalence in human diseases related to miR423. Up to date, miR423 SNP rs6505162 has been implicated in multiple cancers including esophageal, breast, ovarian and colorectal cancer \[[@bib59], [@bib60], [@bib61]\]. Interestingly, there is a significant interaction between rs6505162 and age in esophageal cancers and the protective effect was more than twice among younger patients than older ones \[[@bib59]\]. Except for the above summarized four well studied miR SNPs and several others, biological and clinical significance for the majority of these genetic variants located in microRNA precursor regions remain unclear. Most interesting SNPs among these include multiple common SNPs located in a single microRNA locus (e.g. SNPs for miR6891, miR5189, miR4719, miR7150, miR6839, miR548AJ2 and miR5700) and SNPs showing high minor allele frequencies (e.g. rs2663345, rs71974827). Whether they influenced the maturation, expression of microRNAs or targets selection need to be revealed. From past studies, we could imagine these SNPs may also show population, gender, age and cancer type specific effects and researches on these could lead to significant progression in precision medicine. 4. Isomirs {#sec4} ========== microRNAs were typically annotated as a single defined sequence before Landgraf et al. \[[@bib62]\] have found that the same microRNA could have several variants which differ in length and sequence. The development of deep sequencing has enabled the identification of isomirs, which are microRNA variants that are heterogeneous in length or sequence. Importantly and interestingly, isomirs are not rare and was estimated to contribute to half of the miRnome in human cells and for each microRNA the contribution of isomirs to the expression levels varies from 0% to nearly 100% \[[@bib63],[@bib64]\]. Some isomirs may affect stability of microRNAs and also target selection \[[@bib65]\]. However, up to date, the functionality and biological and clinical significance of majority of these isomirs remain unclear and is yet to be revealed in the future. Isomirs could be classified to 4 main classes: 5′ isomiRs, 3′ isomiRs, polymorphic isomiRs and mixed type isomiRs and they were generated through multiple Drosha or Dicer cleavage, 5′ or 3′ trimming, nucleotide addition, removal or RNA editing \[[@bib66]\]. It was found that \[[@bib67]\] some isomiRs expression are more abundant and show superior performance as biomarkers than their canonical forms in cancer. Scientists then found that isomer expression could be used to classify breast cancer subtypes and some isomirs could be used as biomarkers in breast cancer and even showed improved performances compared with the traditional gene expression profiling \[[@bib68]\]. Recently, scientists used 50 5′ isomir as a panel and showed the expression of these isomirs could effectively classify tumors from 32 different tumor types with high sensitivity \[[@bib69]\]. Individual isomir have been validated to be involved in cancer progression and also have diagnostic values in cancer. For example, 5' isomir of miR140-3p suppress breast cancer proliferation and migration \[[@bib70]\], while miR196a isomirs are significantly different between cancer and normal tissues and is diagnostic in laryngeal cancer patients \[[@bib71]\]. Isomirs also showed gender or race specific expression patterns \[[@bib72]\]. One study from triple negative breast cancer revealed that a total of 21 isomiRs are differentially expressed between white patients and black patients including miR-183-5p isomiRs and they then verified each isomir has distinct effect in cancer cell through microarray analyses of the transcriptome \[[@bib73]\]. Later, they performed a more through analyses of isomirs and corresponding mRNA targets in triple negative breast cancer and discovered that genes from key metastasis linked pathways associated with isomirs in a race dependent manner \[[@bib74]\]. In another study which analyzed isomirs in prostate cancer TCGA cohort that contains 562 samples from 472 patients, scientists detected a total of more than 3000 isomiRs, of which 524 are differential between tumor and normal and a large portion (1519, \~50%) were specifically abundant in white patients \[[@bib75]\]. Aberrant expression of isomirs could be potential cancer biomarkers. In one study, it was reported that some isomirs could be predictive of prostate cancer metastasis \[[@bib76]\] and in another, isomirs expression in the urine extracellular vesicles could be used as a non-invasive method to detect prostate cancer \[[@bib77]\]. 5. Somatic mutations in microRNAs {#sec5} ================================= Somatic mutations also lead to single nucleotide variations in microRNAs. At present, there is one database SomamiR which included data from more than 600 samples in 50 different type of cancers and mapped \~2500 somatic mutations in microRNAs including \~1800 located in pre-microRNA regions, 644 in mature regions and 181 in seed regions \[[@bib78],[@bib79]\]. Up to date, there is still limited understanding of these mutations which could be due to overall low mutation rate and there are still no case control studies which contain enough number of samples to reach statistical significance for these mutations. One recently published paper in lung cancer \[[@bib80]\] analyzing the TCGA cohort which contain \~500 lung adenocarcinomas and \~500 lung squamous cell carcinomas identified more than 1000 mutation affecting \~500 microRNAs. They identified 10 significantly over mutated hotspot microRNA genes including miR379. Future studies which include more samples or focus on other cancer types will lead to more comprehensive understanding of these somatic mutations in microRNAs. 6. Concluding remarks {#sec6} ===================== In this review, I summarized the current knowledge of single nucleotide variations in human microRNAs including the single nucleotide polymorphisms with minor allele frequencies higher than 5%, the isomiRs, and somatic mutations in microRNA genes. A diagrammatic summary was shown in [Fig. 1](#fig1){ref-type="fig"}. In general, this field is not fully explored with too much emphasis on only several star SNPs. Majority of the common SNPs in microRNA regions are not studied, while it is only at the starting phase for the studies on isomiRs and microRNA somatic mutations. Considering the important role of microRNAs in gene regulation and their significance in human diseases, further studies are in need. This review also provided some hint for future studies in this field. For example, researches in the future should take into consideration whether there is age, gender, race/ethnic, tissue/cancer type or other relative factor(s) specific role for a given variation. Knowledge obtained from these studies will surely improve the understanding of microRNAs in human diseases and aid in further development of precision medicine.Fig. 1Diagrammatic summary of single nucleotide variations (SNVs) in microRNAs. SNVs can be classified into three categories: single nucleotide polymorphisms (SNPs), somatic mutations (somamiRs) and isomiRs. SNPs in pre-miRs, mature miRs or within the seed sequences may (1) influence the biogenesis/maturation of the mature miRs and cause change in 5p/3p ratios, (2) strengthen or reduce the binding affinity for the mature miRs to bind to their targets or (3) change miRs\' targetome through loss or gain of binding sites. SomamiRs, which are somatically mutated microRNAs, could be considered as a combination of all possible SNPs. IsomiRs are bona fide microRNA variants or isoforms that result from either addition, depletion/trimming or microRNA editing and usually show race, gender, population, and disease subtype dependencies. All of these SNVs modulate the intracellular microRNA-mRNA interaction network.Fig. 1 Declaration of competing interest ================================= There is no competing interest. Appendix A. Supplementary data {#appsec1} ============================== The following is the Supplementary data to this article:Multimedia component 1Multimedia component 1 Supplementary data to this article can be found online at <https://doi.org/10.1016/j.ncrna.2020.02.003>. [^1]: Present address: Department of Cancer Biology, Wake Forest University School of Medicine, Winston Salem, NC 27157, US.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1} =============== Oral squamous cell carcinoma represents a large, worldwide health burden as the eleventh most common cancer type with approximately 274,000 cases diagnosed annually worldwide \[[@B1], [@B2]\]. Carcinogenesis is a multistage process consisting of initiation, promotion, and progression phases \[[@B3]\]. Thus, the multistage sequence of events has many phases for prevention and intervention. Despite therapeutic advances in this disease, OCSCC is associated with severe morbidity with a 5-year overall survival rate of less than 50% \[[@B4]\] and a high occurrence of second primary tumors (20% to 30%) \[[@B5]--[@B7]\]. The overall 5-year survival rate of OCSCC has not significantly improved in the past 30 years \[[@B2]\]. Smoking and alcohol consumption are major inducers of OCSCC. The prevalence of OCSCC in smokers is 4--7 times higher than in nonsmokers, and if alcohol or chewing tobacco is augmented to the cigarette usage, the incidence of OCSCC is increased by 19- and 123-fold, respectively \[[@B8]\]. In recent years, the human papilloma virus (HPV), particularly HPV type 16, has also been suggested as an etiologic factor for OCSCC, especially among younger patients who are nonsmokers with no or minimal alcohol consumption \[[@B9], [@B10]\]. Patients diagnosed with OCSCC who have been continually exposed to smoking and alcohol are at a high risk for recurrences and second primary tumors (SPTs) \[[@B2]\]. Chemoprevention, a notion introduced by Sporn et al. \[[@B11]\], is the use of natural products or synthetic chemicals for the reversal, suppression, or prevention of the premalignant transformation of cells to a malignant form. In this respect, 13-cis- retinoic acid has been reported to be effective in the prevention of transformation of oral leukoplakia and dysplasia \[[@B12]\] and to stop the development of second primary tumors in patients with head and neck squamous cell carcinoma \[[@B5], [@B6]\]. However, critical problems with retinoids include toxicity and the recurrence of the leukoplakia after the cessation of treatment \[[@B12]\]. Thus, the investigation for a better chemopreventive agent is warranted. An ideal chemopreventive agent must have (i) little or no toxicity, (ii) a high efficacy in multiple sites, (iii) the capability for oral consumption, (iv) a known mechanism of action, (v) low cost, and (vi) human acceptance \[[@B13]\]. Currently, natural products, especially the antioxidants present in common food and beverages, have obtained great attention for cancer prevention owing to their various health benefits, noticeable lack of toxicity/side effects, and the limitations of other chemotherapeutic agents \[[@B14]\]. Several studies have shown that natural products composed of a wide spectrum of biologically active phytochemicals, including phenolics, flavonoids, carotenoids, alkaloids, and nitrogen, suppress the early and late stages of carcinogenesis \[[@B15]\]. Green tea is the most widely consumed beverage in the world and is especially popular in the Far East. Because of its abundant, scientifically proven, beneficial effects on human health, green tea has received considerable attention \[[@B16]\]. The polyphenols found in green tea have been documented to inhibit a variety of processes associated with cancer cell growth, survival, and metastasis \[[@B17]\]. Here, we highlight the evidence of green tea\'s role in oral cancer chemoprevention with regard to multiple molecular mechanisms proposed in various in vitro, in vivo, and clinical trials. 2. Green Tea and Polyphenols {#sec2} ============================ Green teas are derived from Camellia sinensis and represent the most consumed beverages in the world, next to water \[[@B18]\]. Green tea contains polyphenols, glycosides, leucoanthocyanins, and phenol acid \[[@B19]\], with the polyphenols constituting 36% of the fresh green tea leaf dry weight \[[@B20]\]. In a randomized controlled trial (RCT), green tea was documented to significantly reduce the level of total serum cholesterol and low density lipoprotein (LDL) cholesterol in the green tea group, which was treated with 150 mg green tea catechins and 150 mg of other tea polyphenols a day for 3 months \[[@B21]\]. A meta-analysis showed that an increase in tea consumption of three cups (711 mL/day) decreased the risk of a myocardial infarction by 11% \[[@B22]\]. Epidemiological studies have also shown that the intake of green tea may reduce the risk of malignancy \[[@B23], [@B24]\]. In 1997, Imai and colleagues published a prospective cohort survey of more than 8,552 Japanese individuals that demonstrated the ingestion of more than 10 daily cups (120 mL each) of green tea decreased the development of cancer \[[@B25]\]. Additionally, in 2009, Myung and colleagues performed a meta-analysis of epidemiologic studies that found green tea consumption to have a preventative effect on stomach cancer \[[@B26]\]. Seely et al. also reported that green tea consumption may possibly help prevent breast cancer recurrence in early-stage (I and II) cancers \[[@B27]\]. In many mouse skin tumor models, protection against UVB-induced skin carcinogenesis and the inflammatory response was obtained with topical application or oral consumption of green tea \[[@B28]--[@B31]\]. Green tea contains four major polyphenols: epicatechin (EC), epigallocatechin (EGC), epicatechin-3-gallate (ECG), and epigallocatechin-3-gallate (EGCG), composing 1--3%, 3--6%, 3--6%, and 7--13%, respectively, of the fresh green tea leaf dry weight \[[@B20], [@B32]\]. Numerous studies have reported that green tea polyphenols have plentiful beneficial effects on human health, including not only having antiviral, anti-inflammatory, and antiallergic effects, but also aiding in weight control \[[@B33], [@B34]\]. It has also been documented that polyphenols inhibit tumorigenesis in a variety of organs, including skin, lung, oral cavity, esophagus, stomach, small intestine, colon, liver, pancreas, ovary, and mammary gland \[[@B35]--[@B39]\]. The polyphenol EGCG has specifically been documented to have chemopreventive effects against various cancers \[[@B40]--[@B45]\], and one cup of green tea holds 200 mg of EGCG \[[@B46]\]. EGCG has been reported to have multiple biologic functions, including preventing the formation of blood vessels (angiogenesis) and organizing blood vessel permeability to hinder the blood supply to cancerous cells \[[@B47], [@B48]\]. Adding EGCG to cells leads to decreased phosphorylation of the epidermal growth factor receptor (EGFR), Akt, and Stat3, without regulating protein levels \[[@B44], [@B49]\]. EGCG has also been shown to inhibit telomerase activity and lead to telomerase fragmentation \[[@B50]\]. Green tea polyphenols regulate multiple points controlling cancer cell growth, survival, and metastasis, with effects at the DNA, RNA, and protein levels \[[@B18]\]. Desirable green tea ingestion is known to be 3 to 5 cups per day (up to 1200 mL/day), providing a minimum of 250 mg/day catechins \[[@B51]\]. 3. The Chemopreventive Effect of Green Tea Polyphenols on Oral Cancer: In Vitro Studies {#sec3} ======================================================================================= Numerous in vitro studies have shown the antiproliferative effect and apoptosis inducing capability of green tea polyphenols in oral cancer cell lines. Jeffrey et al. have reported that the antiproliferative effects of green tea polyphenols and EGCG were more noticeable towards oral cancer cell lines (CAL27, HSC-2, and HSG1) than normal fibroblasts (GN56 and HGF-1) \[[@B52]\]. The application of EGCG to cell culture medium resulted in the formation of hydrogen peroxide. This result was consistent with EGCG performing as a pro-oxidant because tumor cells are known to be more vulnerable to oxidative forces than normal cells \[[@B52], [@B53]\].An in vitro, multistage tumorigenesis model for oral cancer showed that cell proliferation was arrested at all stages of carcinogenesis by EGCG with the best efficacy demonstrated in dysplastic cells \[[@B54]\]. A study by Masuda and coworkers \[[@B49]\] demonstrated that treatment with EGCG increased the proportion of cells in the G~1~ phase of the cell cycle and induced apoptosis. Moreover, in cells treated with EGCG, a reduction in the cyclin D1 protein and in the hyperphosphorylated form of pRB was observed \[[@B49]\]. Liu et al. showed that green tea extract and epigallocatechin-3 gallate inhibited the growth of 3 squamous cell lines (CAL-27, SCC-25, and KB) via S and G~2~/M phase arrest. The major signaling cascades influenced by green tea extract and EGCG were shown to be the EGFR and Notch pathways, which in turn, affected cell-cycle-related networks \[[@B55]\]. Hsu et al. reported that treatment of caspase 3 wild-type oral carcinoma cell lines with EGCG resulted in a gradual decrease of mitochondrial function down to an insignificant level, but caspase 3 null cells did not undergo apoptosis. This result demonstrates that green tea polyphenol-associated apoptosis is mitochondria targeted and caspase 3 dependent \[[@B56]\]. EGCG was also found by cDNA microarray to upregulate p21WAF1 in the OSC2 oral cancer cell line, which may facilitate caspase 3-mediated apoptosis \[[@B57]\]. The overexpression of HER-2 (neu/erbB2) is correlated with a poor prognosis in patients with breast cancer or oral squamous cell carcinoma, presumably due to an increased metastatic ability and resistance to various cancer chemotherapies \[[@B58], [@B59]\]. Masuda et al. \[[@B45]\] reported that treatment of head and neck squamous cell carcinoma cell lines (HNSCC) (YCU-H891, YCU-N861) with EGCG resulted in a 50% inhibition of growth and a marked inhibition of HER-2 phosphorylation. This outcome was connected with the prevention of Stat3 activation as well as the inhibition of c-fos and cyclin D1 promoter activity. Recent studies have also demonstrated that RECK methylation is associated with enhancing metastasis and invasion in human cancers \[[@B60]--[@B62]\]. Chen et al. illustrated that the invasion, motility, migration, and secretion of MMP-2 and u-PA in SCC-9 oral cancer cells, through attenuation of p-FAK and p-Src, could be significantly inhibited by epigallocatechin-3 gallate \[[@B63]\]. Long et al. reported that hypermethylation of the RECK gene is associated with a poor prognosis in oral squamous cell carcinoma \[[@B64]\]. A study by Kato et al. showed the treatment of oral cancer cells with EGCG partially reversed the hypermethylation status of the RECK gene and significantly increased the expression level of RECK mRNA \[[@B65]\]. EGCG was also found to reduce the expression of MMP-2 and MMP-9 \[[@B16], [@B65]\]. Taken together, these outcomes demonstrate that green tea polyphenols may reduce the invasion and migration of human oral cancer cells. The concept of combination chemoprevention pursues the ability to enhance the chemopreventive efficacy of both agents (synergism), while reducing side effects by dose reduction. Synergistic interactions were shown between beta-carotene and alpha-tocopherol in an in vivo hamster cheek pouch tumorigenesis model \[[@B66]\] as well as between beta-carotene and anticancer alkylating agents in an in vitro study that utilized tongue squamous carcinoma cells \[[@B67]\]. Khafif et al. documented an interactive synergistic effect of EGCG and curcumin treatment. Specifically, EGCG arrested malignant human oral epithelial cells in G~1~, whereas curcumin blocked cells in S/G~2~M \[[@B68]\]. The potency of the cytotoxic and apoptotic effects on human tongue squamous carcinoma cells was greater with a combination of lacroferrin and tea polyphenols than with polyphenols alone \[[@B69]\]. A study by Amin et al. showed that EGCG and erlotinib (epidermal growth factor receptor tyrosine kinase inhibitor) had a synergistic growth-inhibitory effect in a nude mouse xenograft model of squamous cell carcinomas of the head and neck via inhibition of the nuclear factor-κB signaling pathway \[[@B70]\]. EGCG was found to protect normal salivary gland cells from the effects of gamma irradiation and the chemotherapy drug cis-platinum(II)diammine dichloride (CDDP), suggesting that the combination of green tea consumption with chemotherapy or radiotherapy could be a promising avenue of treatment \[[@B71]\]. 4. The Chemopreventive Effect of Green Tea Polyphenols on Oral Cancer: In Vivo Studies {#sec4} ====================================================================================== Experimental carcinogenesis models are valuable tools to investigate the multistep characteristics of carcinogenesis and to study various modulations that intervene in the development of cancer. These models also have many advantages over simple in vivo studies. The hamster buccal pouch carcinogenesis model (HBP) has been extensively employed to investigate the chemopreventive effectiveness of medical plants and dietary agents \[[@B72]\]. Treatment with 7,12-dimethylbenz\[a\]anthracene (DMBA) induced HBP carcinoma and upregulated genotoxicity. Inhibition of DMBA-induced HBP carcinomas by green tea polyphenols was observed, and polyphenol treatment was associated with a significant reduction in phase I enzymes, regulation of lipid peroxidation, and increased antioxidant and phase II enzyme activities in this model \[[@B73]\]. DMBA-induced oral carcinogenesis was arrested by EGCG in hamsters showing a lower risk of dysplasia and oral carcinoma \[[@B74]\]. 4-nitroquinoline 1-oxide (4-NQO) is also known to induce multistep carcinogenesis \[[@B75]\]. Srinivasan et al. reported a significant reduction in the number of tumors, tumor volume, and oral squamous cell carcinomas in green tea polyphenol treated rats relative to 4-NQO induced animals. Additionally, application of green tea polyphenols in this study was shown to enhance the activity of phase II enzymes (glutathione-S-transferase and UDP-glucuronyl transferase) and reduce the activity phase I enzymes (cytochrome b5, cytochrome P450, cytochrome b5 reductase, cytochrome P450 reductase, aryl hydrocarbon hydroxylase, and DT-diaphorase) \[[@B76]\]. EGCG was also sufficient to stop phorbol-12-myristate-induced cell invasion and matrix metalloproteinase-9 expression, as demonstrated by the inhibition of tumor growth observed with EGCG treatment in SCC-9 cells in vivo via a cancer cell xenograft nude mouse model \[[@B63]\]. Recent studies have also reported that amyloid precursor protein (APP) is upregulated in pancreatic cancer cells and SW837 colon carcinoma cells both in vitro and in vivo \[[@B77], [@B78]\]. Ko et al. demonstrated that APP was significantly upregulated in MBN-induced HBP carcinomas but was significantly decreased by tea intake \[[@B79]\]. Moreover, APP expression and secretion from oral squamous cell carcinoma was arrested by the application of green tea polyphenols in a dose-dependent manner \[[@B79]\]. 5. The Chemopreventive Effect of Green Tea Polyphenols on Oral Cancer: Clinical Trials and a Prospective Cohort Study {#sec5} ===================================================================================================================== Schwartz et al. performed a pilot study in which green tea total extracts (2000--2500 mg/day) were administrated in drinking water to smokers for four weeks \[[@B80]\]. The study found that during the course of green tea administration, smoking-induced DNA damage was reduced, cell growth was inhibited, the percentage of cells in S phase was decreased, cells accumulated in G~1~ phase, DNA content became more diploid and less aneuploid, and markers of apoptosis were upregulated \[[@B80]\]. A double-blind intervention trial was performed in patients with oral mucosa leukoplakia using mixed tea extracts \[[@B81]\]. After 6 months, the oral leukoplakia was reduced in size in 40% of the patients who ingested tea extracts. In a phase II randomized, placebo-controlled clinical trial of green tea extract (500, 750, or 1,000 mg/m^2^) in high risk oral premalignant lesions (OPL), the OPL clinical response rate was higher in all green tea extract (GTE) arms versus placebo but did not reach statistical significance \[[@B82]\]. However, higher-dose GTE arms displayed higher responses and demonstrated a dose-dependent effect \[[@B82]\]. Moreover, stromal VEGF and cyclin D1 expressions were reduced in clinically responsive GTE patients and increased in nonresponsive patients at 12 weeks \[[@B82]\]. Ide and colleagues conducted a prospective nationwide, large-scale cohort study in Japan. A total of 20,550 men and 29,671 women aged 40--79 years without any history of oral cancer were included \[[@B83]\]. During a mean follow-up period of 10.3 years, 37 oral cancer cases were identified. For women, the hazard ratios (HRs) of oral cancer for a green tea consumption of 1-2, 3-4, or 5 or more cups per day, were 0.51, 0.60, and 0.31, respectively, when compared with individuals who drank less than one cup per day. However, for men, no such trends were identified \[[@B83]\]. 6. Conclusion {#sec6} ============= This paper summarized the chemopreventive efficacy of green tea polyphenols in various in vitro and in vivo oral cancer models. All in vitro and in vivo studies demonstrated that green tea can regulate numerous molecular pathways involved in cancer promotion and progression. The epidemiology and multistep carcinogenesis of oral cancer suggests that it is a preventable cancer. Prevention is the most ideal strategy to save lives and reduce the morbidity of radical surgery. Although small trials have shown promising results, chemoprevention clinical trials with green tea performed in oral cancer are very limited. Substantial clinical trials are required to investigate the chemopreventive effectiveness of green tea polyphenols against oral cancer either alone or in combination with other chemopreventive agents. To seek a fundamental association between the consumption of green tea and a reduced rate of oral cancer development or mortality, more randomized clinical trials (RCTs) and cohort studies are needed. [^1]: Academic Editor: A. Sapino	{ "pile_set_name": "PubMed Central" }
急性髓系白血病（AML）具有较明显的异质性，基因突变导致细胞增殖、分化及凋亡途径的改变是AML的发病基础[@b1]。FLT3（Fms-related tyrosine kinase 3）是Ⅲ型酪氨酸激酶受体家族成员之一，在早期造血祖细胞的增殖、分化方面起重要作用[@b2]。FLT3有两个有临床意义的突变形式：近膜区的内部串联重复（internal tandem duplication，ITD）和酪氨酸激酶结构域（tyrosine kinase domain，TKD）活化环内的点突变。FLT3-ITD突变在成人AML中的发生率约为30%，FIT3-ITD突变阳性AML患者通常具有外周血白细胞计数及骨髓原始细胞比例较高、临床预后差、易复发等临床特征[@b3]，与AML的不良预后相关[@b4]。FLT3-ITD阳性AML患者allo-HSCT后3年总生存（OS）率可达58%[@b5]。但FLT3-ITD阳性AML患者allo-HSCT后很快复发，生存时间较短。研究显示，FLT3-ITD突变容易诱发DNMT3A突变的产生，而且这种双突变的AML患者诱导化疗的完全缓解（CR）率明显偏低，复发率较高[@b6]。在临床中，我们对于这种双突变的AML患者通常在强烈诱导化疗得到CR后尽快行allo-HSCT，争取尽量延长生存时间和改善生存质量。但这种双基因突变的AML患者能否通过allo-HSCT达到延长生存时间和改善生存质量的目的，鉴于目前对于这种基因双突变AML的移植预后情况鲜有报道，我们回顾性分析了该类患者的临床资料，报道如下。病例与方法 {#s1} ========== 1．病例：本研究纳入2010年1月30日至2014年12月31日在苏州大学附属第一医院和宁波市鄞州人民医院接受allo-HSCT且保留初诊骨髓标本的AML患者206例（M~3~和使用分子靶向药物索拉菲尼的患者除外），其中男104例，女102例，中位年龄38（3\~63）岁。FAB分型：M~0~ 6例，M~1~ 24例，M~2~ 56例，M~4~ 39例，M~5~ 63例，M~6~ 6例，不能分类12例。所有患者均知情同意。 2．突变基因组套检测：患者骨髓标本采用Ficoll分离骨髓单个核细胞，采用Progma公司genomic DNA Purification Kit抽提DNA。ABI公司Hot startaq酶PCR扩增c-kit基因8、17号外显子、NPM1基因12号外显子、FLT3基因20号外显子、DNMT3A基因第882位突变热点区域、CEBPA基因全部外显子。PCR条件：95 °C 5 min；95 °C 30 s，58 °C 30 s，72 °C 90 s，35个循环；72 °C 7 min。PCR产物采用ABI-3730型基因分析仪测序进行突变检测（由苏州大学附属第一医院血液病研究所完成）。 3．化疗：所有患者入院后均接受"3+7"标准诱导方案化疗：去甲氧柔红霉素8\~12 mg·m^−2^·d^−1^第1\~3天，阿糖胞苷（Ara-C）100 mg·m^−2^·d^−1^第1\~7天。诱导化疗获得CR后，对于年龄\<60岁患者，予FA（氟达拉滨30 mg·m^−2^·d^−1^第1\~5天，Ara-C 1\~2 g·m^−2^·d^−1^第1\~5天）或含中剂量Ara-C（1\~2 g/m^2^每12 h 1次第1\~4天）为基础的强化方案巩固化疗1\~2个疗程，有合适供者的患者行allo-HSCT；对于年龄≥60岁的患者，以含标准剂量Ara-C（75\~100 mg·m^−2^·d^−1^×5\~7 d）为基础的方案巩固4\~6个疗程，有合适供者的患者行allo-HSCT。复发/难治患者选用FLAG（氟达拉滨+Ara-C+G-CSF）、MAE（米托蒽醌+ Ara-C+依托泊苷）等挽救性化疗方案，获得CR后如有合适供者尽快行allo-HSCT。 4．allo-HSCT：单倍型造血干细胞移植（haplo-HSCT）采用骨髓加外周血（或脐血）干细胞移植，亲缘全相合以及无关全相合采用外周血干细胞移植。输注细胞数量：单个核细胞（6\~8）×10^8^/kg，CD34^+^细胞（3\~4）×10^6^/kg。16例存在重要脏器功能障碍的患者应用减低强度预处理方案，其余患者均使用改良Bu/Cy预处理方案。接受同胞HLA全相合移植的患者使用甲氨蝶呤（MTX）联合环孢素A（CsA）预防GVHD。接受无关HLA全相合移植和haplo-HSCT的患者使用抗胸腺细胞球蛋白（ATG）、MTX、CsA及霉酚酸酯（MMF）预防GVHD。 5．随访：采用门诊定期随访、问卷及电话随访，随访截止至2015年12月31日。206例患者中位随访时间25（1\~58）个月。随访研究终点为无白血病生存时间（LFS）及OS时间。 6．统计学处理：四组病例临床资料的比较采用卡方检验和方差分析，OS、LFS的比较采用Kaplan-Meier法，累积复发率的比较采用R软件处理。P值\<0.05为差异有统计学意义。结果 {#s2} ==== 1．初诊时突变基因的检出情况：接受allo-HSCT的206例AML患者初诊时突变基因检出结果见[表1](#t01){ref-type="table"}。按突变基因检出率排序，前5位依次为FLT3-ITD^+^（12.6%），CEBPA双突变（duCEBPA）（9.7%），DNMT3A R882^+^FLT3-ITD^+^NPM1^+^（6.8%），DNMT3A R882^+^（6.8%），CEBPA单突变（siCEBPA）^+^（4.4%），其中DNMT3A中的突变点全部在R882位点，未发现DNMT3A非R882突变，其中82例患者未检出任何已知突变。 ###### 接受allo-HSCT的206例AML患者初诊时突变基因的检出情况基因突变例数（%） ---------------------------------- ------------ FLT3-ITD 26(12.6) duCEBPA 20(9.7) DNMT3A R882+FLT3-ITD+NPM1 14(6.8) DNMT3A R882 14(6.8) siCEBPA 9(4.4) FLT3-ITD+NPM1 7(3.4) FLT3-ITD+duCEBPA 4(1.9) FLT3-TKD 4(1.9) DNMT3A R882+FLT3-ITD 4(1.9) c-kit 3(1.5) NPM1 3(1.5) duCEBPA+c-kit 2(1.0) DNMT3A R882+NPM1 2(1.0) siCEBPA+NPM1 2(1.0) DNMT3A R882+NPM1+ FLT3-TKD 2(1.0) FLT3-ITD+siCEBPA 1(0.5) duCEBPA+ FLT3-TKD 1(0.5) NPM1+ c-kit 1(0.5) DNMT3A R882+ FLT3-TKD 1(0.5) FLT3-TKD+ NPM1 1(0.5) DNMT3A R882+NPM1+siCEBPA 1(0.5) DNMT3A R882+FLT3-ITD+NMP1+c-kit 1(0.5) DNMT3AR882+NPM1+siCEBPA+FLT3-TKD 1(0.5) 阴性组 82（39.8）注：duCEBPA：双突变CEBPA；siCEBPA：单突变CEBPA；阴性组：未检出任何已知突变 2．基因突变情况及临床转归：将全部206例患者根据初诊时的突变基因情况分为4组：①FLT3-ITD^+^ DNMT3A R882^+^组19例（A组）；②FLT3-ITD^+^ DNMT3A R8 82^−^ 组3 8例（B组）；③FLT3-ITD^−^ DNMT3A R882^+^组21例（C组）；④FLT3-ITD^−^ DNMT3A R882^−^组128例（D组）。截至随访终点，A组有12例（63.2%）复发，12例（63.2%）死亡，其中5例（26.3%）死于疾病复发，4例（21.1%）死于重度GVHD，2例（10.5%）死于重症感染，1例（5.3%）死于意外事故；B组有14例（36.8%）复发，13例（34.2%）死亡，其中6例（15.8%）死于疾病复发，3例（7.9%）死于重症感染，3例（7.9%）死于重度GVHD，1例（2.6%）死于重症肝炎；C组有6例（28.6%）复发，6例（28.6%）死亡，其中4例（19.0%）死于疾病复发，2例（9.5%）死于重度GVHD；D组有26例（20.0%）复发，28例（21.5%）死亡，其中15例（11.7%）死于疾病复发，6例（4.7%）死于重症感染，4例（3.1%）死于重度GVHD，2例（1.6%）死于重症肝炎，1例（0.8%）死于全身脏器功能衰竭。各组性别、年龄、初诊时白细胞计数、FAB分型、移植前疾病状态、染色体核型、供者类型、预处理方案及GVHD发生率差异无统计学意义（[表2](#t02){ref-type="table"}）。 ###### 不同基因突变急性髓系白血病患者的临床资料指标 A组（19例） B组（38例） C组（21例） D组（128例）统计量 P值 ----------------------------------- ------------------- ------------------ ------------------ ------------------ ------------------ ------- 性别（例，男／女） 8／11 15／23 7／14 74／54 7.598（χ^2^值） 0.055 年龄［岁，M（范围）］ 41(21～59) 40(15～58) 47(27～60) 35(15～57) 1.703（F值） 0.087 初诊WBC［×10^9^／L，M（范围）］ 38.0(10.8～279.0) 35.0(1.0～403.0) 30.0(1.9～121.0) 13.0(0.7～307.0) 2.069（F值） 0.051 FAB分型（例） 20.388（χ^2^值） 0.311 M~0~ 0 2 0 4 M~1~ 0 3 0 21 M~2~ 5 9 7 35 M~4~ 4 9 4 22 M~5~ 8 12 10 33 M~6~ 0 0 0 6 未分类 2 3 0 7 移植前疾病状态（例） 1.680（χ^2^值） 0.641 CR~1~ 14 27 18 98 ≥CR~2~ 5 11 3 30 染色体核型（例） 7.713（χ^2^值） 0.052 中危 19 35 20 105 高危 0 3 1 23 供者类型（例） 1.261（χ^2^值） 0.738 全相合 11 23 14 87 单倍型 8 15 7 41 预处理方案（例） 0.390（χ^2^值） 0.942 改良Bu／Cy 17 35 19 119 RIC 2 3 2 9 急性GVHD（例） 0.223（χ^2^值） 0.974 Ⅰ＋Ⅱ度 4 11 5 31 Ⅲ＋Ⅳ度 3 6 3 16 慢性GVHD（例） 0.488（χ^2^值） 0.921 局限型 3 5 2 16 广泛型 1 4 1 9 注：A组：FLT3-ITD^+^ DNMT3A R882^+^；B组：FLT3-ITD^+^ DNMT3A R882^−^；C组：FLT3-ITD^−^DNMT3A R882^+^；D组：FLT3-ITD^−^DNMT3A R882^−^。CR：完全缓解；CR~1~：第1次完全缓解；≥CR~2~：第2次及以上完全缓解。Bu/Cy：白消安+环磷酰胺；RIC：减轻强度预处理 3．伴有FLT3-ITD及DNMT3A R882双突变AML患者的临床特征：FLT3-ITD^+^ DNMT3A R882^+^组共19例，初诊时中位WBC 38.0（10.8\~279.0）×10^9^/L；FAB分型：M~2~ 5例，M~4~ 4例，M~5~ 8例，不能分类2例；18例染色体核型正常，1例染色体核型为47,XY,+21\[3\]/46,XY\[7\]；除4例为FLT3-ITD+DNMT3A R882双突变外，14例合并NPM1突变（FLT3-ITD+DNMT3A R882+NPM1），1例合并NPM1+c-kit突变（FLT3-ITD+DNMT3A R882+NPM1+c-kit）；第1疗程诱导化疗达CR的有10例；移植前状态：CR~1~ 14例，CR~2~ 5例；allo-HSCT后2例患者发生Ⅲ度急性GVHD，1例发生Ⅳ度急性GVHD。allo-HSCT后6个月内复发7例，12个月内复发9例。allo-HSCT后6个月内死亡5例，1年内死亡9例。 4．各基因突变组复发及生存比较：A组2年累积复发率\[（72.2±2.6）%\]高于B组\[（38.6±0.6）%\]、C组\[（36.8±1.6）%\]（P=0.018，P=0.026），2年OS率\[（30.9±13.3）%\]低于B组\[（67.5±7.8）%\]、C组\[（61.4±12.4）%\]（χ^2^=6.704，P=0.010；χ^2^=4.860，P=0.027），2年LFS率\[（11.3±10.20）%\]亦低于B组\[（47.9±8.4）%\]、C组\[（56.8±12.5）%\]（χ^2^=4.926，P=0.026；χ^2^=6.469，P=0.011）。B组、C组的2年累积复发率均高于D组\[（38.6±0.6）%对（27.8±0.1）%，P=0.032；（36.8±1.6）%对（27.8±0.1）%，P=0.044\]，2年OS率低于D组\[（67.5±7.8）%对（80.1±3.7）%，χ^2^=4.606，P=0.032；（61.4±12.4）%对（80.1±3.7）%，χ^2^=4.301，P=0.044\]，LFS率亦低于D组\[（47.9±8.4）%对（79.7±3.6）%，χ^2^=17.605，P\<0.001；（56.8±12.5）%对（79.7±3.6）%，χ^2^=5.538，P=0.019\]。各组生存及复发曲线见[图1](#figure1){ref-type="fig"}。 ![FLT3-ITD与DNMT3A R882双突变对急性髓系白血病allo-HSCT后复发（A）、总生存（B）和无白血病生存（C）的影响\ A组：FLT3-ITD^+^ DNMT3A R882^+^；B组：FLT3-ITD^+^ DNMT3A R882^−^；C组：FLT3-ITD^−^DNMT3A R882^+^；D组：FLT3-ITD^−^DNMT3A R882^−^](cjh-39-07-552-g001){#figure1} 讨论 {#s3} ==== FLT3-ITD和DNMT3AR882都是AML预后不良的因素，allo-HSCT可以改善分别伴有这两种基因突变的AML的预后[@b7]--[@b8]，然而allo-HSCT是否可以改善伴有FLT3-ITD和DNMT3AR882双突变的AML的预后，目前鲜有报道。为此我们分析了19例接受过allo-HSCT的双突变的AML患者的临床资料，发现双突变组移植后无论OS率还是LFS率均较低，而累积复发率率高，移植预后恶劣。 FLT3是Ⅲ型酪氨酸激酶受体家族成员之一，在早期造血祖细胞的增殖、分化方面起重要的作用，具有FIT3-ITD突变的AML临床预后较差[@b9]。FLT3-ITD阳性AML患者诱导化疗CR率与FLT3-ITD阴性AML患者相似[@b10]，但获得CR后即使给予高剂量Ara-C巩固化疗也易发生早期复发[@b11]。Lin等[@b7]报道，伴有FLT3-ITD阳性AML患者接受allo-HSCT后可以获得与FLT3-ITD阴性AML患者相似的OS率，不接受allo-HSCT者OS时间明显缩短。然而Ahn等[@b12]报道，FLT3-ITD阳性组AML患者的移植预后较FLT3-ITD阴性组差。本研究结果也提示FLT3-ITD^+^ DNMT3A R882^−^组移植后2年OS率及LFS率均低于FLT3-ITD^−^ DNMT3A R882^−^组，2年累积复发率高于FLT3-ITD^−^ DNMT3A R882^−^组。可见FLT3-ITD突变是AML预后不良的指标。 DNMT3A是近年来在AML患者中新发现的基因突变。DNMT3A阳性提示AML预后不良[@b13]。以往研究结果显示，诱导化疗获得CR后桥接allo-HSCT DNMT3A阳性AML患者的3年OS率、无病生存率优于强化巩固化疗组[@b8]。Ribeiro等[@b14]研究显示，DNMT3A是AML不良预后的独立因素。 DNMT3A的热点突变是R882，少部分还包括移码突变、无义突变等，这些非R882突变的临床预后与R882突变存在差异，Gaidzik等[@b15]报道R882突变和非R882突变有相反的临床预后，R882突变提示较低的无复发生存率，然而非R882突变提示更好的OS率。Ahn等[@b12]比较了DNMT3A R882阳性组和非R882 DNMT3A组及野生型DNMT3A三组患者allo-HSCT后的预后情况，发现DNMT3A R882阳性组与另外两组比较，allo-HSCT后5年OS率、无事件生存率较低，复发率较高。本研究DNMT3A R882阳性组（C组）与DNMT3A R882阴性组（D组）比较，allo-HSCT后2年OS率及LFS率较低，2年累积复发率较高。R882位点突变导致AML不良预后的具体机制尚不清楚，可能的原因是DNMT3A R882对DNA甲基化转移酶的抑制作用更强[@b16]。 FLT3-ITD阳性AML患者中DNMT3A突变的检出率明显增高，有报道显示在白血病复发时，表观遗传学修饰基因突变（DNMT3A）容易诱发FLT3-ITD突变[@b17]；还有报道FLT3-ITD阳性AML患者中DNMT3A突变的检出率达41.7%，而FLT3-ITD阴性AML患者中DNMT3A突变的检出率只有23.7%，而且这种双突变AML患者诱导化疗的CR率明显偏低，复发率较高，allo-HSCT的疗效也很差[@b3]。Ahn等[@b12]对115例诱导化疗获CR后桥接allo-HSCT的AML患者进行了回顾性研究（其中伴有DNMT3A R882和FLT3-ITD双突变11例），结果显示DNMT3A R882和FLT3-ITD双突变是AML患者移植后不良预后的独立因素，相比其他突变，这种双突变患者移植后的5年OS率（9.1%）、无事件生存率（9.1%）明显降低，复发率（81.8%）明显升高。本研究中FLT3-ITD和DNMT3A R882双突变AML患者移植后2年OS率、LFS率、累积复发率分别为（30.9±13.3）%、（11.3±10.2）%、（72.2±2.6）%，相比FLT3-ITD和DNMT3A R882单突变组差异有统计学意义。综上所述，本研究显示，伴有FLT3-ITD和DNMT3A R882双突变的AML患者allo-HSCT后2年累积复发率偏高、OS率和LFS率偏低，有待扩大样本量、延长随访时间继续进行探索。	{ "pile_set_name": "PubMed Central" }
Introduction ============ Enterohemorrhagic Escherichia coli O157:H7 (EHEC O157:H7) is a zoonotic enteric pathogen that causes hemorrhagic colitis and hemolytic uremic syndrome (HUS) in humans ([@B24]; [@B5]). EHEC O157:H7 can intimately adhere to intestinal epithelial cells and form characteristic attaching and effacing (A/E) lesions, and antibiotic treatment can potentially increase HUS risk ([@B7]; [@B33]). It is presumed that a vaccine that prevents EHEC from colonizing the intestinal tract would be the most effective strategy for preventing infection ([@B22]). Currently, two vaccines are used for the same purpose in cattle, but neither has been approved for humans ([@B20]). For achieving this goal many hurdles need to be addressed, such as the type or subset of antigens, adjuvant, and the delivery route. The virulence of EHEC O157:H7 can be largely attributed to its toxins, type III secretion system (T3SS) proteins, and surface fimbrial and afimbrial adhesions ([@B13]). The T3SS delivers espA, espB, and espD directly to the host cells, which are essential for signal transduction and A/E lesion formation ([@B11]). The major component, espA forms a bridge to host cells for direct delivery of other virulence factors ([@B21]). Tir is one of the first proteins translocated during infection and functions as a receptor for the EHEC outer membrane protein, intimin ([@B9]). The T3SS proteins, espA and Tir thus play key roles in EHEC adhesion, pedestal formation, and pathogenicity. They may also serve as immunogens for the disruption of bacterial-host cell interaction and prevent the formation of A/E lesions. Lactic acid bacteria (LAB) such as Lactobacillus acidophilus (L. acidophilus), Lactococcus lactis, and Bifidobacterium have been given a "generally regarded as safe" (GRAS) designation and are widely used in human food production. Additionally, LAB's are considered as a useful tool for the development of novel oral vectors and have largely replaced attenuated pathogens with mucosal delivery strategies ([@B6]). Moreover, LAB can also be used as delivery vectors for subunit vaccines, which circumvents costly purification processes ([@B32]). Vaccines employing a mucosal delivery system can elicit both antigen-specific secretory immunoglobulin A (sIgA) and effective systemic immune responses. The relatively low cost, ease of administration, and degree of mucosal tissue protection make LAB more advantageous as vaccine delivery vectors ([@B27]). Lactococcus lactis is most often used for EHEC and other pathogen vaccines. Lactococcus lactis-expressing EspB induces protective immunity against EHEC O157:H7 and recombinant Lactococcus lactis induces protection against C. difficile ([@B2]; [@B16]). However, there are few reports describing the use of L. acidophilus as a vaccine delivery vector that has been approved for use in infant and baby food. L. acidophilus is particularly promising as an oral vaccine vector because it is acid- and bile-tolerant, can express mucus-binding proteins that enable association with intestinal mucosa and regulates immature dendritic and T cell functions ([@B18]). To date, there have been no studies of recombinant L. acidophilus producing the EHEC O157:H7 protective antigen as a candidate vaccine. In this study, we developed a candidate EHEC O157:H7 vaccine by inserting the bivalent antigen espA-Tir-M (ET), which is composed of espA and the Tir central domain (Tir-M), into L. acidophilus. We evaluated the specific immune responses elicited in mice and protection against EHEC O157:H7 challenge both in vitro and in vivo. Materials and Methods {#s1} ===================== Bacteria, Plasmids, and Cells ----------------------------- Lactobacillus acidophilus ATCC4356 (American Type Culture Collection, Manassas, VA, USA) was cultured in de Man, Rogosa, and Sharpe (MRS) agar or broth at 37°C in an atmosphere of 5% CO~2~. EHEC O157:H7 strain EDL933 (our laboratory stock) was grown at 37°C in Luria--Bertani (LB) broth supplemented with 5 g/l streptomycin. The pMG36e plasmid (our laboratory stock) was maintained in DH5α cells and cultured in LB broth with 200 μg/ml erythromycin, and pMG36e-based plasmids were maintained in MRS agar or broth with 0.1 μg/ml erythromycin. The LoVo human colonic cancer cell line (our laboratory stock) was cultured in Roswell Park Memorial Institute (RPMI) 1640 medium containing 10% fetal bovine serum (FBS) at 37°C in a humidified atmosphere of 5% CO~2~. Construction of Recombinant L. acidophilus Strain --------------------------------------------------- To construct the recombinant plasmid pMG36e-espA-Tir-M, the espA-Tir-M (ET) fragment was amplified by overlap extension PCR from espA (GenBank accession no. KJ549678.1) and Tir-M (GenBank accession no. NC002655.2). EspA and Tir-M fragments were joined with a linker (GGA GGC GGA AGT GGA GGA GGT AGC). Both espA and Tir-M were PCR-amplified from E. coli O157:H7 EDL933 using primer pairs with the following sequences: P1 (forward), 5′-A AAA [CTGCAG]{.ul} GAT GGA TAC ATC AAA TGC A- 3′ and P2 (reverse), 5′-GCT ACC TCC TCC ACT TCC GCC TCC TTT ACC AAG GGA TAT TGC TG-3′ for espA; P3 (forward), 5′-GGA GGC GGA AGT GGA GGA GGT AGC AGC CCA ACC ACG ACC GAC-3′ and P4 (reverse), 5′-CCC [AAGCTT]{.ul} TTA GGC TTG CTG TTT GGC CTC TT-3′ for Tir-M; and P1/P4 for espA-Tir M. Restriction enzyme sites (PstI and HindIII) are underlined. The fusion gene espA-Tir-M was inserted into plasmid pMG36e. The recombinant plasmid pMG36e-espA-Tir-M was transformed into L. acidophilus cells according to the manufacturer's instructions for the expression system in Lactococcus lactis (MoBiTec, Germany) with appropriate adjustment. Briefly, cells were cultured in 4 ml of MRS broth with 0.05% cysteine-HCl medium at 37°C for 48 h. When the optical density at 600 nm (OD~600~) reached 0.6, 4 ml of culture was diluted in 100 ml of MRS broth with 0.5 M sucrose and 0.05% cysteine-HCl, followed by incubation for approximately 24 h until the OD~600~ was 0.8. The culture was cooled for 10 min and centrifuged at 5,000 rpm for 20 min at 4°C, then washed twice in 20 ml of 0.5 M sucrose buffer. The cells were re-suspended in 5 ml of pre-cooled transformation buffer composed of 10 mM ammonium and 0.5 M sucrose (pH = 6.0), then centrifuged at 5,000 rpm for 20 min at 4°C and re-suspended in 400 μl of transformation buffer. The recombinant plasmid was transformed into L. acidophilus cells by electroporation using a Pulse Controller apparatus (Bio-Rad, Hercules, CA, USA) at 2.5 kV and 25 μF. Transformed bacteria were re-suspended in MRS broth and cultured at 37°C for 1 h, and were plated on MRS agar (1.5%, w/v) with 0.1 μg/ml erythromycin at 37°C for 48 h. Positive colonies of transformed bacteria were identified by PCR-amplifying the target gene (espA-Tir-M) and sequencing. The empty vector pMG36e transformed into L. acidophilus cells (LA) served as a control. Sodium Dodecyl Sulfate Polyacrylamide Gel Electrophoresis (SDS-PAGE) and Western Blot Analysis ---------------------------------------------------------------------------------------------- The recombinant L. acidophilus (LA-ET) was cultured in MRS broth with 0.1 μg/ml erythromycin at 37°C for 24 h. The supernatant was precipitated with trichloroacetic acid followed by two washes with acetone. A 20 μl volume of sterile distilled water was added to dissolve the supernatant proteins, which were separated by 12% (w/v) sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and analyzed by western blotting using an anti-EHEC O157:H7 polyclonal antibody (EterLife, Birmingham, UK) according to the manufacturer's instructions. Cytotoxicity Analysis of LA-ET ------------------------------ LA-ET or LA at a multiplicity of infection (MOI) of 500:1 was incubated with LoVo cells for 12 h, after which these cells were harvested and washed with PBS. The cells were re-suspended in 500 μl of binding buffer containing 5 μl of FITC-Annexin V and 5 μl of PI. The cells were gently vortexed and incubated for 15 min at room temperature in the dark. Finally, the cells were detected using flow cytometry according to the manufacturer's instructions. Inhibition of EHEC O157:H7 by Colonized LoVo Cells -------------------------------------------------- Inhibition of EHEC O157:H7 by LA-ET by the colonized LoVo cells was evaluated using the following assays as previously described ([@B29]). Briefly, 1 × 10^5^ LoVo cells per well were cultured in 24-well plates. For the exclusion assay, the LoVo cells were pre-inoculated with LA-ET or LA (MOI = 500:1) for 90 min and then, EHEC O157:H7 (MOI = 100:1) was added for co-incubation for 90 min at 37°C. For the competition assay, LA-ET or LA and EHEC O157:H7 were added to LoVo cells, followed by incubation at 37°C for 3 h. LoVo cells inoculated with EHEC O157:H7 (MOI = 100: 1) for 90 min served as a positive control. To measure the total number of adherent EHEC O157:H7 bacteria, cells were lysed and the cell suspension was plated in triplicate on LB agar and incubated for 24 h at 37°C, after which the number of colony-forming units (CFUs) was counted. Each experiment was repeated three times. Results were calculated as a relative adhesion percentage \[100 × (number of adherent bacteria in the experimental group)/(number of adherent bacteria in the positive control group)\]. LA-ET Inhibits EHEC O157:H7-Induced Attaching and Effacing Lesions ------------------------------------------------------------------ Fluorescence actin staining (FAS) was performed to detect the formation of A/E lesions. Briefly, LoVo cells were pre-incubated with LA-ET (MOI = 500:1) for 3 h at 37°C and then EHEC O157:H7 (MOI = 100:1) cells were added for co-incubation for 3 h at 37°C in 5% CO~2~. After incubation, cells were washed three times with PBS (pH 7.4) and fixed in 4% paraformaldehyde in PBS (pH 7.0) for 10 min. Following a 30 s wash in PBS, cells were permeabilized in 0.5% Triton X-100 in PBS for 5 min. The cells were washed once and then 200 μl of 100 nM rhodamine phalloidin (Cytoskeleton, USA) was added and incubated in the dark for 30 min. The cells were washed three times and the DNA was counterstained for 30 s with 200 μl of 100 nM DAPI in PBS. The samples were examined with a confocal laser scanning microscope. Oral Immunization of Mice with Recombinant L. acidophilus Strains ------------------------------------------------------------------- Specific pathogen-free 6-week-old female BALB/c mice were obtained from the Southern Medical University Laboratory Animal Center (Guangzhou, China). All experimental and animal handling procedures were approved by the Institution Animal Care Committee of Southern Medical University (permit no. 44002100006397) and were performed in accordance with the approved relevant guidelines. The mice were sacrificed by cervical dislocation at the end of the experiment. LA-ET strains were grown as described above, then collected and washed twice with sterile PBS. Mice were randomly divided into the following three groups (n = 15): PBS, LA and LA-ET. The PBS group was orally administered with 100 μl sterile PBS; the LA group was orally administered with 1 × 10^9^ CFU LA; and the LA-ET group was orally administered with 1 × 10^9^ CFU LA-ET. All mice were immunized on days 0, 3, 7, 10, 21, and 24 (the day of first immunization regarded as 0 day) and serum samples were collected on days 0, 7, 21, and 35 and stored at -80°C until use. Fecal extracts were collected on day 35 for sIgA measurement by enzyme-linked immunosorbent assay (ELISA) as previously described ([@B3]). Determination of Antibody and Cytokine Levels by ELISA ------------------------------------------------------ Total serum IgG (total IgG) and fecal IgA were measured by indirect ELISA as previously described ([@B12]). ELISA plates were coated with 100 μl of 10 μg/ml purified espA-Tir-M protein (our laboratory stock). Serum and fecal extracts were prepared as described above and diluted 1:50 and 1:10, respectively. In addition, 5 μl serum (collected on day 34) was assayed for murine interferon (IFN)-γ and interleukin (IL)-4 and IL-10 by quantitative ELISA using a mouse ELISA kit (Elabscience, Hubei, China) according to the manufacturer's instructions. ELISA results were obtained by measuring the OD~450~ using an EL9800 ELISA microplate reader (BioTek, Winooski, VT, USA). Survival Analysis of Immunized Mice ----------------------------------- At 10 days after the last immunization, mice in each group were randomly divided into two subgroups, of which one subgroup (n = 10) received streptomycin (5 g/l) in the drinking water from 3 days before infection until the end of the experiment to clear intestinal flora and to enhance EHEC O157:H7 colonization ([@B31]), whereas the other subgroup (n = 5) was not treated with streptomycin. Mice were challenged by oral inoculation of 1 × 10^10^ CFU of EHEC O157:H7 in PBS. Fecal shedding of EHEC O157:H7 was monitored at 2-day intervals as previously described ([@B3]). The number of surviving mice in each group was recorded daily. All remaining mice were sacrificed on day 15 post-challenge. Histopathology -------------- Mouse tissues were fixed by immersion in 10% neutral formalin, embedded in paraffin, and cut into sections that were stained with hematoxylin and eosin and was further microscopically evaluated by two expert pathologists who were blinded to the experimental groups. Statistical Analysis -------------------- Statistical analysis was performed using SPSS v.21.0 software. Differences in antibody and cytokine levels were evaluated by one-way analysis of variance with the least significant difference test. The duration of fecal shedding among groups were evaluated by one-factor repeated measures ANOVA analysis. The number of adherent bacteria was analyzed using Student's t-test. P \< 0.05 was considered statistically significant. Results ======= LA-ET Expressing espA-Tir-M Fusion Protein in the Culture Supernatant --------------------------------------------------------------------- The espA-Tir-M fusion gene was amplified by overlap extension PCR, yielding the expected 816-bp product, which was visualized by agarose gel electrophoresis (data not shown). The fragment was cloned into the pMG36e vector, and the recombinant pMG36e-espA-Tir-M plasmid was transformed into L. acidophilus cells. The supernatant proteins, LA-ET or LA were analyzed by SDS-PAGE and western blotting. A 36-kDa band was observed in the culture supernatant of the LA-ET lysate but not in the supernatant of the LA lysate (Figure [1A](#F1){ref-type="fig"}), which indicates that the LA-ET-expressed espA-Tir-M fusion protein was mainly a secreted protein. The secretion of espA-Tir-M was confirmed by reaction with the anti-EHEC O157:H7 polyclonal antibody with western blotting (Figure [1B](#F1){ref-type="fig"}). ![Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and western blot analysis of proteins released by LA-ET in the supernatant. (A) SDS-PAGE of LA-ET; M represents the molecular size marker in kDa (Bio-Rad); Lane 1, supernatant of LA lysate; Lane 2, supernatant of LA-ET lysate; Lane 3, culture supernatant of LA; Lane 4, culture supernatant of LA-ET; (B) western blot analysis of LA-ET; Lane 1, culture supernatant of LA. Lane 2, culture supernatant of LA-ET.](fmicb-08-00417-g001){#F1} LA-ET Was Safe in a Cell Model ------------------------------ Comparisons of the cytotoxic effect in LoVo cells showed no obvious difference in the LA-ET group compared to the normal group and LA group (Figure [2](#F2){ref-type="fig"}). ![Flow cytometry to determine the cytotoxicity of LA-ET. (A) Normal LoVo cells as control; (B) LoVo cells incubated with LA; (C) LoVo cells incubated with LA-ET.](fmicb-08-00417-g002){#F2} Inhibition of EHEC O157:H7 Colonization and A/E Lesion Formation in LoVo Cells ------------------------------------------------------------------------------ The ability of LA-ET to interfere with EHEC O157:H7 adhesion to LoVo intestinal epithelial cells was investigated using exclusion and competition assays. For the former, both LA-ET and LA excluded EHEC O157 from LoVo cells at a rate of almost 94% (P = 0.000). For the latter, the relative adhesion was 35.8% ± 6.83% for LA-ET and 39.4% ± 8.26% for LA (P = 0.501), and both LA-ET and LA excluded EHEC O157 from LoVo cells by about 60% (P = 0.000) (Figure [3A](#F3){ref-type="fig"}). ![Inhibition of EHEC O157:H7 colonization and A/E lesion formation in LoVo cells. (A) Exclusion and competition adhesion inhibition assays in LoVo cells. Results are expressed as relative adhesion relative to the control without LA or LA-et. Data are shown as a percentage ± SD (n = 9). ^∗∗^P \< 0.001 (one-way analysis of variance with least significant difference test). (B) FAS detected A/E lesion formation. (a) Normal LoVo cell; (b) LoVo cells were incubated with EHEC O157:H7 (MOI = 100:1) for 3 h; White arrows show the A/E lesion and green arrows show EHEC O157:H7; (c) LoVo cells were incubated with LA-ET (MOI = 500:1) for 3 h; (d) LoVo cells were pre-incubated with LA-ET (MOI = 500:1) for 3 h and then EHEC O157:H7 (MOI = 100:1) was added for co-incubation for 3 h.](fmicb-08-00417-g003){#F3} Fluorescent-labeled phalloidin staining of F-actin in LoVo intestinal epithelial cells showed that EHEC O157:H7 challenge of LoVo cells induced bacterial attachment to cells (green arrows), effaced microvilli, and recruited F-actin into pedestals to form A/E lesions (white arrows) (Figure [3Bb](#F3){ref-type="fig"}). By contrast, cells pretreated with the LA-ET no longer demonstrated EHEC O157:H7 adhesion and A/E lesions (Figure [3Bd](#F3){ref-type="fig"}). The LoVo cells (Figure [3Ba](#F3){ref-type="fig"}) and LoVo cells incubated with LA-ET (Figure [3Bc](#F3){ref-type="fig"}) were normal. Serum IgG and Fecal sIgA Levels ------------------------------- Significant IgG responses were elicited in mice immunized with LA-ET as compared to controls (Figure [4A](#F4){ref-type="fig"}). In addition, serum antibody titer increased after each immunization with LA-ET (0.24 ± 0.09 on day 7, 0.35 ± 0.10 on day 21, and 0.68 ± 0.14 on day 35), reaching a peak on day 35 (P = 0.000). sIgA was also measured in fecal samples (Figure [4B](#F4){ref-type="fig"}). Consistent with systemic antibody responses, sIgA levels were increased in mice immunized with LA-ET (0.26 ± 0.08) as compared to control mice (P = 0.000). ![Detection of specific IgG in serum and sIgA in fecal samples. Antibody levels were determined by ELISA. The fusion protein espA-Tir-M was used as a coating antigen. Horseradish peroxidase-labeled anti-mouse IgG or IgA was used as the test antibody. (A) Serum IgG levels were determined 0, 7, 21, and 35 days after the first immunization. (B) sIgA antibody in fecal samples collected on day 35 after the first immunization. Data are shown as the mean ± SD (n = 15). ^∗^P \< 0.05, ^∗∗^P \< 0.001 (one-way analysis of variance with least significant difference test).](fmicb-08-00417-g004){#F4} IL-4, IL-10, and IFN-γ Levels in Immunized Mice ----------------------------------------------- Cytokines play an important role as part of the immune response against infections. Serum IL-4 concentrations were significantly higher in the LA-ET group (285.2 ± 110.3 vs. PBS, P = 0.000; vs. LA, P = 0.000) (Figure [5A](#F5){ref-type="fig"}). Serum IL-10 level was higher in the LA-ET (276.4 ± 138.1, P = 0.000) and LA (127.8 ± 31.1; P = 0.048) groups than in the PBS group (51.4 ± 19.5) (Figure [5B](#F5){ref-type="fig"}). Moreover, the serum IFN-γ level was higher in the LA-ET group (102.2 ± 63.0) than in the control LA (34.0 ± 20.9, P = 0.001) and PBS (31.9 ± 16.0, P = 0.000) groups (Figure [5C](#F5){ref-type="fig"}). ![Serum levels of IL-4, IL-10, and IFN-γ. Cytokines were detected in serum collected 35 days after the first immunization. (A) The serum level of IL-4. (B) The serum level of IL-10. (C) The serum level of IFN-γ. Data are reported as the mean ± SD (n = 10). ^∗^P \< 0.05, ^∗∗^P \< 0.001 (one-way analysis of variance with least significant difference test).](fmicb-08-00417-g005){#F5} Immunization with LA-ET Protects Mice against EHEC O157:H7 Colonization ----------------------------------------------------------------------- The protection efficiency of LA-ET was evaluated based on survival, fecal shedding, and histopathology. Within the streptomycin-treated immunization groups, 90% of mice were susceptible to EHEC O157:H7 challenge and died between days 1 and 3 post-infection in the PBS control group (Figure [6A](#F6){ref-type="fig"}), whereas 60% of mice in the LA group died between days 1 and 5. Clinical symptoms included weight loss, hunched posture and lumbering gait. The post-infection survival data revealed that the protection rate was 80% in LA-ET. ![Survival analysis of immunized mice. (A) Protection rates in EHEC O157:H7-challenged mice. The number of mice that died was monitored for 15 days post-infection. (B,C) Changes in fecal shedding of EHEC O157:H7 in mice. Mice (n = 10) were orally challenged with 10^10^ CFU EHEC O157:H7 either under oral streptomycin treatment condition (n = 10) (B) or without streptomycin treatment (n = 5) (C) after the last immunization and fecal shedding was monitored for 15 days. The limit of detection for plating was 100 CFU/100 mg feces. Data represent the mean ± SD. Data points in rectangles indicate a significant difference between groups. ^∗∗^P \< 0.001 (one-factor repeated measures ANOVA analysis).](fmicb-08-00417-g006){#F6} In addition, the duration of fecal shedding shown by the LA-ET group was significantly shorter than that in the LA group (P = 0.000); one vaccinated mouse had stopped shedding on day 9, and all mice were negative by day 13. In contrast, control mice shed bacteria throughout the 15-day time course (Figure [6B](#F6){ref-type="fig"}). No mice died in the non-streptomycin-treated immunization groups when challenged with EHEC O157:H7. Comparisons of the durations of EHEC O157:H7 fecal shedding showed a significantly shorter duration in the LA-ET group (8.2 ± 0.89 days) compared to the LA and PBS groups (11.4 ± 0.89 and 13.4 ± 0.89 days, respectively) (P = 0.000). Additionally, the shorter duration of fecal shedding shown by the LA group in comparison to the PBS group was significant (P = 0.007) (Figure [6C](#F6){ref-type="fig"}). Histopathological Findings -------------------------- The intestine was dissected out from the mice sacrificed 15 days after challenge. The pathological changes of non-immunized infected mice included major intestinal injury (epithelial cell necrosis) accompanied by damage to the kidney (hyperemia of mesenchyme capillaries), liver (mesenchyme hyperemia), and spleen (splenic sinusoid dilation and ecchymosis) (data not shown). However, the colon of immunized mice infected with EHEC O157:H7 appeared normal, and these animals produced well-formed stools with no obvious mucosal thickening, similar to non-infected mice (Figure [7](#F7){ref-type="fig"}). ![Histopathological analysis of the colon of immunized mice challenged with EHEC O157:H7. Tissue was obtained from mice sacrificed 15 days after infection. Sections of paraffin-embedded tissue were stained with hematoxylin and eosin and examined by light microscopy. (A) Normal mice; (B) PBS group; (C) LA group; and (D) mice immunized with LA-ET group.](fmicb-08-00417-g007){#F7} Discussion ========== Enterohemorrhagic Escherichia coli O157:H7 is a major cause of gastroenteritis. Most infected individuals develop HUS, which is prevalent in children ([@B23]). EHEC O157:H7 adheres to host epithelial cells as the first step of the infection process, leading to pedestal formation and effacement of microvilli, which characterize the A/E lesions. Hence, inhibiting pathogen adhesion to epithelial cells may prevent colonization and limit opportunistic infection ([@B17]). It has been suggested that probiotics are effective in preventing adhesion and invasion of enteric pathogens ([@B10]). Probiotics may reduce intestinal infections by competing with pathogens for binding sites on the intestinal wall as well as for nutrients, and also by producing antibacterial compounds and lactic acid, or via immunomodulation ([@B28]; [@B8]; [@B4]). In this study, LA and LA-ET were both shown to exclude EHEC O157 from LoVo cells in exclusion and competition assays, which highlights the important role of probiotics in interrupting progressive infection with EHEC O157:H7. This finding was consistent with studies in which probiotics reduced intestinal infection ([@B25]; [@B26]). Furthermore, the FAS test showed that LA-ET has the ability to inhibit A/E lesions. It is obvious that LA-ET inhibits EHEC O157:H7 to induce the formation of A/E lesions and adhesion to cells. However, recombinant anti-EspA antibodies block EHEC O157:H7-induced A/E lesions in vitro but do not affect EHEC O157:H7 adhesion to host cells ([@B19]). Therefore, the effect of LA-ET in excluding EHEC O157:H7 from adhering to LoVo cells may be attributed to the probiotics competing with pathogens for binding sites. LA-ET used for oral immunization can prevent gastric infection and allows direct contact between the antigen and the immune system. Significant IgG and sIgA responses were elicited in mice immunized with LA-ET and protective effects were also confirmed. A previous study showed that attenuated EIS-producing recombinant Salmonella induced significant specific IgG in serum and sIgA in feces to protect mice from a challenge of EHEC O157:H7 ([@B15]). These findings indicate that antibody-mediated immunity participates in the prevention of EHEC O157:H7 infection. In addition, the rate of protection was consistent with antibody and cytokine levels in vaccinated mice. LAB strains for vaccines can be specifically selected according to their function to modulate dendritic cells and induce Th1, Th2, or mixed Th1/Th2 responses ([@B32]). The major cytokines associated with Th1 cells are IL-2, TNF-α and IFN-γ, which can enhance T cell cytotoxicity and immune responses. Th2 cells secrete IL-4, IL-5, and IL-10, which mainly promote antibody production and mediate humoral immune responses ([@B34]). We observed that LA-ET enhanced the production of IFN-γ, IL-4, and IL-10, indicating that a mixed Th1/Th2 response was induced. Thus, immunizing mice with LA-ET induced humoral and cellular immunity, consistent with previous findings ([@B2]). Reduced intestinal colonization is regarded as an important criterion for protection against EHEC O157:H7 infection. Oral immunization of mice with LA-ET resulted in a reduction of EHEC O157:H7 fecal shedding in two infection models. The infection model under the oral streptomycin treatment condition emphasized the protective effects of humoral and cellular immune responses induced by LA-ET. LA-ET blocks adhesion of EHEC O157:H7 to intestinal tract, which may be associated with the sIgA. The sIgA antibodies play an important role in blocking the attachment of EHEC O157:H7 to epithelial cells ([@B14]). The infection model without streptomycin treatment took into account the probiotic properties of LA-ET. Probiotics have been shown to protect mice against EHEC O157:H7 ([@B30]). In this study, we found that LA elevated IL-10 secretion, reduced EHEC O157:H7 colonization (without streptomycin treatment), and increased the survival rate in mice, suggesting that L. acidophilus strains play an important role in protecting mice against EHEC O157 infection and may be useful vaccine carriers for combating enteric pathogens in humans. The histopathology indicates that administration of LA-ET has the ability to reduce or inhibit A/E lesions, which suggests that LA-ET could reduce the colonization of EHEC O157:H7 in the intestinal tract and inhibit A/E lesions. Moreover, the mechanism utilized by LA-ET against EHEC O157:H7 is different from that of antibiotics. LA-ET was safe in a cell model, but EHEC O157:H7 released toxins when treated with antibiotics, which could cause toxin-induced systemic injury. Thus, LA-ET is a promising candidate vaccine against EHEC O157:H7, especially for children and the elderly. Although oral immunization with LA-ET induced protective immunity, an important deficiency should be noted. First, the antigen dose provided by LA-ET in vivo cannot be measured. In addition, the expression level of the antigen in LAB is weak and the enzymes in the gastrointestinal tract may digest the proteins ([@B16]), which may reduce the effects of antigen. Furthermore, [@B1] previously constructed a recombinant L. lactis strain expressing the EHEC antigen EspB and this strain expressed rather low levels of antigen EspB. However, they optimized the expression of EspB in L. lactis through secretion of EspB either under constitutive or nisin-inducible control, and the strains successfully induced immune responses ([@B2]). Collectively, we speculated that an oral vaccine would be more effective, than LAB vaccines that are engineered to co-express antigens with cytokines or made inducible in different mutants, or new strains that are selected to have enhanced adjuvant or targeting capacities ([@B27]). In our study, the antigen expressed by LA-ET was not strong, which was shown by SDS-PAGE analysis. This is the first report of the use of L. acidophilus expressing the EHEC O157:H7 antigen. Thus, the optimal condition for LA-ET that can enhance stimulation of strong immune responses still needs to be explored. High expression of L. lactis vector is promising for vaccine development. Conclusion ========== LA-ET was capable of inhibiting A/E lesions and bacterial adhesion in vitro, and oral immunization strain induced both humoral and cellular immune responses in vivo and protected against EHEC O157:H7 colonization and infection in mice. These findings suggest that oral LA-ET is a promising candidate vaccine against EHEC O157:H7 infection. Author Contributions ==================== RL did the experiments with gene construction, cells and mice, analyzed data, and wrote the manuscript. YZ did the experiments with mice and analyzed data. BL designed the experiments and contributed to revising the manuscript. YL and YW did the experiments with gene construction and mice. SD and BZ did the experiments with mice. XW and HF provided overall directions and contributed to revising the manuscript. Conflict of Interest Statement ============================== The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Funding. This research project was supported by grants from the Natural Science Foundation of China (grant 31300762), the Medical Scientific Research Foundation of Guangdong Province of China (grant 2012B031800132 & 2013B021500012), and the Science and Technology Planning Project of Guangzhou of China (grant 2014J4100146). We thank Professor Weisen Zeng for kindly providing advice for constructing recombinant L. acidophilus. [^1]: Edited by: José Roberto Mineo, Federal University of Uberlandia, Brazil [^2]: Reviewed by: Mithilesh Kumar Singh, Indian Veterinary Research Institute, India; Ruchi Tiwari, DUVASU, India [^3]: This article was submitted to Microbial Immunology, a section of the journal Frontiers in Microbiology	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#SEC1} ============ Polyploidy plays a significant role in plant evolution and formation of the important agriculture traits such as high yield and stress resistances in crops ([@B1]). Polyploidy crops own higher plasticity and environmental adaptability ([@B4]) with specific gene expression patterns for the organ development and resistances to abiotic and biotic stresses ([@B1],[@B5]). Recent studies have illustrated that nuclear architecture is important for gene expression and recombination ([@B6]), and the distribution and condensation of chromatin fiber could be influenced by various environmental and physical factors ([@B9],[@B10]). Most eukaryotic genomes are organized at hierarchical levels ([@B11]). First, each chromosome has a preferred but not fixed position named chromosome territory (CT) ([@B14]), which consist of several megabase-scale genomic compartments ([@B15],[@B16]). These compartments are categorized into A-type compartment rich for euchromatic and highly transcribed regions and B-type compartment rich for heterochromatic and gene-poor regions ([@B17]). Further, within the genomic compartments, topologically associating domains (TADs) were identified as clusters of chromatin interactions with their boundaries associated with architecture proteins like CTCF and cohesin, and activating histone markers ([@B18]). Moreover, the smaller scale structures like chromatin loops at 10 kb to 1 Mb can appear within the topological domains ([@B21],[@B22]). Chromatin looping could bring cis-regulatory elements and promoters together to affect gene transcription ([@B23]). Studies on the three-dimensional (3D) genome organization have proposed that Arabidopsis exhibited no large local interactive domains ([@B24]). The chromosome territories of Arabidopsis are organized as chromosome-arm territories which set centromeres as strong insulators that block the interactions across chromosome arms ([@B25]). Arabidopsis does not form obvious TADs, which is different from animals and other large genome plants ([@B26]). In addition, the Arabidopsis chromosome arms are partitioned into loose structural domain (LSD) and compacted structural domain (CSD) which are equivalent to A- and B-type compartments in animals ([@B29]). The studies on the relationships between chromatin packing and epigenetic modifications in mutants defective in the specific epigenetic pathway suggested that changes in the epigenome could correlate with chromatin interaction patterns ([@B30]). The chromatin packing in diploid and allotetraploid cottons with large genomes was shown to be modulated by the density of chromatin and chromosome number ([@B26]). Genome doubling in rice resulted in changes of DNA methylome of transposable elements and transcriptome ([@B31]). However, little is known about the relationship between genome organization and gene regulation in autopolyploid plants. To address the spatial arrangement of the doubled genome and the relationship between chromatin structure and transcriptional regulation, we performed High-throughput Chromosome Conformation Capture (Hi-C) analysis of the autotetraploid (genome-duplicated) and diploid Arabidopsis. We showed that the genome duplication resulted in increased inter-chromosome interactions and decreased intra-chromosome arm interactions compared with its diploid progenitor. In conjunction with RNA-seq and ChIP-seq, we found that the transcription of a group of genes is specifically affected by the chromatin interactions with epigenome dynamics. These results shed light on the 3D genomic organization and transcriptional regulation of the autotetraploid plants. MATERIALS AND METHODS {#SEC2} ===================== Plant materials {#SEC2-1} --------------- Arabidopsis thaliana accession Columbia (Col-0) was used as wild type. Autotetraploid Arabidopsis (4 × Col-0) generated by colchicine using Col-0 as the progenitor ([@B32]) was a kind gift from Prof. Daiyin Chao ([@B4]), SIPPE. Here the autotetraploid Arabidopsis represents the genome-duplicated line. Plants were grown at 23°C in long day condition (16 h light and 8 h dark cycles) on Murashige & Skoog (MS) medium supplemented with 1% sucrose and 0.8% agar. After stratification at 4°C for 3 days, the plants were transferred to the long day condition for 10 days, and then the aerial parts of these seedlings were harvested for Hi-C, ChIP-seq and RNA-seq library preparation. Flow cytometry {#SEC2-2} -------------- Ten-day-old Col-0 or 4 × Col-0 Arabidopsis seedlings were collected into a clean and pre-cooled plate, and then chopped with a new razor blade to release the nuclei in the sterile lysis buffer (45 mM MgCl~2~·6H~2~O, 30 mM sodium citrate (trisodium), 20 mM MOPS, 1% Triton-100, pH 7.0) for 3--5 min until the buffer turns green. Transfer the mixture into a 40 μm strainer. The filtrates were added DAPI solution with final concentration of 1 ng/ml, then incubate for 30 min in the dark on ice. The DAPI-treated filtrates were then loaded to the flow cytometer (Beckman Coulter Moflo-XDP, USA) to measure the ploidy levels. Flowering time and rosette leaf number measurement {#SEC2-3} -------------------------------------------------- After stratification, the seeds of Col-0 and 4 × Col-0 were sown on the MS plates, and grown for 10 days. The seedlings were then transferred to the soil under long day condition until flowering. The days of flowering and rosette numbers were scored. Hi-C library preparation {#SEC2-4} ------------------------ Hi-C experiments were performed essentially as described ([@B29]) with some modifications. Briefly, 2.5 g aerial parts of the seedlings were fixed and grounded into powder in liquid nitrogen. The extracted nuclei were digested with 600 U HindIII restriction enzyme by incubating overnight at 37°C, then the digested chromatin was blunt-ended with 1μl 10 mM dATP, dTTP, dGTP and 25 μl 0.4 mM biotin-14-dCTP and 100 U Klenow fragment for 45 min at 37°C. Next, the ligation reaction was performed in 10 time volume of ligation production buffer under constant shaking with 745 μl 10× ligation buffer and 10% Triton X-100, 80 μl 10 mg/ml BSA and ATP and 100 Weiss U T4 DNA ligase. The ligation reaction was performed at 16°C for 6 h. After ligation, the nuclei were reverse-crosslinked with proteinase K at 65°C overnight. Subsequently, the extracted chromatin was fragmented into a mean size of 300 bp using a sonicator (Covaris S220). Hi-C libraries were constructed with NEBNext Multiplex Oligos kit and KAPA Hyper Prep Kit. The final library was subject to sequencing on an Illumina HiSeq 2000 instrument with 2 × 150-bp reads. Hi-C sequencing data processing {#SEC2-5} ------------------------------- After removing the adapter, the clean Hi-C reads of Col-0 and 4 × Col-0 were aligned to Arabidopsis reference genome (TAIR10). Following with HiC-Pro and Juicer software ([@B33],[@B34]), valid pairs of wild type and autotetraploid were used to create interaction matrixes with bin size 50 kb for further analysis. The interaction matrixes were normalized with KR method from Juicer ([@B34]). The reproducibility of two biological replicates was tested with Pearson correlation coefficient from the interaction matrixes ([@B35]). After excluding the pericentromeres as reported ([@B29]), the first principal component was used to identify chromatin structure domains (SDs) with Juicer. We set the positive Eigenvalues associated with loose structure domains (LSD) and negative Eigenvalues with compacted structural domains (CSD). Calculation of chromatin interaction and interaction decay exponents {#SEC2-6} -------------------------------------------------------------------- To analyze the difference of interaction matrixes between Col-0 and 4 × Col-0, the normalized interaction matrix from Col-0 was divided by the normalized interaction matrix from 4 × Col-0, with all the zeros in the matrixes replaced with 1% quintiles from the non-zero elements in each matrix. Log2 transformation and median normalization were used to standardize the difference matrix. To understand the interaction frequency changes dependent on the genome distance, interaction decay exponents (IDEs) were calculated ([@B29]) for chromosomes, pericentromeres and telomeres. RNA-seq and data analysis {#SEC2-7} ------------------------- Total RNAs were extracted with RNeasy Plant Mini kit (Qiagen), and the libraries were constructed according to a standard protocol (Illumina). Sequencing reads were aligned against the reference genome (TAIR10) using Feature Counts and STAR with default parameters ([@B36]). The differential expression analysis was run using the classical normalization method DESeq2 R package ([@B37]) with a 0.05 p-value, 0.05 false discovery rate, and cutoff of 1 log-fold change. Gene ontology (GO) terms were extracted from clusterprofiler R package ([@B38]). All of the data were from three biological replicates. ChIP-seq and data analysis {#SEC2-8} -------------------------- The ChIP assay was performed according to previously reported ([@B39]). Briefly, 2.5 g fresh 10-day-old seedlings were cross-linked in the cross-linking buffer (0.4 M sucrose, 10 mM Tris--HCl (pH 8.0), 1 mM PMSF, 1 mM EDTA, 1% formaldehyde) for 3 × 5 min using vacuum infiltration and the reaction was terminated in 2 M glycine. Chromatin was sheared to an average size of 300 bp with a sonicator (Bioruptor, Diagenode). Then the sonicated samples were immunoprecipitated with 5 μg anti-H3K27me3 (Millipore 07-449), anti-H3K4me3 (Millipore 07-473) and anti-H3 (Abcam ab1791) antibodies. After incubation at 4°C overnight, the antibodies were recovered with 20 μl Protein A/G magnetic beads (Millipore 16-663). After reverse cross-link, ChIP-ed DNA was extracted with phenol-chloroform method and sequencing library was constructed according to the standard Illumina protocol. The library was sequenced using an Illumina HiSeq 2000 instrument. The sequencing reads were aligned to the reference genome with BWA ([@B40]). The mapped reads were analyzed using Macs2 with default parameters for peak calling ([@B41]). Reads from anti-H3 samples were set as the control. For ChIP-PCR, the immunoprecipitated DNA template and primers covering promoter, exon and intron of FLC ([Supplementary Table S6](#sup1){ref-type="supplementary-material"}) were used to quantify histone modifications by real-time PCR. Real-time PCR data of H3K27me3, H3K4me3 and H3 enrichment at FLC locus were normalized to 1% Input. Histone modification quantifications on CSD and LSD {#SEC2-9} --------------------------------------------------- To quantify the genome wide histone modification levels in all CSDs and LSDs, the means of H3K4me3 and H3K27me3 RPKM (reads per kilobase per million mapped reads) were calculated in CSD and LSD regions in 50 kb bins in Col-0 and 4 × Col-0. Because the SDs were identified with PCA1 (Principal Component Analysis 1), a relatively broad parameter which could not reflect the more complicated structures. To analyze the relationship between histone modification and SD transition more precisely, K-means clustering was used to remove histone modification noise signals. In this method, we set K = 10 to cluster the same kind of histone modification on a given SD. Histone modification peaks in the SD were called and the RPKM in these called peaks was used to compare the differences between these SDs. Jaccard index was used to indicate the correlation between histone modification and SD. Fluorescence microscopy and image processing {#SEC2-10} -------------------------------------------- The binary vector of HTR12-GFP ([@B42]) was used to obtain transgenic plants in Col-0 and 4 × Col-0 backgrounds for centromere visualization under microscopy. 10-day-old seedling leaves were subjected to microcopy analyses. Images of nuclei were acquired with the DeltaVision Personal DV system (Applied Precision). Data collection and image processing were performed according to Fang et al. ([@B43],[@B44]). Fluorescence foci were subjected to statistical analysis using the Spot Pattern module of Imaris software (Bitplane AG). Bootstrapping analysis {#SEC2-11} ---------------------- In the bootstrapping strategies ([@B45]), 1000 groups (n = 1000 times) of equal number randomly selected no-regulated genes were subjected to the same analysis to determine the percentage of those groups fallen in differentially interaction bins. The percentile of the test sample lie above the top 5 percentile of the control distribution was considered confidently. Promoter-Promoter and Promoter-Enhancer analysis {#SEC2-12} ------------------------------------------------ The 2 kb upstream of TSS was selected as the promoter. The distribution of promoters was then analyzed for each bin, and the bins that contained more than the median number of promoters was defined as promoter-enriched bins. The interaction frequency among these promoter-enriched bins was calculated as the genome-wide promoter-promoter interaction frequency. For analyzing the interactions of promoter-enhancers, the published candidate enhancer library ([@B46]) was used to identify enhancer-enriched bins. The bins which contain the number of enhancers more than the median were defined as enhancer-enriched bins. The interaction frequency between enhancer-enriched bins and promoter-enriched bins was calculated as the genome-wide promoter-enhancer interaction frequency. Chromosome Conformation Capture (3C) test {#SEC2-13} ----------------------------------------- Two grams of 10-day-old seedlings were cross-linked and nuclei were isolated as previously reported ([@B29]). Digestion was performed with 400 U BamHI and BglII at 37°C overnight. After inactivating the enzyme with SDS, the ligation reaction was performed by incubation at 16°C for at least 6 h with 100 U T4 DNA ligase. Reverse cross-link was performed at 65°C by treatment with Proteinase K overnight. Relative interaction frequency was assessed by real-time PCR ([@B47]). RESULTS {#SEC3} ======= Phenotypic and transcriptomic changes caused by Arabidopsis genome doubling {#SEC3-1} --------------------------------------------------------------------------- The wild type (Col-0, Columbia) and autotetraploid Arabidopsis (4 × Col-0) were confirmed by flow cytometry ([Supplementary Figure S1](#sup1){ref-type="supplementary-material"}). Compared with Col-0, autotetraploid Arabidopsis exhibited enlarged leaves, flowers and whole plant size (Figure [1](#F1){ref-type="fig"}). It was known that autotetraploid has increased trichome branch numbers, enlarged cells, stomata and seed sizes ([@B48]), and also provided enhanced tolerances to high salinity ([@B4]) and Cu^2+^ stress ([@B51]). We found that 4 × Col-0 plants flower 5 days later under long day condition (Figure [1A](#F1){ref-type="fig"} and [B](#F1){ref-type="fig"}), and have about two more rosette leaves during the vegetative stage than Col-0 (Figure [1C](#F1){ref-type="fig"} and [D](#F1){ref-type="fig"}). ![Comparison of morphology and transcriptome between Col-0 and autotetraploid (4 × Col-0) Arabidopsis. (A) Phenotypes of Col-0 (left) and 4 × Col-0 Arabidopsis (right) grown for 25 days on soil after stratification. (B) Flowering times of Col-0 and 4 × Col-0. Error bars represent means ± SEM (standard error of mean, n = 3 biological replicates). Statistical significance was analyzed by t-test, \\\*p \< 0.001. (C) Phenotypes of rosette leaves of Col-0 and 4 × Col-0 Arabidopsis during vegetative stage. (D) Rosette leaf numbers of Col-0 and 4 × Col-0. Error bars represent means ± SEM (n = 3 biological replicates). Statistical significance was analyzed by t-test, \\\*p \< 0.001. (E) Volcano plot for statistical significance against gene fold change between Col-0 and 4 × Col-0. Each gene was marked as a dot. Blue dots represent up-regulated genes, red dots represent down-regulated genes and green dots represent the other genes. (F) Numbers of up-regulated and down-regulated genes (\\|log~2~fold change\\| \> 1) in 4 × Col-0 compared to Col-0.The data from three biological replicates were combined.](gkz511fig1){#F1} We then examined the transcriptomes of 10 day Col-0 and 4 × Col-0 seedlings by RNA-sequencing. Transcriptome size-based normalization was performed to calculate the relative gene expression levels between Col-0 and 4 × Col-0. Compared with Col-0, we identified 436 up-regulated genes and 307 down-regulated genes (cutoff threshold, 2-fold) in autotetraploid (Figure [1E](#F1){ref-type="fig"}--[F](#F1){ref-type="fig"} and [Supplementary Table S1](#sup1){ref-type="supplementary-material"}). Gene ontology (GO) analysis revealed that these differentially expressed genes associate with responses to starvation, metabolic process and stimuli ([Supplementary Table S2](#sup1){ref-type="supplementary-material"}). Increased inter-chromosome and decreased intra-chromosome arm interactions in autotetraploid Arabidopsis {#SEC3-2} -------------------------------------------------------------------------------------------------------- To elucidate the effect of genome doubling on chromosome arrangement in living Arabidopsis plants, we transformed Col-0 and 4 × Col-0 Arabidopsis with HTR12-GFP, a centromere-specific histone H3 marker, using flower dipping ([@B52],[@B53]). Transgenic plants expressing HTR12-GFP showed no obvious phenotypic changes compared to their corresponding non-transgenic lines, which indicated that the plants expressing this fusion protein have functional centromeres and normal cell cycles. For the analysis of centromere organization, three independent transgenic lines were used to determine the centromere distribution patterns in Col-0 and 4 × Col-0 Arabidopsis plants expressing HTR12-GFP. Small and bright HTR12 fluorescence foci could be detected in nuclei (Figure [2A](#F2){ref-type="fig"}). In wild type Arabidopsis, a range of 8--10 bright spots (with average number 8.5) were usually detected in each nucleus of guard cells, and ∼16--20 spots (with average number 16.7) were visualized in one of the paired tetraploid guard cells (Figure [2B](#F2){ref-type="fig"}). These results showed that the average number of centromere fluorescence foci in an autotetraploid nucleus were almost twice as much as those in diploid, indicating that centromeres in autotetraploid did not merge with each other after chromosome doubling and each chromosome territory (CT) ([@B54]) might be relatively independent. ![Comparison of 3D chromatin interaction profiles between Col-0 and 4 × Col-0 Arabidopsis. (A) HTR12-GFP labeled centromeres in guard cells of Col-0 and 4 × Col-0. Scale bar = 5 μm. (B) The numbers of centromeres in the guard cell nuclei of Col-0 and 4 × Col-0. The data were calculated by Imaris software from 105 Col-0 and 4 × Col-0 guard cell nuclei. (C) Differential chromatin interaction heatmap between Col-0 and 4 × Col-0 at 50 kb resolution. Chromosomes stacked from bottom left to up right were chr1, chr2, chr3, chr4 and chr5. The pericentromere of each chromosome was labeled in dark red bar. (D) Ratios of trans/cis interaction frequency were calculated across the whole genome in Col-0 (brick red) and 4 × Col-0 (blue). The p value was tested with Wilcoxon rank sum test. (E) Interaction decay exponents along with genomic distance at 50-kb resolution of Col-0 (brick red) and 4×Col-0 (blue). (F) Interaction decay exponents along pericentromeric and telomeric regions (5% of the ends) of 4 × Col-0 and Col-0. (G) Boxplots showing the intra-chromosome arm interaction frequencies between Col-0 (brick red) and 4 × Col-0 (blue). \\\\p \< 3.83e--09, the data was tested by Wilcoxon\'s rank sum test. (H) Boxplots showing inter-chromosome arm interaction frequencies between Col-0 (brick red) and 4 × Col-0 (blue). \\\\p \< 5.05e--05 (Wilcoxon\'s rank sum test). (I) Boxplots showing inter-chromosome interaction frequencies between Col-0 (brick red) and 4 × Col-0 (blue). \\\\p \< 2.79e--05 (Wilcoxon\'s rank sum test).](gkz511fig2){#F2} To better understand the chromosomal architecture of autotetraploid Arabidopsis, we applied a Hi-C approach to compare the spatial organization of genomes in diploid and autotetraploid seedlings. After filtering and alignment to Col-0 reference genome (TAIR10), we obtained total 22 million and 32 million valid Hi-C reads from wild type and autotetraploid Arabidopsis, respectively ([Supplementary Table S3](#sup1){ref-type="supplementary-material"}), with a high reproducibility between the biological replicates ([Supplementary Figure S2A and B](#sup1){ref-type="supplementary-material"}). Differential analysis between Hi-C replicates of Col-0 or 4 × Col-0 was also performed to assess the stochastic variation of the experiments, the results showed no obvious interaction differences between the replicates of Col-0 or 4 × Col-0 ([Supplementary Figure S2C and D](#sup1){ref-type="supplementary-material"}). We then calculated the interaction difference matrix from the reads with the same sequencing depth between Col-0 and 4 × Col-0 to reveal the chromosome clustering traits caused by genomic doubling. Genome-wide chromatin interaction difference matrix revealed that autotetraploid Arabidopsis has increased inter-chromosomal interactions, and reduced intra-arm interactions compared to Col-0 (Figure [2C](#F2){ref-type="fig"}). The ratio of trans/cis interaction confirmed this phenomenon at 50 kb binning resolution (Figure [2D](#F2){ref-type="fig"}), in which autotetraploid Arabidopsis showed a high ratio of trans/cis (inter-/intra) interaction frequency. To better understand the distribution of intra-chromosomal contacts, we used Interaction Decay Exponents (IDEs) with power-law curves to reveal the difference of chromatin fiber characters between autotetraploid and diploid. The IDE could characterize chromatin packing by calculating a linear fit to the average interaction frequencies observed at given genomic distances. The results displayed that interaction frequency decreases as the genomic distance increases both in autotetraploid and wild type Arabidopsis. The intra-chromosomal interaction in autotetraploid showed decreased contact probabilities compared with Col-0 as a function of increasing genomic distance (Figure [2E](#F2){ref-type="fig"}). Beside, power-law curves of individual chromosome indicated that each chromosome showed the same interaction pattern with the whole genome in autotetraploid ([Supplementary Figure S3](#sup1){ref-type="supplementary-material"}). In addition, we observed the decreased intra-pericentromeric interactions (chromatin interactions within the pericentromeric region of a chromosome) and intra-telomeric interactions (chromatin interactions within a telomere of a chromosome) in autotetraploid (Figure [2F](#F2){ref-type="fig"} and [Supplementary Figures S4 and S5](#sup1){ref-type="supplementary-material"}). However, increased inter-pericentromeric interactions (chromatin interactions among pericentromeres of all chromosomes) and decreased inter-telomeric interactions (chromatin interactions among telomeres of all chromosomes) were observed in autotetraploid plants ([Supplementary Figure S6A and B](#sup1){ref-type="supplementary-material"}). To better understand the chromatin packing in autotetraploid Arabidopsis, we compared the interaction frequencies within an arm of a chromosome (intra-arm), between two arms of a chromosome (inter-arm) or among different chromosomes (inter-chromosome). We found that the intra-arm interactions in 4 × Col-0 tend to have lower frequency compared with Col-0 (Figure [2G](#F2){ref-type="fig"}), and inter-arm interactions have no obvious interaction changes (Figure [2H](#F2){ref-type="fig"}), suggesting that autotetraploid Arabidopsis have a relative less compactness in chromosome arms compared to Col-0. In addition, the inter-chromosome interactions presented higher strengths in autotetraploid than those in Col-0 according to the interaction frequencies between a specific chromosome and the remaining chromosomes (brick red and blue boxplots in Figure [2I](#F2){ref-type="fig"}). We also found that the increased inter-chromosomal interactions for each chromosome are distributed largely uniformly across all other chromosomes ([Supplementary Figure S6C](#sup1){ref-type="supplementary-material"}). In diploid Arabidopsis, a nuclear heterochromatic island structure was identified and known as KNOT ([@B29],[@B30]). We found that autotetraploid also has the KEEs (Knot Engaged Elements)/IHIs (Interactive heterochromatic islands) regions in all chromosomes with slight weaker chromosome interaction frequency among them and flanking regions ([Supplementary Figure S6D](#sup1){ref-type="supplementary-material"}). In order to understand the spatial chromosome arrangement of the duplicated genome in nucleus, we calculated the log-ratio of the observed inter-chromosome interaction frequencies and the expected interaction frequencies as described ([@B55]). The matrix showed that autotetraploid has equal interactions among all chromosomes to those of Col-0, indicating that genome duplication has a limited effect on the spatial interactions of chromosome territories ([Supplementary Figure S6E](#sup1){ref-type="supplementary-material"}). Specific chromatin structure domains were changed during autopolyploidization {#SEC3-3} ----------------------------------------------------------------------------- Principal Component Analysis (PCA) of the distance-normalized and correlated intra-chromosomal interactions ([@B17],[@B29]) was previously used to determine the arrangements of discrete structure domains. To determine whether there are any differences in the chromatin structure domain organization between Col-0 and 4 × Col-0, we compared the structure domains through the first principal component at 50 kb resolution. We divided Arabidopsis diploid and autotetraploid genomes into two types of distinct structure domains, one for the compacted structure domains (CSD) with negative values in the first eigenvector and the other for the loose structure domains (LSD) with positive values in the first eigenvector (Figure [3A](#F3){ref-type="fig"} and [Supplementary Figure S7](#sup1){ref-type="supplementary-material"}) ([@B29]). Our results indicated that the autotetraploid and diploid have similar interaction matrix patterns of the discrete structure domains (Figure [3A](#F3){ref-type="fig"} and [Supplementary Figure S7](#sup1){ref-type="supplementary-material"}), with certain regions showing changes in structure domains from CSD to LSD or LSD to CSD (Figure [3B](#F3){ref-type="fig"}). Compared with Col-0, 48.09% and 39.5% of the genome showed conserved LSD and CSD in autotetraploid Arabidopsis (Figure [3C](#F3){ref-type="fig"}), respectively. Interestingly, we found that 6.08% of structure domains in Col-0 genome converted from CSD to LSD, and 6.25% converted from LSD to CSD in autotetraploid (Figure [3C](#F3){ref-type="fig"}). To better characterize SD switching, we quantified the chromosome preferences. We found that the compartment transitions from LSD to CSD occur mostly on chromosome 1 (Chr1) and Chr4, and CSD to LSD mainly on Chr1 and Chr2 (Figure [3D](#F3){ref-type="fig"}). Quantification of chromatin interactions in LSD and CSD revealed that there is a higher interaction frequency in CSD and lower in LSD both in 4 × Col-0 and Col-0 ([Supplementary Figure S7E](#sup1){ref-type="supplementary-material"}). ![Comparison of chromatin structure domains between Col-0 and 4 × Col-0 Arabidopsis. (A) Pearson correlation coefficient matrix and its respective first eigenvector of chr1. The dark regions with positive values in the first eigenvector represent loose structure domains (LSD) and the gray regions with negative values represent compact structure domains (CSD). Genomic bin size: 50 kb. (B) First eigenvector of chr1 right arm (18 000 000--30 427 671 bp). CSDs are presented in blue and LSDs in orange. The dark block highlights the transition region from CSD in Col-0 to LSD in 4 × Col-0 (28 850 000--29 550 000 bp). (C) Pie chart representing the percentages of genomic structure domain changes between Col-0 and 4 × Col-0. LSD to LSD represents the loose chromatin domains in both Col-0 and 4 × Col-0. CSD to CSD represents the compact chromatin domains in both Col-0 and 4 × Col-0. LSD to CSD denotes loose chromatin domains in Col-0 but compact in 4 × Col-0, whiles CSD to LSD denotes compact chromatin domains in Col-0 but loose in 4 × Col-0. (D) Bar graph showing the statistics of structure domain changes in all chromosomes between Col-0 and 4 × Col-0. The annotations of CSD to LSD and LSD to CSD are the same as in (C).](gkz511fig3){#F3} We also analyzed the genome elements in the chromosome regions converted from CSD to LSD with RepeatMasker. The results showed that the CSD to LSD regions contain retroelements (1.89%), DNA transposons (1.36%), small RNA (0.08%), satellite DNA (0.06%), low complexity DNA (0.41%) and simple repeat sequence (1.29%). Then we annotated the genes in the CSD to LSD switched regions with ARAPORT (Arabidopsis Information Portal). We found that there were 1749 coding genes (83%), 310 non-coding genes (14.7%) and 48 pseudogenes (2.3%). Similarly, we found that LSD to CSD regions contain retroelements (3.75%) and DNA transposons (2.51%), small RNAs (0.14%), low complexity DNA (0.38%) and simple repeat sequence (1.29%). The same gene annotation was also performed in these regions. The results showed that there were 939 coding genes (83.3%), 160 non-coding genes (14.2%) and 28 (2.5%) pseudogenes. The specific chromatin structure domain correlates with histone modifications H3K4me3 and H3K27me3 in autotetraploid. {#SEC3-4} --------------------------------------------------------------------------------------------------------------------- The changes of chromatin structure domains in the autotetraploid prompted us to test whether histone modifications associate with these alterations. To examine the epigenomic dynamics between autotetraploid and diploid Arabidopsis, we analyzed H3K4me3 and H3K27me3 histone modifications by ChIP-seq. At the genome-wide scale, we found that the levels of H3K4me3 and H3K27me3 in 4 × Col-0 are only slightly lower than those in Col-0 (Figure [4A](#F4){ref-type="fig"} and [B](#F4){ref-type="fig"}) with high Pearson\'s correlation coefficients between the two replicates of each experiment ([Supplementary Figure S8](#sup1){ref-type="supplementary-material"}). However, there were no obvious differences of H3K4me3 and H3K27me3 on each individual chromosome ([Supplementary Figure S9A and B](#sup1){ref-type="supplementary-material"}). These results suggested that the global levels of H3K4me3 and H3K27me3 were not obviously affected by the genome duplication. We then analyzed the H3K4me3 and H3K27me3 modifications on chromatin structure domains. The results showed that the CSD compartments in both Col-0 and 4 × Col-0 had a higher level of H3K27me3 and lower level of H3K4me3 compared to LSD (Figure [4C](#F4){ref-type="fig"} and [D](#F4){ref-type="fig"}), and LSD compartments in both Col-0 and 4 × Col-0 had a lower level of H3K27me3 and higher level of H3K4me3 compared to CSD (Figure [4C](#F4){ref-type="fig"} and [D](#F4){ref-type="fig"}). ![The relationship between chromatin status and histone modifications in Col-0 and 4 × Col-0 Arabidopsis. (A) H3K4me3 levels in genomes of Col-0 and 4 × Col-0 Arabidopsis, plotted as the ratio of H3K4me3 to H3 ChIP-seq. (B) H3K27me3 levels in genomes of Col-0 and 4 × Col-0 Arabidopsis, plotted as the ratio of H3K27me3 to H3 ChIP-seq. (C, D) H3K4me3 and H3K27me3 levels in CSD and LSD in Col-0 (C) and 4 × Col-0 (D). CSD regions have a higher level of H3K27me3 RPKM (reads per kilobase per million mapped reads) and lower level of H3K4me3 RPKM in both Col-0 and 4 × Col-0, whereas LSD regions have a higher level of H3K4me3 RPKM and lower level of H3K27me3. (E) Boxplot showing the ratio of H3K4me3 in 4 × Col-0 to Col-0 at the converted or no-converted structure domains. (F) Boxplot showing the ratio of H3K27me3 in 4 × Col-0 to Col-0 at the converted or no-converted structure domains. The p values were tested by Wilcoxon--Mann--Whitney test.](gkz511fig4){#F4} To gain insight into the correlation between histone modification and chromatin domain, we analyzed the distribution of histone modifications in the chromatin regions with converted and unconverted structure domains. Compared to the unconverted compacted structure domain (CSD to CSD in Figure [4E](#F4){ref-type="fig"}), we found that the level of H3K4me3 in 4 × Col-0 was higher in the chromatin regions where CSDs in Col-0 were converted to LSDs in 4 × Col-0 (CSD to LSD in Figure [4E](#F4){ref-type="fig"} and [Supplementary Figure S9C](#sup1){ref-type="supplementary-material"}). Similarly, compared with the unconverted loose structure domain (LSD to LSD in Figure [4E](#F4){ref-type="fig"}), we observed a lower level of H3K4me3 in 4 × Col-0 in the regions where LSDs in diploid were converted to CSDs in 4 × Col-0 (LSD to CSD in Figure [4E](#F4){ref-type="fig"} and [Supplementary Figure S9C](#sup1){ref-type="supplementary-material"}). Similarly, we compared the H3K27me3 modification level between converted and unconverted structure domains. The results showed that the level of H3K27me3 in 4 × Col-0 was lower in the CSD to LSD converted regions (Figure [4F](#F4){ref-type="fig"} and [Supplementary Figure S9D](#sup1){ref-type="supplementary-material"}) compared to unconverted CSD (CSD to CSD in Figure [4F](#F4){ref-type="fig"}), and the level of H3K27me3 in 4 × Col-0 was higher in the LSD to CSD converted regions (Figure [4F](#F4){ref-type="fig"} and [Supplementary Figure S9D](#sup1){ref-type="supplementary-material"}) compared to unconverted LSD (LSD to LSD in Figure [4F](#F4){ref-type="fig"}). We also checked the correlation between SD and the changes in H3K4me3 and H3K27me3 with Jaccard index. The results showed that the histone modifications correlate with the switched SDs ([Supplementary Table S4](#sup1){ref-type="supplementary-material"}). Together, these results indicated that the loose structure domains tend to enrich in the active histone modification H3K4me3 and the compacted structure domains are associated with the repressive histone modification H3K27me3. When the chromatin status in some loci changes upon genome duplication, the histone modifications in these domains altered accordingly. We also analyzed the histone modification in the unconverted regions. Interestingly, we found that the CSD to CSD regions in the 4 × Col-0 had a higher H3K27me3 level than that of Col-0, and the LSD to LSD in the 4 × Col-0 had higher H3K4me3 level and lower H3K27me3 level than that of Col-0 ([Supplementary Figure S9C and D](#sup1){ref-type="supplementary-material"}), which implicated that the compacted chromatin regions might become more condensed and the loose chromatin regions become more decondensed after genome duplication. The transcriptions of a population of genes in autotetraploid Arabidopsis are specifically regulated by chromatin interaction {#SEC3-5} ----------------------------------------------------------------------------------------------------------------------------- Differential expression analyses showed that the transcription levels of 743 genes were obviously changed in autotetraploid. First, we analyzed whether the changes in gene expressions were associated with histone modifications by comparing the H3K4me3 and H3K27me3 modification profiles between 4 × Col-0 and Col-0. The differentially-expressed genes did not show obvious differences in patterns and levels of H3K4me3 and H3K27me3 in autotetraploid Arabidopsis compared with these genes in wild type ([Supplementary Figure S10A--D](#sup1){ref-type="supplementary-material"}). Similar to all genes, the H3K4me3 was observed to be enriched in transcription-start site (TSS) ([Supplementary Figure S10A](#sup1){ref-type="supplementary-material"} to B) in both up- and down-regulated genes. Interestingly, compared to all genes, H3K27me3 shows elevated levels only in the bodies of down-regulated genes but not in up-regulated genes both in Col-0 and 4 × Col-0 ([Supplementary Figure S10C](#sup1){ref-type="supplementary-material"} to D). In addition, the relative amounts of H3K4me3 and H3K27me3 in differentially expressed genes were similar in Col-0 and 4 × Col-0 with slight down-regulation of H3K27me3 for up-regulated genes ([Supplementary Figure S10E and F](#sup1){ref-type="supplementary-material"}). To further address the mechanism underlying the gene regulation during genome duplication, we asked whether the observed differential gene expressions between autotetraploid and diploid associated with chromatin packing. We aligned the differentially expressed genes with the genes localized in the altered interaction frequency chromatin bins. We found that most of genes (539/743, about 72.5%) in the differentially expressed gene library localized in the differential interaction bins (Figure [5A](#F5){ref-type="fig"} and [Supplementary Table S5](#sup1){ref-type="supplementary-material"}). Bootstrapping randomized analysis was then performed to analyze the confidence of this phenomenon. To this end, we randomly selected 1000 group of equal number (743) of no-regulated genes to determine the percentage of those genes fallen into the differentially interacting bins. The result showed that the top 5 percentile of randomly selected control genes is \<60%, which is lower than the percentage of the differentially expressed genes localized in the differential interaction bins (72.5%) (Figure [5B](#F5){ref-type="fig"}), indicating that the gene expression in autotetraploid Arabidopsis associated with chromatin interaction. We also found that these 539 genes were related to starvation and stimulus response pathways (Figure [5C](#F5){ref-type="fig"}), consistent to the GO analysis of the differentially expressed genes between diploid and autotetraploid ([Supplementary Table S2](#sup1){ref-type="supplementary-material"}). ![The relationship between chromatin structure and gene transcription in Col-0 and 4 × Col-0 Arabidopsis. (A) Venn diagrams showing numbers of genes in differential interaction bins (dark blue) and differentially expressed genes (yellow) in Col-0 and 4 × Col-0 Arabidopsis. The 539 genes were overlapped from the two conditions. (B) Histogram of randomly selected no-regulated genes in differentially interaction bins (n = 1000). The top 5 percentile of randomly selected control genes was labeled in red. (C) GO analysis of overlapped genes between the differentially expressed genes and those in differential interaction bins. (D) Expression levels of genes in the CSD (brick red) and LSD (blue) in Col-0 and 4 × Col-0 Arabidopsis. (E) Boxplot of the log2 fold change of expression levels from genes residing in the regions with chromatin structure domain transition between LSD and CSD or without domain transition (no converted). (F) Promoter-promoter interaction frequencies in Col-0 and 4 × Col-0 Arabidopsis. (G) Enhancer-promoter interaction frequencies in Col-0 and 4 × Col-0 Arabidopsis. The p values were tested by Wilcoxon rank-sum test.](gkz511fig5){#F5} To uncover the structure domains (SD) in the transcriptional regulation of the genome-doubled Arabidopsis, we analyzed the normalized RNA-seq RPKM in LSDs and CSDs, and found that genes in the LSD compartments display higher expression levels than those in CSD compartments for both Col-0 and 4 × Col-0 (Figure [5D](#F5){ref-type="fig"}). We then analyzed the differentially-expressed genes at the regions where the SD transitions occurred. We found that the differentially expressed genes in the regions of CSD to LSD tend to be obviously up-regulated (with 37 up-regulated genes and 7 down-regulated genes) (Figure [5E](#F5){ref-type="fig"}), and differentially expressed genes in the regions of LSD to CSD in autotetraploid (Figure [5E](#F5){ref-type="fig"}) tend to be slightly down-regulated (with 16 down-regulated genes and 13 up-regulated genes). We also found that a number of differentially expressed genes (657 genes) localize in the un-converted SD regions (Figure [5E](#F5){ref-type="fig"}), indicating that the chromatin interaction, but not the SD transition, is the major factor for the gene regulation upon genome duplication in Arabidopsis. To gain more insight into the transcription regulation, we tested the genome-wide promoter-mediated interactions to address possible mechanisms through long-range interactions. We aligned all the promoters into the interaction bins and calculated the numbers of interactions among these bins. The results indicated that the promoter-promoter interaction frequency in the autotetraploid was higher than that in wild type (Figure [5F](#F5){ref-type="fig"}), implicating that the promoters in the polyploid have more intense interactions among them. Next we examined the interaction frequencies of other cis-regulatory elements like enhancers using the candidate enhancer library identified from DNase I hypersensitive site-sequencing data ([@B46]). We found that the enhancer-promoter interaction frequency in autotetraploid is also higher than that in diploid Arabidopsis (Figure [5G](#F5){ref-type="fig"}), implicating that long-range chromatin interaction might play an important role in the regulation of the doubled genome. Arabidopsis genome duplication remodeled the FLC gene loop {#SEC3-6} ------------------------------------------------------------ It was reported that autotetraploid confers a series of physiological consequences upon genome duplication. An obvious phenotype of the autotetraploid Arabidopsis plants is later flowering which might be regulated by the differentially expressed genes. We noticed that the transcript level of a floral repressor gene FLOWERING LOCUS C (FLC) was higher in autotetraploid than that in Col-0 in RNA-seq data ([Supplementary Table S1](#sup1){ref-type="supplementary-material"}). Recent studies revealed that the N- and C-terminal flanking regions of FLC form a gene loop which is affected by the chromatin-remodeling factor BAF60 ([@B47],[@B56],[@B57]). To compare the gene loop in FLC locus between autotetraploid and Col-0, we performed Chromatin Conformation Capture (3C) experiments. The results showed that the relative amount of FLC loop increased in autotetraploid plants (Figure [6A](#F6){ref-type="fig"}). ![Comparisons of gene loop and histone modifications in FLC locus between Col-0 and 4 × Col-0 Arabidopsis. (A) Chromatin interaction frequency of FLC in wild type and autotetraploid as revealed by 3C. A schematic representation of the FLC locus is shown at the top, and the arrow indicates the position of the transcription start site. Exons are represented as black boxes and introns as black lines, dotted lines indicate BamH1 or BglII restriction sites, and the primer locations were marked by black triangles. Roman numbers represent the FLC gene regions. IF means interaction frequency. (B--D) ChIP-PCR assays to analyze the H3K27me3 (B), H3K4me3 (C) and H3 (D) enrichments at FLC locus. A schematic representation of the FLC locus is presented at the top of (B). The positions of nine primer pairs ([@B1]) used for ChIP-quantitative PCR are indicated. Values came from three biological replicates each with three technical repeats, and were presented as ratio to 1% Input.](gkz511fig6){#F6} We then assessed the levels of H3K27me3 and H3K4me3 by chromatin immunoprecipitation (ChIP-PCR) using primers covering promoter, exon and intron of FLC ([Supplementary Table S6](#sup1){ref-type="supplementary-material"}) ([@B57]). We found that the level of H3K27me3 in the autotetraploid decreased across the full length of gene locus compared with Col-0 (Figure [6B](#F6){ref-type="fig"} and [C](#F6){ref-type="fig"}). In contrast, we observed no changes in the levels of H3K4me3 and H3 which was served as a control (Figure [6D](#F6){ref-type="fig"}). The elevated levels of gene loop and the loss of H3K27me3 modification in FLC locus in autotetraploid implicated the role of chromatin status in the flowering regulation. DISCUSSION {#SEC4} ========== Polyploid with two or more sets of genomes is more common in plants. Most researches focused on the role of polyploid in eukaryote evolution ([@B5],[@B58],[@B59]). Comparing with allopolyploid, autopolyploid having homologous genomes received less attention. In fact, the rate of autopolyploid formation is higher than allopolyploid in nature ([@B32]). Our results showed that genome-doubled Arabidopsis acquires morphologic traits like, more and larger rosette leaves and elevated biomass which are similar to other polyploidy plants ([@B31],[@B60]). We also noticed that the genes related to physiological and ecological tolerances were changed in autotetraploid, which might relate to the phenotypes of autotetraploid with environmentally adaptive benefits. In this work, we investigated how genome doubling affects the chromatin architecture and the relationship between chromatin structure and transcription regulation. Recent studies have illustrated that polyploidy genome is dynamic and undergoes structural alterations ([@B61]). The separated centromeres marked by Arabidopsis CenH3 (HTR12)-GFP in 4 × Col-0 indicated that the homologous chromosomes do not merged together after genome duplication (Figure [2A](#F2){ref-type="fig"}), implicating that each chromosome occupies a relatively separated space. Also, our results showed that the ratio of observed and expected chromosome interaction frequency was similar in wild type and autotetraploid Arabidopsis, which suggested that each chromosome forms a relatively independent chromosome territory in autotetraploid ([Supplementary Figure S3A](#sup1){ref-type="supplementary-material"}). Our comparative analysis of the genome-wide chromatin interaction between 4 × Col-0 and Col-0 by Hi-C revealed that autotetraploid Arabidopsis has its specific conformation with decreased intra-arm interactions and increased inter-chromosome interactions (Figure [2](#F2){ref-type="fig"}). It was reported that the volume of autotetraploid nuclei increased approximately only by a factor of 1.7 but not 2 compared with diploid, the relatively smaller nuclear volume indicated the higher average chromatin density ([@B64]). Thus, one postulation is that the increased inter-chromosome interactions in autotetraploid lead to higher chromosomal compaction, resulting in a smaller nuclear size. We found that the genomes of autotetraploid and diploid showed similar compacted structure domains (CSDs) and loose structure domains (LSDs) with ∼12% structural domains switched between CSD and LSD (Figure [3C](#F3){ref-type="fig"}), indicating that genome duplication can remodel the chromatin structures in some loci. Analysis of switched domains showed that the majority of transition events occur on chr1, chr2 and chr4. We proposed that the big and gene-rich chromosome 1 might have a higher intensity of intra-chromosome communication. Previous studies reported that the entire left arm of Arabidopsis chromosome 4, including pericentromere regions and NAD (nucleolus associated chromatin domains), associates with nucleolus, and the nucleolus organization region 4 (NOR4) is actively transcribed. In contrast, the short arm of chromosome 2 is excluded from nucleolus and NOR2 is inactive ([@B65],[@B66]). In autotetraploid, major chromatin domain switches of CSD to LSD occur on chr2, which might result from the relocation of chr2 to nucleolus and decondensation of the chromatin, resulting in the active transcription. For chr4, the chromatin domain switches of LSD to CSD might implicate that NOR4 is not associated with nucleolus in autotetraploid, and the transcription of NOR4 rRNA or NAD-related genes failed to be intensely activated. Further studies are necessary to test these hypotheses. Histone modifications of H3K4me3 and H3K36me3 on euchromatin were known to correlate to LSDs, whereas the polycomb-associated H3K27me3 is connected with heterochromatin that contains CSDs ([@B17],[@B25],[@B29]). In contrast, H3K27me1, H3K18ac and H3 showed weak correlations with the chromatin packing status ([@B29]). We assessed the relationships between H3K27me3/H3K4me3 and chromatin status in autotetraploid and wild type. Our results showed that the global H3K4me3 and H3K27me3 did not show obvious difference after genome doubling, which might be due to that autotetraploid has two identical genomes. In addition, we found that LSDs (loose structure domains) tends to have a higher level of the active marker (H3K4me3) and a lower level of repressed marker (H3K27me3), whereas CSDs have a lower level of H3K4me3 and higher level of H3K27me3 in both Col-0 and genome-duplicated Col-0. Importantly, we found that H3K27me3 and H3K4me3 could reflect the genome structure domain transformation. The CSD to LSD converted regions in the autotetraploid present a higher level of H3K4me3 and lower level of H3K27me3, while the LSD to CSD converted regions showed contrary tendency (Figure [4E](#F4){ref-type="fig"}, [F](#F4){ref-type="fig"} and [Supplementary Figure S9C and D](#sup1){ref-type="supplementary-material"}), indicating that chromatin structure domain switches are correlated with the changes of histone modifications upon genome duplication. It was known that H3K4me3 is an active marker while H3K27me3 a repressive marker of gene expression ([@B67]). In autotetraploid Arabidopsis, we found that the H3K4me3 profile did not show obvious differences both in up- or down-regulated genes compared with the total genes (Supplementary S10A, B). However, although the modification pattern of H3K27me3 in the up- and down-regulated genes were similar both in Col-0 and 4 × Col-0, the up-regulated genes have lower levels of H3K27me3 while the down-regulated genes have higher levels of H3K27me3 compared with the total genes (Supplementary S10C, D), implicating that H3K27me3 might play a role in defining chromatin identity for the transcriptional regulation of the related genes during genome doubling. Further quantification analysis showed that the total level of H3K4me3 or H3K27me3 in up- or down- regulated genes did not reveal obvious differences between Col-0 and 4 × Col-0 ([Supplementary Figures S10E and F](#sup1){ref-type="supplementary-material"}), suggesting that the transcription changes in autotetraploid Arabidopsis are not tightly linked to the H3K4me3 and H3K27me3 modifications, similar to the previous study that changes in H3K27me3 were not significantly correlated with gene expression at shoot apical meristems during the transition to flowering in Arabidopsis ([@B68]). To better understand the gene regulation underlying genome duplication in the autotetraploid, we addressed the relationship between chromosome structure and gene transcription. Our results showed that 72.5% of the differentially expressed genes localized in the differential interaction bins, which indicated that the chromatin interaction might play a role in the regulation of the doubled genome. In addition, we found that a large amount of these differentially expressed genes associate with and responses to stimuli, suggesting that the chromatin interaction might be important for the plant to acclimation (Figure [5C](#F5){ref-type="fig"}). Although some differentially expressed genes localize to the converted structure domains where the genes tend to be up-regulated in the CSD to LSD regions, most of the differentially expressed genes localize in the unconverted regions, which implied that the SD transition is not a major factor in gene regulation upon genome duplication. According to these results, we propose that the chromatin interaction is important in regulating gene expression in autotetraploid. At the specific gene locus of FLC, we found that there are higher intensities of loops around FLC gene in autotetraploid Arabidopsis compared with diploid. This structure might help to recruit more RNA polymerase complexes to activate the FLC expression to repress flowering. The elevated promoter-promoter and enhancer-promoter interaction frequencies in autotetraploid confirmed the elevated level of long-range interactions in autotetraploid. We also found that the level of H3K27me3 modification but not H3K4me3 decreased on FLC gene in autotetraploid Arabidopsis, which promoted the expression of FLC in another way. Together, our results suggested that H3K4me3 and H3K27me3 modifications are not the major players in the global regulation of autotetraploid genomes. For some specific loci, they may synergistically regulate gene expression together with chromatin structure in autotetraploid. Our work thus revealed a new perspective regarding the genome organization and gene regulation in autotetraploid plants. DATA AVAILABILITY {#SEC5} ================= The raw and processed Hi-C, ChIP-seq and RNA-seq datasets have been submitted to NCBI Gene Expression Omnibus (GSE114950). Supplementary Material ====================== ###### Click here for additional data file. Author contributions: H.Z. and Y.D.F. conceived and designed the experiments, H.Z. conducted Hi-C and ChIP-seq experiments, Y.Z. for RNA-seq experiment, R.Q.Z., Y.L.W., P.H., Y.P.F., G.L.L., H.Z. and Y.D.F. analyzed and interpreted the data. Y.D.F. and G.L.L. for critical manuscript modification. All authors read and approved the final manuscript. SUPPLEMENTARY DATA {#SEC6} ================== [Supplementary Data](https://academic.oup.com/nar/article-lookup/doi/10.1093/nar/gkz511#supplementary-data) are available at NAR Online. FUNDING {#SEC7} ======= National Key Research and Development Program of China \[2016YFD0100902\]; National Natural Science Foundation of China \[31871230 to Y.F., 31771402 and 91440114 to G.L., 31501076 to Y.P.F.\]. Funding for open access charge: Yuda Fang from grant of National Natural Science Foundation of China \[31871230\]. Conflict of interest statement. None declared.	{ "pile_set_name": "PubMed Central" }
Introduction ============ Inflammatory bowel diseases (IBD) are chronic inflammatory disorders of the gastrointestinal tract with increasing incidence worldwide ([@ref-37]). Inflammation develops because of an exaggerated immune response to luminal antigens derived from the gut microbiota or from infecting pathogens in genetically predisposed individuals ([@ref-38]; [@ref-5]). Genetic and environmental factors also contribute to IBD development, although the exact atiology remains to be fully disclosed. Inflammation leads to profound alterations of the intestinal architecture, which influence intrinsic neuronal circuitries constituting the enteric nervous system (ENS). Neuronal adaptation includes morphology, excitability and synaptic changes, with consequent alterations of sensory, motor and secretory functions contributing to the development of IBD-associated symptoms ([@ref-12]; [@ref-16]; [@ref-4]). The cross-talk occurring among different cell populations of the enteric microenvironment (i.e., neurons, enteric glia, interstitial cells of Cajal, immunocytes), the saprophytic microbial flora and infiltrating inflammatory cells may account for the structural and functional changes occurring in enteric circuitries in different physiological and pathological conditions, including inflammation ([@ref-21]; [@ref-32]; [@ref-10]). For example, neuro-immune interactions may help explain the occurrence of damage at non-inflamed gastrointestinal sites, distant from the inflammatory process ([@ref-12]; [@ref-16]). It is, thus, particularly important to unveil possible molecular pathways sustaining neuronal degeneration during inflammatory states, both on site and, more distantly, along the gastrointestinal tract. Indeed, research in this field opens an exciting potential scenario where new molecules displaying high efficacy in the treatment of IBD may became available in association and/or substitution of conventional anti-inflammatory, immunosuppressive and biologic drugs ([@ref-39]; [@ref-36]; [@ref-44]; [@ref-40]). In the present study, we investigated possible changes in the expression of homeoprotein transcription factors, orthodenticle homeobox protein 1 and 2 (OTX1 and OTX2), in the rat myenteric plexus after an experimentally-induced colitis. OTX1 and OTX2 are transcription factors derived from homeobox-containing genes located, in humans, on chromosome 2p15 and 14q22, respectively ([@ref-26]). OTX1 and OTX2 homeobox genes are the vertebrate orthologues to the Drosophila orthodenticle homeobox genes and are essential for the specification, regionalization and terminal differentiation of the rostral part of the central nervous system during development ([@ref-1]; [@ref-29]). OTX2 is an essential factor influencing the maturation of various neuronal subpopulations, including oculomotor, thalamic glutamatergic progenitors and midbrain dopaminergic neurons ([@ref-42]; [@ref-41]). Recent evidence suggest for OTX2 a fundamental role in regulating neuronal plasticity in the visual cortex ([@ref-8]). During development in both rodent and human, OTX1 is prevalently expressed in proliferative zones of the neocortex and is described to have a fundamental role in determining the volume and proportion of the visual cortex ([@ref-2]). Despite their central role in neuronal development and maturation, there are lines of evidence suggesting that both OTX1 and OTX2 are expressed also in adult tissues and may participate to the development of pathological conditions linked to inflammation ([@ref-24]; [@ref-3]). In addition, the involvement of both factors in tumor growth, such as in medulloblastoma, the most frequent malignant brain tumor in children, and in breast cancer has been demonstrated ([@ref-17]; [@ref-45]). Furthermore, OTX1 is involved in the epithelial damage promoting colorectal cancer progression ([@ref-47]). Recently, both OTX1 and OTX2 were found to be expressed in normal adult rat small intestine myenteric ganglia: OTX1 has been predominantly found in enteric glial cells and in few myenteric neurons, while OTX2 was uniquely expressed in the soma of a relatively small percentage of myenteric neurons ([@ref-19]). Expression of both orthodenticle homeobox proteins was upregulated during intestinal ischemia/reperfusion injury and correlated with alterations of the intestinal neuromuscular function in this pathophysiological condition ([@ref-19]). Such changes involved an interplay between both transcription factors and enteric nitrergic neurons, sustaining the development of altered motor responses involving NO production ([@ref-19]). In our study, we aim for the first time to evaluate possible changes of OTX1 and OTX2 expression in rat colon myenteric ganglia, after dinitro-benzene sulfonic (DNBS) acid-induced colitis, by means of molecular biology and morphological approaches. In view of the possible adaptive changes induced by the inflammatory insult at remote sites from the injury, we also investigated the consequences of DNBS treatment on both OTX1 and OTX2 expression in small intestine myenteric plexus. Materials and Methods ===================== Animals ------- Male Sprague--Dawley rats (weight 250--300 g, Envigo, Udine, Italy), were handled following principles of good laboratory animal care, in accordance with specific national and international laws and regulation. Animals were maintained at a regular 12/12 h light/dark cycle, under controlled environmental conditions (temperature 22 ± 2 °C; relative humidity 60--70%), with free access to a standard laboratory tap water and chow. The protocol was approved by the Animal Care and Use Ethics Committee of the University of Pavia (n. 3/2011). DNBS-induced colitis -------------------- The model of DNBS-induced experimental colitis in rats was chosen since the inflammatory response develops rapidly, reaching the maximum grade in 6 days, and shares many features with the human IBD response ([@ref-18]). A single dose (30 mg) of 2,4-dinitro-benzene-sulfonic acid (DNBS, ICN Biomedicals, CA, USA), dissolved in 0.25 ml of 50% ethanol, was administered to isofluorane anesthetized rats by means of a polyethylene (PE-60) catheter into the colon 8 cm proximal to the anus. Ethanol, owing to its ability to break the mucosal barrier thus permitting DNBS penetration into the bowel wall, was administered (0.25 ml) to control animals as a vehicle. The dose of DNBS used in the study induces adequate inflammation, without causing unnecessary distress, suffering or mortality to the animals. DNBS-treated and control rats were kept separated during the study. After 6 days, animals were euthanized and the small intestine and distal colon was removed and washed with a physiological Tyrode's solution (in mM: 137 NaCl; 2.68 KCl; 1.8 CaCl~2~·2H~2~O; 2 MgCl~2~; 0.47 NaH~2~PO~4~; 11.9 NaHCO~3~; 5.6 glucose). Throughout the treatment period, animals were monitored to evaluate possible physiological and behavioral changes (i.e., changes in body weight, respiration, occurrence of diarrhea, alterations of posture and in the appearance of the coat) as marks of suffering and distress. Assessment of colonic damage ---------------------------- The severity of intestinal inflammation was evaluated macroscopically and histologically. Criteria for the macroscopic evaluation included the presence of adhesions between the colon and other intra-abdominal organs, the extension of hyperemia and macroscopic mucosal damage, the thickening of the colonic wall and the consistency of colonic fecal material, all attributed with a score ranging from 0 (no damage) to 6 (maximum damage) [@ref-18]. For the microscopic histological evaluation, standard hematoxylin and eosin (HE) staining was carried out on serial sections (7 μm) of small intestine and colonic segments, from control and DNBS-treated rats, which were first fixed for 24--48 h with 4% formaldehyde in acetate buffer 0.05 M and successively embedded in paraffin. Preparations were observed under a light microscope (Nikon Eclipse Ni; Nikon, Tokyo, Japan) and data were recorded using a DS-5M-L1 digital camera system (Nikon Corporation, Tokyo, Japan). Myeloperoxidase activity ------------------------ To evaluate the development of the inflammatory state caused by neutrophil infiltration, myeloperoxidase (MPO) was measured as previously described by [@ref-19]. Briefly, intestinal segments deprived of the mucosal layer were homogenized (50 mg/mL) with a solution of ice cold potassium phosphate buffer (50 mm, pH 6.0) containing 0.5% hexadecyl trimethylammonium bromide (HTAB). After centrifugation (14,000 rpm, 20 min, 4 °C), an aliquot of the supernatant fraction was mixed with HTAB-phosphate buffer containing O-dianisidine dihydrochloride with hydrogen peroxide before spectrophotometrically recording at 460 nm changes in the rate of absorbance. MPO activity was expressed in units (U)/wet tissue weight, where U defined the amount of enzyme that degrades 1 μmol/min of hydrogen peroxide at 25 °C. Experiments were performed six times for each experimental group. Immunofluorescence ------------------ Paraffin-embedded tissue sections and whole-mount preparations were processed for the immuno-localization of OTX1 and OTX2 on small intestine and colon of control and DNBS-treated rats, as previously described in [@ref-10] and [@ref-15]. Paraffin sections ----------------- Seven μm paraffin cross sections incubated with CD45, as a marker of inflammatory infiltration, or either with OTX1 or OTX2 with the pan neuronal marker HUC/D, for double staining. Optimal dilutions of antibodies are reported in [Table 1](#table-1){ref-type="table"}. Coverslips were mounted with Citifluor mounting medium and then observed with fluorescent microscopy (Nikon Instruments, Melville, NY, USA). 10.7717/peerj.8442/table-1 ###### Primary and secondary antisera and their respective dilutions used for Western Blot (WB) assay and immunohistochemistry (HC). ![](peerj-08-8442-g010) Antiserum Dilution (WB) Dilution (HC) Source Host species ----------------------------------------------- --------------- --------------- -------------------------------------- -------------- Primary antisera OTX1 1:100 Invitrogen (PA5-67901) Rabbit OTX1 1:200 1:100 Santa Cruz (sc133872) Rabbit OTX2 1:100 -- R&D systems (AF1979) Goat OTX2 Alexa Fluor 594 conjugated -- 1:100 Bioss Antibodies (bs-11597R-A594) Rabbit HUC/D biotin -- 1: 100 Invitrogen (16A11) Mouse β-actin 1:1,000 -- Cell Signalling Technology (8H10D10) Mouse nNOS -- 1:200 Millipore (AB1529) Sheep nNOS -- 1:50 Santa Cruz (sc648; R-20) Rabbit iNOS -- 1:50 Santa Cruz (sc8310; H-174) Rabbit S100 -- 1:100 Merck Millipore (MAB079-1) Mouse CD45 -- 1:100 Merck Millipore (MAB079-1) Mouse Secondary antisera & streptavidin complexes Anti-rabbit Alexa Fluor 488 -- 1:200 Molecular Probes (A21206) Donkey Anti-mouse Alexa Fluor 488 -- 1:200 Molecular Probes (A21202) Donkey Cy3-conjugated streptavidin -- 1:500 Amersham (PA43001) FITC-conjugated straptavidin -- 1:200 Molecular Probes (SA1001) Anti-rabbit IgG HRP peroxidase conjugated 1:5,000 -- Santa Cruz (sc2004) Goat Anti-goat IgG HRP peroxidase conjugated 1:10,000 -- Santa Cruz (sc2020) Donkey Anti-mouse IgG, HRP-linked 1:1,000 -- Cell Signalling Technology (\#7076) Horse Note: Supply companies: Amersham, GE Healthcare, Buckinghamshire, UK; Bioss antibodies, MA, USA; Cell Signaling Technology, Danvers, MA, USA; Invitrogen, Thermo Fisher, MA, USA; Merck Millipore Burlington, MA, USA; Molecular Probes, Thermo Fisher, MA, USA; R&D systems, Minneapolis, CA, USA; Santa Cruz Biotechnology, CA, USA; Sigma--Aldrich, Milano, Italy. Whole-mount preparations ------------------------ Whole-mount preparation immunolabeling was performed on the longitudinal muscle with the attached myenteric plexus (LMMP) proceeding with a double-staining with optimally diluted primary and secondary antibodies ([Table 1](#table-1){ref-type="table"}). Preparations were mounted onto glass slides with a mounting medium (Vectashield with DAPI; Vector Lab, Burlingame, CA, USA) and analyzed with Image J NIH image software (<http://imagej.nih.gov/ij>). Total neuron number per ganglion area was expressed as the ratio between the number of neurons positive for HuC/D and the total ganglion area (µm^2^). Areas of 15--19 myenteric ganglia from LMMP preparations of 5 vehicle-treated CTR and 5 DNBS-treated animals were measured at 40× magnification by tracing boundaries around stained cell somas (HuC/D) ([@ref-19]). Neuronal cell body area was measured with Image J. The number of OTX1 or OTX2 immunoreactive neurons that co-localized with HuC/D were counted from a total of 10--15 ganglia (5 animals for each experimental group). The proportion of OTX1 and OTX2 immunoreactive myenteric neurons, was expressed as percentage of the total number of HuC/D positive neurons ([@ref-19]). To evaluate negative controls and interference control staining, both primary and secondary antibodies were omitted, incubating the colonic whole-mounts with non-immune serum from the same species in which the primary antibodies were raised. In all these conditions, no specific signal was detected. Pictures from preparations, collected with a Leica TCS SP5 confocal laser scanning system (Leica Microsystems GmbH, Wetzlar, Germany) were then processed with Adobe-Photoshop CS6S software. Real time quantitative RT-PCR ----------------------------- To evaluate the influence of DNBS-induced colitis on OTX1, OTX2, TNFα, pro-IL1β, IL6, HIF1α, VEGFα, mRNA levels, total RNA from rat small intestine and colon LMMPs was extracted with TRIzol (Invitrogen, Carlsbad, CA, USA) and treated with DNase I (DNase Free, Ambion) to remove possible traces of contaminating DNA. 2.5 μg of total RNA was retrotranscribed using the High Capacity cDNA synthesis kit (Applied Biosystems, Life Technologies, Grand Island, NY, USA). Quantitative RT-PCR was performed on the Abi Prism 7000 real-time thermocycler (Applied Biosystems, Foster City, CA, USA) with Power Sybr Green Universal PCR Master Mix (Applied Biosystems, Foster City, CA, USA) following manufacturer's instructions. Primers were designed to have a similar amplicon size and similar amplification efficiency as required for the utilization of the 2^−ΔΔCt^ method to compare gene expression ([@ref-9]), using Primer Express software (Applied Biosystems, Foster City, CA, USA) on the basis of available sequences deposited in public database ([Table 2](#table-2){ref-type="table"}). For quantitative RT-PCR a final concentration of 500 nm for each primer was used. Experiments were performed at least 7 times for each experimental group. 2^−ΔΔCt^ values obtained from the comparison between normalized Ct values of DNBS-treated samples with those obtained from control samples were used to evaluate the effect of DNBS-induced colitis on OTX1 and OTX2 mRNA levels in the small intestine and colon. 10.7717/peerj.8442/table-2 ###### Sequence of primers used in the study for the analysis of qRT-PCR. ![](peerj-08-8442-g011) Gene Sequence --------------------------------- --------------------------------- β-actin F 5′-AGGCCCCTCTGAACC-3′ R 5′-GGGGTGTTGAAGGTC-3′ OTX1 F 5′-GCGAGGAGGTGGCTCTCA-3′ R 5′-GGCTCGGCGGTTCTTGA-3′ OTX2 F 5′-CCCAATTTGGGCCGACTT-3′ R 5′-GCGTAAGGCGGTTGCTTTAG-3′ TNFα F 5′-CCCAGACCCTCACACTCAGAT-3′ R 5′-TTGTCCCTTGAAGAGAACCTG-3′ pro-IL1β F 5′-CCCTGCAGCTGGAGAGTGTGG-3′ R 5′-TGTGCTCTGCTTGAGAGGTGCT-3′ IL6 F 5′-GTGCAATGGCAATTCTGATTGTA-3′ R 5′-CTAGGGTTTCAGTATTGCTCTGA-3′ VEGFα F 5′-GCTGTGTGTGTGAGTGGCTTA-3′ R 5′-CCCATTGCTCTGTACCTTGG-3′ HIF1α F 5′-AAGCACTAGACAAAGCTCACCTG-3′ R 5′-TTGACCATATCGCTGTCCAC-3′ Western immunoblot analysis --------------------------- Purified membrane fractions obtained after successive centrifugations of homogenized intestinal LMMPs preparations were used to analyze OTX1 and OTX2 protein level as described elsewhere ([@ref-23]). Membrane incubation with OTX1 and OTX2 primary antibodies was performed by incubation with a horseradish peroxidase-conjugated secondary antisera ([Table 1](#table-1){ref-type="table"}). The signal of antibody/substrate complex was visualized by chemiluminescence using an enhanced chemiluminescence kit (ECL advance Amersham Pharmacia Biotech, Cologno Monzese, Italy), and then evaluated by densitometric analysis using Image J NIH image software. The effect of DNBS-induced colitis on OTX1 and OTX2 protein levels was expressed as the percentage variation of the optical density (expressed in arbitrary units) of OTX1 and OTX2 signals normalized to the respective β-actin, used as protein loading control, in LMMP preparations obtained from DNBS-treated animals compared to controls. Experiments were performed at least five times for each experimental group. Specificity of OTX1 and OTX2 primary antibodies was evaluated by testing their selectivity in spontaneous immortalized human Müller cell line (MIO-M1) and in rat hippocampus, respectively (data not shown) ([@ref-19]). Negative controls were performed by omitting the primary antibody. Statistical analysis -------------------- All data are expressed as mean ± S.E.M. Statistical significance was calculated by either Student's t test or by one-way ANOVA with Tukey's post hoc test, where appropriate, using GraphPad Prism (version 5.3; GraphPad Software, San Diego, CA, USA). Differences among groups were considered significant when P value were 0.05 or lower. Results ======= Assessment of colitis --------------------- ### Macroscopic assessment Body weight was significantly reduced in DNBS-treated rats compared to non-inflamed controls at day 6 after treatment (DNBS-treated: 280 ± 6.50 g, n = 7; vehicle-treated CTR: 306 ± 2.3 g, n = 7, P = 0.027 by Student's t test). Six days after DNBS administration, the distal colon was thickened and ulcerated with evident regions of transmural inflammation, adhesions between the colon and other intra-abdominal organs were often present and the bowel was occasionally dilated. Macroscopic damage score significantly increased in the colon of DNBS-treated animals in comparison with vehicle-treated controls (DNBS-treated: 4.80 ± 021, n = 10; CTR: 0.3 ± 0.08, n = 14 P \< 0.0001, by Student's t test). No significant gross morphology changes were observed in the small intestine. This was reflected by low macroscopic damage scores in both experimental groups (DNBS-treated: 0.11 ± 0.08, n = 10; CTR: 0.09 ± 0.05, n = 14, by Student's t test). These findings are in agreement with other reports, showing that after induction of an experimentally-induced colitis, the occurrence of gross macroscopic changes is obvious only on the site of injury, preserving distant sites along the gastrointestinal tract ([@ref-35]). Histological assessment ----------------------- A further set of experiments was carried out on intestinal cross sections to evaluate the impact of DNBS--induced colitis on the architecture of the rat small intestine and distal colon by means of standard HE staining, and the degree of inflammatory infiltrate by CD45 immunofluorescent staining. Small intestine and distal colon cross-sections obtained from control animals showed normal histological features, with compact myenteric ganglia, formed by healthy neuron and glial cells, laying between the circular and longitudinal muscle of the muscularis propria ([Figs. 1A](#fig-1){ref-type="fig"}, [1B](#fig-1){ref-type="fig"}, [1K](#fig-1){ref-type="fig"} and [1L](#fig-1){ref-type="fig"}). CD45 staining, indicative of inflammatory cell infiltration, was almost negligible in both small intestine and colon control cross sections ([Figs. 1C](#fig-1){ref-type="fig"} and [1M](#fig-1){ref-type="fig"}). At day 6 after DNBS treatment, small intestine specimens did not show prominent histological abnormalities ([Figs. 1F](#fig-1){ref-type="fig"} and [1G](#fig-1){ref-type="fig"}). In contrast, both mucosa and serosal epithelium of the distal colon obtained from DNBS-treated rats, displayed morphological abnormalities ([Figs. 1P](#fig-1){ref-type="fig"}--[1R](#fig-1){ref-type="fig"}). The colonic mucosal surface was irregular and crypt architecture was profoundly altered, the muscularis mucosae, submucosa and muscularis propria layers were thickened ([Fig. 1P](#fig-1){ref-type="fig"}). Prominent spaces between smooth muscle cells and leukocyte infiltration were also observed ([Fig. 1Q](#fig-1){ref-type="fig"}). In the small intestine of DNBS-treated animals myenteric neurons had a normal morphology comparable to that observed in control preparations as shown [Fig. 1G](#fig-1){ref-type="fig"}. In contrast, colonic myenteric ganglia underwent important degenerative changes, with neurons displaying cytoplasm vacuolization and irregular nuclear and cellular membrane. Large spaces between muscle cells were also evident ([Fig. 1R](#fig-1){ref-type="fig"}). After DNBS-treatment, CD45 staining significantly increased both in the small intestine and in the colon ([Figs. 1H](#fig-1){ref-type="fig"} and [1S](#fig-1){ref-type="fig"}). ![Histological evaluation, OTX1 and OTX2 staining in small intestine and colonic cross sections after DNBS-induced colitis.\ Hematoxylin-eosin (HE) staining highlights the well-preserved cellular and structural morphology of small intestine (A) and distal colon (K) cross-sections of control (vehicle-treated, CTR) rats (bar: 50 µm). Detail of a myenteric ganglion in cross-sections obtained from small intestine (B) and colon (L) of CTR rats (bars: 10 µm). Both in the small intestine (C) and distal colon (M) of CTR animals, CD45 staining, as an index of inflammatory infiltrate, was slight (bars: 10 µm). After DNBS treatment no significant morphological changes were evident in the small intestine (F), while important structural changes were observed in the distal colon (P), including increased thickness of the muscularis mucosae, submucosal layer and muscularis propria layers (bars: 50 µm). Detail of a myenteric plexus ganglion in small intestine (G) and distal colon (R) cross-sections obtained after DNBS treatment (bars: 10 µm). In particular, HE staining highlights prominent distortions in a colonic myenteric plexus ganglion (R). (Q) Shows the presence of spaces among smooth muscle cells and leukocytes infiltrates (arrowheads) in a colonic cross-section after DNBS treatment (bar 10 µm). After DNBS treatment, CD45 significantly increased both in the small intestine (H) and in the colon (S). Co-staining with the pan neuronal marker HuC/D, showed few myenteric neurons labeling for both OTX1 and OTX2 (arrows) in the small intestine (D and N) and colon (E and O) of CTR rats, inserts show details of myenteric ganglia (bars: 10 µm; inserts 5 µm). After DNBS treatment, both OTX1 and OTX2 immunoreactivity increased in myenteric neurons of both regions (small intestine: (I and T); distal colon: (J and U); bars: 10 µm, inserts 5 µm). Arrows indicate myenteric plexus ganglia. CM, circular muscle; LM, longitudinal muscle; MP, myenteric plexus; M, mucosa; SM, submucosa; Mus, Muscularis Propria.](peerj-08-8442-g001){#fig-1} We then focused our histological investigations on myenteric ganglia by carrying out immunofluorescence staining of small intestine and distal colon LMMP whole-mount preparations with the pan neuronal marker HuC/D. In this set of experiments changes in myenteric neuron number were observed in both gastrointestinal regions after DNBS treatment ([Figs. 2A](#fig-2){ref-type="fig"}--[2F](#fig-2){ref-type="fig"}). In particular, the number of myenteric neurons staining for HuC/D was significantly reduced (P \< 0.05, by one way-ANOVA with Tukey's post hoc test) with respect to values obtained in respective control preparations in both regions ([Fig. 2E](#fig-2){ref-type="fig"}). Myenteric neuron number was reduced to a significantly greater extent (P \< 0.05, by one way-ANOVA with Tukey's post hoc test) in the colon than in the small intestine ([Fig. 2E](#fig-2){ref-type="fig"}). In addition, in the colon, but not in the small intestine, DNBS-induced inflammation was associated with a significant reduction of myenteric neuron soma area (P \< 0.001, by one way-ANOVA with Tukey's post hoc test) compared to control preparations ([Fig. 2F](#fig-2){ref-type="fig"}). As expected, in accordance with previous studies, in control preparations, the number of myenteric neurons was significantly lower in the colon (P \< 0.05, by one way-ANOVA with Tukey's post hoc test) than in the small intestine ([Fig. 2E](#fig-2){ref-type="fig"}). Additionally, in the distal colon the neuronal soma area was significantly larger (P \< 0.05, by one way-ANOVA with Tukey's post hoc test) than in the small intestine ([Fig. 2F](#fig-2){ref-type="fig"}). Overall, these observations suggest that after DNBS-induced colitis, major histological changes involving myenteric ganglia, occur in the distal colon. However, also the small intestine may be influenced by the injury, as indicated by the increased infiltration of inflammatory CD45^+^ cells and by the reduction of myenteric neuron number. ![DNBS treatment-induced changes in myenteric neurons.\ (A--D) HuC/D staining of myenteric neurons in CTR and DNBS-treated whole-mount LMMP preparations. (E) Myenteric neuron number normalized per ganglion area in small intestine and distal colon LMMP whole-mount preparations obtained from DNBS-treated (solid bar) and control animals (vehicle-treated CTR, empty bar). (F) Mean myenteric neuron area calculated in small intestine and distal colon whole-mount preparations obtained from DNBS-treated (solid bar) and control animals (vehicle-treated CTR, empty bar). Values are expressed as mean ± S.E.M. N = 5 rats per group. \*P \< 0.05 and \\\*P \< 0.001 vs. values obtained in CTR animals by One Way ANOVA with Tukey's post hoc test. °P \< 0.05 vs. CTR small intestine. ^§^P \< 0.05 vs. DNBS-treated small intestine.](peerj-08-8442-g002){#fig-2} Biomolecular assessment ----------------------- The occurrence of an inflammatory response in LMMP preparations, on site and distantly from the injury, was further investigated by evaluating the levels of the mRNA of different cytokines, which are considered standard markers of intestinal inflammation, and by measuring MPO activity. In both intestinal regions, after DNBS treatment, a significant enhancement of the expression of inflammatory cytokines TNFα (P \< 0.001 in both regions, by one-way ANOVA with Tukey's post hoc test), pro-IL1β (P \< 0.01 in the ileum and P \< 0.001 in the colon, by one way-ANOVA with Tukey's post hoc test), IL6 (P \< 0.001 in both regions by one-way ANOVA with Tukey's post hoc test), and of HIF-1α (P \< 0.05 in both regions, by one-way ANOVA with Tukey's post hoc test) and VEGFα (P \< 0.01 in both regions, by one-way ANOVA with Tukey's post hoc test), was evidenced with respect to control animals ([Figs. 3A](#fig-3){ref-type="fig"}--[3E](#fig-3){ref-type="fig"}). As regards the pro-inflammatory cytokines, enhancement of TNFα mRNA was similar in the colon and in the small intestine, while pro-IL1β and IL-6 mRNA levels were significantly higher in the colon with respect to the small intestine (P \< 0.05 and P \< 0.001, respectively, by one-way ANOVA with Tukey's post hoc test) ([Figs. 1A](#fig-1){ref-type="fig"}--[1C](#fig-1){ref-type="fig"}). HIF1α and VEGFα mRNA levels after DNBS treatment was not significantly different in the small intestine with respect to the distal colon ([Figs. 1D](#fig-1){ref-type="fig"} and [1E](#fig-1){ref-type="fig"}). ![Expression of Inflammatory markers in the ileum and colon after DNBS-induced colitis.\ (A--E) qRT-PCR quantification of TNFα, pro-IL1β, IL6, HIF1α, VEGFα mRNA levels obtained in the small intestine and distal colon of control animals (vehicle-treated, CTR) and after DNBS treatment (solid bars). The relative gene expression was determined by comparing 2^−ΔΔCt^ values in CTR and DNBS-treated samples normalized to β-actin. Values are mean ± S.E.M. of at least 7--9 experiments. \*P \< 0.05, \\P \< 0.01 and \\\*p \< 0.0001 vs. values obtained in control animals; ^§^P \< 0.05 and ^§§§^P \< 0.001 vs. DNBS-treated small intestine. Significance was evaluated by one-way ANOVA with Tukey's post hoc test. (F) MPO activity measured in mucosa-deprived small intestine and distal colon segments obtained from DNBS-treated (solid bar) and control animals (vehicle-treated, CTR). Values are expressed as mean ± S.E.M. of 6 experiments. \\\*P \< 0.001 vs. values obtained in control animals, by one-way ANOVA with Tukey's post hoc test.](peerj-08-8442-g003){#fig-3} MPO activity significantly increased in rat small intestine (P \< 0.001, by one way-ANOVA with Tukey's post hoc test) and colonic (P \< 0.001, by one way-ANOVA with Tukey's post hoc test) segments compared to control animals, suggesting the occurrence of inflammation-induced neutrophil infiltration on site and distally from the injury ([Fig. 1F](#fig-1){ref-type="fig"}). Distribution of OTX1 immunoreactivity in small intestine and colon LMMP whole-mount preparations from normal and DNBS-treated animals ------------------------------------------------------------------------------------------------------------------------------------- In view of the adaptive changes observed after DNBS-induced inflammation in the rat myenteric plexus, we subsequently evaluated if this adaptation encompasses alterations in the distribution of homeobox proteins involved neuronal plasticity, OTX1 and OTX2, in cross sections and LMMP whole-mounts of the small intestine and distal colon. In cross-sections of small intestine and distal colon obtained from vehicle-treated control rats, myenteric ganglia displayed faint OTX1 immunoreactivity ([Figs. 1D](#fig-1){ref-type="fig"} and [1N](#fig-1){ref-type="fig"}). In control LMMP whole-mount preparations OTX1 antibody stained few myenteric neurons in both the small intestine (3.79 ± 1.00%, n = 15) and distal colon (4.77 ± 1.43%, n = 15) ([Fig. 4A](#fig-4){ref-type="fig"}). In both intestinal regions, OTX1 antibody stained the soma and nucleus of large and medium size myenteric neurons with either a round or an ovoidal shape ([Figs. 5A](#fig-5){ref-type="fig"}--[5C](#fig-5){ref-type="fig"} and [5G](#fig-5){ref-type="fig"}--[5I](#fig-5){ref-type="fig"}). In small intestine and distal colon cross-sections obtained from DNBS-treated animals, a significant increase of OTX1 staining was observed within the myenteric plexus with respect to control preparations ([Figs. 1I](#fig-1){ref-type="fig"} and [1T](#fig-1){ref-type="fig"}). After DNBS treatment, a significant increase in the number of OTX1 immunoreactive myenteric neurons was observed in whole-mount preparations of both small intestine (9.86 ± 0.86%, n = 15, P \< 0.0001, by one-way ANOVA with Tukey's post hoc test) and colon (40.25 ± 1.2%, n = 15 P \< 0.0001, by one-way ANOVA with Tukey's post hoc test) with respect to control preparations ([Figs. 4A](#fig-4){ref-type="fig"}, [5D](#fig-5){ref-type="fig"}--[5F](#fig-5){ref-type="fig"} and [5J](#fig-5){ref-type="fig"}--[5L](#fig-5){ref-type="fig"}). In DNBS-treated colonic whole mount preparations, the percentage of OTX1 immunoreactive myenteric neurons was significantly higher than in the small intestine ([Fig. 4A](#fig-5){ref-type="fig"}). In LMMP whole-mount preparations of both intestinal regions, OTX1 antibody co-localized with the glial marker S100β, and co-staining was particularly intense after DNBS-induced inflammation ([Figs. 5M](#fig-5){ref-type="fig"}--[5R](#fig-5){ref-type="fig"}). In small intestine and colonic LMMPs obtained from control and DNBS-treated animals, myenteric neurons staining for OTX1 were also immunopositive for iNOS, as shown in [Figs. 6A](#fig-6){ref-type="fig"}--[6L](#fig-6){ref-type="fig"}. Both in control and DNBS-treated LMMP whole-mount preparations, OTX1 immunostaining did not co-localize with nNOS ([Figs. 6M](#fig-6){ref-type="fig"}--[6O](#fig-6){ref-type="fig"}). These results suggest that OTX1 is expressed both in myenteric neurons, where it is highly co-expressed with iNOS, and in enteric glial cells of the rat small intestine and distal colon. After DNBS-induced colitis the number of myenteric neurons expressing OTX1 is highly upregulated both on site and distantly. ![Percentage of OTX1 and OTX2 neurons in the rat small intestine and distal colon after DNBS-treatment.\ (A) Percentage of OTX1 immunoreactive myenteric neurons with respect to the total of HuC/D immunoreactive neurons per ganglion in small intestine and distal colon LMMPs whole-mount preparations obtained from DNBS-treated (solid bar) and control animals (vehicle-treated CTR, empty bar). (B) Percentage of OTX2 immunoreactive myenteric neurons with respect to the total of HuC/D immunoreactive neurons per ganglion in small intestine and distal colon whole-mount preparations obtained from DNBS-treated (solid bar) and control animals (vehicle-treated CTR, empty bar). Values are expressed as mean ± S.E.M, N = 5 rat per group. \\\*P \< 0.001 vs. values obtained in CTR animals, ^§§§^P \< 0.001 vs. DNBS-treated small intestine by one-way ANOVA with Tukey's post hoc test.](peerj-08-8442-g004){#fig-4} ![OTX1 localization in the rat small intestine and distal colon myenteric plexus of CTR and DNBS-treated rats.\ Co-localization of OTX1 with the neuronal marker HuC/D in LMMPs whole-mount preparations of the small intestine and distal colon obtained from vehicle-treated CTR animals (A--C; G--L, respectively) and DNBS-treated animals (D--F; J--L, respectively). In CTR preparations few neurons displayed OTX1 immunoreactivity (arrow) (A--C; G--I). In both regions, the number of OTX1-IR neurons was higher in DNBS-treated preparations (D--F; J--L). Photomicrographs of small intestine (M--O) and colonic (P--R) LMMP whole-mount preparations showing OTX1 immunoreactivity in enteric glial cells (asterisk) as evidenced by co-staining with the glial cell marker, S100β. Bar 50 µm.](peerj-08-8442-g005){#fig-5} ![OTX1 and iNOS co-localization in the rat small intestine and distal colon myenteric plexus of CTR and DNBS-treated rats.\ Photomicrograph showing few myenteric neurons (arrow) co-staining OTX1 and iNOS in CTR LMMPs whole-mount preparations of the small intestine (A--C) and distal colon (G--I). After DNBS treatment, the number of iNOS immunoreactive myenteric neurons (arrow) increased and co-expressed OTX1 both in small intestine (D--F) and distal colon (J--L). Both in CTR and DNBS-treated group, co-staining between OTX1 and nNOS was not detected in myenteric ganglia of both small intestine and distal colon. A colonic whole-mount LMMP preparation obtained from DNBS-treated animals is reported as an example (M--O, bar: 50 µm) bar: 50 µm.](peerj-08-8442-g006){#fig-6} Distribution of OTX2 immunoreactivity in small intestine and colon LMMP whole-mount preparations from normal and DNBS-treated animals ------------------------------------------------------------------------------------------------------------------------------------- In control sections of the rat small intestine and colon, faint OTX2 staining was found in myenteric ganglia ([Figs. 1E](#fig-1){ref-type="fig"} and [1O](#fig-1){ref-type="fig"}). In LMMPs whole-mount preparations obtained from control animals OTX2 specific antibody stained the soma of few myenteric neurons both in the small intestine and in the distal colon ([Figs. 7A](#fig-7){ref-type="fig"}--[7C](#fig-7){ref-type="fig"} and [7G](#fig-7){ref-type="fig"}--[7I](#fig-7){ref-type="fig"}, respectively). Some neurons had a large cell body, while other smaller neurons with an ovoidal shape were, prevalently, localized in the periphery of the ganglion. The percentage of OTX2-immunoreactive neurons in control small intestine and distal colon myenteric plexus was 7.07 ± 1.68%, n = 15 and 8.93 ± 2.11%, n = 15, respectively ([Fig. 4B](#fig-4){ref-type="fig"}). In small intestine and colon cross-sections obtained from DNBS-treated animals, OTX2 staining significantly increased in myenteric ganglia with respect to control preparations ([Figs. 1J](#fig-1){ref-type="fig"} and [1U](#fig-1){ref-type="fig"}). Accordingly, after DNBS treatment, the percentage of OTX2 immunoreactive myenteric neurons in LMMP whole-mount preparations obtained from the small intestine and colon significantly increased (47.95 ± 3.20%, n = 15; 41.47 ± 5.12%, n = 15, respectively P \< 0.001, by one-way ANOVA with Tukey's post hoc test) with respect to control values ([Figs. 4B](#fig-4){ref-type="fig"}, [7D](#fig-7){ref-type="fig"}--[7F](#fig-7){ref-type="fig"} and [7J](#fig-7){ref-type="fig"}--[7L](#fig-7){ref-type="fig"}). The number of OTX2 immunoreactive neurons was not significantly different in the two regions ([Fig. 4B](#fig-4){ref-type="fig"}). In small intestine and colonic myenteric ganglion obtained from control and DNBS-treated animals, the majority of OTX2 immunoreactive neurons stained for nNOS ([Figs. 8A](#fig-8){ref-type="fig"}--[8L](#fig-8){ref-type="fig"}). There was no evidence of co-staining between OTX2 and iNOS, as shown in [Figs. 8M](#fig-8){ref-type="fig"}--[8O](#fig-8){ref-type="fig"}. In small intestine and distal colon LMMPs obtained from control animals 31.94 ± 3.33%, n = 10 and 30.83 ± 10.83%, n = 10, respectively, of OTX2-IR myenteric neurons stained also for OTX1. In both regions, after DNBS treatment, 25.82 ± 5.78%, n = 10 and 25.63 ± 2.17% n = 10, respectively, of OTX2-IR were also immunoreactive for OTX1. These findings suggest that OTX2 is exclusively expressed in myenteric neurons and displays a high co-localization with nNOS. Experimentally-induced colitis upregulates the number of myenteric neurons expressing OTX2 both on site and distantly. ![OTX2 localization in the rat small intestine and distal colon myenteric plexus of CTR and DNBS-treated rats.\ Co-localization of OTX2 with the neuronal marker HuC/D in LMMPs whole-mount preparations of the small intestine and distal colon obtained from vehicle-treated CTR (A--C; G--I, respectively) and DNBS-treated animals (D--F; J--L, respectively). In CTR preparations, few neurons displayed OTX2 immunoreactivity (arrow). In both regions, after DNBS treatment, the number of OTX2 immunoreactive neurons significantly increased. Bar 50 µm.](peerj-08-8442-g007){#fig-7} ![OTX2 and nNOS co-localization in the rat small intestine and distal colon myenteric plexus of CTR and DNBS-treated rats.\ In myenteric neurons of both small intestine and distal colon from CTR and DNBS-treated rats, OTX2 co-localized with nNOS (arrows). Some OTX2 immunoreactive neurons, however, did not stain for nNOS (asterisk) (A--L, bars: 50 μm). Co-staining between OTX2 and iNOS was not detected, in myenteric ganglia of both small intestine and distal colon of both experimental groups. A colonic whole-mount LMMP preparation obtained from DNBS-treated animals is reported as an example (M--O, bar: 50 µm).](peerj-08-8442-g008){#fig-8} DNBS-induced colitis upregulates OTX1 mRNA and protein in LMMPs of the rat small intestine and distal colon ----------------------------------------------------------------------------------------------------------- In order to confirm the immunofluorescence data reflecting an up-regulation of OTX1 and OTX2 in the rat small intestine and colon myenteric plexus after DNBS-induced colitis, we evaluated the levels of mRNA and proteins of both homeobox proteins in LMMP preparations. Rat small intestine and colonic OTX1 mRNA levels obtained in the different experimental groups are described in [Fig. 9A](#fig-9){ref-type="fig"}. In LMMPs of both regions, DNBS treatment induced a significant increase of OTX1 mRNA with respect to values obtained in control preparations (small intestine: P \< 0.05; distal colon: P \< 0.001, by one-way ANOVA with Tukey's post hoc test). OTX1 mRNA enhancement in colonic specimens was significantly higher than in the small intestine (P \< 0.001, by one-way ANOVA with Tukey's post hoc test). In rat small intestine and colonic LMMP preparations the specific OTX1 antibody revealed one band at 37 kDa ([Fig. 9B](#fig-9){ref-type="fig"}). In both gut regions, after DNBS treatment, OTX1 protein expression significantly increased (small intestine: P \< 0.05; distal colon: P \< 0.01, by one-way ANOVA with Tukey's post hoc test). These biomolecular data confirm that an experimentally-induced colitis may up-regulate both the transcript and protein of OTX1 in the myenteric plexus both on site and distantly from the injury. ![qRT-PCR and western blot analysis of OTX1 and OTX2 mRNA and protein levels in the rat small intestine and colon after DNBS-induced colitis.\ RT-PCR quantification of OTX1 (A) and OTX2 (C) transcripts in preparations of the small intestine and distal colon obtained from control (CTR, empty bars) and DNBS-treated animals (solid bars). The relative gene expression was determined by comparing 2^−ΔΔCt^ values in CTR and DNBS-treated samples normalized to β-actin. Values are mean ± S.E.M. of 7--13 experiments. \*P \< 0.05, \\\*P \< 0.001 vs. CTR animals; ^§§§^P \< 0.001 vs. DNBS-treated small intestine by one-way ANOVA with Tukey's post hoc test. (B) OTX1 and (D) OTX2 protein expression analyzed in LMMPs preparations of the small intestine and colon obtained from CTR (empty bars) and DNBS-treated animals (solid bars). Blots representative of immunoreactive bands for either OTX1 or and β-actin in the different experimental conditions are reported on top of each panel. Numbers at the margins of the blots indicate relative molecular weights of the respective protein in kDa. Samples (200 μg) were electrophoresed in SDS-8% polyacrylamide gels. Values are expressed as mean ± S.E.M. of 5--6 experiments of the percentage variation of the normalized optical density (O.D.) obtained from DNBS-treated preparations with respect to values obtained in control samples. \*P \< 0.05 ad \\P \< 0.01 vs. CTR by one-way ANOVA with Tukey's post hoc test.](peerj-08-8442-g009){#fig-9} DNBS-induced colitis upregulates OTX2 mRNA and protein in LMMPs of the rat small intestine and distal colon ----------------------------------------------------------------------------------------------------------- OTX2 mRNA levels measured in rat small intestine and colonic LMMPs preparations obtained from control and DNBS-treated animals are described in [Fig. 9C](#fig-9){ref-type="fig"}. In both regions, DNBS treatment induced a significant increase of OTX2 mRNA with respect to values obtained in control preparations (small intestine: P \< 0.05; distal colon: P \< 0.001, by one-way ANOVA with Tukey's post hoc test). OTX2 mRNA levels in small intestine specimens were not significantly different than those obtained in the distal colon. In rat small intestine and colon LMMP preparations the specific OTX2 antibody revealed one band at 32 kDa ([Fig. 9D](#fig-9){ref-type="fig"}). In both gut regions, OTX2 protein expression significantly increased (P \< 0.01, by one-way ANOVA with Tukey's post hoc test) after DNBS treatment. After DNBS-induced colitis, analogously to OTX1, up-regulation of OTX2 in the rat myenteric plexus was evidenced also for the transcript and protein, by means of qRT-PCR and western immunoblotting. Discussion ========== In myenteric neurons, intestinal inflammation induces adaptive changes, which may be responsible for both acute and long-lasting alterations of the gastrointestinal functions, underlying disease symptoms ([@ref-12]). The molecular mechanism/s of enteric neuronal adaptation to inflammation are still largely to be uncovered. In this study, we show that the expression of orthodenticle homeoproteins, OTX1 and OTX2, which participate to neuroplastic changes both in physiological and pathological conditions ([@ref-43]), is up-regulated in the myenteric plexus of the rat small intestine and distal colon after an experimentally induced colitis with DNBS acid. In analogy with the results recently obtained in the rat small intestine myenteric plexus ([@ref-19]), in the present study OTX1-immunoreactivity (IR) is predominantly found in enteric glial cells and in few neurons of the rat distal colon myenteric plexus, while OTX2-IR is uniquely detected in the soma of a relatively small percentage of myenteric neurons. In pathophysiological conditions, that is, after in vivo-induced ischemia/reperfusion (I/R) injury as well as in mild inflammatory conditions induced in sham-operated animals OTX1 and OTX2 expression in rat small intestine myenteric ganglia significantly enhanced ([@ref-19]). In good agreement, in the present study we show a significant increase of both OTX1 and OTX2 mRNA and protein levels in LMMP preparations of the rat small intestine and distal colon myenteric plexus after DNBS-induced colitis. In addition, the number of myenteric neurons staining for both transcription factors in the two intestinal regions significantly increases, suggesting that the inflammation influences the expression of both homeoproteins in myenteric neurons, on site and distantly from the injury. This adaptive response is not associated with major histopathological alterations in the small intestine myenteric plexus. In line with previous data, DNBS treatment, induces important morphological and histological changes, including distortion of myenteric ganglia with signs of neuronal degeneration, in the colon, but not in the small intestine, where, however, the number of myenteric neurons significantly decreases. Interestingly, MPO activity and the number of CD45 positive cells, which are indicative of inflammatory infiltration, are elevated both in small intestine and distal colon segments. In addition, after DNBS-induced colitis, in rat colonic and small intestine LMMPs the levels of pro-inflammatory cytokines, such as TNFα, pro-IL1β and IL6, and of factors involved in active intestinal inflammation, such as HIF1α and VEGFα, are up-regulated, as observed along the rat gastrointestinal tract after trinitrobenzene sulfonic (TNBS) acid- and DNBS-induced colitis by other groups ([@ref-6]; [@ref-7]). Overall, our observations suggest that important molecular changes, including upregulation of OTX1 and OTX2 pathways, may occur in the myenteric plexus along the gastrointestinal tract in response to an inflammatory challenge. Such changes may be transmitted from the site of injury to distant sites, where overt structural changes are not evident, and may account for the functional alterations observed in the proximal intestine in both IBD patients and animal models, including altered motility ([@ref-12]; [@ref-16]; [@ref-11]). We cannot exclude that inflammation-induced OTX1 and OTX2 up-regulation in rat small intestine and distal colon myenteric neurons may influence the neuromuscular function by regulating enteric neurotransmitter pathways. Indeed, TNBS-induced colitis in the rat was associated with a decrease of neurotransmitter release, not only in the inflamed colon, but also distally, in the small intestine resulting in the reduction of small intestinal transit ([@ref-11]). Nitric oxide (NO), may represent a putative enteric neurotransmitter involved in OTX1 and OTX2 inflammation-induced upregulation. Noteworthy, a preferential involvement of enteric nitrergic pathways underlays development of dysmotility after deletion of a homeobox gene phylogenetically related to OTX, such as Ncx/Hox11L.1 ([@ref-27]). In addition, the occurrence of an interplay between enteric nitrergic pathways and OTX transcription factors has been suggested to sustain nitric oxide (NO)-mediated dysmotility after I/R injury in the gut ([@ref-19]). In these conditions, NO was shown to favor OTX1 and OTX2 up-regulation in rat small intestine myenteric ganglia. After I/R, NO derived from iNOS promoted OTX1 up-regulation more, while nNOS more closely related to OTX2 up-regulation. During gut inflammation and I/R injury, iNOS and nNOS have different roles on enteric neuronal homeostasis, retaining a neurodamaging and neuroprotective role, respectively ([@ref-49]; [@ref-19]; [@ref-13]). Activation of the inducible isoform and downregulation of nNOS, are correlated with derangement of the neuromuscular function and slowing of the gastrointestinal transit ([@ref-49]; [@ref-22]; [@ref-20]). In view of the relationship between the two NO synthases and OTX1 and OTX2 these effects on intestinal motility may involve corresponding activation of molecular pathways downstream of OTX1 and OTX2, respectively ([@ref-19]). Accordingly, in the present study, a substantial involvement of iNOS and nNOS in colitis-induced upregulation of OTX1 and OTX2, respectively, is suggested by the superimposition of OTX1 with iNOS and OTX2 with nNOS immunostaining, in myenteric ganglia of both regions studied. It is possible that activation of the iNOS-OTX1 pathway both in neurons and glial cells may sustain inflammation-induced alterations of the enteric neuromuscular function, as observed after I/R injury ([@ref-19]). Interestingly, overexpression of OTX1 sustained human colorectal cancer cell proliferation and invasion in vitro and tumor growth in vivo, suggesting a pathogenic role for this homeoprotein in colon cancer ([@ref-47]). By contrast, up-regulation of OTX2 in nNOS^+^ myenteric neurons may have a protective role, preventing inflammation-induced neuronal cell death, as suggested for OTX2 expression in photoreceptors ([@ref-24]). Different hypotheses may be put forward to explain the mechanism/s underlying NO-mediated up-regulation of OTX1 and OTX2. This latter may depend on increased levels of NO-derived reactive oxygen species (ROS), whose production in myenteric neurons enhances as a consequence of several gut pathophysiological conditions ([@ref-14]; [@ref-46]). In this context, OTX2 up-regulation was recently observed in bovine in vitro embryos as a result of increased oxidative stress inducing ROS production ([@ref-30]). We cannot, however, exclude that OTX1 and OTX2 up-regulation in our model may depend upon neuroimmune interactions between different subpopulation of enteric neurons and immunocytes ([@ref-16]). Neuronal cells in the ENS are located in close proximity to mucosal immunocytes and may regulate one another's functions by releasing a complex set of cytokines, neurotransmitters and hormones ([@ref-28]). OTX1 and OTX2 expression in the colonic myenteric plexus may be influenced by inflammatory mediators, such as VEGFα, which is positively correlated to OTX2 expression in retinal pigment epithelial cells during inflammation ([@ref-3]). More recently, TNFα has been proposed as a modulator of OTX2 expression in in vitro models of chronic subretinal inflammation ([@ref-33]). In addition, a recent genomewide study has identified a positive correlation between inflammatory cytokines, particularly IL6, and OTX1, in the pathogenesis of foot-and-mouth viral disease in animals ([@ref-48]). Circulating cytokines and inflammatory mediators may influence myenteric neurons even at sites distant from the lesion, thus sustaining OTX1 and OTX2 up-regulation in the small intestine myenteric plexus ([@ref-31]). Changes in myenteric neuron responses during inflammation may be conveyed also via neuronal pathways constituting local or extrinsic reflex arches, as demonstrated in gastroparesis associated with TNBS-induced ileitis in rats ([@ref-34]). OTX proteins may be secreted and internalized by live cells, including neurons, and a retrograde transfer of these proteins from colonic myenteric neurons to those of the small intestine by peculiar nonconventional mechanisms is also feasible ([@ref-25]). The pathophysiology of intestinal inflammation reflects a balance between mucosal injury and repair mechanism and may have important consequences on the integrity of intrinsic neuronal circuitries constituting the ENS. Several studies are focused on molecular pathways underpinning damage in different cellular populations, including myenteric neurons. Development of new therapeutic strategies in this field comprise the modulation of inflammation-induced cell death pathways ([@ref-5], [@ref-4]). In this view our data, by providing new hints on possible regulatory genes involved in myenteric neurons response to an inflammatory injury may contribute to the development of novel strategies for the treatment of gastrointestinal diseases with important social and clinical impact, such as IBD. Conclusions =========== In this study we provide evidence that DNBS treatment in rats may elevate inflammatory markers not only in the site of inflammation, but also distally, in the small intestine. Such response is associated with changes in myenteric neuron number in both regions, although more severe damage occurs in the distal colon. After DNBS treatment the expression of two homeobox transcription factors, OTX1 and OTX2, is upregulated in the myenteric plexus of the small intestine and distal colon. Supplemental Information ======================== 10.7717/peerj.8442/supp-1 ###### Raw data. ###### Click here for additional data file. 10.7717/peerj.8442/supp-2 ###### Full-length gel of OTX2 and b-actin. ###### Click here for additional data file. 10.7717/peerj.8442/supp-3 ###### Full-length gel of beta actin for OTX1. ###### Click here for additional data file. 10.7717/peerj.8442/supp-4 ###### Full-length gel of OTX1. ###### Click here for additional data file. The Authors wish to thank Mr. Antonio Pelizzoli for the technical support. Dr. Fabrizio Bolognese, Dr. Ivan Vaghi, are kindly acknowledged for the excellent assistance in the acquisition of confocal images. Michela Bistoletti, Giovanni Micheloni and Annalisa Bosi are PhD students of the "Experimental and Translational Medicine" course at the University of Insubria. Nicolò Baranzini is a PhD student of the Biotechnology, Biosciences and Surgical Technology" course at the University of Insubria. Additional Information and Declarations ======================================= The authors declare that they have no competing interests. [Michela Bistoletti](#author-1){ref-type="contrib"} conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Giovanni Micheloni](#author-2){ref-type="contrib"} conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Nicolò Baranzini](#author-3){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Annalisa Bosi](#author-4){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Andrea Conti](#author-5){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Viviana Filpa](#author-6){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Cristina Pirrone](#author-7){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Giorgia Millefanti](#author-8){ref-type="contrib"} performed the experiments, prepared figures and/or tables, and approved the final draft. [Elisabetta Moro](#author-9){ref-type="contrib"} performed the experiments, analyzed the data, prepared figures and/or tables, and approved the final draft. [Annalisa Grimaldi](#author-10){ref-type="contrib"} conceived and designed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the paper, and approved the final draft. [Roberto Valli](#author-11){ref-type="contrib"} conceived and designed the experiments, authored or reviewed drafts of the paper, and approved the final draft. [Andreina Baj](#author-12){ref-type="contrib"} conceived and designed the experiments, analyzed the data, authored or reviewed drafts of the paper, and approved the final draft. [Francesca Crema](#author-13){ref-type="contrib"} conceived and designed the experiments, authored or reviewed drafts of the paper, and approved the final draft. [Cristina Giaroni](#author-14){ref-type="contrib"} conceived and designed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the paper, and approved the final draft. [Giovanni Porta](#author-15){ref-type="contrib"} conceived and designed the experiments, authored or reviewed drafts of the paper, and approved the final draft. The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): The protocol was approved by the Animal Care and Use Ethics Committee of the University of Pavia (Approval number n. 3/2011). The following information was supplied regarding data availability: Raw data and images of representative western blot are available in the [Supplemental Files](#supplemental-information){ref-type="supplementary-material"}.	{ "pile_set_name": "PubMed Central" }
The thermal conductivity of bulk materials can be engineered by various means such as the use of composite materials[@b1] or metamaterials[@b2]. This allows the manipulation of heat flow with precision and inspires various functionalities, like thermal inverter, thermal concentrator, thermal cloak and so on[@b3][@b4][@b5][@b6][@b7][@b8][@b9][@b10][@b11]. However, the realization of such concepts at the nanoscale is difficult because engineering the thermal conductivity of materials at small dimensions is much more difficult. The thermal conductivity of nano-sized material systems can be tuned by phonon-boundary scattering[@b12][@b13], interfacial thermal resistance[@b14], and/or by exploiting phononic crystals[@b15][@b16], but to engineer the thermal conductivity locally with precision is challenging from both fabrication and theoretical-prediction aspects. In particular, in the case of Si nanowires---a versatile building block for nanoelectronic[@b17], photovoltaic[@b18], thermoelectric[@b19] and photonic[@b20] devices---it has been shown experimentally that the thermal conductivity of an individual Si nanowire can be reduced by shrinking its diameter[@b13][@b21] and by introducing surface roughness[@b19][@b22][@b23]. However, these prior works only demonstrated the tuning and measurement of the thermal conductance of a nanowire as a whole between its ends. Significantly, in terms of mechanical robustness for practical device design, ultra-thin or highly roughened silicon nanowires are extremely fragile, apart from the fact that making reliable ohmic contacts to such silicon nanowires is also a major issue[@b24]. Moreover, the use of physical structuring means that post-fabrication tuning of thermal conductance by this method is not possible. Another approach to engineer the thermal conductivity of a Si nanowire without affecting its surface morphology is through ion implantation. By introducing impurity atoms into the crystalline silicon nanostructures, the thermal conductivity can be tuned due to enhanced phonon-defect scattering and the damage can be annealed away (at CMOS process-compatible thermal budgets) as well[@b25]. Even though heavy-ion irradiation (for example, using gallium and uranium ions) has been observed to change the thermal conductivity in nanostructures by defect engineering[@b26][@b27][@b28], point defect generation is inevitably accompanied by cluster amorphization along the ion trajectory. For less heavy ions, like silicon ions, the process of continuously elastic collision would make it difficult to generate a spatially uniform defect profile, and a complicated region containing simple defects as well substantial cluster formation results[@b25][@b29]. It is thus desirable to adopt a technique that can produce different concentrations of simple point defects in a well-controlled manner to establish unequivocally the role of such defects on the thermal conductivity. In this paper, we show that the thermal conductivity of an individual Si nanowire can be locally changed by irradiating it with helium ions with well-defined doses at different positions along its length (see Methods section for detailed sample preparation). Because of the small nanowire diameter (∼160 nm) and moderately high helium ion energy (30--36 keV), the helium ions pass through the nanowire with minimal forward scattering and energy loss, and their effect is reasonably uniformly distributed over the nanowire cross-section, unlike the case of a much thicker Si substrate. The thermal conductivity of each irradiated nanowire segment was then measured by an electron beam heating technique that is capable of spatially resolving the thermal conductivity along the nanowire's length[@b30] (Methods section present the detailed measurement and calculation procedure). Using this technique, the effect of different helium ion dose can be measured conveniently along a single nanowire, thereby eliminating potential discrepancies that could arise from sample-to-sample variations in the nanowire diameter and from uncertainty in the thermal contact resistance between the nanowire ends and the thermometers. From the measurement, we observed a reproducible thermal conductivity versus ion irradiation dose curve, from which a clear crystalline-to-amorphous transition is seen. The effect of defects created by helium ions below the transition dose threshold on the nanowire thermal conductivity is discussed with theoretical support from the kinetic theory of phonon gas model and non-equilibrium molecular dynamics (NEMD) simulations. We also show that the thermal conductivity of the irradiated nanowire can be further tuned by annealing. The ability to tune the thermal conductivity along silicon nanowires with high spatial resolution by a well-controlled helium ion irradiation, and to measure the local thermal conductivity thereafter, provides a new platform to study nanoscale thermal transport. Results ======= Thermal conductivity measurement -------------------------------- The silicon nanowire is suspended between two temperature sensors comprising platinum (Pt) loops on silicon nitride membranes, each of which is suspended by six nitride beams. [Figure 1a](#f1){ref-type="fig"} shows the measurement result for Sample \#1 for the damaged portion irradiated with the highest dose (7.5 × 10^16^ cm^−2^). From this figure, we can see that as the electron beam scans from the left to right across the damaged portion, the temperature rise of the left (right) sensor, ΔT~L~ (ΔT~R~), undergoes obvious decrease (increase), indicating that the thermal resistivity of the damaged portion is much larger than that of the intrinsic portion. This is further confirmed by the increase in slope of the R~i~(x) curve within the damaged portion, where R~i~(x) is the cumulative thermal resistance from the left sensor to the heating spot. It is observed that the power absorption of the nanowire from the electron beam, (ΔT~L~+ΔT~R~)/R~b~, where R~b~ is the equivalent thermal resistance of the suspension beams, does not vary significantly at the damaged portion, indicating negligible material removal from inside the nanowire. The R~i~(x) curves for all the portions with different doses are plotted in [Supplementary Fig. 3](#S1){ref-type="supplementary-material"}. The thermal conductivity of each portion was calculated as where A is the cross-sectional area of silicon nanowire with diameter d=160 nm as measured in the TEM. To minimize the error, was obtained by linearly fitting the R~i~(x) curve of the damaged portion and the intrinsic portion, with the beginning and ending 50 nm of these portions stripped off to avoid the non-uniformity in the dose near the boundaries arising from ion forward scattering. As the thermal conductivity is derived from the gradient () of each portion, one can conveniently obtain the local thermal conductivity of each irradiated/non-irradiated portion in a single nanowire. Moreover, the need to contend with the unmeasurable thermal contact resistance between the ends of the nanowire and the sensors, which is a long standing problem of the conventional thermal bridge method, is avoided with our technique. Plotted in [Fig. 1b](#f1){ref-type="fig"} are the thermal conductivity versus dose curves for the eight samples measured at room temperature. The error bar is calculated by error propagation due to the uncertainty in during linear fitting and the uncertainty in the diameter (±5 nm), with the former being one to two orders smaller than the latter. The intrinsic thermal conductivity of the nanowire is 50.2±2.3 W m^−1^ K^−1^, obtained by linearly fitting several undamaged portions of the measured samples and taking the average. The amorphous limit of ∼1.7 W m^−1^ K^−1^ (at 300 K) is taken from the literature[@b31][@b32]. Clearly two regimes can be identified in [Fig. 1b](#f1){ref-type="fig"}---in the regime of relatively low dose (\<1.5 × 10^16^ cm^−2^), the thermal conductivity drops rapidly initially and then tapers off as the dose increases, whereas in the regime of dose ≥3 × 10^16^ cm^−2^, the thermal conductivity approaches an asymptotic value. Of particular note is a steep jump in thermal conductivity between these two regimes, as shown in the inset of [Fig. 1b](#f1){ref-type="fig"}. Low-dose regime---point defect creation and agglomeration --------------------------------------------------------- The as-implanted damage was calculated by Monte Carlo simulations based on a binary collision approach (TRIM/SRIM[@b33], the detailed simulation conditions and results are described in [Supplementary Fig. 2](#S1){ref-type="supplementary-material"} and [Supplementary Note 1](#S1){ref-type="supplementary-material"}). The simulation shows that one helium ion can create a large number of displaced atoms (or Frenkel pairs, each of which is composed of a vacancy and a nearby interstitial) along its track; moreover, this as-produced damage is discrete and distributes relatively uniformly across the nanowire cross-section due the large range of the light and high energy (30--36 keV) incoming helium ions compared to the sample thickness (d=160 nm); last, from simulation results, 98% of the helium ions penetrate the sample, and one helium ion can create 33 defects on average, while the number of residual (embedded) helium atoms is much smaller than the number of damaged lattice sites. However, at room temperature, the as-produced point defects are not stable[@b34]. They undergo extensive interstitial-vacancy recombination, defect clustering and pairing with impurities (for example, carbon and oxygen) during and after the irradiation process at room temperature. As a result, only 1--10% of the defects survive at room temperature for bulk materials[@b35]. For the case of nanowires, annihilation at the surface may also enhance this annealing effect given the high surface-to-volume ratio[@b36]. An upper bound for the concentration of defects can be roughly estimated by assuming that 10% of the Si vacancies created by helium ions survive. For calculation purposes, we take it that the Si nanowire is circular with a diameter of 160 nm, based on our experimentally imaged and measured cross-section. The number of vacancies remaining is a function of dose and this relation is finally described as 2.0 × 10^5^ (cm^−1^) × dose (cm^−2^), with details discussed in detail in [Supplementary Note 1](#S1){ref-type="supplementary-material"}. As there are 5 × 10^22^ Si atoms per cm^3^, for a dose of 1 × 10^16^ cm^−2^, around 4% of the Si atoms are displaced, acting as scattering centres for phonons. This value provides an upper bound of the phonon scattering centres because we use the upper bound of the survival rate (10%). Moreover, there is a consensus view that the stable defects left in the room-temperature silicon are predominantly divacancies[@b37], which are formed by two nearby vacancies. It has also been shown experimentally that at room temperature, divacancy is the dominant defect for the near-surface region of the helium ion damaged Si[@b38]. Regarding its impact on the thermal conductivity, this means two displaced sites combine and act as one phonon scattering centre. Besides the vacancy-type defect, there is also a small amount of helium ions left; however, the quantity is limited as most of the helium ions penetrate the nanowire during irradiation. There may also be interstitial or interstitial-like defects in the nanowire, because although Si interstitials can follow athermal migration even at 4.2 K and the migration distance is long in high-purity material, they can also form clusters or complexes which are stable at room temperature[@b34]. The inverse phonon life time (phonon relaxation rate) contributed by substitutional point defects for a simple cubic lattice is , with where c is the point defect concentration per atom, ΔM the mass difference between the substitutional atom and the host atom, M the mass of host atom, a the lattice constant and v the phonon group velocity[@b39][@b40]. For the case of vacancy defect, as it removes the mass of one atom and also the force constants of two atomic sites, it can be treated as[@b41][@b42] . For simplicity and better comparison with previous studies, we use [Equation 1](#eq6){ref-type="disp-formula"} for the substitutional point defect calculation in silicon nanowires, since this formula captures the essence of phonon scattering rates due to vacancy defects, and it is indeed valid for complex lattices[@b43][@b44]. Hence, compared with the substitutional point defects, the vacancy defect has a much stronger phonon scattering effect and is highly effective in reducing the lattice thermal conductivity. Indeed, experiments by Pei et al.[@b45] showed that a small amount of vacancy defects (3%) in the In~2~Te~3~ lattice introduced by alloying with InSb can substantially reduce the lattice thermal conductivity (by ∼80%). To investigate the effect of point defects on the phonon scattering rate, we first consider a simple Debye formula for the thermal conductivity: , where C is the specific heat, v is the average group velocity assuming a linearized phonon dispersion model and τ is the phonon relaxation time. We further assume that τ^−1^ can be calculated from Matthiessen's rule, which implies that different phonon scattering mechanisms are independent of each other. As a result, the thermal resistivity (1/κ) is additive from various scattering mechanisms, including , the phonon-defect scattering rate. Thus 1/κ should be linearly related to the concentration of point defects, as depends linearly on the concentration of point defects ([Eq. 1](#eq6){ref-type="disp-formula"}). However in the experiment what can be obtained is the 1/κ curve as a function of dose ([Fig. 2a](#f2){ref-type="fig"}), from which we can see that in the low-dose regime, 1/κ first increases linearly and then tends towards saturation. It is possible to circumvent the linearized phonon dispersion assumption and extract the value of parameter D by fitting the measured thermal conductivities within the framework of the kinetic theory of phonon gas[@b46]. The fitting model, utilizing the full phonon dispersion of Si, includes anharmonic phonon--phonon scattering events and the scattering of phonon by isotopes, boundaries and point defects as well (a detailed description can be found in [Supplementary Note 2](#S1){ref-type="supplementary-material"}). The optimal D obtained from the best fit is plotted as a function of dose as shown in [Fig. 2b](#f2){ref-type="fig"}, where the possible influence from boundary scattering is considered separately. The exclusion of boundary scattering term makes the parameter D increase by a factor of 4 compared with that of the case where boundary scattering is included. Nevertheless the overall trend for D with dose is qualitatively similar---the parameter D shows a linear dependence on dose in the low-dose regime and becomes saturated in the high-dose regime. The relationship between D and dose can be well fitted by a simple function, , where the dose is small, the parameter D scales linearly with dose, namely, . Considering the fact that the phonon-defect scattering rate depends linearly on the concentration of point defects ([Eq. 1](#eq6){ref-type="disp-formula"}), we may infer that for doses less than 1 × 10^15^ cm^−2^ in our experiment, the point defects created by helium ion irradiation is linearly proportional to the helium ion dose. However, as the dose increases the parameter D converges gradually to some asymptotical value, which suggests that the effective number of scattering centres does not increase significantly with further increases in dose. One possible reason is that for higher dose (from 2 × 10^15^ to 1 × 10^16^ cm^−2^), the point defects agglomerate due to room temperature annealing. Last, the parameter D is about 1--3 orders of magnitude larger than the corresponding parameter for isotope scattering[@b46], which indicates the dominant role of point defects in suppressing the thermal conductivity of Si nanowire. To further examine the role of point defect scattering, and to incorporate the effect of phonon-boundary scattering, NEMD simulation was carried out for qualitative comparisons with the experimental results. In the simulation, Si nanowires with various cross-sections of 3 × 3, 6 × 6 and 9 × 9 unit cells (0.543 nm per unit cell) and fixed length of 20 unit cells were considered ([Supplementary Fig. 6](#S1){ref-type="supplementary-material"}). Point defects are in the form of vacancies modelled by randomly removing some Si atoms (up to 20%) and then removing the single-bonded atom pairs ([Supplementary Fig. 7](#S1){ref-type="supplementary-material"}). The details of the NEMD simulation are given in [Supplementary Note 3](#S1){ref-type="supplementary-material"}. Compared to the pristine Si nanowires, the simulated thermal conductivity of the defective Si nanowires is reduced, and decreases with the increase of vacancy concentration due to the enhanced phonon scattering by vacancy ([Fig. 2c](#f2){ref-type="fig"}). At low vacancy concentration (smaller than 5%), the thermal conductivity of defective Si is further reduced by boundary scattering, while it is almost the same for different diameters at high vacancy concentration (greater than 15%), indicating the dominant role of vacancy scattering in this regime. We also plotted the 1/κ curve as a function of vacancy concentration ([Fig. 2d](#f2){ref-type="fig"}). As the Si nanowire diameter used in the simulation increases, 1/κ shows no trend of saturation as the isolated vacancy concentration increases; on the other hand this corroborates the model of point defect agglomeration in the real situation under helium ion irradiation at room temperature. From the experimental and simulation results, we can see that in the low-dose regime, point defects are effective in changing the thermal conductivity of the helium-irradiated Si nanowire. First, the thermal conductivity initially decreases drastically as the dose (and the corresponding density of point defects) increases; however further increasing the dose yields diminishing return on reducing thermal conductivity, that is, to obtain the same marginal decrease of thermal conductivity, one has to at least double the dose (inset of [Fig. 1b](#f1){ref-type="fig"}). Second, at most 4% displaced Si atoms (corresponding to a dose of 1 × 10^16^ cm^−2^) is capable of decreasing the nanowire thermal conductivity by nearly 70%. It should be noted that under our experimental conditions of high helium ion energy, thin sample and same dose rate and temperature, the thermal conductivity for a particular dose is repeatable for all the samples measured. Effect of annealing ------------------- It has been shown in the literature that point defects that are stable at room temperature such as divacancies can be annealed out at higher temperatures[@b37][@b47]. This annealing effect on the thermal conductivity was investigated using Sample \#3. Upon annealing, the total thermal resistance (R~T~) of the nanowire became progressively smaller. The results from spatially resolved thermal conductance measurement show that this decrease of R~T~ is due to the increase of the irradiated portion's thermal conductivity, while the thermal conductivity of the non-irradiated portion remained the same ([Fig. 3a](#f3){ref-type="fig"}) as expected. The before- and after- annealing results are plotted in [Fig. 3b](#f3){ref-type="fig"}. From this figure, we can see that for annealing at 120 °C, the thermal conductivity of the irradiated portion increases noticeably after 20 h. To further improve the nanowire thermal conductivity at 120 °C, a long annealing time, for example, 160 h, is required. On the other hand, merely 2 h of annealing at 300 °C improved the thermal conductivity substantially, and for the lowest dose (2.5 × 10^14^ cm^−2^), the thermal conductivity had almost recovered to that of the non-irradiated portions. This is further illustrated by [Fig. 3c](#f3){ref-type="fig"}, in which the R~i~(x) curve becomes a straight line, and the irradiated and non-irradiated potions are indistinguishable from each other. It has been shown that the predominant point defects---divacancies---become mobile only at 200--300 °C, but at lower temperatures, divacancies could be annihilated by interstitials that have dissociated from clusters and diffused through the lattice. It was previously found that by annealing a proton-irradiated Si substrate at 150 °C for ∼60 h, the divacancy concentration can be reduced by 54% (ref. [@b48]). This annihilation of divacancies by interstitials explains the increase in the thermal conductivity upon annealing at 120 °C. On the other hand, at 300 °C, divacancies become mobile and migrate to sinks (primarily to the surface in the case of a nanowire), and if their concentration is high enough, they also agglomerate to form small clusters that are stable up to 500 °C (ref. [@b37]). This explains the experimental results of the 300 °C annealing at low doses (2.5 × 10^14^ cm^−2^), nearly all the divacancies migrate to sinks and disappear, so that the thermal conductivity has practically fully recovered, while at higher doses, divacancies agglomerate, forming clusters which continue to act as scattering centres, so that the thermal conductivity of these portions could not be recovered. [Figure 3b](#f3){ref-type="fig"} shows that the thermal conductivity of the irradiated Si nanowire can be further tuned through annealing. High-dose regime---crystalline-amorphous transition --------------------------------------------------- There is a steep drop in thermal conductivity when the dose further increases from 2.5 × 10^16^ cm^−2^ (inset of [Fig. 1b](#f1){ref-type="fig"}). This demarcation between the low- and high-dose regimes indicates a phase transition in the material. We believe this phase change is the crystalline-amorphous transition of the Si nanowire arising from helium ion irradiation. For defect clusters and amorphous region in crystalline silicon nanowires, the phonon would be scattered differently. The defect clusters, which can be assumed to possess inner boundary if their size is large enough, have a relatively weak scattering for the phonon transport and the projected areas of defect clusters towards the heat-flow direction usually overlap, leading to the relative insensitive dependence of phonon transport on cluster size[@b49]. Hence the thermal conductivity tapers off as the dose increases (for dose below the critical value of amorphous region creation). In an amorphous silicon region, the phonon would be scattered significantly by disorder. For amorphous silicon, the majority of heat carriers are diffusions, which are non-propagating and delocalized modes, and their contribution is predicted to be ∼1 W m^−1^ K^−1^, according to the calculation by Allen-Feldman theory[@b50][@b51]. The rest of the contribution is from the propagating propagons, which account for the remaining half of the thermal conductivity, due to their small proportions (around 3%). Thus, the existence of an amorphous region in defect-clustered crystalline silicon, if any, would greatly impede phonon transport. In the low-dose regime, the disorder created by helium ions is mainly point defects, which agglomerate as the dose increases and the continuous aggregation of defects would change the crystalline lattice structure gradually in the silicon nanowires. When the dose goes beyond a critical value, the high energy ion-induced disorder can build up in crystalline silicon until its local free energy overcomes the energy barrier between the crystalline and amorphous phases. At this point, the defective crystalline lattice can collapse into amorphous phases[@b34]. Once an amorphous region is formed, it promotes the growth of damage in surrounding regions which in turn leads to further amorphization[@b52]. Finally, the entire irradiated region is amorphized. This amorphization process can be visualized from the diffraction pattern of an irradiated Si nanowire prepared separately on a TEM grid. [Figure 4a](#f4){ref-type="fig"} shows the bright-field image and diffraction pattern of an individual Si nanowire irradiated at two doses (7.5 × 10^16^ cm^−2^ and 2.5 × 10^16^ cm^−2^), using a focused electron beam with a spot size of 200 nm. The non-irradiated portion of the nanowire shows diffraction spots of single-crystal Si (inset(c)); for the portion with irradiation dose of 7.5 × 10^16^ cm^−2^, the nanowire is fully amorphous, leaving only short-range order among Si atoms, as indicated by the Debye-Scherrer rings[@b25] (inset(b)); for the irradiated portion with dose of 2.5 × 10^16^ cm^−2^, both Debye-Scherrer rings and diffraction spots can be seen (inset(d)), indicating the coexistence of both amorphous and crystalline phases. Similar diffraction patterns for Sample \#2 can also be seen in [Supplementary Fig. 4a](#S1){ref-type="supplementary-material"}. The coexistence of both amorphous and crystalline phases explains why the thermal conductivity for doses below 1 × 10^16^ cm^−2^ is much higher than that for larger doses at the dose of 2.5 × 10^16^ cm^−2^, where both crystalline and amorphous phases coexist. However, the amorphous region rapidly takes over the entire volume for higher doses, such as the doses of 3.5 × 10^16^ cm^−2^ and 4 × 10^16^ cm^−2^, as shown in [Supplementary Fig. 4b](#S1){ref-type="supplementary-material"}, where the Debye-Scherrer rings become more obvious. At the dose of 5 × 10^16^ cm^−2^, the thermal conductivity is close to the amorphous limit. Our experiment also demonstrates that helium ion irradiation can amorphize the Si nanowire without changing its morphology. Moreover, the interface between the irradiated area (for dose of 7.5 × 10^16^ cm^−2^) and non-irradiated area is particularly sharp and distinguishable, which corresponds well to the thermal resistance measurement by the e-beam technique. This clear boundary confirms the well-controlled dose irradiation with definition of a few nanometres. The interface thermal resistance of these abrupt boundaries can be measured, and can be exploited to further reduce the thermal conductivity of an intrinsic nanowire by incorporating multiple such boundaries, but this is beyond the scope of the present study. [Figure 4e](#f4){ref-type="fig"} shows the clear interface and insets (f) and (g) are the FFTs for the crystalline and amorphous regions, respectively. Discussion ========== In this paper, we have demonstrated that the thermal conductivity of an individual Si nanowire can be changed by selective helium ion irradiation. A single Si nanowire was irradiated at different positions with well-controlled helium ion doses, and an electron beam heating technique was used to measure the local thermal conductivity along the nanowire, which was then related to helium ion dose. We observed a clear transition from crystalline Si to amorphous phase at a dose between 1.5 × 10^16^ and 2.5 × 10^16^ cm^−2^ from the thermal conductivity versus dose curve. This result suggests a novel method to amorphize a Si nanowire without affecting its morphology. Moreover, within the dose regime in which only point defects are created, the Si nanowire thermal conductivity decreases drastically as the dose increases, and merely ∼4% defects could reduce the thermal conductivity by ∼70%, indicating a strong phonon scattering effect by the point defects. Within this regime, the effective scattering centres, inferred from parameter D, initially increase linearly with dose, and then saturates for larger dose. Finally, we observed that annealing could improve the thermal conductivity of the damaged nanowire, and at 300 °C for 2 h, the thermal conductivity of a damaged portion irradiated with a dose of 1 × 10^15^ cm^−2^ could be recovered to the value of the non-irradiated case. Methods ======= Sample preparation ------------------ Silicon nanowires with diameter ∼160 nm (Sigma-Aldrich 730866) grown along the \[111\] direction dispersed in 2-propanol solution were drop-casted onto a SiO~2~/Si substrate. A single Si nanowire was picked up from the substrate by a nano-manipulator (Kleindiek MM3A-EM) with a sharp tungsten probe and placed on a prefabricated suspended METS device. The METS device comprises two silicon nitride membranes, each of which is suspended by six long suspension beams for thermal isolation. Integrated on top of the silicon nitride platforms are Pt loops acting as resistance temperature sensors. The nanowire was positioned so as to bridge the two sensors, with the two ends fixed onto the two sensors by the electron beam-induced deposition (EBID) of Pt-C composite. The EBID process was carried out using a focused electron beam with acceleration voltage of 30 kV and current of 1.3--5 nA. The on-device Si nanowire was then cleaned using an Evactron RF plasma cleaner attached to the SEM chamber operated at a power of 14 W in 0.4 Torr of air for 2 h, after which it was put into a helium ion microscope (Zeiss Orion Plus) chamber and pumped overnight. It was then irradiated by helium ions (∼30--36 keV, 0.2--0.7 pA) with different doses at different positions ([Fig. 5a](#f5){ref-type="fig"}). This was achieved with the help of a nanometre pattern generation system. Eight samples were irradiated under similar helium ion energy and current (or dose rate) at room temperature. The SEM images of samples \#2 to \#8 are shown in [Supplementary Fig. 1](#S1){ref-type="supplementary-material"} and the irradiation details are shown in [Supplementary Table 1](#S1){ref-type="supplementary-material"}. The highest dose (7.5 × 10^16^ cm^−2^) chosen is such that there is no significant sputtering of Si atoms, and a change in the Si nanowire diameter is not observable in a transmission electron microscope (TEM). The dose was then reduced for each irradiated position. The irradiation length and irradiation distance are different for all the measured silicon nanowires, the details of which are summarized in [Supplementary Table 1](#S1){ref-type="supplementary-material"}. Sample \#3 was annealed in air at 120 °C for cumulative periods of 20, 40, 80 and 160 h, with measurement by an electron beam heating technique carried out for each interval. After this, it was put into a tube furnace and annealed in forming gas at 300 °C for 2 h. To check for possible oxidation during annealing, HRTEM and diffraction pattern measurements of the post-annealed nanowire were carried out and the results are shown in [Supplementary Fig. 5](#S1){ref-type="supplementary-material"}. The nanowire is observed to remain single-crystalline and an unavoidable intrinsic oxide layer of ∼2.5 nm thickness is seen, which is similar to or even slightly smaller than that found in electroless-etching silicon nanowires[@b19]. Thermal conductivity measurement by electron beam technique ----------------------------------------------------------- In the electron beam heating technique, a focused electron beam is used as a localized heating source. By moving the electron beam along the nanowire and measuring the temperature rise (ΔT~L~ and ΔT~R~) of the left and right sensors corresponding to the position of the electron beam x as measured from the left sensor, the cumulative thermal resistance R~i~(x) between the electron beam location and the left sensor is[@b30]: where α~i~(x)=ΔT~L~(x)/ΔT~R~(x), and R~b~ is the equivalent thermal resistance of the six beams that link each suspended membrane to the environment as measured by the conventional thermal bridge method in which a 10 μA DC current was passed through the left Pt loop, raising its temperature (ΔT~L0~) by ∼8.5 K, and the temperature rise of both sensors (ΔT~L0~ and ΔT~R0~) were measured from the resistance change of the Pt loops. Finally, α~0~=ΔT~L0~/ΔT~R0~. From the R~i~(x) curve, the spatially resolved thermal conductivity of the nanowire can be calculated as: , where A=πd^2^/4 and d is the diameter of the nanowire. Data availability ----------------- The data that support the findings of this study are available from the corresponding authors upon request. Additional information ====================== How to cite this article: Zhao, Y. et al. Engineering the thermal conductivity along an individual silicon nanowire by selective helium ion irradiation. Nat. Commun. 8, 15919 doi: 10.1038/ncomms15919 (2017). Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary Material {#S1} ====================== ###### Supplementary Information The authors thank Dr Rongguo Xie for useful discussions. This work is funded by grant MOE2011-T2-1-052 from the Ministry of Education, Singapore, and grant NRF-CRP002-050 from the National Research Foundation, Singapore. The authors declare no competing financial interests. Author contributions Y.Z., D.L. and J.T.L.T. designed the experiment; Y.Z. and D.L. made the samples and performed the measurement; Y.Z., D.L., B.L. and J.T.L.T. wrote the paper; J.C. and L.Z. conducted the simulations; Y.Z., A.B., O.S.O., M.J.B., S.K., H.H. and D.S.P. carried out the irradiation works; Y.Z. and R.R.U. performed the TEM characterization; B.L. and J.T.L.T. directed the project. ![Thermal conductivity measurement by e-beam technique.\ (a) Measurement result for the damaged portion with highest dose (7.5 × 10^16^ cm^−2^) in Sample \#1. The solid black line is the cumulative thermal resistance R~i~(x) from the electron beam position x to the left sensor. The dashed red (blue) line is the temperature rise of the left (right) sensor, ΔT~L~ (ΔT~R~). The solid green line is ΔT~L~+ΔT~R~, which is proportional to the power absorption from the electrons. (b) Measured thermal conductivity (κ) of Samples \#1 to \#8 versus dose. Inset: the same data plotted on a logarithmic scale. The solid black square denotes the thermal conductivity of intrinsic nanowires (namely, with zero dose), that is derived from averaging calculated ones by fitting several intrinsic portions for the eight samples.](ncomms15919-f1){#f1} ![Analysis thermal conductivity dependence on dose and vacancy concentration.\ (a) Measured thermal resistivity (1/κ) as a function of dose for low-dose regime. For dose \<1 × 10^15^ cm^−2^, thermal resistivity increases linearly with dose; whereas as the dose further increases, the thermal resistivity tends to saturate. (b) Extracted parameter D as a function of dose with and without considering boundary scattering. Open blue circles and red squares correspond to optimal D fitted from measured thermal conductivities with and without considering the boundary scattering, respectively. D can be well fitted into an exponential curve , denoted by the red and blue solid line; while the black dashed line represents the linearly fitted curve of D as a function of dose in the low-dose regime: . (c) Thermal conductivity as a function of percentage of removed Si atoms from NEMD simulations. The blue arrow indicates the trend with increasing cross-sectional area. (d) Thermal resistivity (1/κ) as a function of percentage of removed Si atoms from NEMD simulations. From the calculation in [Supplementary Note 1](#S1){ref-type="supplementary-material"}, dose of 1 × 10^16^ cm^−2^ corresponds to 4% removed Si atoms. The legend is the same as that in c.](ncomms15919-f2){#f2} ![Thermal conductivity enhancement due to annealing.\ (a) Cumulative thermal resistance (R~i~) as a function of position, before and after annealing. The damaged portion is irradiated by helium ions at a dose of 5 × 10^15^ cm^−2^. The dashed R~i~ curve is for a portion that is not irradiated. (b) Thermal conductivity versus dose under different annealing conditions. (c) After annealing at 300 °C for 2 h, the thermal conductivity of the damaged potion with 2.5 × 10^14^ cm^−2^ (shown in the red dotted circle in (b)) was recovered.](ncomms15919-f3){#f3} ![TEM characterization.\ (a) Bright field TEM image and diffraction patterns of a Si nanowire irradiated by helium ions. Inset (b) is the diffraction pattern of the portion with helium ion dose of 7.5 × 10^16^ cm^−2^. The Debye-Scherrer rings indicate the fully amorphous status. (d) Diffraction pattern of the portion with helium ion dose of 2.5 × 10^16^ cm^−2^. Both Debye-Scherrer rings and diffraction spots can be seen, indicative of the coexistence of both amorphous and crystalline phases. (c) Diffraction pattern of non-irradiated portion; (e) HRTEM for the selected area marked in red dashed line in a. Inset (f,g) are FFT for crystalline and amorphous regions. The distinguishable interface between irradiated- and non-irradiated areas illustrates the well-controlled dose irradiation. Scale bar, for (a,e) are 300 nm and 10 nm, respectively.](ncomms15919-f4){#f4} ![Silicon nanowire on METS device with helium ion irradiation.\ (a) Scanning electron microscope (SEM) image of Si nanowire damaged by helium ions (Sample \#1). The portions coloured orange (with length of L~1~) denote the parts damaged by helium ions; the uncoloured portions (with length of L~2~) denote the intrinsic Si nanowire. Scale bar, 1 μm. (b) Schematic showing the spatially resolved measurement of thermal conductivity of the damaged Si nanowire using electron beam heating technique. A focused electron beam (purple cone) is used as a heating source, and the temperature rise of the left and right sensors (ΔT~L~ and ΔT~R~) is measured by the Pt resistance thermometers (blue and purple loops).](ncomms15919-f5){#f5} [^1]: These authors contributed equally to this work	{ "pile_set_name": "PubMed Central" }
Background ========== Adolescence is considered a critical period in human evolution, although it is often not recognized as such by health care workers and parents as well as professionals in adult medicine and pediatric disciplines \[[@B1]\]. The onset of menstruation in adolescence is a phenomenon that signals reproductive maturity and should not be seen as an abnormal condition or disease. Adolescent girls often do not receive accurate information about menstrual health because of culturally specific practices that lead to incorrect and unhealthy behaviors \[[@B2]\]. In recent years, reproductive health care has been a primary concern of the Ministry of Health and Medical Education and the Ministry of Education in Iran. In this regard, the Department of Youth and School Health started a reproductive health care program in January 2005 to accompany the United Nations Population Fund programs. Education regarding menstrual health has changed since 2004 in Mazandaran province. In 2006, in addition to routine education under the supervision of the Deputy of Provincial Health taking place in schools, education about hygiene during menstruation was introduced to target groups including students, parents and school personnel. Subsequently, the education administration of Mazandaran province began conducting health promotion projects focused on puberty. The puberty health education is running in some schools now. Research worldwide has addressed issues related to menstrual health. Studies have revealed the effectiveness of health education interventions conducted among high school, guidance school and even primary school girls \[[@B3]-[@B6]\]. A study in Nepal \[[@B7]\] reported that knowledge and practices related to menstruation were not satisfactory. These researchers suggested that health education by teachers, parents and the media is important in addressing misconceptions about menstrual health. Changes in the health education system to promote reproductive health have become a priority for the Ministry of Health and Medical Education and the Ministry of Education. This is likely to result in a change in hygiene patterns during menstruation for girls participating in the program. Research related to menstrual hygiene programs in the past decade suggests average to poor menstrual health among Iranian girls \[[@B1],[@B2],[@B8]\]. One study reported that although female adolescents in Mazandaran province had some knowledge of puberty, most of them did not fully understand it and lacked accurate scientific knowledge about their physiological maturity \[[@B9]\]. The present study investigated the effect of education to improve menstrual health through a health promotion project for several schools in Mazandaran province. Methods ======= A quasi-experimental study was conducted to evaluate the personal health and hygiene of high school girls during menstruation. An experimental study is a research design which researcher manipulates one or more variables, and measures any change in other variables. It includes pre-post test, a study group and a control group, and random sampling methods. A quasi-experimental study is lack one or more of these design characters. Ethical approval was obtained from the Ethics Committee of the Mazandaran University of Medical Sciences (code number: 87-10-16-87-92) for this research project. Informed consent was obtained from parents of all teenagers who participated in the study. The health promotion initiative by the education organization in Mazandaran province, in north of Iran, was conducted in selected high schools. The project included 10 two-hour educational sessions using adolescent health resources. The educational manual was developed by adolescent health professionals\' team. Educational topics included the significance of adolescence, physical and emotional changes during adolescence, pubertal and menstruation health and premenstrual syndrome. Participants comprised 689 high school girls (349 in the study group and 349 in the control group) 14 to 18 years of age in urban and rural public high schools with low socio economic status in Mazandaran province. The control group comprised high school students in Mazandaran province who did not participate in the education. As much as possible, the control samples were selected from the nearest neighbor school. Factors matched in the study and control group were school type (rural or urban), grade, age and educational field of study. Inclusion criteria were students in high school, 14 to 18 years of age, parental consent for participating in the project and having had at least one menstrual period. Sampling was done in cooperation with the Ministry of Education in Mazandaran province and related administrations. A questionnaire developed by the researcher was used for data collection. The menstrual health education was a government-sponsored program. The Youth and School Health Department delivered the training and evaluated menstrual health status before the education. Menstrual health was considered poor or average in Mazandaran province. The questionnaire in the present study was administered by the researcher to both the experimental and control group after the education. This self-administered questionnaire comprised 71 items in five sections: demographic characteristics, behaviors during menstruation, menstrual-related patterns, sources of information about menstruation and questions related to personal health. We determined the validity and reliability of the questionnaire as follows. Face and content validity were determined based on viewpoints of the adolescent girls, health sciences specialists, and experts in the community. Test-retest reliability was examined using a 10-14 day interval with 20 qualified and available female students. These students were not part of the study. The questionnaire was reviewed and analyzed for repeatability and internal consistency aspects. Cronbach\'s alpha coefficient was used to assess internal consistency. Repeatability was estimated using the intra-class correlation coefficient (ICC). Cronbach\'s alpha coefficient and mean ICC were 0.88, and 0.90, respectively. Based on ICC reliability ranges of less than 0.4 (poor), 0.4-0.7 (fair to good), 0.6-0.8 (good) and 0.8-1 (excellent), the reliability of the questionnaire was considered excellent \[[@B10]\]. Menstrual health was measured by investigating variables such as the pad material used, daily use of a clean pad, number of pad replacements during 24 hours, changing the pad at night or when in school, tendency to stay at home or not exercise, ability to properly perform daily tasks, negative impact of menstruation on studying, frequency of school absences during menstruation, noting the first day of the monthly menstruation, noticing typical period occurrence, paying attention to sudden changes in menstruation, changes in diet during menstruation including eating more protein and less salt and sugar, reduced consumption of snacks, sweets or pickles and increased consumption of vegetables, fruits and grains, avoiding certain foods, iron supplements during menstruation and hygiene habits including bathing the genitals. The socio-economic level was categorized into four groups, based on the occupation of the father of the family: the higher class, including major landowners, merchants and manufacturers; the middle class, including government administrators, teachers, minor landowners, army officers, clergymen, individuals with personal professions and professionals; the working class, including professional and skilled workers; and the lower class, including the unemployed and the unskilled workers. It is noteworthy that scores 0 and 1 were allocated to each of the questions of menstrual health and individual health status. Adolescent were arrived at percentile 0-25 (poor), 25-50 (average), 50-75 (good), and 75-100 (excellent). The data collected were analyzed using the Statistical Package for the Social Sciences version 16 for Windows (SPSS Inc., Chicago, IL, USA). Descriptive and analytical indicators were determined. T-tests and chi square tests were employed for analysis, and the level of significance was set at 0.05. Results ======= The average age of the girls in this study was 15.7 years (SD = 1.08). A total of 35.9% were from urban areas, and 95.8% were from Mazandaran province originally. The socioeconomic level was lower class for 86.2% of the participants. For 53.2%, their family had more than 5 members. Regarding parental education, 77.5% of fathers did not have a diploma, and 11.6% held an academic degree. Of the mothers, 87.6% did not have a diploma, and 4.1% had an academic degree. As shown in Table [1](#T1){ref-type="table"}, the level of menstrual health of subjects who participated in the puberty health promotion program was significantly better than that of the control group (p = 0.013). ###### Comparison of menstrual health between the experimental group and control group Menstrual Health Experimental Group n (%) Control Group n (%) -------------------- -------------------------- --------------------- Good and Excellent 30(8.6) 17(4.9) Average 299 (85.7) 295 (84.5) Poor 20 (5.7) 37 (10.6) Total 349 (100) 349 (100) Among the most significant results was the impact of participation in the classes on bathing and genital hygiene. A total of 61.6% in the experimental group compared with 49.3% in the control group engaged in usual bathing during menstruation (p = 0.002). There was a statistically significant difference between those who bathed in the first two days of menstruation in the experimental and control groups, 19.5% and 12.5%, respectively (p = 0.01). Only 19.7% of study participants compared with 25.3% of controls always or sometimes avoided washing their genitals after urinating during menstruation (p = 0.02). Regarding educational field of study, 12.3% of students in the experimental sciences had good genital hygiene during the menses compared to 6% for math and physics students and 5.5% for students in the humanities (p = 0.007). Regarding the relationship between menstruation patterns and level of menstrual health, premenstrual syndrome (PMS) and menstrual cycle regularity were statistically among the most significant factors. PMS was reported by 9% of the participants. Among the students with irregular menstrual cycles, 10.4% had poor menstrual health compared with 5.5% among students with regular cycles (p = 0.001). The students with PMS had poorer menstrual health (17.5%) than those not suffering from PMS (7.2%, p = 0.01). Individual health status was also statistically related to menstrual health. Those with better individual health behaviors had better menstrual health (p = 0.001) (Table [2](#T2){ref-type="table"}). ###### Relationship between menstrual health and individual health status Individual Health Status Menstrual Health -------------------------- ------------------ ------------ ----------- Good and Excellent 18 (38.3) 60 (10.1) 2 (3.5) Average 24 (51.1) 355 (59.8) 32 (65.1) Poor 5 (10.6) 179 (30.1) 23 (40.4) Total 47 (100) 594 (100) 57 (100) Attitude towards menstruation was related to menstrual health. A total of 78.7% of participants with good and excellent menstrual health considered menstruation a natural phenomenon and an indication of mental and physical health. For 6.4% of participants, it was troublesome, disturbing and agonizing, and 8.5% expressed both views. The remaining participants expressed neither opinion. These four views (i.e., positive, troublesome, both and no opinion) among the participants with average menstrual health represented 50.8%, 13%, 19.4% and 16.9%, respectively, and among those with poor menstrual health 33.3%, 15.8%, 35.1% and 15.8%, respectively (p = 0.0001). A total of 38.3% with high, 22.2% with average and 24.6% with poor menstrual health felt positive about menstruation; 40.4% with high, 39.7% with average and 28.1% with poor menstrual health had no opinion; and 10.6% with high, 32.2% with average and 40.4% with poor menstrual health felt it was troublesome (p = 0.01.). Sources of information about menstruation were mainly mothers (24.5%), health care professionals at clinics (9.3%), friends (8.7%), elder sisters (7.7%), other family members (3%), books and magazines (2.7%), mass media (1%) and other. Discussion ========== The results showed a statistically significant difference between menstrual health of the experimental group compared with the control group, providing support for participation in adolescent health care programs \[[@B6],[@B11],[@B12]\]. Poor menstrual hygiene is a risk factor for reproductive tract infections \[[@B11]\]. Thus, health care during menstruation is a vital element of any health program for girls during puberty. Health programs on television, programs by health care professionals in schools and trained parents can play a major role in transmission of knowledge related to good menstrual health. Chang et al. \[[@B6]\] conducted a similar study of primary school girls in grades 5 and 6 and showed that educational programs in schools for students and their parents are quite effective in increasing menstrual health. A study in Egypt \[[@B4]\] showed the efficacy of a menstrual education program for first and second year girls at a secondary school and recommended expanding the program to elementary, preparatory and other secondary schools. The above studies confirm that health education programs in schools can intervene effectively to improve menstrual health. Research studies in Iran, in contrast to international studies that consider menstrual health satisfactory or somewhat poor, reported menstrual health as average or poor prior to 2004. Researchers in Iran investigated 250 high school girls in the city of Karaj in 1998 and found that only 6% practiced menstrual hygiene such as bathing and using hygienic materials. Forty-eight percent avoided any physical activity or exercise during menstruation. A total of 67% reported having taken medicine for pain without consulting a doctor. Furthermore, their background knowledge had no impact on their menstrual health because it was insufficient and not completely accurate \[[@B2]\]. According to our results and similar studies, adolescents get most of their information about menstruation from their mothers \[[@B1],[@B12]\]. Salari et al. \[[@B1]\] investigated knowledge, attitudes and practices related to menstrual health of high school girls in the city of Urmia, Iran in 1999-2000. The results showed that 68.8% got menstruation information from their mothers and sisters, and 64.1% reported a need for further knowledge. Statistically, there was a significant relationship between knowledge and menstrual health of the subjects and the mothers\' occupation. The results of a similar study done in the north and south of Tehran showed that menstrual health is not related to socioeconomic level of the family. Having an elder sister in the family had no significant relationship with the level of menstrual health. We propose that the absence of health education in schools is an important factor affecting menstrual health in adolescent girls \[[@B13]\]. Various studies have shown that health education increases knowledge and positive attitudes towards puberty, and facilitates acceptance of pubescent changes as a natural physiological phenomenon \[[@B12]\]. Conclusion ========== The present study confirms that the health promotion project in Mazandaran province has been quite effective in promoting menstrual health among high school girls, especially in deprived regions that are the target groups in any health promotion project. Similar studies and educational programs sponsored by the Ministry of Education and health clinics would be of value. In addition, future studies to support the results of the present study regarding the benefits of direct education intervention are recommended. Competing interests =================== The authors declare that they have no competing interests. Authors\' contributions ======================= All four authors contributed to the development of ideas and design of the study. MF wrote the first draft of the manuscript, which was critiqued by the other authors. The final version of the manuscript was critically reviewed by ZH. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2458/12/193/prepub> Acknowledgements ================ This study was funded and supported by Mazandaran University of Medical Sciences, Grant No. 87-9-6-87-92. We would like to thank all the adolescents and their parents for their participation.	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== Among secondary tumors of the liver, metastases from colorectal adenocarcinomas are the easiest to identify histopathologically. In most cases, they exhibit glands strongly resembling those of the primary colorectal cancers (CRCs), their lumens lined by tall columnar cells and filled with necrotic debris ("dirty necrosis") \[[@CR1]\]. The histopathological grade and differentiation capacity of a primary colorectal cancers can be even recapitulated in xenografted tumor organoids \[[@CR2]\]. Given the strikingly different environments in which the primary and metastatic (or xenografted) tumor cells are forced to grow, this phenotypic robustness is remarkable. It might reflect the existence of a genetic--epigenetic program that remains more or less unchanged, even when the CRC cells migrate to the liver and establish a new tumor there. This hypothesis is consistent with reports of high genomic concordance between some primary CRCs and their metastases \[[@CR3], [@CR4]\] and with claims that the DNA-methylation-based epigenetic profile of liver metastases of unknown origin can reliably reveal the lesions' primary cancer source \[[@CR5]\]. However, the genetic and epigenetic profiles of the CRC liver metastases will never be 100% identical to those of the primary tumor growing in the gut. Among other things, some anticancer drugs cause genetic and/or epigenetic changes that favor the selection and emergence of drug-resistant cellular clones \[[@CR6]\]. Some degree of divergence from the primary tumor is thus inevitable, and its magnitude is probably characterized by substantial inter-tumor and inter-tumor-type variation \[[@CR7], [@CR8]\]. Investigation of this divergence requires powerful omics tools capable of exploring most if not all of the human genome. Genome-wide analysis of DNA methylation, for example, can be particularly informative in this setting. During embryogenesis, the DNA methylome undergoes profound remodeling, with the removal and addition of methyl groups at cytosine bases, primarily those making up CpG dinucleotides. These changes are associated with markedly modified gene-expression (although the mechanisms underlying this association are still incompletely understood) \[[@CR9]--[@CR13]\], and play pivotal roles in the processes of cellular differentiation and organ specification \[[@CR14], [@CR15]\]. Once the large intestine methylome is established, however, it is chemically stable and faithfully maintained by mitotic inheritance for the life of the individual. A major exception to this rule occurs with the onset of neoplasia in the gut. During colorectal tumorigenesis (as well as that occurring in other organs), many normally unmethylated CpG islands located in gene promoters become heavily methylated. For a few of these promoters, the hypermethylation has been strongly linked to gene silencing with potentially crucial roles in tumorigenesis. Paradigmatic is the epigenetic inactivation of the DNA mismatch repair gene MLH1, which leads to the emergence of post-replicative DNA mutations and microsatellite instability (MSI) \[[@CR16]\]. The hypermutator phenotype of these cancers has distinctive effects on their prognosis and their sensitivity to chemotherapy \[[@CR17]--[@CR19]\]. They account for about 15% of all CRCs, and are presently classified as consensus molecular subtype 1 in the gene-expression-based CRC classification \[[@CR20]\] and as MSI/CIMP-H (CpG island methylator phenotype-high) in the TCGA classification of gastrointestinal cancers \[[@CR21]\]. We recently investigated the DNA methylomes of precancerous and cancerous colorectal lesions \[[@CR13]\]. In addition to our extensive characterization of the hypermethylation phenotype of gene regulatory genomic regions in both types of tumor, we also confirmed the presence in these lesions of another classical tumor-associated change in the methylome: widespread hypomethylation. These findings were obtained with bisulfite sequencing of DNA subjected to targeted enrichment for regions containing CpG islands since whole-genome sequencing is associated with well-known constraints in terms of costs, data storage, and analysis. In the present study, we used methyl-CpG-binding domain (MBD) sequencing to explore the methylomes of normal and neoplastic colon tissues, with the aim of discovering whether the process of colorectal tumor-associated epigenetic remodeling continues to evolve in CRC cells that have migrated to and established metastases in the liver. We also compared the performances of the MBD- and targeted-enrichment methods in the characterization of the CRC methylome. Methods {#Sec2} ======= Tissues {#Sec3} ------- Tissues were prospectively collected at the University Hospital of Zurich (Switzerland) with institutional research ethics committee approval. Donors provided written consent to tissue collection, testing, and data publication. Samples were numerically coded to protect donors' rights to confidentiality. Immediately after resection, samples of normal colon mucosa, primary CRCs, and CRC liver metastases were frozen in liquid nitrogen and stored at − 80 °C. They included six primary tumors, three of which were accompanied by patient-matched samples of normal mucosa from the same gut segment (\> 2 cm from the tumor), and twelve liver metastases (none of which were from the primary CRC donors) (Table [1](#Tab1){ref-type="table"}). Nine patients were diagnosed with a single metastasis. For the three who developed multiple metastases, the largest lesion was included in the study. All of the primary and secondary lesions were microsatellite-stable and CpG island methylator phenotype-negative. Table 1Characteristics of tissues included in the studyPatient numberTissueSiteLesion size (mm)Stage and grade of primary cancer ^a^C1cprimary CRCR30 × 25T4aN2bR0cM1 / G2C1nnormal mucosaRC2cprimary CRCS27 × 10T3N0M0R0 / G3C2nnormal mucosaSC3cprimary CRCR25 × 25T2N1bM0 / G2C3nnormal mucosaRC4primary CRCT18 × 15T4aN0M0R0 / G2C5primary CRCS50 × 20T3N2acM1 / G2C6primary CRCA30 × 25T4aN2bcM1 / G3M1CRC metastasisliver13 × 7pT3pN2M0 / G3M2CRC metastasisliver25 × 13pT4pN0M0 / G3M3CRC metastasisliver10 × 6pT3pN0cM1 / G3M4CRC metastasisliver20 × 7pT1pN1M0 / G2M5CRC metastasisliver15 × 10pT4pN1M1 / G2M6CRC metastasisliver10 × 5pT3N0M0 / G2M7CRC metastasisliver15 × 15pT1pN1M0 / G2M8 ^b^CRC metastasisliver10 × 8pT3N1M0 / G2M9 ^b^CRC metastasisliver20 × 20pT3N0M0 / G2M10 ^b^CRC metastasisliver25 × 15pT3N2M0 / G3M11 ^b^CRC metastasisliver10 × 8pT3pN1M0 / G2M12 ^b^CRC metastasisliver15 × 8pT3pN1M0 / G2Abbreviations: CRC colorectal cancer, R rectum, S sigmoid, T transverse colon, A ascending colon^a^ Cancer TNM and grade classification in Sobin LH, Wittekind C. TNM classification of malignant tumors. 6th ed. New York, NY: Wiley-Liss, 2002.^b^ Patients who received chemotherapy 1--20 months before resection of the liver metastasis. All other donors were chemo-naïve when sampled tumors were resected Laser tissue microdissection {#Sec4} ---------------------------- Laser tissue microdissection was done with a CellCut system (Molecular Machines & Industries, Glattbrugg, Switzerland). Briefly, 10 μm-thick sections were cut with a Hyrax C60 cryostat (Zeiss, Feldbach, Switzerland) from frozen tissues embedded in Tissue-Tek O.C.T. (i.e., optimum cutting temperature formulation; Sakura, Alphen aan den Rijn, The Netherlands). The sections were placed on membrane slides (Molecular Machines & Industries), fixed in propanol for 45 s, and covered with one drop of Mayer's hematoxylin (MHS128, Sigma-Aldrich, Buchs, Switzerland) for 45 s. After vigorous washing, the sections were sequentially immersed in 0.1% ammonia (10 s), propanol (45 s), and xylene (45 s) and dried for 5 min. Stained tissues were then subjected to laser microdissection using special tubes with caps to which the dissected sections adhered (Molecular Machines & Industries, Glattbrugg, Switzerland). Epithelial cells from normal or neoplastic crypts were selectively collected on the cap, with care taken to minimize stromal-cell contamination. Between 10 and 25 × 10^6^ μm^2^ of dissected epithelium (= ≈ 100,000 to 250,000 epithelial cells) was collected from each sample. Immediately after dissection, DNA was extracted with the QIAmp DNA Micro kit (Qiagen, Hombrechtikon, Switzerland) and quantified with a Qubit fluorometer and dsDNA HS Assay kit (Thermo Fisher Scientific, Reinach, Switzerland) (total yield: 100--500 ng DNA). MBD-peptide-based capture of DNA for deep sequencing {#Sec5} ---------------------------------------------------- Methylated DNA for sequencing was isolated using an MBD-capture protocol \[[@CR22]\]. Reaction volumes were scaled down to successfully process the small volumes of genomic DNA obtained with laser tissue microdissection. Briefly, 100 ng of input DNA was fragmented to an average length of 200 bp in a Covaris (Brighton, UK) S2 ultrasonicator. Recombinant MBD2 protein-mediated enrichment (MBDE) for methylated DNA was carried out with the MethylMiner kit (ThermoFisher, Waltham, MA, USA) using a single elution step with 2000 mM NaCl elution buffer. Multiplex Illumina libraries were prepared with the NEBNext Ultra DNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA), and their sizes and concentrations were evaluated with the Agilent (Santa Clara, CA, USA) 2200 TapeStation system. Libraries were sequenced on an Illumina 2500 system (Illumina, San Diego, CA, USA) (on average 50-bp single-end reads, 40 million reads per sample). Raw methylome data are deposited in ArrayExpress (accession number: E-MTAB-8232). Detection of differential methylation {#Sec6} ------------------------------------- MBD-sequencing reads were aligned to the GRCh37/hg19 human reference genome using bwa (version 0.7.12-r1039) and the BWA-MEM algorithm \[[@CR23]\] and de-duplicated with Picard (Picard Toolkit, version 2.13.2; 2018. Broad Institute, GitHub Repository: <http://broadinstitute.github.io/picard/>). The R-package csaw (version 1.14.1) \[[@CR24]\] was used to identify regions that were differentially methylated in neoplastic tissues (primary and/or metastatic) vs. normal mucosa or in metastatic vs. primary cancers. The number of reads per sample was counted in consecutive overlapping windows (length: 200-bp, overlap: 190 bp). Windows with minimum count sums of 30 across samples were kept. Additional filtering was used to exclude windows with average log counts per million that were below − 1. Reads aligned to the X or Y chromosome were excluded from analysis. The csaw package uses methods from edgeR (version 3.22.3) to identify differential binding \[[@CR25]\] (i.e., NB GLM model to fit read counts). Each filtered window was tested for differential methylation (i.e., differential binding of the methyl-binding protein) associated with disease status (CRC, primary or metastatic, vs. normal mucosa) or disease stage (primary vs. metastatic CRC). After testing, windows separated by no more than 50 bp were merged to form regions, and P-values were calculated for each region using the Simes method, as described in the package. Differentially methylated regions (DMRs) were classified as hypermethylated or hypomethylated based on the direction of the methylation change in the majority of windows included in the region. Regions were annotated using the annotatr R-package \[[@CR26]\], which overlays regions of interest with predefined annotations from external sources. Annotations were grouped to create three "intragenic" categories: 1) regulatory: genomic regions 5 kb upstream from the TSS, and 5'UTRs; 2) gene body: exons and introns (from the end of the 5'UTR to the beginning of the 3'UTR); and 3) the 3'UTRs. Everything outside these regions was considered "extragenic" (Fig. [1A](#Fig1){ref-type="fig"}). Regions were also classified based on their CpG density: CpG islands and the CpG shores and shelves flanking them were grouped together and referred to as an sSISs (shelf--shore-island-shore-shelf) region, and everything outside these regions was classified as extra-sSISs (Fig. [1D](#Fig1){ref-type="fig"}). These regions were overlaid onto all windows tested for differential methylation. Fig. 1Genomic distribution and functional classification of MBD-captured regions within the genome. a. Schematic showing components of an \"intragenic\" genomic region. Areas outside such regions were classified as \"extragenic\". b. MBD-captured regions (total: 322,551) classified as extragenic vs. intragenic. c. Distribution of the 206,258 intragenic MBD-captured regions by subregions: gene regulatory (including 5\'UTRs and regions 5 kb upstream from the TSS); gene body (including exonic and intronic sequences between the end of the 5\'UTR and the beginning of the 3\'UTR); and the 3\'UTRs. d. Schematic showing components of an "sSISs" region (i.e., a CpG [i]{.ul}sland and flanking CpG [s]{.ul}hores and [s]{.ul}helves). Areas outside such regions were defined \"extra-sSISs\". Schematics in this panel and in panel A were adapted from diagrams contained in the R-package annotatr vignette.e. MBD-captured regions classified as sSISs vs. extra-sSISs. f. Distribution of the 53,521 sSISs MBD-captured regions by subregions (CpG islands, shores, and shelves). N.B. After exclusion of the X and Y chromosomes, the human genome contains 26,361 canonical CpG islands, 15,423 of which were covered by MBDE. (See Discussion). Visualization was performed using the ggplot2 (version 3.0.0) \[[@CR27]\] and UpSet (version 1.3.3) \[[@CR28]\] R-packages. Analyses were performed with R versions 3.6.0. Estimated CpG density {#Sec7} --------------------- Each CpG site (CpGs) (Team TBD 2014; BSgenome.Hsapiens.UCSC.hg19; R package version 1.4.0.) was centered within a 200-bp window, and the ratio of the observed to expected numbers of CpGs (O:E CpG ratio) \[[@CR29]\] was calculated for each window, as a measure of its CpG density (low - O:E CpG ratios \< 0.3; intermediate \> = 0.3 but \< 0.6; high \> = 0.6 -- high density). Comparison of MBDE- and targeted bisulfite-sequencing in colorectal cancer {#Sec8} -------------------------------------------------------------------------- In a previous study \[[@CR13]\], we used targeted bisulfite sequencing (SeqCapEpi CpGiant protocol, Roche, Basel, Switzerland) to assess DNA methylation in three CRCs (all microsatellite-stable and CpG island methylator phenotype-negative) and matched samples of normal mucosa (ArrayExpress accession number: E-MTAB-6949). The results of those experiments were compared with those obtained in the present study with MBD-capture sequencing, to identify potential strengths and weaknesses of the two enrichment methods for CRC methylome characterization. For this analysis, we considered CpGs covered by the targeted enrichment (TE) procedure, those covered by MBD enrichment (MBDE), and those covered by both enrichment methods. Because nearby CpGs are frequently co-methylated \[[@CR30], [@CR31]\], the log-fold change (logFC) and P-value for each CpG site in a given MBD-captured region were assumed to be identical to those calculated for the region as a whole. For CpGs covered by the TE procedure, differences in methylation proportions and P-values were calculated with the BiSeq R package (Hebestreit K, Klein H, 2018. BiSeq: Processing and analyzing bisulfite sequencing data. R package version 1.22.0), by modeling the methylation level within a beta regression and estimating the group effect, in this case primary CRC vs. normal tissue. CpGs covered by both methods were classified according to the CpG density of the region in which they were located (as specified above) and the direction of the differential methylation identified at the site in primary CRCs (vs. normal mucosa): hypermethylated (TE: difference in methylation proportions \> 0; MBDE: logFC \> 0 and P-value \< 0.05); hypomethylated (TE: methylation proportion difference \< 0; MBDE: logFC \< 0 and P-value \< 0.05); isomethylated (TE: methylation proportion difference = 0; MBDE: logFC = 0 and P-value = \> 0.05). The strength of the linear association between the two methods was calculated with the Pearson correlation coefficient. DMRs identified from TE data with the BiSeq R package were overlaid with DMRs detected from MBDE reads, and regions were classified as overlapping if they shared a sequence of at least 1 bp. Code implemented in the analysis is available at <https://github.com/sorjuela/livermetastasis_MBDseq_paper>. Results {#Sec9} ======= DNA extracted from the 21 laser-microdissected, colorectal tissue samples was subjected to MBD-capture sequencing (Table [1](#Tab1){ref-type="table"} and Supplementary Fig. [1](#MOESM1){ref-type="media"} in Additional Files). Fig. [1](#Fig1){ref-type="fig"} shows the genomic locations of the 322,551 methylated regions isolated with this technology. Most (64%) were located in the "intragenic" genome, which contains coding genes, their regulatory segments, and 3'UTRs (Fig. [1](#Fig1){ref-type="fig"}a and b), and 73% of these intragenic regions were situated within a gene body (Fig. [1](#Fig1){ref-type="fig"}c). As for CpG statuses (Fig. [1](#Fig1){ref-type="fig"}d), 17% of all the captured regions were located in sSISs sequences (Fig. [1](#Fig1){ref-type="fig"}e), predominantly (73%) in CpG shores or shelves rather than islands (Fig. [1](#Fig1){ref-type="fig"}e). Few of the 322,551 MBD-captured regions displayed significant differential methylation (defined by an adjusted P-value cutoff of 0.05) in the cancer tissues (n = 2155 in the primary CRCs; n = 3223 in the liver metastases) in comparison with normal mucosal samples (Fig. [2](#Fig2){ref-type="fig"}, Supplementary Table 1 in Additional file [5](#MOESM5){ref-type="media"}), and no DMRs were identified when primary and metastatic cancers were compared. Five of the 12 liver metastases came from patients who had received chemotherapy 1--20 months before liver resection (Table [1](#Tab1){ref-type="table"}). No significant differences were found between their methylomes and those of the seven metastases from chemo-naïve patients. For this reason, all 12 metastatic lesions were considered as a single group in the statistical analysis. Fig. 2Genomic regions in primary CRCs or CRC liver metastases displaying differential methylation relative to that in normal colon mucosa (NM). a and b. Differentially methylated regions (DMRs) characterized by hypermethylation and hypomethylation (vs. NM) present in primary CRCs, metastatic CRCs, or both. c, d, e, and f. Distributions of hypermethylated and hypomethylated DMRs in the extragenic vs. intragenic genomes; among the intragenic genome components; between the sSISs and extra-sSISs genomic segments, and among the sSISs components, respectively. (See Figure 1 for topography of genomic segments.) The vast majority of DMRs identified in tumor tissues (94% of those in the primaries, 70% in metastases) were hypermethylated relative to the normal mucosa. In contrast, hypomethylated DMRs were generally less common, and almost all of them were found in liver metastases rather than in the primary tumors (Fig. [2](#Fig2){ref-type="fig"}a and b). Hypomethylated DMRs were thus a distinctive feature of the metastatic lesions (Fig. [2](#Fig2){ref-type="fig"}b), whereas most hypermethylated DMRs were found in both types of neoplastic tissue (Fig. [2](#Fig2){ref-type="fig"}a). In both primary CRCs and metastases, hypermethylated DMRs were far more abundant in intragenic than extragenic regions (Fig. [2](#Fig2){ref-type="fig"}c), in regulatory segments than in gene bodies or 3'UTRs (Fig. [2](#Fig2){ref-type="fig"}d), in sSISs than in extra-sSISs segments (Fig. [2](#Fig2){ref-type="fig"}e), and in CpG islands than in CpG shores/shelves (Fig. [2](#Fig2){ref-type="fig"}f). In contrast, hypomethylated DMRs were equally represented in the extragenic and intragenic genomes (Fig. [2](#Fig2){ref-type="fig"}c). Within intragenic regions, they were more frequent in gene bodies (than in regulatory segments or 3'UTRs) (Fig. [2](#Fig2){ref-type="fig"}d). Hypomethylation was more common in the extra-sSISs genome (Fig. [2](#Fig2){ref-type="fig"}e), but those located within sSISs areas displayed similar frequencies in CpG islands, shores, and shelves (Fig. [2](#Fig2){ref-type="fig"}f). The distribution of DMR lengths among the genomic segments discussed above is shown in the Additional Files (Supplementary Fig. [2](#MOESM2){ref-type="media"}). As detailed in the Methods section, we then compared the primary CRC methylomes characterized with MBDE with the results obtained in other primary CRCs using TE \[[@CR13]\]. As shown in Fig. [3](#Fig3){ref-type="fig"}a, MBDE covered a larger portion of the genome with a higher CpG content (249.5 Mb containing \~ 7,6 million CpGs vs. \~ 79.1 Mb with \~ 2.8 million CpGs with TE), and around 1,1 million CpGs were captured by both methods. Both methods preferentially covered CpGs in regions where the density of these dinucleotides was relatively high (Fig. [3](#Fig3){ref-type="fig"}b), and 1,105,282 (96.5%) of the CpGs covered by both methods were located in regions with an intermediate- or high-CpG density (Fig. [3](#Fig3){ref-type="fig"}c). Fig. 3Performance of MBD enrichment (MBDE) and targeted enrichment (TE) in identifying differentially methylated CpGs or regions (DMRs) in primary CRCs (vs. normal mucosa). a. Compared with TE, MBDE captured 3 times more genomic base pairs (including 33.4% of TE-captured base pairs) and almost 3 times more CpGs (including 41.3% of TE-captured sites). b. First three density plots, from left: Compared with CpGs missed by both methods (mean O:E CpG ratios: 0.35 for MBDE, 0.39 for TE), the CpGs captured by each method were preferentially located in genomic areas with relatively high CpG densities (mean O:E CpG ratios: 0.58 for MBDE, 0.65 with TE). Fourth density plot: CpGs captured by MBDE only, by TE only or by both methods tended to be located in genomic areas with similar CpG densities (mean O:E CpG ratios: 0.56 for MBDE, 0.65 for TE, 0.66 for both methods). c. Methylation changes (log fold change in MBDE, beta-values in TE) at CpGs captured by both methods in low-, medium-, and high-CpG-density areas. (See Methods for calculation of differential methylation with each method and for calculation and classification of CpG density.) Intermethod correlation values are shown for each density area. For most of the CpGs captured by both methods (gray bars), no methylation differences in primary CRCs (vs. normal mucosa) were identified with either MBDE or TE. The number of hypermethylated CpGs detected by both methods (red bars) steadily increased with increasing CpG rates. In addition, in medium- and high-CpG-rate areas of the genome, MBDE captured hypermethylated CpGs that were not found to be significantly hypermethylated with TE (blue vs. green bars). Hypomethylated CpGs were more frequently identified by TE than MBDE (yellow vs. orange bars). d. The CpG site trends were confirmed by analysis of DMRs. Over 400 hypermethylated DMRs were identified with both methods (in addition to the 1584 identified only by MBDE and the 827 detected only with TE). However, the overlap of commonly identified hypermethylated DMRs is likely underestimated because of the use of different analytical packages to identify them with MBDE or TE (see example in Supplementary Figure [3](#MOESM3){ref-type="media"} C). Hypomethylated DMRs were in contrast identified almost exclusively by TE (examples in Supplementary Figure [3](#MOESM3){ref-type="media"} D and E). Cancer-associated changes in methylation levels identified with the MBDE and TE displayed a moderately strong positive correlation (r: from 0.59 to 0.65, Fig. [3](#Fig3){ref-type="fig"}c, upper panel). However, the methods differed substantially in calling hyper- or hypomethylation at individual CpGs (Fig. [3](#Fig3){ref-type="fig"}c, lower panel). Of the 1,144,600 CpGs covered by both methods, 57,077 (5%) were concordantly identified as hypermethylated and 30,525 (2.7%) as hypomethylated. More frequently, however, the two methods yielded discordant results: indeed, 163,841 (14.3%) of the CpGs were identified as significantly hypermethylated only with MBDE, and a similar proportion (n = 167,545 \[14.6%\]) emerged as significantly hypomethylated only with TE. Accordingly, hypermethylated DMRs were more frequently identified in primary CRCs with MBDE than with TE, whereas hypomethylated DMRs were almost exclusively identified with TE (Fig. [3](#Fig3){ref-type="fig"}d, and examples of DMRs in Additional Files, Supplementary Fig. [3](#MOESM3){ref-type="media"}). The lower number of hypermethylated DMRs identified with TE was due in part due to the lack of SeqCapEpi CpGiant probes targeting many of the regions found to be hypermethylated with MBDE (Supplementary Fig. [3](#MOESM3){ref-type="media"}B) and in part to the different computational approaches used to analyze the MBDE and TE data sets (Supplementary Fig. [3](#MOESM3){ref-type="media"}C). As for the difference involving hypomethylated DMRs, it probably reflects the fact that these regions were generally shorter than their hypermethylated counterparts and displayed less markedly altered methylation levels (hypo beta-value mean: 0.21, hyper beta-value mean: 0.31; see also Supplementary Table [1](#MOESM5){ref-type="media"}, Supplementary Figs. [2](#MOESM2){ref-type="media"} and [3](#MOESM3){ref-type="media"} D-E, and Fig. [1](#Fig1){ref-type="fig"}e of our previous study \[[@CR13]\]). Discussion {#Sec10} ========== We used a high-stringency MBD-protein-based enrichment protocol to obtain methylated DNA for deep sequencing from CRCs (primary and liver metastases) and normal colon mucosa. The portion of the genome sequenced (7.8%) included \~ 27% of the \~ 28 million CpGs found therein (\~ 7.6 × 10^6^). Small fractions of the MBD-isolated regions were differentially methylated in primary (2155 regions including 279,441 CpGs) or metastatic CRC (3223 regions containing 312,723 CpGs) samples relative to the unmatched samples of normal colon mucosa we tested. Importantly, the hypermethylation phenotype of the liver metastases closely resembled that of primary CRCs, in terms of both the identity and location of the hypermethylated DMRs. We have previously shown that this phenotype is already evident in precancerous colorectal lesions (i.e., sessile serrated lesions and, to a lesser extent, adenomatous polyps), and it becomes increasingly obvious in CRCs \[[@CR13]\]. Our present findings suggest that this progression reaches a plateau before CRC cells seed the liver. In contrast, the spread of CRC-associated hypomethylation continues after the tumor cells metastasize to the liver. The extent of this progression requires further investigation with whole-genome analysis of the methylome (see below), but it is tempting to speculate that late increases in hypomethylation might contribute to metastasis-specific alterations in the gene expression, genomic stability, and/or drug susceptibility of CRCs \[[@CR32]--[@CR36]\]. Our study is the first to use genome-wide deep-sequencing to compare the methylomes of primary CRCs and CRC liver metastases. To our knowledge, only three previous studies \[[@CR37]--[@CR39]\] have analyzed this issue at the genome-wide level. In two studies, freshly collected, patient-matched tissue pairs (nine in one case \[[@CR37]\], three in the other \[[@CR38]\]) were analyzed using a methylated CpG island amplification microarray approach involving methylation-specific, restriction-enzyme digestion of defined CpGs in approximately 6000 gene promoters \[[@CR40], [@CR41]\]. Both found that the hypermethylation phenotypes of CRC liver metastases closely resembled those of their primary cancer counterparts, leading the investigators to conclude ---as we have--- that most of the DNA hypermethylation associated with colorectal tumorigenesis probably occurs before the disease spreads to the liver. Our in silico analysis of previously published llumina Infinium 450 microarray data on six patient-matched tissue pairs \[[@CR39]\] also confirmed that the methylomes of primary and metastatic colorectal cancers are similar (Additional Files, Supplementary Fig. [4](#MOESM4){ref-type="media"}). This conclusion is also supported by findings of a recent analysis of 70 pairs of formalin-fixed, paraffin-embedded tissues, which revealed concordance between primary CRCs and matched metastases taken from different organs in the CpG island hypermethylation phenotype at five gene promoters \[[@CR42]\]. The tissues we tested were prospectively collected to obtain, from fresh samples, high-quality DNA for MBD capture, and this markedly reduced our chances of obtaining matched samples of colorectal cancers and their corresponding liver metastases in the timeframe of the study. (The clinical management of patients with colorectal cancer---before and after detection of liver metastases---is usually a fairly long process marked by multiple surgical and chemotherapeutic interventions, often carried out in different hospitals.) Although this is undeniably a limitation, the strikingly similar hypermethylation phenotypes observed in the unmatched primary and metastatic tumors suggests that similar or even greater concordance would probably be evident in matched samples. Our use of laser-capture tissue microdissection probably played a key role in reducing the variability between primary and metastatic epithelial tumors by eliminating normal and tumor-related stromal cells from their respective microenvironments, an important aspect that recently emerged in a gene-expression study of primary and metastatic colorectal cancers \[[@CR43]\]. Our findings on hypomethylation are also in agreement with the previous observations of Hur et al. \[[@CR35]\], who found hypomethylation of long interspersed nuclear element-1 (LINE-1) sequences in 77 formalin-fixed, paraffin-embedded samples of primary CRCs (vs. normal mucosa), and this alteration was even more evident in matched samples of liver metastases. The fact that some of the hypomethylated LINE-1 sequences were found to be located within the intronic regions of proto-oncogenes whose expression was increased in liver metastases points intriguingly to possible functional consequences of the late increase of hypomethylation in cancer cells seeding in the liver \[[@CR35]\]. The present study was conducted exclusively on microsatellite stable/non-CIMP CRCs, which are far more common than CRC with MSI/CIMP-H phenotype. The decision to focus our work on the more frequently encountered phenotype was motivated by the difficulties we encountered in the prospective collection of samples (see above) and the costs of the genome-wide analysis of the DNA methylome. In addition, our previous work \[[@CR13]\] has shown that the DNA methylome of primary microsatellite stable/non-CIMP CRCs differs from that of MSI/CIMP-H primaries. Data in the literature are lacking on the possible evolution of the MSI/CIMP-H CRC methylome during metastasis. Genome-wide analysis of the methylome should therefore be extended to this molecular type of CRCs to determine whether hyper−/hypomethylation changes between primary and metastatic tumors are CRC-type specific. We found no evidence that chemotherapy significantly alters the methylomes of CRC liver metastases. This is an important issue in view of the emergence of drug-resistant clones that might exhibit clinically-relevant epigenomic changes \[[@CR44], [@CR45]\], and our finding obviously requires further and more in-depth investigation. The timing of chemotherapy relative to primary and metastatic tumor resections varied widely in our study, as did the drugs administered. All, however, were cytotoxic agents, and it is important to extend the investigation to include the possible effects of more recently introduced targeted approaches, such as anti-EGFR antibodies. Bisulfite sequencing is still considered the gold standard technique for analyzing DNA methylomes. However, bisulfite conversion of unmethylated cytosines causes substantial DNA damage \[[@CR46], [@CR47]\], which can be a major concern when the amount of input DNA extracted from clinical samples is limited (e.g., the laser-microdissected sections of frozen tissues used in our study). Using MBDE, we obtained high-quality methylome data with only 100 ng of input DNA per sample, but reliable results with this enrichment method can reportedly be obtained with volumes as small as 15 ng \[[@CR48]\]. Furthermore, owing to cost considerations and computational constraints, bisulfite sequencing analysis is usually limited to a genome-wide selection of regions, such as that obtained with the targeted enrichment step we used for our bisulfite-sequencing analysis of the methylomes of normal, precancerous, and cancerous colorectal tissues \[[@CR13]\]. Metastatic CRCs were not included in that study, but comparison of the data it generated on primary cancers and normal mucosa with those obtained here provided insights into the pros and cons of the two pre-sequencing enrichment protocols (Fig. [3](#Fig3){ref-type="fig"}). As expected, MBDE covered a larger portion of the genome and more CpG dinucleotides than TE (Fig. [3](#Fig3){ref-type="fig"}a). The probes used for TE, which were designed a priori, target specific genomic loci consisting mainly of CpG islands in regulatory regions \[[@CR13]\]. In contrast, MBDE relies on the binding of the MBD polypeptide to any of the numerous methylated regions in the genome---extragenic as well as intragenic, and CpG shores and shelves as well as the islands they flank (Fig. [1](#Fig1){ref-type="fig"}b-c). Therefore, MBDE allowed us to recover more genomic information than TE. Our analysis also confirmed that both MBDE and TE preferentially cover CpG-dense regions (Fig. [3](#Fig3){ref-type="fig"}b), but the mean O:E CpG ratio of those covered by TE was slightly higher than that of MBDE-covered regions (difference between means = 0.075). This small increase is consistent with the fact that TE detected 25,291 (96%) of the 26,361 "canonical" CpG islands located in non-sex chromosomes, as opposed to only 15,423 (59%) detected with MBDE. Most CpG islands are unmethylated in human tissues \[[@CR49], [@CR50]\] and will therefore be missed by the MBD polypeptide, which binds to methylated regions \[[@CR22]\]. In contrast, however, the shores and shelves flanking canonical CpG islands are not missed by MBDE since they are usually methylated. Indeed, by considering a broader window that includes shelves and shores, as well as the island they flank (i.e., sSISs), MBDE actually covered 93.5% of these regions (24,644, Fig. [1](#Fig1){ref-type="fig"}f). As for the CpG sites covered by both enrichment technologies, previous comparisons have shown high concordance in the methylation levels identified, in non-colorectal cells or tissues, by MBDE and bisulfite-sequencing approaches (reduced representation bisulfite sequencing \[[@CR51]\] and whole genome bisulfite sequencing \[[@CR52], [@CR53]\]). Unlike these studies, our work included both diseased tissues (i.e., cancers) and normal colorectal tissues, with the latter as a reference. We could thus focus on methylation level alterations in primary CRCs, as compared with basal levels observed in normal colorectal mucosa, and we also assessed the performance of MBDE vs. TE in regions differing in CpG-density (Fig. [3](#Fig3){ref-type="fig"}c). On the whole, the methylation changes identified with the two methods were concordant, with correlation values ranging from 0.59 to 0.65 (Fig. [3](#Fig3){ref-type="fig"}d). Significant changes in methylation were observed in high-, intermediate, and low-CpG-density regions. Interestingly, however, MBDE identified hypermethylation in high-density regions that was not detected by TE, and TE consistently detected more hypomethylation, regardless of the regions' CpG density (Fig. [3](#Fig3){ref-type="fig"}c-d). The latter difference was likely due to the limited magnitude of the changes involving hypomethylation and their tendency to occur in relatively short stretches of DNA. Both factors reduce the odds that these alterations will be detected by MBDE, which cannot quantify methylation levels at single CpGs \[[@CR51], [@CR52]\]. Despite the evidence for concordance between MBDE and TE, the differences (e.g., those related to cost, coverage achieved, and resolution) should also be considered when choosing between the two types of enrichment protocols (see \[[@CR51], [@CR54]\] for formal comparisons). In general, for the study of regional methylation, single-CpG resolution might not be necessary, and a lower-cost method of enrichment like MBDE can be used. Targeted enrichment, however, is more suitable when there is a need for CpG resolution of methylation levels (e.g., identification of methylation at transcription-factor binding sites). Conclusions {#Sec11} =========== While tumor-associated hypermethylation at specific genomic regions---promoters in particular---appears to be a progressive phenomenon that plateaus before CRC cells disseminate to the liver, the progression of CRC-associated hypomethylation seems to continue in metastatic lesions, with low-magnitude changes that are nonetheless evident throughout the genome. At this stage, the clinical implications of our findings are unknown. One can reasonably speculate, however, that the relatively fixed nature of the CRC hypermethylome could render liver metastases less "flexible" and consequently more vulnerable to certain medical treatments, whereas the ongoing extension of hypomethylation after liver seeding might lead to the emergence of adaptive changes favoring drug resistance. Further study of this issue, using methods capable of characterizing whole-genome DNA methylation and gene expression, is therefore important since it might well reveal new prospects for the treatment of patients with late-stage CRC. Supplementary information ========================= {#Sec12} Additional file 1: Supplementary Fig. 1. MDS (multidimensional scaling) plot of DNA methylation levels in the 21 samples included in the study. While normal mucosa samples (blue circles) were clustered together and separated from tumors, there is no obvious separation between CRCs (red circles) and liver metastases (green circles). Additional file 2: Supplementary Fig. 2. DMR length distribution. Length in base pairs of hypermethylated and hypomethylated DMRs in: A. the extragenic vs. intragenic genomes; B. among the intragenic genome components; C. the sSISs and extra-sSISs genomic segments; and D. among the sSISs components. NM: normal mucosa, CRC: primary cancer, Met: metastasis. Additional file 3: Supplementary Fig. 3. Integrative Genomics Viewer snapshots showing examples of significantly hyper- or hypomethylated DMRs detected with MBDE, TE, or both enrichment methods. Abbreviations: DMR: differentially methylated region; MBDE: MBD enrichment; TE: targeted enrichment; NM: normal mucosa; CRC: colorectal cancer. Tracks corresponding to NM samples, primary CRCs, and CRC liver metastases are shown in blue, red, and green, respectively. For MBDE, track-landscape heights indicate read depth; for TE, track-bar heights indicate the methylation level (%) at a given CpG site. A: Large, hypermethylated DMR overlapping the CpG island in the EYA4 promoter, detected by both methods. B: Four consecutive, hypermethylated DMRs detected by MBDE. The fourth one was also detected by TE, with a probe directed specifically at CpG island no. 37 at this genomic locus. C: This DMR was found to be significantly hypermethylated only with MBDE. At the adjusted P-value cutoff used in the analysis of TE data, the differential methylation in this region was not significant, although methylation levels at many CpG sites (indicated by bar heights) are clearly higher in primary CRCs than in NM samples. D and E: Examples of hypomethylated DMRs detected only with TE: like most of the hypomethylated DMRs we found, these two are shorter than the hypermethylated DMRs. In general, the hypomethylated DMRs were also characterized by relatively small differences with respect to the methylation levels in NM. (See Results and Discussion.) Additional file 4: Supplementary Fig. 4. In-silico analysis of Gene Expression Omnibus DataSet GSE53051 (submitted by Timp et al., reference \[[@CR39]\]). For each sample, the normalized average beta values per probe were downloaded. A. MDS (multidimensional scaling) plot of the beta values for 6 patients with paired normal mucosa (NM), primary cancer (CRC) and liver metastasis (Met). The limma package (reference \[[@CR25]\]) was used to identify the 1000 probes displaying the highest variability and to plot the MDS. Consistent with the findings of our own study (Supplementary Fig. [1](#MOESM1){ref-type="media"}), normal mucosa samples from the DataSet clustered together and were appreciably separated from tumors, whereas there was no obvious separation between CRCs and liver metastases. B. Scatter plots of the mean beta values per tissue (Met vs NM, CRC vs NM, CRC vs Met) for all the probes (Genome, top 3 plots), and for probes located in CpG Islands (Islands, bottom 3 plots). CRCs and Mets had similar methylomes, both with Genome and Island probes (top and bottom panels on the right, respectively), while skewed profiles towards hypomethylation (Genome probes) or hypermethylation (Islands probes) in tumors (vs NM) were detected (the four panels on the left and in the middle). C. The similarity of the CRC and Met methylation patterns is also reflected by the mean beta values per sample across all probes (Genome, top) and the probes in CpG Islands (Islands, bottom). Additional file 5: Supplementary Table 1. CpGs : CpG sites CIMP : CpG island methylator phenotype CRC : Colorectal cancer DMRs : Differentially methylated regions FDR : False discovery rate logFC : Log-fold change MBD : methylCpG binding domain MBDE : MBD protein-mediated enrichment MSI : Microsatellite instability O:E CpG ratio : Ratio of the observed to expected numbers of CpGs sSISs : Shelf-shore-island-shore-shelf TE : Targeted enrichment Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information ========================= Supplementary information accompanies this paper at 10.1186/s12885-020-06777-6. We thank Hadi Garibi for performing preliminary analysis of the data, Hannah Parker for the production of data reported in reference \[[@CR13]\], Susanne Dettwiler for technical assistance, and Marian Everett Kent for editing the manuscript. S.O. and M.D.R. analyzed data; M.M. and G.M. performed research; P.S. and P.K. collected tissue samples and clinical data; S.O. and G.M. wrote the paper. All authors have read and approved the manuscript. Swiss National Science Foundation grant no. 310030--179477 to S.O. (PhD salary) and G.M. (principal investigator of the research financed by the Foundation); Swiss Cancer League grant no. 3397-02-2014 to M.M. (postdoc salary). M.D.R. acknowledges support from the University Research Priority Program Evolution in Action at the University of Zurich (costs of deep sequencing). The datasets generated and analyzed during the current study are available in the ArrayExpress repository (accession numbers: E-MTAB-8232). Ethics approval was received from the Zürich university hospital ethics committee, Switzerland (Nr. EK2011--0275). Donors provided written consent to tissue collection, testing, and data publication. Samples were numerically coded to protect donors' rights to confidentiality. Not applicable. The authors declare that they have no competing interests.	{ "pile_set_name": "PubMed Central" }
Objective ========= To develop a modular approach to infectious disease genomic analysis that can easily integrate with public health analytics systems. Using dynamic approaches to genomic sequence analysis, relevant whole genome data can be quickly and accurately visualized and correlated, using a minimum of computational resources. We propose to develop visualization modules that integrate disparate data sources including integrate geospatial location metadata with associated epidemiological factors to enable faster outbreak identification and enhance surveillance. Introduction ============ Whole-genome sequencing of disease-causing organisms provides an unabridged examination of the genetic content of individual pathogen isolates, enabling public health laboratories to benefit from comparative analyses of total genetic content. Combining this information with sample metadata such as temporal, geospatial, morbidity, and mortality can greatly increase the efficacy of genomics analysis. However, with the vast amount of data generated by such techniques, meaningful, rapid, and accurate analysis that interprets and correlates nucleotide polymorphisms for public health practice presents many challenges. To this end we have created a modular genomics analysis toolkit that can easily integrate diverse data streams and couple analysis with an array of visualization platforms. Methods ======= Using open source tools we have assembled an analysis package that automatically processes next generation sequencing (NGS) data from the ubiquitous Illumina MiSeq. FastQ files are uploaded, filtered, trimmed and assembled. The largest contiguous DNA assemblies are BLASTed (Basic Local Alignment Search Tool) to determine closest reference genome match in RefSeq to identify the species of any isolate sequenced. After reference determination, a custom gene by gene typing algorithm calculates the core genome alignment required for phylogenetic evolutionary analysis. This approach is based on the whole genome multiple sequence typing (wgMLST) approach that was developed to define a rapid universal identification and typing scheme for pathogens. Alternative genomic methods used to process NGS data for evolutionary analysis rely on first calculating high quality single nucleotide polymorphisms (SNPs) for all sequenced isolates with respect to the reference genome and then creating a phylogeny. These approaches however can be computationally expensive as the number of sequenced isolates increases. Our algorithm attempts to overcome these computational bottlenecks through the more efficient gene by gene typing approach. Additionally, a key component of our algorithm is a rapid tree construction module where we calculate the minimal set of genes that can effectively recreate the ideal (core genome) phylogeny at a user accepted threshold of consensus identity. Results ======= This toolkit provides an automated analysis suite for processing isolate sequencing data directly from the Illumina MiSeq. Utilizing a minimal core genome algorithm simplifies the data sets and reduces overall compute time for even large data sets. Additional modules being developed utilize open source tools and common sequence formats to integrate evolutionary analysis results from quality scored whole genome sequences with geographical data in order to provide geospatial visualization of distinct and related isolates in an outbreak. Output data from the PERL modules is seamlessly integrated into open source C++ Qt libraries prepackaged to perform geospatial visualization and relatedness clustering using multidimensional scaling (MDS) approaches. Platform independent Qt libraries provide a cross-platform application framework for easy integration of these "genomic surveillance" modules into existing surveillance applications. The virtual overlay of phylogenetic relationships onto isolate maps provides population structure in epidemiological studies and provides a mechanism for rapid real time analysis of transmission chains and effective retrospective analysis of pathogen evolutionary trends. Conclusions =========== Utilizing and analyzing raw whole genome sequence data directly from the Illumina MiSeq moves current capabilities one step closer to real-time infectious disease characterization. Minimal core gene alignment analysis allows for computation on systems commonly available to infectious disease laboratories, circumventing the need for computationally expensive analysis. These genomic methods, if implemented within existing public health laboratory response programs, promise to revolutionize the ability of the laboratory to provide information and evidence on the evolution, transmission and virulence for pathogenic organisms.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#sec1-1} ============ Cadmium contamination has serious potential implications for soil, water, and human health\[[@ref1]\]. Unlike organic pollutants, heavy metals are not degradable and may persist for relatively long periods of time in aquatic and terrestrial environments\[[@ref2]\]. Amongst the methods that remove heavy metals from environment, biosorption approaches have received increased attention because of their environmentally friendly and cost-effective feature, as well as their superior performances. Due to these properties, the biosorption is usually preferred to physical or chemical remediation technologies. Biosorption takes advantage of naturally occurring biomaterials for heavy metal removal\[[@ref3]\]. Biomaterials such as bacteria, yeasts, algae, and some plants have been successfully applied for pollutant adsorption, and it has been shown that these materials demonstrate a high capacity in removing large amounts of heavy metals from wastewaters. Other benefits of using biomaterials include lower cost of investment, lower energy consumption, and high removal efficiency of the heavy metals\[[@ref4]\]. However, natural biomaterials have neither a high selectivity nor a high capacity for the adsorption of heavy metals. Surface-engineered bacteria solve these problems by expressing the large amounts of selective sorbent motifs, such as metallothioneins (MTs), on their surfaces. MTs are low-molecular-weight (6--7 kDa), cysteine-rich, metal-binding proteins found in animals, higher plants, eukaryotic microorganisms, and some prokaryotes\[[@ref5]\]. MTs have the ability to bind to both physiological (such as zinc, copper, and selenium) and xenobiotic (such as cadmium, mercury, silver, and arsenic) heavy metals through the thiol group of its cysteine residues, which represents nearly the 30% of its amino acidic residues\[[@ref6]\]. The main advantages associated with the surface expression of metal adsorbents, in comparison to intracellular expression, include the elimination of the time-consuming and rate-limiting step of crossing the membrane, prevention of interference with redox pathways in the cytosol, the uptake of any heavy metal of interest by expressing its binding peptide on the surface of cell, and recycling of the biosorbents\[[@ref7]-[@ref11]\]. Various bacterial-based systems have been developed for heavy metal removal by using microorganisms over-expressing MTs on their surfaces\[[@ref9],[@ref12]\]. Surface-engineered bacteria show enhanced adsorption of heavy metals and offer a promising strategy with respect to the development of bacterial-based biosorbents for the removal of heavy metal ions from wastewater. Cadmium is considered as one of the most toxic heavy metals, and in fact, cyanobacterial MTs such as SmtA have been shown to have high affinity for Cd (II), Zn (II), and Cu (II) ions\[[@ref13]\]. In our previous research, the Lipoprotein-outer membrane protein A (Lpp'-OmpA) system was used for surface display of cyanobacterial MTs\[[@ref14]\]. The ability of these MTs to be displayed on the bacterial surface has not been evaluated so far. The Lpp'-OmpA system consists of a signal sequence and the first nine N-terminal amino acids of the major E. coli Lpp′ joined to a transmembrane domain (residues 46 to 159) of the outer membrane protein of OmpA\[[@ref15]\]. The purpose of this paper was to study the kinetics and equilibrium isotherm of the Cd (II) biosorption process. We optimized the uptake of Cd (II) by surface-engineered E. coli using the statistical design of experiments. In addition, the present study investigated the impact of desorption agents on the recovery of adsorbed Cd (II). Surface functional groups of bacterial cells involved in cadmium adsorption were also determined using Fourier transform infrared spectrophotometer analysis. MATERIALS AND METHODS {#sec1-2} ===================== Microorganism and media {#sec2-1} ----------------------- The recombinant E. coli strain E, engineered with a cyanobacterial MT, SmtA, using the pET26b-Lpp′-OmpA expression vector was obtained from National Institute of Genetic Engineering and Biotechnology (Tehran, Iran)\[[@ref14]\]. Luria Bertani (LB) was used as the growth medium and supplemented with kanamycin sulfate to a final concentration of 50 mg/mL. Isopropyl β-D-1-thiogalactopyranoside, as an inducer, was added to the culture medium when the cells reached an optical density of 0.6 at 600 nm. After induction period, the culture was incubated at 25ºC for 5 h. The cells were subsequently harvested by centrifugation at 4000 ×g and freeze dried until further use. Preparation of metal solutions {#sec2-2} ------------------------------ Metal solutions were prepared by diluting a 1000 mg/L stock solution of Cd (NO~3~)~2~.4H~2~O, with 0.1 M trisaminomethane-hydrochloric acid (Tris-HCl) to obtain concentrations between 10-110 mg/L. For each solution, the initial Cd (II) concentration and the concentration in the samples following the biosorption treatment process were determined using a flame atomic absorption spectrometer (Perkin Elmer Aanalyst, USA). Cadmium adsorption studies {#sec2-3} -------------------------- Batch adsorption experiments were conducted to study kinetic models, equilibrium isotherms, and the effect of different variables on cadmium adsorption, consisting of pH, temperature, and mass dosage. Each experiment was carried out in 100-mL Erlenmeyer flasks containing 10 mL of Cd (II) solution by shaking at 100 rpm. Then biomass was separated by centrifugation at 4000 ×g and filtered through a Whatman filter paper with a pore size of 25 µm. Filtered samples were then analyzed for residual Cd (II) ion concentration using an analyst 700 atomic adsorption spectrometer (Perkin Elmer Aanalyst 700, USA). A control experiment was also carried out using the same solution and equipment, but in the absence of the bioadsorbent, E. coli E. Solute uptake by the recombinant E. coli E strain can be calculated from the differences between the initial and final quantities of the solute contained in the supernatant as follows: ![](IBJ-21-380-g001.jpg) where Q is the solute uptake (mg/g); C~0~ and C~f~, the initial and equilibrium solute concentrations in solution (mg/L), respectively; V, solution volume (L); M, the mass of the biosorbent (g)\[[@ref16]\]. Biosorption kinetics {#sec2-4} -------------------- The sorption kinetics data provide valuable insights into the reaction pathways, the mechanism of the sorption reaction, and solute uptake\[[@ref16]\]. The pseudo-first-order and pseudo-second-order biosorption models were applied to describe the kinetics of biosorption. The initial Cd (II) concentration was 20 mg/L in Tris-HCl buffer, pH 6.5. The sorption time varied between 5 and 100 min, and temperature was set at 30ºC. At different times, each flask was removed from the shaker, and the biomass was centrifuged as mentioned above and then filtered. Finally, the solutions were analyzed to measure the residual Cd (II) concentration. The pseudo-first order model points that the rate of adsorption sites occupation is proportional to the number of unoccupied sites\[[@ref17]\]. The linear equation for this model is: ![](IBJ-21-380-g002.jpg) where q~e~ and q are the amounts of metal ions adsorbed at equilibrium and at any time (t), respectively (nmol/mg) onto the biosorbent surface, and K~1;~ is the rate constant of the first-order biosorption\[[@ref17]\]. In the pseudo-second order model, it is assumed that the rate of the occupation of adsorption sites is proportional to the square of the number of unoccupied sites\[[@ref17]\]. Linear equation for this model is: ![](IBJ-21-380-g003.jpg) where q~eq~ and q~t~ are the amounts of metal ions adsorbed on the biosorbent at equilibrium and at any time (t), respectively (nmol/mg), and K~2~ is the rate constant of second-order biosorption (mg/nmol min)\[[@ref17]\]. The linear regression curve fitting procedure was performed with Microsoft Excel (version 7). The goodness of fit of the data to the model was evaluated by the coefficient of determination, R^2^, by least-squares method\[[@ref17]\]. Biosorption isotherms {#sec2-5} --------------------- Equilibrium adsorption isotherms are usually used to determine the capacity, surface properties, and affinity of an adsorbent. Among all theoretical models, the Langmuir and Freundlich equilibrium models, the most widely used sorption isotherms, were chosen for the estimation of the adsorption capacity of E. coli E. The linear Langmuir equation is written as follows\[[@ref18]\]: ![](IBJ-21-380-g004.jpg) where q~e~ is the equilibrium biosorption capacity of biomass in nmol Cd (II)/mg of biomass, C~e~ is the equilibrium concentration of Cd (II) ion in nmol/L, q~max~ is the maximum amount of metal sorbed in nmol Cd (II)/mg of biomass, and K~L~ is the constant that is referred to the bonding energy of sorption in nmol/L. Langmuir isotherm refers to homogeneous adsorption, in which each molecule possesses constant enthalpies and sorption activation energy (all sites possess equal affinity for the adsorbate), with no transmigration of the adsorbate in the plane of the surface\[[@ref19]\]. The linear Freundlich equation is written as follows\[[@ref18]\]: ![](IBJ-21-380-g005.jpg) where q~e~ is the equilibrium biosorption capacity of the biomass in nmol Cd (II)/mg biomass, C~e~ is the equilibrium concentration of Cd (II) ion in nmol/L, and K~f~ (in nmol/L) and 1/n are constants related to the sorption capacity and intensity, respectively. Freundlich isotherm is the earliest known relationship describing the non-ideal and reversible adsorption, not restricted to the formation of monolayer. This empirical model can be applied to multilayer adsorption, with non-uniform distribution of adsorption heat and affinities over the heterogeneous surface\[[@ref19]\]. Isotherm experiments were carried out at 30ºC, using 10 mg dried biomass/10 mL of varying initial Cd (II) concentrations in the range of 10--110 mg Cd (II)/L of Tris-HCl buffer (pH 6.5), with constant shaking at 100 rpm and using an equilibrium time of 1 h. The linear regression curve fitting procedure was performed as mentioned above. Response surface methodology (RSM) {#sec2-6} ---------------------------------- One-factor-at-a-time, the classical method of experimental designs, involves changing one independent variable while maintaining all others at a fixed level. This method does not include the interactive effects among the variables. Experimental factorial designs can overcome this problem\[[@ref20]\]. The RSM based on full or factorial design is a powerful tool for optimization, which have been employed extensively to optimize the biosorption of heavy metals\[[@ref21]-[@ref23]\]. In order to obtain the optimum conditions for Cd (II) adsorption, three variables, including pH, temperature, and biomass dosage of solution were selected for the study. The range for these factors was chosen based on preliminary screening experiments according to cell viability on LB agar. For this purpose, cell suspensions of the same weight were adjusted to different pH values using the following buffers (0.1 M): glycine-HCl (pH 3), sodium acetate (pH range of 4-5), Tris-HCl (pH range of 6-8), and glycine-NaOH (pH range of 9-10) and incubated at 30°C for 1 h with shaking at 100 rpm. In another test, cell suspensions of the same weight were incubated at different temperatures (20, 30, 40, 50, 60, 70, and 80ºC) for 1 h at pH 7 with shaking at 100 rpm. Subsequently, equal volumes of the cell suspensions were spread onto LB agar plates (pH 7) and incubated at 37 ºC overnight. Colony forming units were counted and compared to the condition with maximum colony forming units, at pH 7 and 30ºC. The central composite design (CCD) based on RSM was used to optimize the above mentioned factors. There are three types of CCD, which depend on where the axial points are placed. In the face-centered CCD type, the axial points are at the center of each face of the factorial space, so α=±1. The high and low values of the factors are coded as +1 and -1, respectively. The mean value of the factors was assigned to 0 as the central point. The experimental design for the three mentioned factors contained a total of 17 experiments, representing six axial points on cubic surfaces and 2^3^ factorial points on vertices, as well as the central point with 3 replications ([Table 1](#T1){ref-type="table"}). ###### Experimental design based on the central composite design used in this study Trials Point type pH Temprature (ºC) Biomass dosage (mg/mL) -------- ------------ ----- ----------------- ------------------------ 1 Factorial 5.0 20 10 2 Factorial 8.0 20 10 3 Factorial 5.0 40 10 4 Factorial 8.0 40 10 5 Factorial 5.0 20 30 6 Factorial 8.0 20 30 7 Factorial 5.0 40 30 8 Factorial 8.0 40 30 9 Axial 5.0 30 20 10 Axial 8.0 30 20 11 Axial 6.5 20 20 12 Axial 6.5 40 20 13 Axial 6.5 30 10 14 Axial 6.5 30 30 15 Central 6.5 30 20 16 Central 6.5 30 20 17 Central 6.5 30 20 The optimum values of the selected variables were obtained by solving the regression equation and also by analyzing the response surface plots using the Design Expert software (version 7.0.0; Stat-Ease, Inc., USA). The initial Cd (II) concentration was 40 mg/L, and each experiment was carried out in duplicate, with the results being reported as mean values±standard deviation. FTIR analysis {#sec2-7} ------------- The FTIR study was intended to provide a deeper insight into the interaction between the surface functional groups of the biosorbent and the cadmium ions. The biomass was first dried prior and after adsorption using a lyophilizator (Denmark) and was grounded into fine particles using mortar and pestle. Each sample was then mixed with potassium bromide (1 mg in 100 mg of KBr), compressed into a 0.25-mm thickness disk and stabilized under controlled relative humidity before acquiring the spectrum. The FTIR spectrophotometer (Bruker-Vector22) used to record spectra was a Shimadzu IRPrestige in the wave number range of 400 to 4,000 cm^−1^. Desorption {#sec2-8} ---------- Metals desorption is an important process in regeneration of the used adsorbents for their repeated use in water purification systems. It is also necessary to recover the precious metals. In this study, the binding sites of the bioadsorbent were first loaded with metal ions and desorption was then carried out to recover the adsorbed Cd (II). The cells were collected after adsorption under optimum conditions and washed with 0.1 M Tris-HCl. They were then incubated on ice for 15 min with 5 mM EDTA in 0.1M Tris-HCl (pH 8.0) to remove the surface-bound metals. Alternatively, following adsorption and washing, the cells were incubated with 0.1 M CaCl~2~ in 0.1 M Tris-HCl (pH 8.0) at 40ºC for 1 h. A control sample was also incubated in Tris-HCl (pH 8.0) at 40ºC for 1 h. The supernatants resulting from these treatments were then subjected to atomic absorption analysis. RESULTS {#sec1-3} ======= Kinetic studies {#sec2-9} --------------- [Figure 1](#F1){ref-type="fig"} represents the effects of contact times (5 to 100 min) on the biosorption of Cd (II) by E. coli E. A fast rate of Cd (II) adsorption was observed in the first 5 min. The maximum removal of Cd (II) occurred after 40 min, when the uptake of Cd (II) was approximately 74.72%. ![Effect of contact time on the biosorption capacity of Cd (II) (q~e~) by E. coli E. \[Cd (II)\] concentration, 20 mg/L; biomass dosage, 1 mg/mL Tris-HCl, pH 6.5, 30ºC, 100 rpm).](IBJ-21-380-g006){#F1} Biosorption kinetics was studied to understand the adsorption dynamics of Cd (II) onto the E. coli E surface. The models obtained in this study allowed to estimate the amount of Cd (II) adsorbed at the time of processing. Accordingly, two types of kinetic models were applied, the pseudo-first-order and pseudo-second-order. The pseudo-first-order kinetic model is the plot of Log (qe--q~t~) versus time. [Table 2](#T2){ref-type="table"} shows the values of the biosorption rate constant, K~1~, calculated q~e~, and experimental q~e~. The coefficient of determination (R^2^=0.75) showed that linear regression did not fit the experimental data ([Table 2](#T2){ref-type="table"}) and could not predict q~e~ accurately. Therefore, the pseudo-second-order kinetic model was used to analyze the biosorption kinetics of Cd (II). The pseudo-second-order kinetic model is the plot of t/q~t~ versus time. As shown in [Table 2](#T2){ref-type="table"}, according to the R^2^ of 0.99, the pseudo-second-order model could satisfactorily fit the experimental data, where the calculated q~e~ was in acceptable agreement with the experimental data. ###### The biosorption rate constants and the q~e~ values from the pseudo-first-order and pseudo-second-order kinetics for the biosorption of Cd (II) by E. coli E Metal ion Expt. Q Pseudo-first-order kinetics Pseudo-second-order kinetics ----------- ------------ ----------------------------- ------------------------------ ------- ------ -------- ------- Cd (II) 133 ± 3.53 0.75 16.34 0.035 0.99 135.13 0.005 Expt., experimental data of adsorbed metal; Cal., data of adsorbed metal calculated from the model Isotherms {#sec2-10} --------- The biosorption capacity of E. coli E was increased with the initial concentration of Cd (II) ions in solution ([Fig. 2](#F2){ref-type="fig"}), demonstrating the potential of E. coli E, as a biosorbent, to treat wastewater containing high concentrations of metal ions. ![Biosorption of Cd (II) at various initial concentrations (pH 6.5, 1 mg/mL biomass, 30ºC).](IBJ-21-380-g007){#F2} The biosorption process will stop only when equilibrium is reached between the amount of metal ions adsorbed onto the biosorbent and that of solution. The isotherm models were thus investigated at the time of equilibrium. The parameters of equilibrium sorption were used to investigate the surface properties and affinity of the sorbent\[[@ref24]\]. The Langmuir and Freundlich isotherm models were employed for this purpose. The Langmuir adsorption model was applied for the experimental data obtained. A graph of C~e~/q~e~ versus C~e~ is a linear plot of the Langmuir model. The value of the biosorption capacity, q~max~, and the Langmuir constant, K~L~, were obtained from linear regression. The value of R^2^ for Cd (II) was found to be 0.97. Although the Langmuir model fits the experimental data for Cd (II), it was preferred to study the model fitting for the Freundlich model according to the R^2^ value of 0.97. The linear Freundlich plot can be obtained by the Log qe (nmol/mg) versus Log C~e~ (nmol/mL). According to R^2^ value of 0.98, the Freundlich isotherm was in good agreement with the experimental data in this study. The value of the 1/n was obtained as 0.32, indicating there is a chemical interaction between Cd (II) and E. coli E. The adsorption constants estimated from the Freundlich and Langmuir isotherms are summarized in [Table 3](#T3){ref-type="table"}. ###### Langmuir and Freundlich constants for Cd (II) biosorption by E. coli E Langmuir Freundlich ---------- ------------ ------ -------- ------ ------ 0.01 277.77 0.97 30.535 0.32 0.98 Optimization of the biosorption process {#sec2-11} --------------------------------------- The range of variables used in the optimization process was determined according to screening of variables based on cell viability. The t-test indicated a significant difference between pH values 4 and 7, as well as 9 and 7; however, no significant differences were found between pH values 5 and 7, 6 and 7, and 8 and 7 (P\>0.05). Also, it demonstrated a significant difference between the temperatures 50, 60, 70, 80, and 30°C, but no significant difference was detected between temperatures 20, 30, and 40°C (P\>0.05). Consequently, the range of variables chosen to analyze the optimum conditions provided maximum biosorption efficiency using the face-centered CCD, as mentioned in [Table 1](#T1){ref-type="table"}. The Cd (II) uptake rate (q~e~) was measured as the response, and then the results were compared with the predicted values ([Table 4](#T4){ref-type="table"}). Analysis of variance (ANOVA) showed that there is a statistical significance in the quadratic model. The F and P values of the regression model were 91.42 and \<0.0001, respectively, implying the significance of the model. R^2^ was 0.98, which indicates that only 0.28% of the total variable could not be explained by the model. The lack of fit value of 0.05 implied that the lack of fit is not significant relative to the pure error. Both parameters showed that the model could well fit the experimental data. The value of the adjusted R^2^ of 0.96 was in reasonable agreement with the predicted R^2^ of 0.93. In addition, a relatively low value of the coefficient of variation (CV=6.46) indicates the repeatability of the experiments. ###### Observed and predicted values Trial Actual response q~e~ (nmol/ mg) Predicted response ------- --------------------------------- -------------------- 1 195.73±3.34 190.71 2 249.11±2.12 254.52 3 213.52±3.54 208.12 4 284.70±3.54 271.92 5 94.84±2.83 89.00 6 96.86±3.54 101.91 7 97.86±1.41 106.41 8 118.59±2.83 119.32 9 115.66±3.54 126.98 10 160.14±2.40 165.34 11 151.25±3.53 137.46 12 160.14±1.43 154.86 13 213.52±3.53 231.32 14 112.63±3.48 104.16 15 146.80 ±2.83 146.16 16 142.35±2.12 146.16 17 146.80 ±2.83 146.16 Eq indicates the final mathematical model corresponding to the coded factors after eliminating the insignificant terms (P\>0.05), as determined by the Design-Expert software. (6): ![](IBJ-21-380-g008.jpg) where Y shows q (metal uptake, nmol/mg); A, pH; B, temperature (ºC); C, biomass dosage (mg); AC, the interaction of pH and biomass dosage; C^2^, the second order effect of biomass dosage. Interactive effect of biomass dosage and pH of solution on biosorption {#sec2-12} ---------------------------------------------------------------------- [Figure 3A](#F3){ref-type="fig"} shows the effect of the biomass dosage and pH of the solution on Cd (II) uptake. Our results indicated that metal uptake was decreased when the amount of biomass was increased from 1 to 3 mg/mL, and the pH was decreased from 8 to 5. Metal uptake, however, was at a maximum when the biomass concentration and pH value reached 1 mg/mL and of 8, respectively. Therefore, increased biomass dosages and decreased pH values simultaneously led to a reduction in the Cd (II) ion uptake. ![Response surface plot of different factors interactions. (A) The interactive effect of the biomass dosage and pH at constant temperature 30ºC on metal uptake (q); (B) the effect of temperature and pH at constant biomass dosage of 2 mg/mL on the metal uptake (q); (C) the effect of temperature and biomass dosage at constant pH of 6.5 on metal uptake (q).](IBJ-21-380-g009){#F3} Interactive effect of temperature and pH on biosorption {#sec2-13} ------------------------------------------------------- [Figure 3B](#F3){ref-type="fig"} represents the effect of temperature and pH in the solution on Cd (II) uptake. Our results demonstrated the decreased metal uptake after the temperature and pH were lowered from 40 to 20ºC and 8 to 5, respectively. The lowest uptake was observed at a temperature of 20ºC and pH value of 5. Interactive effect of biomass dosage and temperature on biosorption {#sec2-14} ------------------------------------------------------------------- The effect of biomass dosage and temperature on Cd (II) uptake is shown in [Figure 3C](#F3){ref-type="fig"}. It was observed that the metal uptake was decreased when the amount of biomass was increased from 1 to 3 mg/mL, and the temperature was lowered from 40 to 20ºC, demonstrating the highest reduction in metal uptake at the biomass value of 1 mg/mL and a temperature of 40ºC. Results from ANOVA showed that temperature has a significant effect on adsorption (P=0.0001). However, the F value of this factor (7.20) implied that this variable is not highly significant when compared with the biomass dosage (F=384.36). Hence, unlike the biomass dosage, a color range (blue to red) associated with temperature was not present in the response surface plot ([Fig. 3C](#F3){ref-type="fig"}). A study, carried out by Hassan et al.\[[@ref25]\] in 2009 showed that temperatures in a range from 25 to 55ºC have no remarkable effects on the copper biosorption by the brown seaweed Sargassum sp. Effect of the optimization of different variables {#sec2-15} ------------------------------------------------- According to the results shown in [Table 4](#T4){ref-type="table"}, optimization of different variables led to increased cadmium adsorption from 94.84±2.83 to 284.70±3.54 nmol/ mg. This result was compared with the other biosorbents retrieved from selected literatures ([Table 5](#T5){ref-type="table"}). ###### Comparison between cadmium adsorption of selected literatures and this work Biosorbent Operating condition Amount adsorbed (nmol/mg) Reference ------------------------ --------------------- --------------------------- --------------- -------- --------- -------------- Bacillus circulan 7 20 0.5 2 235.764 \[[@ref37]\] Pseudomunas putida 6 30 Not available 1 71 \[[@ref38]\] Pseudomunas stutzeri 5 30 1 30 min 387 \[[@ref25]\] KCCM 34719 5 28+2 1 \>5 284.252 \[[@ref27]\] Bacillus cereus RC-1 Not 37 3.33 2 510 Engineered E.coli available Engineered E.coli 8 40 0.01 1 284.697 this study FTIR analysis results {#sec2-16} --------------------- FTIR analysis allows the identification of the functional groups involved in cadmium bioadsorption by shifting the changes in signal intensity compared to the control. FTIR analysis was carried out on our developed bioadsorbent (E. coli strain E) in the absence and presence of cadmium to determine the differences that are due to interaction of the metal ions with the surface functional groups ([Fig. 4](#F4){ref-type="fig"}). Our results showed that cadmium ions bind to the functional groups, including carboxyl, amino, phosphoryl, thiol, and hydroxyl groups, which results in the shift and changes in the spectra of the bioadsorbent before and after cadmium adsorption. ![The FTIR spectra of our developed bioadsorbent in the absence and presence of cadmium.](IBJ-21-380-g010){#F4} The results attributable to the presence of the specific bands is summerized in [Table 6](#T6){ref-type="table"}. As shown in [Figure 4](#F4){ref-type="fig"}, a weak --SH stretching was observed in the range of 2500-2600 cm^-1^. It has long been known that the SH stretch(s) mode, ν (SH), is generally found in the range of 2500-2600 cm^-1^, being weak in the IR\[[@ref26]\]. The peaks at 2937.2, 1433, 1373 and 922 cm^−1^ might be due to CH~2~ stretching and bending vibrations, while the peaks at 922 and 913 cm^−1^ might be because of P-OR stretching. These groups could be attributed to polysaccharides, phospholipids, or ribose phosphate chain pyrophosphate, which are the main components of the bacterial cell wall\[[@ref27]\]. The band observed at 1543.5 cm^−1^ was assigned to amide the bands of proteins and peptides. ###### The FTIR Spectral Characteristics of Biosorbent before and after Biosorption of Cd Wavelength range (cm^-1^) Biosorbent after biosorption Assignment Functional class --------------------------- ------------------------------ ------------------------- -------------------------------------------------------------------------------- 3550-3600 3200-3550 3562.4 O--H Oxime (=NOH) 3200-3550 3338.1 O--H or N--H Alcohols, phenols or secondary amines 3389.5 OO O--H or N--H Alcohols, phenols or secondary amines 2500-3300 2937.2 C--H or O--H Alkanes(methylene) or carboxylic acids 2600-2550 Very weak S--H Thiols 1630-1680 1655.3 C=C or C=N Alkenes or oxime (=NOH) 1500-1560 1543.5 N=O or N--H Nitro compounds or amides 1350-1470 1433.0 C--H or O--H Alkanes and aldehyde or carboxylic acids 1330-1430 1373.1 C--H or O--H Alkanes or alcohols, phenols 1325±25 1346.7 S=O Sulfonic acid or sulfone 1210-1320 1279.1 C--O or N--O Carboxylic acids or aromatic amine oxide 1210-1320 1238.4 C --O or P=O Carboxylic acids or phosphonate and phosphoramide 1100-1200 1125.2 P=O Phosphine oxide and phosphate 1050-1200 1068.0 C=S Thiocarbonyl 880-1050 992.6 P--OR or =C--H & =CH~2~ Esters or alkenes 900-1050 913.2 P--OR (P--O--C) (Phosphite) esters 600-900 864.2 C--H Alkynes or arenes 600-900 683.6 C--H or O-H cis-RCH=CHR alkynes and arenes or or NH~2~ and N-H alcohols, phenols or amines 500-700 583.4 C--Br and C--Cl Bromoalkanes and chloroalkanes 500-600 551.5 C--Br and C--Cl Bromoalkanes and chloroalkanes Desorption of Cd (II) {#sec2-17} --------------------- Because of the sensitivity of E. coli E to such conditions, no alkaline and acidic elutants were used for Cd (II) desorption. The recovery experiments showed that approximately 68.58% and 56.54% of the Cd (II) ions were recovered when 5 mM EDTA and 100 mM CaCl~2~ were used, respectively. DISCUSSION {#sec1-4} ========== In this study, the bioadsorbent developed in our previous study was evaluated by kinetic and isotherm studies, followed by optimization of adsorption conditions. Study on contact times showed that cadmium adsorption was initially rapid, presumably due to more active-binding sites present on the adsorbent in the beginning. However, the rate of cadmium adsorption became slower after some time, probably because of a lesser number of active-binding sites available on the adsorbent\[[@ref28]\]. The first-order kinetic process was used for reversible reactions, in which equilibrium was established between liquid and solid phases. However, the pseudo-second order kinetic model assumes the chemical adsorption, as a probable rate-limiting step\[[@ref29]\]. Bacterial cell walls contain several functional groups, including carboxyl, phosphonate, amine, and hydroxyl groups\[[@ref30]\]. It is known that the functional groups present on the negatively-charged bacterial cell wall participate in the binding of metal cations, such as Cd (II)\[[@ref30]\]. In addition, the cyanobacterial MT, SmtA, used in this adsorbent as a metal-binding component of this bioadsorbent contains a Zn~4~Cys~9~His~2~ cluster structure that can replace Cd (II) ions in the coordination bonds\[[@ref31],[@ref32]\]. This result might be the main reason why the Cd (II) biosorption reaction by E. coli E has been well-modeled by the pseudo-second order kinetics. A number of studies have shown that the pseudo-second order mechanism is a better model for explaining the kinetics of divalent metal sorption onto heterogeneous sorbents\[[@ref28],[@ref33]\]. According to Giles et al.\[[@ref34]\], the plateau or the beginning of the linear portion above the "knee" must show "first-degree saturation" of the surface ([Fig. 2](#F2){ref-type="fig"}); this observation conforms the conditions that all possible sites in the original surface are filled, and further adsorption would take place only on the new surfaces. For convenience, this degree of coverage may be called the formation of a complete "monolayer," but this does not necessarily mean that it is a close-packed layer of single molecules or ions, as in a compressed monolayer on water\[[@ref34]\]. Based on the R^2^ value, the experimental data was in good agreement with the Freundlich isotherm. It can be due to the presence of different functional groups on the E. coli E surface, especially, SmtA that produces a heterogeneous biosorbent. The Freundlich isotherm can fit various experimental adsorption data and has been found to be in line with data from the highly heterogeneous sorbent systems. For example, Gong et al.\[[@ref35]\] have reported that the process of lead biosorption by Spirulina maxima, a filamentous cyanobacterium that can be used as a food supplement, follows the Freundlich isotherm model. In optimization process, the interactive effect of biomass dosage and pH showed that increased biomass dosages and decreased pH values simultaneously lead to a reduction in Cd (II) ion uptake. This phenomenon can be attributed to cadmium precipitation at higher pH\[[@ref36]-[@ref38]\] or the reconstitution of SmtA molecules with cadmium ions, at higher pH values\[[@ref32]\]. Masoudzadeh et al.\[[@ref39]\] have stated that it might also be due to the fact that the available solute was insufficient to completely cover the available exchangeable sites on the cell surface, thereby resulting in low solute uptake at high biomass concentration. Similar observations have been made in study on Cd (II) biosorption using pretreated Saccharomyces cerevisiae biomass\[[@ref22]\]. Meanwhile, biomass aggregation could interfere with the surface metal-binding sites. At high biomass dosages, the available binding sites are insufficient as they could be masked due to the surface protein interactions and limited metal accessibility\[[@ref40]\]. It is likely that protons will then combine with the metal ions, thereby decreasing the interaction of the metal ions with the cell components\[[@ref41]\]. The study of interaction between temperature and pH ranges on biosorption showed Cd (II) adsorption at a higher pH value. This finding can be related to the reaction between the binding sites of the sorbent and the metal, leading to the breaking of hydrogen bonds and the release of hydrogen ions, which are ultimately substituted by the metal ions\[[@ref42],[@ref43]\]. In the process of physicosorption, weak adsorption interactions between surface and the metal ions was decreased with increased temperature\[[@ref44]\]. However, based on the pseudo-second-order kinetic model, the interaction between E. coli E and Cd (II) ions can be ascribed to chemosorption. Therefore, temperature has a positive effect on the metal uptake capacity of E. coli E over the tested temperature ranges of 20-40ºC. Indeed, the results from this study, regarding the effect of temperature on Cd^2+^ adsroption, are consistent with those observed by Dang et al.\[[@ref45]\]. On the other hand, the study of interaction between effects of biomass dosage and temperature on biosorption showed that although raising the biosorbent dosage caused an increase in the biosorbent surface area and the availability of more adsorption sites, the uptake capacity (q~e~) was decreased under such a condition. This result can be attributed to the unsaturated sites during the adsorption reaction, whereas the number of the sites available for the adsorption increases by raising the adsorbent dosage. FTIR analysis was carried out to determine the functional groups involved in the cadmium bioadsorption. The significant changes in the appearance of new peak positions at specific wavelengths suggested that different compounds such as polysaccharides, phospholipids, proteins, and peptids were involved in the cadmium adsorption on the cell surface. A weak --SH stretching on FTIR spectra could be assigned to the molecules including thiol groups present in the cystein residues in MT on the surface of our developed bioadsorbent. According to the Hard-Soft Acid-Base theory, sulfur is a relatively soft (polarizable) atom. This also explains the tendency of thiol groups to bind to the soft elements/ions such as mercury, lead, or cadmium. Our experiment regarding cadmium desorption using EDTA and CaCl~2~ showed that this kind of adsorption could be due to electrostatic and coordinance interactions occuring between Cd (II) ions and the E. coli E cell surface. The presence of competitor ions, such as calcium, at high concentrations, can release the Cd (II) ions involved in the electrostatic interactions. In addition, because of the affinity of heavy metals for EDTA, as a chelating agent, it could release the Cd (II) ions more efficiently than CaCl~2.~ This result is consistent with that of Hua et al.\[[@ref46]\] who indicated that the presence of EDTA generally decreased the adsorption of Cd to biofilm in natural waters. Our results demonstrated that the presence of the different functional groups and MT present on the surface of the E. coli E caused an increased chemical adsorption on heterogenus surface. Furthermore, the surface adsorption was remarkabaly increased while the critical adsoprtion parameters were optimized. It can be concluded that surface engineering of the surface proteins involving metal adsorption as well as optimization of operational parameters have a great impact on the efficiency of biosorption process. The financial support of this research project by the National Institute of Genetic Engineering and Biotechnology (NIGEB) of Iran is acknowledged. CONFLICT OF INTEREST. None declared.	{ "pile_set_name": "PubMed Central" }
Introduction {#Sec1} ============ Porcine reproductive and respiratory syndrome virus (PRRSV) infection is common worldwide and has become one of the most economically important diseases in the swine industry \[[@CR1]\]. PRRSV can cause reproductive failure in sows and is associated with a respiratory disease complex that can affect pigs of all ages \[[@CR2]--[@CR4]\]. PRRSV was discovered almost simultaneously in the late 1980s on the European and North American continents and was divided into type 1 PRRSV (European genotype) and type 2 PRRSV (North American genotype) \[[@CR5], [@CR6]\]. PRRSV is an enveloped, single-stranded positive-sense RNA virus. Its genome is approximately 15.4 kb in size and has at least 10 open reading frames (ORFs) \[[@CR7]\]. ORF1a and ORF1b encode non-structural proteins, and the other genes encode structural proteins, including GP2a, GP2b, GP3, GP4, GP5, GP5a, matrix protein M and the nucleocapsid protein N \[[@CR8]--[@CR13]\]. Of these, four are glycoproteins: the minor proteins GP2a, GP3 and GP4 and the major protein GP5, which is thought to be important for virus assembly and entry into permissive cells. Researchers have made great efforts towards preventing and controlling PRRSV infection. Currently, vaccination is the main method to control PRRSV infection. Although many commercially available modified live and inactivated PRRSV vaccines have been developed, they fail to provide complete protection against the devastating disease caused by this virus \[[@CR6], [@CR14]--[@CR16]\]. Furthermore, a large amount of literature has reported natural compounds that exhibit activity against PRRSV infection, including flavaspidic acid AB \[[@CR17]\], matrine \[[@CR18]\], 12-deoxyphorbol 13-phenylacetate 20-acetate (dPPA), ouabain, valinomycin and bufalin \[[@CR19]\], and glycyrrhizin \[[@CR20]\]. Additionally, other therapies designed to control PRRSV infection have been reported, such as LiCl \[[@CR21]\], polyinosinic-polycytidylic acid \[[@CR22]\], type I and type III interferons \[[@CR23]\], double-stranded RNA (dsRNA) activated caspase oligomerizer (DRACO) \[[@CR24]\], and microRNAs \[[@CR25]\]. However, none of these aforementioned therapies can effectively treat or prevent PRRSV infection, so there is an urgent need to develop new antiviral strategies. Surfactant protein A (SP-A) belongs to a family of mammalian C-type lectins, and it is an important innate immune molecule that is involved in a range of immune responses. The primary structure of SP-A is composed of four regions, including an N-terminal region, a collagen-like region, a coiled-coil neck peptide, and a carbohydrate-recognition domain (CRD) \[[@CR26]\]. The CRD can interact with sugar moieties that are present on the surface of many pathogens, including viruses, fungi and bacteria, resulting in inhibition of infectivity in a calcium-dependent manner. Previous studies on the antiviral activity of SP-A have been focused on human medicine. In the late 1990s, human SP-A was shown to inhibit respiratory syncytial virus by blocking viral entry and subsequent syncytium formation \[[@CR27]\]. Research on porcine SP-A activity against PRRSV was limited to one report indicating that recombinant porcine SP-A could exert anti-PRRSV activity \[[@CR28]\]. Unfortunately, no additional studies on the anti-PRRSV potential and mechanism of SP-A were performed. However, studies on the antiviral activity of SP-D, which has a primary structure that resembles that of SP-A, have been reported more frequently. SP-D had an inhibitory effect on HIV infection through inhibition of viral entry by affecting gp120--CD4 interactions \[[@CR29]\]. Marine and colleagues demonstrated that recombinant porcine SP-D prevented the attachment of human seasonal H~1~N~1~ and H~3~N~2~ virus to receptors on epithelial cells of the upper respiratory tract \[? [36]{.ul} ?\]. Notably, porcine ficolin is also a collagenous lectin that was established to reduce PRRSV infection in Marc-145 cells in neutralization assays and to inhibit the replication of infectious viral particles in a GlcNAc-dependent manner \[[@CR30]\]. In the present study, we report the antiviral properties and mechanism of action of recombinant porcine SP-A in Marc-145 cells. Our results indicate that RpSP-A affects virus attachment and blocks the cell-to-cell transmission pathway to suppress PRRSV infection in Marc-145 cells, which indicates that RpSP-A may be a useful antiviral agent for the treatment of PRRSV infection. Materials and methods {#Sec2} ===================== Cells and virus {#Sec3} --------------- Marc 145 cells were grown in Dulbecco's modified Eagle's medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10 % newborn bovine serum (NBS) at 37 °C and 5 % CO~2~. Spodoptera frugiperda cells (Sf9 cells), purchased from Invitrogen, were grown in Grace's insect medium (1×; Gibco, Grand Island, NY, USA) supplemented with 10 % heat-inactivated fetal bovine serum (FBS; ScienCell, CA, USA) for recombinant baculovirus production or were grown in Sf 900 II SFM (Gibco, Grand Island, NY, USA) for protein expression at 27 °C. HP PRRSV strain NJ-a (type 2 PRRSV) was isolated from a pig on a farm with an atypical PRRS outbreak in Nanjing city, Jiangsu province, China, in 2008. It was identified and stored by the National Research Center of Engineering and Technology for Veterinary Biologicals. The NJ-a strain was propagated and titrated by the plaque method in Marc145 cells to prepare the viral stock used in this study. Preparation of recombinant porcine SP-A (RpSP-A) {#Sec4} ------------------------------------------------ RpSP-A was obtained using a Bac-to-Bac baculovirus expression system (Invitrogen, CA, USA). An N-terminal in-frame fusion of the pSP-A (GenBank accession no. NM_214265) gene with an 8-histidine tag was codon-optimized for baculovirus expression, synthesized, and cloned in the pUC57 vector by GenScript (Piscataway, NJ, USA). The recombinant pSP-A gene was subcloned into a pFastBac1 vector under the control of the PH promoter. Once generated in a pFastBac^™^ construct, the plasmid DNA was used to transform MAX Efficiency^®^ DH10Bac^™^ competent E. coli for transposition into a bacmid. Blue/white selection was used to identify colonies that contained a recombinant bacmid. High-molecular-weight RpSP-A bacmid DNA from DH10Bac^™^ E. coli was isolated using a BAC/PAC DNA Isolation Kit (OMEGA, USA). The concentration and purity of the RpSP-A bacmid were estimated using a Thermo Scientific NanoDrop^TM^ 2000 spectrophotometer, which was then used to transfect Sf9 insect cells using Cellfectin II reagent. The recombinant baculovirus was harvested 7 days post-transfection and was amplified twice to obtain higher titer virus stocks. Recombinant P3 viral stocks were used to infect Sf9 cells at an MOI of 1.5 PFU/cell. At 5 days postinfection, cells were harvested and washed twice in phosphate-buffered saline (PBS). Cell pellets were immediately stored at -80 °C. The protein RpSP-A was extracted and purified using a 1-mL-size Ni^2+^-Sepharose HisTrap HP^TM^ affinity column (GE Healthcare, Sweden) according to the instructions provided by the manufacturer. Imidazole in the eluate, which contained RpSP-A, was removed on a HiTrap^™^ desalting column (5 mL) and then exchanged with storage buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 0.1 mM PMSF). Then, the eluate was filter-sterilized through a 0.22-µm membrane filter (Millipore) and analyzed by Western blotting using an anti-His antibody (SunShineBio, China). Protein concentrations were determined using a BCA Protein Assay Reagent Kit (Pierce, Rockford, IL, USA). Cell viability assay {#Sec5} -------------------- The cytotoxic effect of RpSP-A was examined using an MTT assay. Briefly, Marc 145 cells in 96-well plate were exposed to various concentrations of RpSP-A at 37 °C in a 5 % CO2 atmosphere for 48 h. After incubation, the culture medium was replaced with fresh medium containing 20 µL of MTT at a concentration of 5 mg/ml, and the cells were incubated for an additional 4 h. After removal of the supernatant, 150 µL of DMSO was added to each well to dissolve the crystals for 10 min. Absorbance at 490 nm was recorded using an ELISA microplate reader. Plaque assay {#Sec6} ------------ Antiviral activity was detected by plaque assay using Marc 145 cells in 6-well plates. A preliminary study was performed to determine the concentration of infectious virus particles using a 2 × 10^−4^ dilution of virus stock (1.4 × 10^6^ PFU mL^−1^). Subconfluent monolayers of Marc145 cells in 6-well plates were treated with RpSP-A at the indicated concentrations and were simultaneously infected with PRRSV. The cells continued to be cultured at 37 °C for 1 h. Then, the cells were washed five times with PBS and covered with 2 mL of medium containing a 1:1 mixture of 2× DMEM and 2 % low-gelling-temperature agarose (Sigma, USA). After 72 h, the agarose was removed, and cells were stained with 1 % crystal violet dissolved in 10 % formaldehyde at 37 °C for 4 h. After washing gently with water, plaques were counted. Antiviral activity was calculated using the following formula: inhibition rate (%) = $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ 1-\frac{\text{number of plaques}_\text{test}}{\text{number of plaques}_\text{control}} \times 100\; \% $$\end{document}$ Real-time quantitative PCR (real-time qPCR) {#Sec7} ------------------------------------------- Total RNA was extracted from samples using a TaKaRa MiniBEST Universal RNA Extraction Kit (TaKaRa, Osaka, Japan). RNA was reverse transcribed into first-strand cDNA using PrimeScript^TM^ Reverse Transcriptase with random primers in a 20-µL final reaction volume according to the manufacturer's instructions. First, 5 µL of total RNA was added to mixture I containing 1 µL of random primer and 4 µL of dNTPs (2.5 mM), and then the mixture was heated at 65 °C for 5 min and then kept at 4 °C for 5 min. Second, a reaction mixture was prepared by adding the above mixture to mixture II, containing 4 µL of 5× PrimeScript Buffer, 1 µL of RNase inhibitor, 1 µL of PrimeScript Reverse Transcriptase and 4 µL of RNase-free H~2~O. The reaction was performed under the following conditions: 30 °C for 10 min, 42 °C for 30 min, 70 °C for 15 min and, finally, 4 °C for 10 min. Real-time qPCR was performed in a LightCycler480 II System (Roche, Basel, Switzerland) using an EvaGreen 2X qPCR MasterMix-No Dye kit (abm, Canada). Reaction mixtures were prepared according to the manufacturer's instructions and incubated for 10 min at 95 °C, followed by 40 cycles of 15 s at 95 °C and 1 min at 60 °C, followed by 15 s at 95 °C, 1 min at 60 °C, and 15 s at 95 °C. A portion of the ORF5 gene was detected by real-time qPCR, and GAPDH was used as a control. The data were analyzed using the 2^-ΔΔCt^ method. Customized primers are listed in Table [1](#Tab1){ref-type="table"}.Table 1Primers used for real-time PCRPrimerSequence (5'-3')ORF5-FTCTGGACACTAAGGGCAGACTCORF5-RGGAACCATCAAGCACAACTCTCGAPDH-FCCTTCCGTGTCCCTACTGCCAACGAPDH-RGACGCCTGCTTCACCACCTTCT N protein detection by western blotting {#Sec8} --------------------------------------- Total protein was extracted using cell lysis buffer (20 mM Tris, pH 7.5, 150 mM NaCl, 1 % Triton X-100, 0.1 mM PMSF) and quantified using a BCA Protein Assay Reagent Kit (Pierce, Rockford, IL, USA). Then, 20 µg total protein was subjected to 12 % sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinyl difluoride (PVDF) membrane, which was and blocked with 5 % skim milk in TBST buffer at 37 °C for 1 h. Membranes were incubated overnight with gentle shaking at 4 °C with rabbit anti-PRRSV N protein antibody (diluted at 1:2500, donated by Dr. Yuanpeng Zhang) and mouse anti-β-actin antibody (1:200, Boster, China). After washing, membranes were incubated with an appropriate secondary antibody (1:10,000, SunShineBio, China) for 1 h at 37 °C. Antibodies were visualized using enhanced chemiluminescence (ECL) reagent (Boster, China) and images were recorded using an ImageQuant LAS 4000 (GE Healthcare, Sweden). PRRSV attachment assay {#Sec9} ---------------------- The attachment assay was carried out as described previously \[[@CR31]\] with some modifications. Marc 145 cells were seeded into 6-well plates. Cells were washed with cold PBS followed by DMED and then were pre-chilled at 4 °C for 30 min. Cells were incubated with PRRSV at an MOI of 0.01 and purified RpSP-A (30 µg mL^−1^) at 4 °C for 2 h. Cells were washed with cold PBS to remove unbound virus and RpSP-A, and then, fresh medium containing 2 % FBS was added. The cells were then cultured at 37 °C for 24 h, after which the virus titer was measured. PRRSV penetration assay {#Sec10} ----------------------- The penetration assay was performed as described previously with some modifications \[[@CR17], [@CR31]\]. PRRSV could be internalized from the surface of Marc 145 cells within 3 h to 6 h \[[@CR17]\]. Thus, we examined the kinetics of activity against PRRSV penetration using time-of-addition assays. Marc 145 cells were seeded into 6-well plates. Cells were washed with cold PBS, followed by DMED, and were then pre-chilled at 4 °C for 30 min. Cells were incubated with PRRSV at 4 °C for 2 h. Cells were washed with cold PBS. RpSP-A was added at 0 h, 1 h, and 2 h and removed at 3 h following a temperature shift. The time point when the cell culture was shifted to 37 °C was designated as 0 h. Cells were washed with cold PBS to remove unbound virus and RpSP-A, and fresh medium containing 2 % FBS was added. Cells were incubated for 24 h at 37 °C, and the virus titer was measured. Data analysis {#Sec11} ------------- The data are presented as the mean ± standard deviation from three independent experiments. The data were analyzed using Student's t-test with GraphPad Prism 6 software. The differences between the mean values were considered statistically significant when the P-value was less than 0.05. Results {#Sec12} ======= Measurement of anti-PRRSV activity of RpSP-A by plaque assay {#Sec13} ------------------------------------------------------------ Plaque assay results revealed that RpSP-A had potency against PRRSV. As shown in Fig. [1](#Fig1){ref-type="fig"}A, the number of visible plaques was significantly reduced in the presence of 30 µg of RpSP-A per mL. Furthermore, as shown in Fig. [1](#Fig1){ref-type="fig"}B and C, treatment with 30 µg of RpSP-A per mL reduced plaques by 77 % from 1.46 × 10^6^ PFU mL^−1^ to 0.33 × 10^6^ PFU mL^−1^ (P \< 0.0001), whereas treatment with 15 µg of RpSP-A per mL reduced plaques by 43 % from 1.46 × 10^6^ PFU mL^−1^ to 0.83 × 10^6^ PFU mL^−1^ (P \< 0.05). Treatment with 30 µg of RpSP-A per mL in the absence of Ca^2+^ was used as a negative control, and no inhibition was observed.Fig. 1Anti-PRRSV activity of RpSP-A, examined using a plaque assay. A. Visible plaques on Marc 145 cells infected with PRRSV in 6-well culture plates. B. The corresponding virus titer after treatment with RpSP-A. C. Inhibition of infection by RpSP-A. Each value represents the mean of three independent experiments with standard derivation. \, P* \< 0.05; \\, P \< 0.01; \\\, P* \< 0.001 compared to the 0/Ca+ group To rule out the possibility that nonspecific toxicity caused by RpSP-A was affecting PRRSV replication, we investigated its cytotoxicity in Marc-145 cells at concentrations of 0, 10, 20, 30, 40, 50, 60 µg/mL. As shown in Fig. [2](#Fig2){ref-type="fig"}, 48 hours after treatment, the cells cultured in medium containing RpSP-A at concentrations below 60 µg mL^−1^ retained approximately 100 % viability relative to the control.Fig. 2Cytotoxicity of RpSP-A in Marc 145 cells, measured using an MTT assay. Cells were incubated with different concentrations of RpSP-A for 48 h, and a cell viability assay was performed Effects of RpSP-A on virus attachment and penetration {#Sec14} ----------------------------------------------------- To elucidate the mechanism of the antiviral effects of RpSP-A, a virus entry assay was utilized. The data showed that RpSP-A could largely inhibit virus attachment; however, it was relatively inactive against virus penetration. In the virus attachment assay, Marc 145 cells were incubated with PRRSV at an MOI of 0.01 in the presence of RpSP-A at 4 °C, a condition in which viruses attach to cells but do not penetrate. As shown in Fig. [3](#Fig3){ref-type="fig"}, treatment with 30 µg of RpSP-A per mL largely inhibited Marc145 cell attachment. The mean viral titer was 10^5.64^ PFU mL^−1^ in the absence of RpSP-A, but it was reduced to 10^4.03^ PFU mL^−1^ in the presence of 30 µg of RpSP-A per mL with the addition of Ca^2+^ (P \< 0.0001). No significant difference was observed between 0 and 30 µg mL^−1^ of RpSP-A in the absence of Ca^2+^ (P \> 0.05).Fig. 3Effect of RpSP-A on attachment of PRRSV to Marc 145 cells. Cells were incubated for 24 h after treatment with PRRSV alone or together with RpSP-A in the presence or absence of calcium ions, and then virus titer was measured. Data are presented as mean ± S.D. of three independent experiments. \, P* \< 0.05; \\, P \< 0.01; \\\, P* \< 0.001 compared to the control In the penetration assay, we studied the inhibition kinetics of RpSP-A against PRRSV. The mean viral titer in the presence of 30 µg of RpSP-A per mL with the addition of Ca^2+^ was 10^4.56^ PFU mL^−1^ at 0 h (P \< 0.0001), 10^5.3^ PFU mL^−1^ at 1 h (P \< 0.05), and 10^5.46^ PFU mL^−1^ at 2 h (P \> 0.05) compared to 10^5.6^ PFU mL^−1^ in the absence of RpSP-A (Fig. [4](#Fig4){ref-type="fig"}), suggesting that the inhibitory effect of RpSP-A on virus penetration decreased over time, although there was no significant difference at 2 h. Additionally, 30 µg of RpSP-A per mL in the absence of Ca^2+^ did not affect virus penetration.Fig. 4Effect of RpSP-A on Marc 145 cell penetration. Cells were incubated for 24 h after treatment with PRRSV alone or after addition of RpSP-A at 1-h intervals in the presence or absence of calcium ions, and then virus titer was measured. Data are presented as mean ± S.D. of three independent experiments. \, P* \< 0.05; \\, P \< 0.01; \\\, P* \< 0.001 compared to the control Results of RT-PCR and western blotting {#Sec15} -------------------------------------- We next attempted to investigate the kinetics of the RpSP-A-mediated effect on the intracellular viral load. First, Marc 145 cells were incubated with PRRSV at an MOI of 0.01 for 24 h, and the medium was replaced with fresh medium containing 30 µg of RpSP-A per mL. Then, the total RNA in Marc 145 cells was extracted for PRRSV RNA genome detection by real-time RT-PCR after culture for 6 h, 12 h, 24 h or 36 h. As shown in Fig. [5](#Fig5){ref-type="fig"}A, the addition of RpSP-A resulted in a significant reduction in the total RNA level at 12 h, 24 h and 36 h compared with the non-drug-treated cells (P \< 0.0001), and this reduction reached a peak at 24 h after addition of RpSP-A. However, there was no significant difference at 6 h following RpSP-A addition compared with the non-drug-treated group (P \> 0.05). Moreover, we also detected N protein of PRRSV in Marc 145 cells by western blotting after treatment with RpSP-A for 36 h. As shown in Fig. [5](#Fig5){ref-type="fig"}B, western blotting analysis indicated that RpSP-A reduced the level of N protein in cells in a dose-dependent manner.Fig. 5Reduction of the total levels of PRRSV RNA and protein of in Marc 145 cells by treatment with RpSP-A. A. The relative mRNA level was determined by amplifying a portion of PRRSV ORF5 using real-time RT-PCR at different time points. Relative expression (fold change) in comparison to the control group in the absence of RpSP-A (set as 1) is illustrated. Data are presented as mean ± S.D. of three independent experiments. \, P* \< 0.05; \\, P \< 0.01; \\\, P* \< 0.001 compared to the control. B. The viral N protein and β-actin were detected by western blotting. β-actin was used as an internal control Discussion {#Sec16} ========== Surfactant protein A, an antimicrobial defense molecule of the innate immune system, has the potential to be exploited as an antiviral drug in the future. For this reason, we sought to develop it as a recombinant product in this study. Here, RpSP-A was obtained using a Bac-to-Bac baculovirus expression system. We found that RpSP-A expressed by Sf9 insect cells could block virus entry. Viral entry represents an attractive target for chemotherapeutic intervention \[[@CR32]\]. However, drugs that inhibit virus entry prevent infection of cells by cell-free virus particles and also prevent virus transmission between virus-infected and uninfected cells \[[@CR32], [@CR33]\]. The capacity to inhibit both modes of infection by a single drug is one of the reasons why SP-A may be superior to other microbicide drugs. Thus, SP-A might be a candidate agent for preventing PRRSV infection. So far, there have been limited data available regarding the anti-PRRSV activity of porcine SP-A. In China, highly pathogenic PRRSV appears to be widely prevalent, and therefore, the highly pathogenic PRRSV strain NJ-a, a type 2 PRRSV strain, was used in this study. To explore the anti-PRRSV activity of porcine SP-A, further studies need to be performed on type 1 PRRSV. Our data indicate that RpSP-A significantly suppresses PRRSV infection in a dose-dependent manner without causing cytotoxicity in Marc 145 cells. We also found that RpSP-A exerts potent antiviral activity in a Ca^2+^-dependent manner, indicating that RpSP-A may interact with PRRSV via the carbohydrate-recognition domain, which is consistent with previous studies \[[@CR27], [@CR28]\]. Researchers have confirmed that lectins inhibit virus entry \[[@CR27], [@CR29], [@CR34], [@CR35]\], and each step of the viral entry pathway is a potential target for antiviral agents. We conducted virus attachment and penetration assays to assess whether RpSP-A also interfered with PRRSV entry. The virus entry assay showed that RpSP-A strongly inhibits virus attachment, but its effect on virus penetration was, at most, modest, suggesting that RpSP-A most probably blocks virus attachment, like other lectins. It has been shown that hippeastrum hybrid agglutinin (HHA), a mannose-specific lectin, probably interferes with SARS coronavirus attachment. When attachment occurred at 4 °C, HHA was twice as active (EC~50~ = 2.5 µg mL^−1^) as it was in a situation where both attachment and penetration were allowed to occur at 37 °C (EC~50~ = 5.2 µg mL^−1^) \[[@CR34]\]. In another study, recombinant porcine SP-D was shown to inhibit IAV attachment using human trachea tissue, and it fully prevented binding of human seasonal H1N1 and H3N2 IA at higher dose \[[@CR35]\]. Our data indicate that the suppression of PRRSV penetration by RpSP-A gradually weakened and even disappeared as the time between infection and addition of RpSP-A increased, suggesting that RpSP-A might act on PRRSV entry into Marc 145 cells only at the cell surface and that it has no effect once the virus has entered the cell. A better understanding of the underlying mechanisms by which RpSP-A influences viral entry might lead to new antiviral interventions. Taking the previously reported data and our own findings together, we hypothesize that RpSP-A inhibits PRRSV entry by binding to glycoprotein GP2, GP3, GP4 or GP5 on the surface of PRRSV, thereby preventing it from interacting with its receptors on Marc 145 cells. The mechanism of the antiviral activity of lectins has been studied by several researchers. It has been shown that human SP-A can bind to the F (fusion) glycoprotein on the surface of respiratory syncytial virus and may neutralize RSV by directly inhibiting the fusion function of this protein, thereby inhibiting viral entry \[[@CR27]\]. In another study, Marine et al. reported that the interaction between RpSP-D and the viral HA prevented attachment of the virus to target cells and contributed to the neutralizing effects of RpSP-D \[[@CR35]\]. In addition, other lectins such as SP-D and plant lectins have been shown to bind to glycoprotein gp120 of HIV-1 and block access of gp120 to the receptor protein CD4 \[[@CR29], [@CR32], [@CR33], [@CR36]\], thereby blocking viral entry to reduce HIV-1 infection. The nature of the interactions among PRRSV, RpSP-A and cells is not yet known, so this model should be assessed in further studies. Interestingly, our data also indicated that RpSP-A had no effect on the total RNA level when the virus was allowed to undergo only a single cycle of replication but caused the total RNA level to decrease greatly at 12 h, 24 h and 36 h. A possible explanation for this is that virus progeny released from infected Marc145 cells are blocked by RpSP-A and prevented from infecting other cells. Since Marc-145 cells are not of porcine origin but of monkey origin, the mechanism of infection of porcine alveolar macrophages (PAMs) is not identical to that in Marc 145 cells \[[@CR37], [@CR38]\]; furthermore, a previous study showed that some lectins had a stronger inhibitory effect on virus infection when PAMs were treated with lectins than when the virus was treated with lectins prior to infection, in contrast to what was observed with Marc 145 cells, suggesting potentially complex interactions among PAMs, lectins and PRRSV \[[@CR39]\]. More studies will be needed to characterize the antiviral activity of RpSP-A in PAMs, which are the main target cells for PRRSV in vivo. In conclusion, we show that RpSP-A exerts potent activity against HP-PRRSV in Marc 145 cells. We clearly demonstrate that RpSP-A interacts at the level of virus attachment and blocks the cell-to-cell transmission pathway, which has never been observed before. RpSP-A most probably interacts with the glycoproteins of PRRSV, thereby preventing it from binding to host-cell receptors and subsequently interfering with transmission. A better understanding of the mechanisms by which RpSP-A influences viral infection in porcine alveolar macrophages might lead to new antiviral interventions. This study was supported by grants from the Special Fund for Agro-Scientific Research in the Public Interest (No. 201303046) and the Independent Innovation of Agricultural Sciences Program of Jiangsu Province (No. cx (14)2089). This study was supported by grants from the Special Fund for Agro-Scientific Research in the Public Interest (No. 201303046) and the Independent Innovation of Agricultural Sciences Program of Jiangsu Province (No. cx (14)2089). The authors declare that they have no conflicts of interest. This article does not contain any studies with human participants or animals performed by any of the authors.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION ============ Laparoscopic radical prostatectomy (LRP) and retropubic radical prostatectomy (RRP) are well-established procedures for the management of localized prostate cancer \[[@B1],[@B2]\]. Recently, cancer treatment strategies have focused more on patient quality of life after surgery, thus requiring minimally invasive techniques. In 2002, an initial experience with extraperitoneal laparoscopic radical prostatectomy (ELRP) showed similar oncological outcomes to intraperitoneal prostatectomy without intraperitoneal complications \[[@B3]\]. Due to the low perioperative and postoperative mortality rate, postradical prostatectomy incontinence, which interferes with the patient\'s quality of life, has become an issue. One-year continence rates are excellent after RRP and LRP in larger series \[[@B4]\]. However, the early return of continence remains a challenge and has a major impact on the postoperative patient\'s health-related quality of life. Further understanding of the anatomy of adjacent structures of the prostate, such as the bladder neck, urethra, fascia, ligaments, and neurovascular bundle (NBV), led to the development of nerve-sparing ELRP (nsELRP) and the most recent intrafascial nsELRP, which shows a positive surgical outcome of early continence \[[@B5],[@B6]\]. In this study, we report our experience with 50 consecutive patients who underwent intrafascial nsELRP by a single surgeon. MATERIALS AND METHODS ===================== 1. Patient selection criteria and evaluation -------------------------------------------- Among 62 patients, 50 patients with clinically localized prostate cancer at the time of preoperative magnetic resonance image (MRI) with PSA levels less than 10 ng/ml and Gleason scores less than 7 (3+4) were included for intrafascial nsELRP from November 2011 to April 2012. If there were any suspicious lesions with extracapsular extension of tumors or high Gleason scores (7 \[4+3\] to 10), conventional radical prostatectomy and lymph node dissection were performed. A small group of patients underwent intrafascial nsELRP despite their higher PSA levels of more than 10 ng/ml because they had low Gleason scores (≤6). 2. The surgical techniques -------------------------- ### Patient positioning and trocar placement In short, the patient was placed in the supine position with mild head-down tilt. The preparation of the space of Retzius began with a 2-cm sized incision in the infraumbilical crease laterally to the midline, which was then carried down to the posterior rectus sheath where a balloon trocar was inserted and the preperitoneal space was developed. Finally, the camera trocar was placed and further trocars were inserted with attention to the course of the epigastric vessels and peritoneum. Next, the 12-mm assistant port, 12-mm operator port, and 5-mm operator port were placed sequentially. ### Retroperitoneal preparation The fatty tissue surrounding the prostate and pelvic space was removed to expose the landmark structures such as the pubic arch, symphysis, endopelvic fascia, bladder, and prostate. The anterior surfaces of the bladder and prostate as well as the endopelvic fascia became visible to the operator. The removal of preprostatic fibro-fatty tissues facilitates more definite differentiation between the prostate and the bladder. ### Initial approach and bladder neck dissection The first step of the intrafascial procedure was different from conventional ELRP. The endopelvic fascia and pubo-prostatic ligament were not incised and the deep dorsal venous complex was not ligated at the beginning of the procedure. A bilateral incision of the periprostatic fascia was made medial to the puboprostatic ligament (PPL) and directed to the base of the prostate. The right plane of dissection was recognized when the surface of the prostate was completely smooth. As a result, the development of a plane between the prostate and overlaying fascia was possible and the operator could detach the prostate from its enveloping fascia. All lateral periprostatic fascia, endopelvic fascia, and puboprostatic fascia remained intact ([Fig. 1](#F1){ref-type="fig"}). ### Preparation of the seminal vesicles Careful coagulation-free dissection is necessary during bilateral seminal vesicle dissection because the pelvic plexus and NVB run in close proximity to the tip of the seminal vesicle. Five-millimeter titanium clips (Ligaclip, Ethicon Inc., Somerville, NJ, USA) were used to ligate the seminal vesicular artery. When the vas deferens and seminal vesicle were both fully dissected, the Denonvilliers fascia was visualized. The Denonvilliers fascia was stripped down to find the correct plan for the intrafascial dissection of the prostate. After identification of the prostate capsule, the dissection was gradually directed toward the apex of the prostate in the midline to avoid injury of the NBVs. ### Dissection of the prostatic pedicle and NBV The prostate was fully detached from its surrounding fascias but was still attached by the pedicles and the apex. The surface of the prostatic capsule was clearly seen medially and laterally. The assistant took the seminal vesicle and vas deferens and gently elevated them ventrally to allow clear sight of the prostatic pedicle. The prostatic pedicles were then clipped and cut in a step-by-step manner directly on the surface of the prostatic capsule without excessive traction or electrocautery. When dissecting the NBV, care must be taken to avoid damage to the NVB. After the correct plane was opened, the dissection was performed by using cold scissors in an essentially avascular plane. Traction and electrocautery on the prostate and the NVB were also avoided. Meticulous dissection is needed between the lateral side of the prostate and the lateral remnant of the periprostatic fascia toward the apex of the prostate to preserve the accessory NVB ([Fig. 2](#F2){ref-type="fig"}). The assistant retracted the partially mobilized prostate to the opposite side. The dissection was performed on the opposite side of the prostate pedicle and the NVB in the same fashion as that of the primary side. Finally, the prostate was completely detached from the surrounding fascias, bladder, prostatic pedicles, and NVBs. ### Dissection of the urethra Sharp dissection of the prostate from the external sphincter and urethra at the site of the apex was performed. We did not ligate the deep dorsal venous plexus (Santorini plexus). The assistant retracted the prostate to secure a clear view of the apex margin of the prostate. Dissection and division of the external urethral sphincter and prostate were carefully made ([Fig. 3](#F3){ref-type="fig"}). Dissection of the urethra was performed proximally very close to the prostate to preserve the urethral length as much as possible. Vertical dissection was performed with cold scissors alone to achieve complete division of the prostate from the urethra. An endoscopic retrieval bag was used to remove the prostate. 3. Postoperative evaluation --------------------------- All patients underwent cystography on the 4th postoperative day. After identification of no contrast leakage at the anastomosis site, the Foley catheter was removed. Postoperative evaluation of continence was performed by evaluation of the number of pads used per day. Postoperative follow-up was defined at 2 weeks, 6 weeks, and 3 months. Postoperative interviews were also conducted. Complete continence was defined as usage of no pads and patient\'s report of no urinary leakage. We also evaluated erectile function and oncologic outcome after nsELRP. 4. Statistical analysis ----------------------- Statistical analysis was performed by using PASW ver. 18.0 (IBM Co., Armonk, NY, USA), with statistical significance considered at p\<0.05. Spearman\'s correlation test was used to evaluate the factors related to early continence recovery after nsELRP. RESULTS ======= A total of 50 consecutive patients underwent intrafascial nsELRP by one surgeon. Preoperative and perioperative data are summarized in [Table 1](#T1){ref-type="table"}. The rate of conversion to open surgery and the transfusion rate were both 0%. The mean operation time was 149.3 minutes (range, 90 to 240 minutes). The mean hospitalization time was 6.3 days (range, 4 to 19 days), and the mean catheterization time was 5.5 days (range, 3 to 26 days). Postoperative continence results are shown in [Table 2](#T2){ref-type="table"}. At the 2-week follow-up after surgery, 14 patients (28.0%) achieved total continence and on average 2.3 pads were required for the other incontinent patients. A total of 35 patients (70.0%) were continent at 6 weeks after surgery, 8 patients (16.0%) had mild stress incontinence (1 to 2 pads), and 7 patients (14.0%) required \>2 pads per day ([Fig. 4](#F4){ref-type="fig"}). A total of 31 patients were available for follow-up at 3 months after surgery, and 26 patients (83.9%) were pad-free. Compared with conventional nerve-sparing LRP, intrafascial nsELRP showed early continence recovery ([Table 2](#T2){ref-type="table"}). In patients with a relatively old age of above 65 years, preoperative low prostate volume and low Gleason score were related with early continence at 6 weeks after surgery ([Table 3](#T3){ref-type="table"}). Signs of early recovery of potency (morning erection or erection sensation) were reported by 19 patients (38.0%) at 6 weeks after surgery and the rate increased to 54.8% at 3 months after surgery. The average positive surgical margin rate was 34.0%, respectively. But from the 30 cases, the positive surgical margin rate decreased to 9.5% ([Table 4](#T4){ref-type="table"}). Most positive surgical margins were seen at the apex of the prostate. Early postoperative complications were encountered. Two patients (4.0%) with anastomosis site leakage were treated with prolonged catheterization for an additional 1 week. Two patients (4.0%) with acute urinary retention after catheter removal were treated with re-catheterization for 1 week. DISCUSSION ========== Urinary incontinence is potentially the most debilitating complication of radical prostatectomy. The risk of urinary incontinence is not small and is variable among surgeons. It has been shown that the incidence of postoperative incontinence depends on the urologist\'s experience, patient\'s age (increased frequency after 70 years), and whether the operative technique includes minimal distal incision of the endopelvic fascia, preservation of the bladder neck, bilateral nerve-sparing surgery, or preservation of the PPL \[[@B7],[@B8]\]. In the American Urological Association guidelines, the reported risk of urinary incontinence ranges from 3 to 74% for radical prostatectomy \[[@B9]\]. Continence mechanisms involve many structures, including the PPL, Denonvilliers\' fascia, levator muscle, endopelvic fascia, and internal and external sphincters. Ventrally, the proximal prostate is covered by muscle fibers originating from the outer longitudinal bladder muscle and extending over the gland. These fibers constitute a detrusor apron \[[@B10],[@B11]\]. The pubovesical/PPLs (PV/PPLs) are paired fibrous bands originating from visceral endopelvic fascia. They insert on the distal third of the posterior surface of the pubic bone adjacent and anterior to the urethral sphincter \[[@B11]\]. The visceral component of the endopelvic fascia covers the pelvic organs including the prostate, bladder, and rectum, and it is fused with the anterior fibromuscular stroma of the prostate at the upper ventral aspect of the gland \[[@B12]-[@B14]\]. Along the pelvic sidewall at the lateral aspect of the prostate and bladder, the parietal and the visceral components of the endopelvic fascia are fused. As a fascial condensation, this fusion is often recognizable as a whitish line and is named the fascial tendinous arch of the pelvis. It stretches from the PV/PPLs to the ischial spine. During surgery, access to the lateral prostate may be gained by incision of the endopelvic fascia either medial or lateral to this fusion \[[@B10]\]. The PV/PPLs stabilize the prostate, urethra, and bladder to the pubic bone and are considered an important part of the \"suspensory system\" of the continence mechanism \[[@B15]-[@B19]\]. Some authors have suggested that preservation of these ligaments during radical prostatectomy may improve early recovery of urinary continence, but no definitive evidence has yet been established \[[@B17],[@B18]\]. Preservation of the PV/PPLs is facilitated by using the perineal and laparoscopic approach, whereas during open retropubic prostatectomy, the PV/PPLs are more difficult to preserve \[[@B20],[@B21]\]. Some authors have suggested that avoiding incision of the endopelvic fascia during radical prostatectomy, often combined with an intrafascial nerve-sparing procedure, might improve early recovery of urinary continence as well as improve postoperative erectile function, but definitive evidence has yet to be established \[[@B12],[@B22]\]. The parietal endopelvic fascia includes fascia of the levator ani muscle. The incision of this fascia immediately lateral to the fascial tendinous arch incises the levator ani fascia (LAF) and leaves the muscle fibers of the levator ani bare and the LAF adherent to the prostate \[[@B10],[@B19]\]. An incision of the visceral endopelvic fascia medial to the fascial tendinous arch results in a dissection plane that leaves the levator ani muscle covered with its fascia without exposure of its fibers \[[@B12],[@B22]\]. The result is a prostate covered only by prostatic fascia (PF), when present, and not by a layer of LAF \[[@B12],[@B13]\]. This fascia is not a discrete single-layered structure stretching over the lateral surface of the prostate. Laparoscopic surgery may offer an improved identification of these structures, resulting in less damage to the structures around the prostate. We could also identify these structures during the operation, which is related to recovery of postoperative urinary continence. In addition, others have showed that continence rates correlate with differences in the mean functional urethral length and the existing differences in the maximal urethral closure pressure postprostatectomy \[[@B11],[@B15],[@B16],[@B23]\]. Poore et al. \[[@B17]\] reported an earlier return of continence with a PPL-sparing technique versus a nonsparing technique, but the final outcomes were equivalent. Other authors also advocate the latter technique and have reported encouraging results \[[@B11],[@B15],[@B16],[@B23]\]. We believe that during mobilization of the prostate and especially during apical dissection, the intactness of the urethral supporting structures are of paramount importance because this avoids shear stress to the urethra as well as possible denervation. In our study, we observed an earlier return to continence in patients who underwent intrafascial nsELRP compared with previous conventional nerve-sparing LRP. Complete continence was achieved by 70.0% of patients who underwent nsELRP at 6 weeks after surgery and by 83.9% at 3 months after surgery. In the control group, however, 55.2% of patients achieved complete continence at 6 weeks after surgery and 75.0% at 3 months after surgery. Reconstruction methods such as periurethral suspension stitch, bladder neck reconstruction, and posterior reconstruction have also been shown to provide improved early continence recovery, but the results are still controversial \[[@B24],[@B25]\]. Therefore, we focused on the preservation of normal structures rather than performing reconstruction after destruction of these structures. Studies have reported that older age may be the only increasing risk factor for postprostatectomy incontinence \[[@B26]\], but in our study, increased prostate volume and Gleason score were related with post-prostatectomy incontinence in the older aged group (age over 65 years). We speculate that large prostate volume and old age may be related to preoperative bladder and external urethral sphincter dysfunction leading to interference with postprostatectomy recovery of continence. A preoperative urodynamic study may be helpful in proving the relationship between prostate volume and old age with postoperative incontinence. Anatomical studies \[[@B27],[@B28]\] have illustrated the prostatic neuroanatomy in detail, detecting additional neural tissues to the nerve bundles on the anterior midpart and posterior surface of the prostate. Costello et al. \[[@B29]\] recently showed that most of the NVB descends posteriorly to the seminal vesicle. The nerves pass anteriorly and converge at the midprostatic level, and when they approach the apex, they diverge again. The anterior and posterior nerves of the NVB are separated by 3 cm at the level of the base of the prostate. At this anatomic site, the cavernosal nerves are not easily distinguished from the surrounding tissues and care should be taken during urethra-vesical anastomosis. Walsh \[[@B30]\] proposed that the NVB is enclosed within the two layers of the lateral pelvic fascia composed of the lateral layer of the levator fascia and the medial layer of the PF. Kiyoshima et al. \[[@B14]\] proved that the NVB was located on the posterolateral region of the prostate in 48% of their patients. In the remaining patients (52%), the NVB was widely distributed on the entire lateral aspect of the prostate without any specific localization. Thus, the authors proposed performing wide dissection of the lateral aspect of the prostate during radical prostatectomy to preserve the NVB. Therefore, meticulous dissection and disuse of electro-cauterization at the lateral and apex sides of the prostate are required to preserve the NVB by intrafascial nsELRP. Anastasiadis et al. \[[@B1]\] reported potency rates of 30% and 41% at 12 months after LRP (n=230) and open retropubic prostatectomy (n=70), respectively. After preservation of one or both NVBs, the potency rates increased from 37 to 44% with the retropubic approach and from 46 to 53% with LRP, respectively. Patients younger than 60 years who underwent bilateral NVB preservation were reported to be potent in 72% and 81% of cases, respectively. Graefen et al. \[[@B20]\] reported rates of erections of 96.5%, 90.7%, and 84.3% and rates of intercourse of 69.0%, 52.8%, and 37.3% at 12 months after bilateral RRP in men \<55 years, 55 to 65 years, and \>65 years, respectively. In our study, the patients were generally old aged and were not sexually active; thus, we could not obtain potency recovery data. However, 19 patients (38.0%) reported subjective signs of potency recovery (morning erection or erection sensation) at 6 weeks after surgery, and at 3 months after surgery, the percentage was increased to 54.8% of patients. The oncologic data of 1,000 LRPs at the Montsouris Institute revealed positive surgical margin rates of 6.9% for pT2a and 34% for pT3b tumors \[[@B2]\]. In our study, the overall positive surgical margin rate was 34.0%. We consider this high positive margin rate as a \"learning curve.\" In fact, from 30 cases, positive surgical margin rates decreased compared with the first 30 cases (51.7% vs. 9.5%). Most positive margins were seen at the apex of the prostate. CONCLUSIONS =========== Intrafascial nsELRP is a further evolution of the current nsELRP. The initial results are promising, providing favorable functional outcomes and oncologic results similar to other RP techniques. The authors have nothing to disclose. ![Initial approach for intrafascial nerve-sparing extraperitoneal laparoscopic radical prostatectomy. Incision is made medial to the puboprostatic ligament. PPL, puboprostatic ligament; P, prostate; Bl, bladder; Ef, endopelvic fascia.](kju-53-836-g001){#F1} ![Prostatic pedicle dissection and sparing of the neurovascular bundles. Both anterior and posterior neurovascular bundles were preserved by using cold scissors. P, prostate; aNVB, accessory neurovascular bundle; pNVB, predominant neurovascular bundle.](kju-53-836-g002){#F2} ![Dissection of the prostate apex and urethra. Urethral dissection was performed proximally very close to the prostate to preserve urethral length. We did not ligate the deep dorsal venous plexus. The endopelvic fascia, which envelops the levator ani muscle, was preserved during surgery and the longer urethra was preserved. Both puboprostatic ligaments and arcuous tendinosus were also preserved. Dv, deep dorsal venous plexus; U, urethra; P, prostate; Ef, endopelvic fascia; PPL, puboprostatic ligament; At, arcuous tendinosus.](kju-53-836-g003){#F3} ![Continence rates after intrafascial nerve-sparing extraperitoneal laparoscopic radical prostatectomy.](kju-53-836-g004){#F4} ###### Preoperative and perioperative patient characteristics (n=50) ![](kju-53-836-i001) Values are presented as mean±SD (range). PSA, prostate-specific antigen. ###### Continence after intrafascial nsELRP and conventional nsLRP ![](kju-53-836-i002) Values are presented as number (%). nsELRP, nerve-sparing extraperitoneal laparoscopic radical prostatectomy; nsLRP, nerve-sparing laparoscopic radical prostatectomy. ###### Factors related to early continence at 6 weeks after surgery ![](kju-53-836-i003) PSA, prostate-specific antigen. ^a^:Statistical significance. ###### Oncologic outcomes (n=50) ![](kju-53-836-i004)	{ "pile_set_name": "PubMed Central" }
![](edinbmedj73919-0093){#sp1 .389} ![](edinbmedj73919-0094){#sp2 .390} ![](edinbmedj73919-0095){#sp3 .391}	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#s1} ============ Adoptive cellular immunotherapy with antigen-specific cytotoxic T lymphocytes (CTLs), which utilizes the power and specificity of the autologous immune system, has emerged as a promising strategy for the treatment of established malignancies \[[@R1]--[@R5]\]. For effective therapeutic infusion, adequate quantities of in vitro-primed and -expanded high-affinity T cells restricted to tumor-associated antigens are necessary. Antigen-presenting cells (APCs), particularly dendritic cells (DCs), macrophages, and B lymphocytes, are the most potent stimulators of the generation and amplification of tumor antigen-specific CTLs in adequate quantities for clinical use \[[@R6]--[@R8]\]. However, the isolation and generation of sufficient autologous APCs is cumbersome and expensive, requiring a large amount of blood from patients or compatible healthy donors. Moreover, the efficiency of APCs in inducing and expanding antigen-specific T cells is significantly affected by tumor microenvironments \[[@R9]\]. Previous studies have demonstrated that the epitope peptide FMNKFIYEI (AFP~158-166~) is a HLA-A\0201-restricted human hepatocellular carcinoma (HCC)-specific antigen that can stimulate specific T cell responses \[[@R10]\]. The main obstacle limiting the wide application of alpha-fetoprotein (AFP)-specific CTLs in adoptive immunotherapy is the technical challenge of manufacturing them. Currently, the generation of CTLs requires several rounds of activation with antigenic peptide-pulsed DCs and a high dosage of interleukin (IL)-2, which are time-consuming and expensive procedures. Moreover, the efficiency of the specific activation is quite low. Artificial antigen-presenting cells (aAPCs) have been developed as a cost-effective alternative to natural APCs \[[@R11], [@R12]\]. aAPCs are engineered using a wide variety of platforms and are readily prepared as off-the-shelf, standardized, and renewable reagents that deliver the appropriate signals to naive T cells. Cellular aAPCs are commonly based on xenogeneic or allogeneic cells and genetically modified using retrovirus or lentivirus transduction \[[@R13]--[@R16]\]. Also, acellular aAPCs can be synthesized by covalently coupling the MHC/peptide complex and CD28 ligand on various materials \[[@R17]\]. However, in most of the currently available aAPC systems, synthetic antigenic peptides are exogenously loaded on to aAPCs. In general, exogenous peptides are processed by APCs in the endosomes and presented to CD4+ T cells by MHC class II molecules through the class II pathway, while endogenously expressed peptides are processed in the cytoplasm and presented to CTLs by MHC class I molecules through the class I pathway. Although there are interconnections and cross-presentations between the class I and class II pathways, the efficiency of CTL activation and expansion is reduced when exogenous loading methods are used \[[@R18]\]. In the current study, we employed human B cell lymphoma cell line BJAB as the scaffold to develop a cellular aAPC that could efficiently generate AFP-specific CTLs for adoptive immunotherapy of HCC \[[@R19]\]. BJAB cells were chosen because they exhibit a high expression level of adhesion molecules, including CD80 and CD86, which are known as the co-stimulatory signals in T cell activation and are easy to handle for cell proliferation and genetic modification. BJAB cells were genetically engineered to endogenously express both AFP~158-166~ peptide and IL-15 via* lentivirus transduction, which we expected would increase the specific activation rate of AFP~158-166~-specific CTLs. We then conducted a series of function tests on the resulting BA15 cells to evaluate the specific cytotoxicity of CTLs against HCC cells in vitro and in vivo. RESULTS {#s2} ======= Establishment of aAPC expressing peptide-MHC complex, co-stimulatory molecule ligands, and cytokine {#s2_1} --------------------------------------------------------------------------------------------------- HLA-A2, CD80, and CD86 were expressed at high levels in BA15 cells because these cells are based on the BJAB cell line. We found no significant differences in the expression of these proteins in BA15 cells compared with DC and BJAB cells (Figure [1A](#F1){ref-type="fig"}). Real-time quantitative PCR (qRT-PCR) and ELISA showed that the mRNA and secreted protein levels of IL-15 in BA15 cells were significantly higher than those in DC and BJAB cells (Figure [1B](#F1){ref-type="fig"}). The presence of the AFP~158-166~ peptide was biochemically confirmed directly from the HLA-A\02:01 groove on the surface of the BA15 cells. The high-performance liquid chromatography (HPLC) analysis revealed that an eluting peak corresponding to the synthetic peptide was found in acid stripping of BA15 cells, but not in BJAB cells. Mass spectrometry revealed that the molecular weight of the peptide in this eluting peak was the same as that of the synthetic peptide (Figure [1C](#F1){ref-type="fig"}). ![Expression of AFP~158-166~ peptide-HLA-A\02:01 complex, CD80, CD86, and IL-15 in DC, BA15, and BJAB cells\ A. FCM revealed that there were not significant differences in the expression of HLA-A2, CD80, and CD86 among DC, BA15 and BJAB cells. B. qRT-PCR and ELISA showed that the mRNA and secreted protein levels of IL-15 in BA15 cells were significantly higher than those in DC and BJAB cells. C. HPLC showed that the eluting peak corresponding to the synthetic AFP~158-166~ peptide was found in acid-stripped BA15 cells but not in BJAB cells. Mass spectrometry also revealed that the molecular weight of the peptide in this eluting peak was the same as that of the synthetic peptide. Error bars indicate standard deviations. \* indicates P \< 0.05.](oncotarget-07-17579-g001){#F1} Stability of peptide-MHC complex, co-stimulatory molecule ligands, and cytokine expression in aAPCs after γ-ray irradiation {#s2_2} --------------------------------------------------------------------------------------------------------------------------- In BA15 cells, the expression of HLA-A2, CD80, and CD86 were not significantly affected by different dosages of irradiation (Figure [2A](#F2){ref-type="fig"}). ELISA showed that the secretion of IL-15 in BA15 cells decreased after exposure to 30 Gy of radiation but was not significantly affected by irradiation at lower dosages (Figure [2B](#F2){ref-type="fig"}). HPLC showed that the eluting peak corresponding to the synthetic AFP~158-166~ peptide was found in acid-stripped BA15 cells both before and after treatment with 30 Gy of radiation. Mass spectrometry revealed that the molecular weight of the peptide in this eluting peak was the same as that of the synthetic peptide (Figure [2C](#F2){ref-type="fig"}). ![Stability of AFP~158-166~ peptide-HLA-A\02:01 complex, CD80, CD86, and IL-15 expression in BA15 cells after γ-ray irradiation\ A.* FCM revealed that the expression of HLA-A2, CD80, and CD86 were not significantly affected by different dosages of irradiation. B. ELISA showed that the secretion of IL-15 in BA15 cells decreased after exposure to 30 Gy of irradiation but was stable at lower dosages. C. HPLC showed that the eluting peak corresponding to the synthetic AFP~158-166~ peptide was found in acid-stripped BA15 cells both pre- and post-irradiation. Mass spectrometry revealed that the molecular weight of the peptide in this eluting peak was the same as that of the synthetic peptide. Error bars indicate standard deviations.](oncotarget-07-17579-g002){#F2} Inhibiting proliferation and inducing apoptosis of aAPCs by γ-ray irradiation {#s2_3} ----------------------------------------------------------------------------- In our dosage-course experiment using γ-ray irradiation, the MTT assay indicated that the viability of BA15 cells decreased after exposure to 20 Gy and 30 Gy of radiation (Figure [3A](#F3){ref-type="fig"}). The cell counting and carboxyfluorescein succinimidyl ester (CFSE) analyses indicated that BA15 cell proliferation was completely inhibited at doses of 20 Gy and 30 Gy (Figure [3B](#F3){ref-type="fig"} and [3C](#F3){ref-type="fig"}). Apoptosis assays performed every 3 days after irradiation for 12 days revealed that all the cells in the 20-Gy and 30-Gy groups were either in apoptosis or dead after irradiation; all the cells had died within 12 days. There were fewer dead cells in the 20-Gy group than in the 30-Gy group at each time point (Figure [3D](#F3){ref-type="fig"}). Thus, 20 Gy was determined to be the optimal dosage at which the proliferation of BA15 cells was completely inhibited while leaving most of the cells still viable within the frame of 1 round of activation (7 days). Expression of HLA-A2, CD86, CD80, IL-15, and AFP~158-166~ peptide was not significantly affected by radiation at that point. After the activation process, all BA15 cells would have to die to guarantee the clinical safety of adoptive infusion. ![Inhibition of proliferation and induction of apoptosis of BA15 by γ-ray irradiation\ After different dosages of irradiation, the cell viability and proliferation of BA15 cells were analyzed by MTT, cell counting, and CFSE assays. Apoptosis assays were performed every 3 days after irradiation. A. MTT assay indicated that the cell viability of BA15 cells decreased after exposure to 20 Gy and 30 Gy of irradiation. B. Cell counting indicated that the number of BA15 cells decreased after exposure to 20 Gy and 30 Gy of irradiation. C. CFSE labeling revealed that the proliferation of BA15 cells was completely inhibited after exposure to irradiation of 20 Gy and 30 Gy. D. The apoptosis assay revealed that all the cells in the 20-Gy and 30-Gy group were in apoptosis or dead 3 days after irradiation and that all the cells had died by day 12. There were fewer dead cells in the 20-Gy group than in the 30-Gy group at every time point. Error bars indicate standard deviations.](oncotarget-07-17579-g003){#F3} Efficient activation and expansion of AFP~158-166~-specific CTLs by aAPCs {#s2_4} ------------------------------------------------------------------------- CTLs isolated from HLA-A\02:01+ healthy donors were stimulated by co-culturing with different APCs for 3 weekly cycles. Cell counting and CFSE assays showed that BA15 cells efficiently activated CTLs at different APC/lymphocyte ratios (1:10 and 1:20), with maximum efficiency at 1:10 (Figure [4A](#F4){ref-type="fig"} and [4B](#F4){ref-type="fig"}). After 3 weekly rounds of stimulation at this ratio, BA15 cells showed the same activation efficiency as DCs, but AFP~158-166~ MHC Pentamer staining showed that the percentage of AFP-specific CTLs was significantly higher in the BA15 cell population than in the DCs (6.7 ±0.4% vs. 4.5 ±0.3%, P* \< 0.05) (Figure [4C](#F4){ref-type="fig"}). The results indicated that the density of the AFP peptide expressed and presented by the HLA-A\02:01 molecule on BA15 cells was higher than that of the DC-pulsed exogenous AFP peptide (40 mg/mL). Furthermore, BJAB cells exhibited no significant CTL activation function. ![Activation and expansion of functional AFP~158-166~-specific CTLs with BA15 cells\ CTLs isolated from HLA-A\02:01+ healthy donors were stimulated weekly by co-culturing with different APCs for 3 cycles. A. Cell counting showing CTL activation by different APCs and APC/lymphocyte ratios. BA15 cells activated CTLs at maximum efficiency at a ratio of 1:10. After 3 cycles of stimulation at this ratio, BA15 cells and DCs showed the same CTL activation efficiency. B. CFSE assay showing activation of CTLs by different APCs at different APC/lymphocyte ratios. BA15 cells had maximum CTL activation efficiency at a ratio of 1:10. C. AFP~158-166~ MHC Pentamer staining showed that CTLs generated by BA15 cells established a higher proportion of AFP~158-166~-specific CTL population than CTLs generated by peptide-pulsed DCs (6.7 ±0.4% vs. 4.5 ±0.3%, respectively, P \< 0.05). D. Secretion of IFN-γ by antigen-specific CTLs activated by different APCs at different APC/lymphocyte ratios. A ratio of 1:10 displayed the highest levels of antigen-specific secretion of IFN-γ. The mean number of SFU per million was 810.1 (±58.3) in BA15-activated CTLs versus 660.4 (±54.2) in CTLs obtained with AFP-pulsed DCs. E. Comparison of the specific lysis rate of CTLs activated by different APCs toward HepG2. The CTLs activated by BA15 cells had significantly higher lysis rates toward HepG2 cells than did the CTLs obtained from AFP-pulsed DCs. F. Comparison of the specific lysis rate of AFP-positive (HepG2 and T2+AFP) and AFP-negative (SW480, MCF7, T2, and T2+HER2/neu) cells. AFP-positive cells were specifically lysed by antigen-specific CTLs stimulated with BA15 and peptide-pulsed DCs, but AFP-negative cells were not. E: T Ratio indicates effector to target ratio. Error bars indicate standard deviations. \* indicates P \< 0.05. NS: not significant.](oncotarget-07-17579-g004){#F4} Functional tests of AFP~158-166~-specific CTLs expanded by aAPCs {#s2_5} ---------------------------------------------------------------- To confirm the functionality of the antigen-specific CTLs, we assessed the IFN-γ secretion of CTLs that had been expanded by different APCs. The numbers of IFN-γ-secreting T cells differed across the APC-stimulated CTL groups and corresponded with the results determined by the AFP~158-166~ MHC Pentamer labeling. The antigen-specific CTLs activated by BA15 at the APC/lymphocyte ratio of 1:10 exhibited the highest rates of antigen-specific secretion of IFN-γ (Figure [4D](#F4){ref-type="fig"}). The mean number of spot-forming units (SFU) per million was 810.1 (±58.3) in BA15-activated CTLs versus 660.4 (±54.2) in CTLs obtained with AFP-pulsed DCs. When we examined the cytotoxicity of the AFP-specific CTLs we generated, we found that AFP-positive cells (HepG2 and T2+AFP) but not AFP-negative cells (SW480, MCF7, T2, and T2+HER2/neu) were specifically lysed by antigen-specific CTLs stimulated with BA15 and peptide-pulsed DCs. This result suggests that the generated CTLs had sufficiently high antigen-specific T-cell receptor avidity to recognize AFP+ HCC cells. Furthermore, the CTLs activated by BA15 also had a higher specific lysis rate toward HepG2 cells than did the CTLs obtained from AFP-pulsed DCs. These results show that the CTLs generated by BA15 were able to recognize AFP-positive HCC cells and possessed potent cytotoxic functions (Figure [4E](#F4){ref-type="fig"} and [4F](#F4){ref-type="fig"}). AFP~158-166~-specific CTLs expanded by aAPCs suppress tumor growth in tumor-bearing NOD /SCID mice {#s2_6} -------------------------------------------------------------------------------------------------- When we examined the cytotoxicity of the AFP-specific CTLs in in vivo mouse models, we found that AFP-specific CTLs expanded by DC and BA15 displayed significant tumor suppressive functions when compared with unspecific activated CTLs and control group treated with PBS. CTLs expanded by BA15 were the most efficient in these groups (Figure [5A](#F5){ref-type="fig"}). Hematoxylin and eosin (HE) staining showed that significantly more tumor cells exhibited morphological changes indicative of apoptosis and necrosis in the BA15 group than others. The infiltration level of CD8+ CTLs in the BA15 group was also significantly higher than that in other groups (P \< 0.01) (Figure [5B](#F5){ref-type="fig"}). ![Suppression of tumor growth in vivo by AFP~158-166~-specific CTLs generated by BA15 cells\ HepG2 cells were used to establish xenograft tumors in NOD /SCID mice. Nonspecific activated CTLs, DCs, or BA15-induced AFP-specific CTLs (1 × 10^8^ in 0.2 mL PBS) were transferred into mice of different groups via intravenous injection at days 0, 7, and 14. Black arrows indicate the treatment times. PBS-treated mice were used as a control group. A. Treatment with AFP-specific CTLs expanded by DCs and BA15 displayed significantly higher tumor-suppressive functions than treatment with nonspecific activated CTLs or PBS. CTLs expanded by BA15 were the most efficient of these groups. B. HE staining showed that there were significantly more tumor cells exhibiting morphological changes indicative of apoptosis and necrosis in the BA15 group than others. The infiltration of CD8+ CTLs in the BA15 group was also significantly higher than that in other groups (P \< 0.01). Scale bar = 100 μm; magnification, 400 ×. Error bars indicate standard deviations. \* indicates P \< 0.05.](oncotarget-07-17579-g005){#F5} DISCUSSION {#s3} ========== In this study, we have reported on the development of a novel AFP-specific aAPC based on the human B lymphoma BJAB cell line. We contend that this system could provide a robust platform for the generation of functional AFP-specific CTLs for adoptive immunotherapy of HCC. The aAPC we developed is a stably engineered cell line and can be generated in a standardized manner. aAPCs have been developed as potent tools for generating and expanding antigen-specific CTLs for adoptive immunotherapy \[[@R20]--[@R23]\]. Although acellular synthetic aAPCs can provide good control of signal delivery and efficiently induce high T cell expansion rates, their application is hampered by several limitations \[[@R24]--[@R26]\]. The surface-bound antibodies of aAPCs are not only different from those provided by natural ligands but also technically difficult to stably link to the platform. Control of local release of cytokines is also difficult in acellular aAPC systems. Cellular aAPCs derived from genetically engineered cell lines provide even easier control of signal delivery than do natural APCs and can be stored for an extended time as a readily accessible source of cells for immunotherapy. The optimal design of aAPCs for practical use in adoptive immunotherapy is still being determined. A general paradigm for the design of an optimal aAPC must include 3 stimulating signals, as does the aAPC described here. (1) Activation signal 1 is mediated by the antigenic peptide-bearing MHC that bonded with the T cell receptors specific to the corresponding epitope \[[@R27]--[@R28]\]. BA15 cells were genetically engineered to endogenously express AFP~158-166~ peptide and to achieve a higher specific activation rate of antigen-specific CTLs than would exogenously loaded peptide. (2) Activation signal 2 is conducted by co-stimulators, such as the B7 family proteins B7.1 and B7.2 \[[@R29]--[@R31]\]. BA15 cells can express levels of CD80 and CD86 as high as that of their BJAB backbone (3) Cytokines synthesized and secreted by APCs function as activation signal 3 and are essential for T cell survival and shaping of immune responses \[[@R32]--[@R34]\]. The BA15 cells in our study were genetically engineered to endogenously express IL-15. The aAPC we developed also met the safety requirements of for clinical administration of immunotherapy. Gamma-ray irradiation at 20 Gy completely inhibited the proliferation of BA15 cells while keeping most of them viable for 1 round of activation. All the cells died within 12 days. Importantly, we found that the expression of activation signals was not significantly affected at this dose of radiation. In this way, irradiated BA15 cells would not cause cancer to spread and would preserve their ability to activate T cells. IL-15, a member of the common cytokine receptor γ-chain family, plays a more important role in the activation and proliferation of CTLs than does IL-2. It can promote expansion with an early differentiation phenotype and may allow greater expansion and prolonged in vivo persistence. A previous study reported that IL-15 enhanced the biological activity of CD8+ T cells and induced memory CD8+ T cells with high levels of CD28 \[[@R35]--[@R38]\]. Similarly, we found that BA15 cells also stably expressed and secreted high levels of IL-15. Although the production of IL-15 was not adequate to completely replace exogenous IL-2 to support the proliferation of T cells, it nonetheless did enhance the biological activity of the CTLs. Our study has some limitations. First, the expression of the AFP~158-166~ epitope peptide was only analyzed qualitatively; this aspect of our study was not quantitative. We could only indirectly conclude that the presented AFP~158-166~ peptide level in BA15 cells was higher than that of DCs loaded exogenously from the observation that the BA15-activated CTLs had a significantly higher percentage of AFP~158-166~-specific CTLs than did the DC-activated CTLs. Second, our study lacked control groups to establish the advantages of IL-15 in generating and expanding AFP-specific CTLs. In summary, we have generated BJAB-based aAPCs expressing an endogenous AFP~158-166~ epitope peptide-HLA-A\02:01 complex, CD80, CD86, and IL-15 that successfully expanded functional CTLs specific to the AFP~158-166~-restricted epitope. The CTLs generated by the genetically engineered BA15 cell line established a higher proportion of AFP-specific CTL populations and exhibited higher cytolytic activity against AFP-positive HCC cells than did CTLs obtained from peptide-pulsed DCs. MATERIALS AND METHODS {#s4} ===================== Cell culture {#s4_1} ------------ Blood samples were obtained from healthy donors under the supervision of the Institutional Review Board of Tianjin Medical University. Donors provided written informed consent, in accordance with the Declaration of Helsinki. We selected cells with the HLA-A\02:01 phenotype using flow cytometry (FCM) and sequence-specific primed PCR. Peripheral blood mononuclear cells (PBMCs) were isolated from HLA-A\02:01 samples. The monocytes were then purified using CD14 microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) and cultured in RPMI 1640 medium supplemented with 10% FBS (Invitrogen, Frederick, MD), IL-4, GM-CSF, and TNF-α (PeproTech, Rocky Hill, NJ) to generate mature DCs. The human HCC cell line HepG2 (AFP+, HLA-A\02:01), colorectal adenocarcinoma cell line SW480 (AFP−, HLA-A\02:01), breast adenocarcinoma cell line MCF7 (AFP−, HLA-A\02:01), lymphoblastoid cell line T2 (AFP−, HLA-A\02:01), B-cell lymphoma cell line BJAB (AFP−, HLA-A\02:01), and embryonic kidney cell line 293T were purchased from ATCC (Rockville, MD). HepG2, SW480, MCF7, and 293T cells were cultured in Dulbecco\'s modified Eagle medium (DMEM), while T2 and BJAB cells were cultured in RPMI 1640 (Invitrogen) supplemented with 10% FBS. Plasmid and lentivirus vector construction {#s4_2} ------------------------------------------ Plasmid JA15, derived from an IL-15 expression plasmid that had been constructed and kept by our lab, was modified to code and synergistically express AFP~158-166~ epitope peptide and IL-15. The coding sequence was cloned into a lentiviral expression plasmid. 293T cells were then transfected with the lentiviral expression plasmid and Lenti-Pac HIV packaging plasmids (GeneCopoeia, Rockville, MD) to produce lentiviral particles. Establishment of aAPC {#s4_3} --------------------- A stable AFP~158-166~-specific aAPC line, named BA15, was established by infecting BJAB cells with the lentiviral particles, followed by puromycin selection for 3 weeks. ### Phenotypic analysis {#s4_3_1} We employed FCM to detect the phenotypes of APCs on DC, BA15, and BJAB cells. We used 3 fluorochrome-conjugated antibodies: anti-human CD80, CD86, and HLA-A2 (eBioscience, San Diego, CA). ### qRT-PCR {#s4_3_2} To determine the IL-15 expression of DC, BA15 and BJAB cells by measuring mRNA levels, we used a mirVana Isolation Kit (Ambion, Grand Island, NY) to extract the total RNA from the APCs. Reverse transcription was performed with SuperScript II Reverse Transcriptase (Invitrogen). A TaqMan Gene Expression Assay (Applied Biosciences, Grand Island, NY) was used to detect and quantify IL-15. Relative expression was normalized to GAPDH as an endogenous control using the 2^−ΔΔCt^ method. ### ELISA {#s4_3_3} To determine IL-15 secretion levels, we incubated APCs in complete medium for 24 hours. Supernatants were collected, centrifuged to remove cellular debris, and stored at −20°C. An ELISA assay for IL-15 expression was performed according to the manufacturer\'s protocol (R&D Systems, Minneapolis, MN). ### Extraction of cell surface peptides {#s4_3_4} To biochemically confirm the presence of the AFP~158-166~ peptide, we washed BA15 and BJAB cells (5 × 10^9^) twice with cold PBS and exposed them to a citrate buffer (0.13 mol/L citric acid, 0.06 mol/L Na~2~HPO~4~, pH 3.0) for 5 minutes. The eluted extracts were spun and the peptide-containing supernatant was filtered with a 0.22-μm membrane filter and frozen at −20°C for storage pending further analysis. ### Liquid chromatograph mass spectrometry of AFP~158-166~ peptide {#s4_3_5} To determine the presence of the AFP~158-166~ peptide, we analyzed the filtered peptide extract samples from the previous step using a Single Quadrupole Liquid Chromatograph Mass Spectrometer (LCMS-2020, Shimadzu, Kyoto, Japan). The peptide solutions (20 μL) were injected on a C~18~ column (5 μm, 4.6 × 150.0 mm) (Shimadzu), with a flow rate of 0.2 mL/minute at 50°C. Mobile phase A was 0.1% trifluoroacetic acid in water, and phase B was 0.1% trifluoroacetic acid in acetonitrile, with a gradient starting at 90% A to 5% A in 6 minutes and back to 90% in 8 minutes. The peptide fraction corresponding to the synthetic AFP~158-166~ peptide was collected and analyzed with mass spectrometry. Spectra were obtained using the entire 0.2-mL/minute column effluent over a mass-to-charge ratio range of 100:1 to 2000:1. Irradiation of aAPCs {#s4_4} -------------------- To ensure the safety of future clinical applications in adoptive immunotherapy, we irradiated the BA15 cells with γ-rays before use. The optimal dose of irradiation was determined using 2 criteria. First, early apoptosis had to be induced in the aAPCs had to be induced so proliferation would be inhibited but the cells kept alive for a short period. Second, irradiation had to have minimal effects on the expression of HLA, co-stimulatory molecule ligands, AFP~158-166~ epitope peptide, and IL-15. BA15 cells were irradiated with ^60^Co γ-rays at 0, 10, 20, and 30 Gy before further assays. The expression of AFP~158-166~-HLA-A\02:01 complex, CD80, CD86, and IL-15 were analyzed as described in previous steps. ### Cell viability and proliferation assay {#s4_4_1} To assay the proliferation of BA15 cells after irradiation, CFSE-stained cells were cultured in complete medium and analyzed by FCM every day for 7 days. The BA15 cells were also counted and analyzed for viability by an MTT assay every day after irradiation during the 7-day culture period. ### Apoptosis assay {#s4_4_2} The apoptosis rates of the irradiated BA15 cells were analyzed every 3 days using the Annexin V Apoptosis Detection Kit (eBioscience) according to the manufacturer\'s protocol. Generation of AFP~158-166~-specific CTL clone {#s4_5} --------------------------------------------- To generate the AFP-specific CTL clone, we enriched CTLs from PBMCs using CD8 microbeads (Miltenyi Biotec). The mature DCs were incubated with the HLA-A\02:01-restricted AFP~158-166~ peptide (40 mg/mL) (SBS Genetech, Beijing, China) for 4 hours in FBS-free RPMI 1640 medium. BA15 cell and peptide-pulsed autologous DCs were used as APCs to induce AFP-specific CTLs at different APC/lymphocyte ratios in the presence of 10 U/mL IL-2 (PeproTech). The CTLs were restimulated with the APCs at days 7 and 14. The CTLs were also activated with HER2/neu~369-377~ (KIFGSLAFL) peptide-pulsed DCs and BJAB cells as the control group. The activated CTLs were then stained with HLA-A\02:01-restricted AFP~158-166~ MHC Pentamer (Proimmune, Oxford, UK) and anti-CD8 antibody. The CD8 and Pentamer double-positive populations were AFP-specific CTLs. Proliferation of AFP~158-166~-specific CTL clone {#s4_6} ------------------------------------------------ After 3 rounds of stimulation, CFSE-stained activated CTLs were cultured in complete medium for 7 days, and their proliferation was analyzed using FCM. The number of CTLs was counted after each round of activation with APCs. Human IFN-γ enzyme-linked immunospot assay {#s4_7} ------------------------------------------ To compare IFN-γ production among the AFP-specific CTLs induced with the AFP peptide-pulsed DCs, BA15, and BJAB cells, we performed an enzyme-linked immunospot (ELISPOT) assay (R&D Systems). The control cells were either untreated or treated with phytohemagglutinin. ELISPOT data were expressed as the total IFN-γ spots/10^6^ PBMCs. Lactate dehydrogenase cytotoxicity assay {#s4_8} ---------------------------------------- To evaluate the specific cytotoxic effects of in vitro-generated AFP-specific CTLs on AFP+ HCC cells, we used a lactate dehydrogenase cytotoxicity assay. HepG2, SW480, MCF7, T2, T2 pulsed with AFP~158-166~, and T2 pulsed with HER2/neu~369-377~ were used as the target cells. CTLs activated by the different APCs were used as effector cells and were added to the target cells at different effector-to-target ratios (10:1; 20:1; 40:1). After incubation for 6 hours, cell supernatants were collected and the amount of lactate dehydrogenase in the culture medium was assessed using the CytoTox 96 non-radioactive cytotoxicity assay (Promega, Madison, WI), according to the manufacturer\'s instructions. Adoptive cell transfer {#s4_9} ---------------------- To evaluate the specific cytotoxic effects of AFP-specific CTLs on AFP+ HCC cells in vivo, we used HepG2 cells to establish xenograft tumors in NOD /SCID mice. Mice were maintained at the Animal Center of Tianjin Medical University (Tianjin, China). Animal experiments were carried out according to protocols approved by the Ethics Review Committee for Animal Experimentation. HepG2 cells (5 × 10^6^ in 0.1 mL PBS) were inoculated subcutaneously into the right armpit of mice. When the tumor volumes reached 100 mm^3^, the mice were randomized into 4 groups with 8 mice in each group to receive adoptive cell transfer. Nonspecific activated CTLs, DCs, or BA15-induced AFP-specific CTLs (1 × 10^8^ in 0.2 mL PBS) were transferred into the mice via* intravenous injection on days 0, 7, and 14. PBS was used to treat the control group. On day 21, 2 mice from each group were sacrificed, and their tumors were excised for pathological analysis. The tumors were fixed in paraformaldehyde and embedded in paraffin. To evaluate histopathologic changes, 5-μm sections of tumor specimens were stained with HE. Tumor-infiltrating lymphocytes were analyzed by immunohistochemical (IHC) staining with anti-human CD8 antibody (Santa Cruz Biotechnology, Santa Cruz, CA). The other mice were sacrificed before the mean diameter of the tumor reached 20 mm, according to institutional guidelines for the termination of animal research. Statistical analysis {#s4_10} -------------------- Data are presented as means ± standard deviations. A 2-tailed Student\'s t test was performed to compare the differences between groups. P values \< 0.05 were considered statistically significant. SPSS version 17.0 (SPSS Inc., Chicago, IL) software was used for all analyses. This work was supported by the National Natural Science Foundation of China (grant 81172167 to X.H.). CONFLICTS OF INTEREST The authors declare no conflicts of interest.	{ "pile_set_name": "PubMed Central" }
Background {#Sec1} ========== Calf mortality has been an important problem in the dairy industry for more than a hundred years, and the causes of death are multifactorial, from environmental and infectious agents, to host phenotype. Although the knowledge about neonatal diseases in calves has increased in recent years, the mortality rate is still rather high \[[@CR1]\]. One of the major reasons for high mortality rate among neonatal calves is ruminal bloat. Ruminal bloat occurs when gas produced during fermentation builds up in the rumen and is unable to escape. It is usually a secondary problem in newborn calves. Ruminal bloat can become life threatening within a few hours and often requires medical attention \[[@CR2]\]. The most common cause of ruminal bloat in calves that solely consume milk, is failure of esophageal groove closure \[[@CR3]\]. For the first two weeks after birth, a calf is monogastric, a simple stomached animal, using only the abomasum to digest the milk or milk replacer. When the calf suckles milk, milk bypasses the rumen and reticulum to enter into the abomasum, where digestion and absorption takes place. Milk entering into the rumen and reticulum is both wasteful and dangerous to the newborn calf; hence the importance of the esophageal groove in diverting milk from the esophagus into the abomasum \[[@CR4]\]. The physiology of esophageal groove closure was studied by many scholars. For instance, in 1826, Tiedeman and Gmelin were the first workers to report that milk passed directly to the abomasum in young lambs and calves \[[@CR5]\]. Since then, a couple of physiologists had investigated what triggers the closure of the esophageal groove and have come up with very diverging opinions. Colin \[[@CR6]\] concluded that the "esophageal groove closed when boli was swallowed during rumination and that this was the main route for passage of solid matter from the rumen to the omasum and abomasum". Schalk and his colleague \[[@CR7]\], and a non-peer reviewed compilation Costello, \[[@CR8]\] and Wise \[[@CR9]\] suggested that if calves suckled milk from a rubber nipple it usually passed into the abomasum, while it often passed to the rumen if it was drunk from a bucket. They concluded that the closure of the esophageal groove is trigged when the calf directly suckled the milk from a dam. Larry \[[@CR10]\] stated that the esophageal groove closure depends upon the liquid ingested which stimulates the nerve receptors in the mouth. Other studies suggest that esophageal groove closure and dilatation of the omaso-abomasal canal is initiated by the stimulation of the vagus nerve through contact with sensory receptors in the oral cavity and pharyngeal area \[[@CR11]\]. Gradually (after a few weeks of weaning), this response fades so that the groove is no longer functional. Dysfunction of the esophageal groove results in leakage of fluid into the forestomach. Spillage from the esophageal groove may result from either a complete failure of groove closure or sequential opening and closure during drinking. According to Gentile \[[@CR12]\] pathological conditions (diarrhea, phlebitis of jugular vein, cough, otitis and anorexia), irregular feeding (irregular feeding times, forceful feeding, bucket feeding of milk, abnormal milk temperature) and stress factors (long distance transportation) are some of the causes of esophageal groove dysfunction. However, many studies \[[@CR3], [@CR8], [@CR11]\] indicate that esophageal groove dysfunction is unusual in calves that suckle directly from the dam. This report presents a single clinical case of ruminal bloat associated with a putative esophageal groove dysfunction in a 10-day-old calf. We believe that the therapeutic intervention made in the field was a better management approach in respect to the area's lack of facilities to conduct a laboratory investigation. Case presentation {#Sec2} ================= A 10-day-old male crossbred (Frisian x local indigenous) calf presented with a severely distended abdomen (Fig. [1](#Fig1){ref-type="fig"}). Due to the distention the paralumbar fossa, especially on the left, was not visible. The calf was reluctant to suckle from the dam, unable to walk, exhibited rapid and shallow breathing, and had visibly congested mucus membranes. Percussion of the left abdomen revealed a drum-like gaseous sound. On auscultation of the left abdomen, a dull fluid sound was detected. The anamnesis indicated that the calf had been dribbling urine continuously, unable to defecate, or had irregularly voided very little, hard, and pasty feces. General physical examination revealed no esophageal obstruction, but the calf was weak and with an abnormal gait. The calf was suckling its dam twice in a 12 h interval (at morning and evening) and had not started feeding the hay/roughage/concentrate or the calf starter at the moment. The physiological parameters of the calf were as follows: Rectal temperature = 39.8 degree Centigrade (°C), Pulse =175 beats/minute, Respiration =60 breaths/minute.Fig. 1A Photo showing ruminal bloat in 10 days old crossbred calf Differential diagnosis {#Sec3} ---------------------- Abomasal bloat and choke. Treatment approach {#Sec4} ------------------ In order to release trapped gases and check the patency of the esophagus, a flexible stomach tube coated with mineral oil was inserted into the esophagus, and advanced down into the rumen. A fermented watery-like fluid accompanied by some clots of milk and gases was released from the rumen through the stomach tube. Procaine penicillin (Pen Aqueous; Zoetis Canada), 10 ml (ml), 10,000 international unit per milliliter (Iu/ml of solution) mixed with 0.25Liter (L) of mineral oil was administered orally for 3 days, while milk was withheld to reduce the microbial burden and coalescence of gas. An isotonic solution containing 0.9% Sodium Chloride (Jiangsu HFQ Bio-Technology Co., Ltd), 8.4% Sodium Bicarbonate (Vet One, Nova-Tech, Grand Island, USA) and 5% Dextrose in water 1000 ml injection (Addis Pharmaceutical factory) was administered intravenously (IV) at a rate of 100 ml/kilogram (kg) over 3--5 h for 2 days. Before administration of IV fluid, the calf was sedated using Xylazine hydrochloride, 20 mg/ml (xylazine® immunological LTD, Hyderabad, India) intramuscularly (IM). This was administered during every fluid therapy, and the calf was tied up with rope in a lateral recumbent position. The hair around the jugular groove of the neck was clipped and the area was cleaned and disinfected using diluted 70% Ethanol (Addis Pharmaceutical factory). The superficial jugular vein was catheterized using 20 Gage, 0.8 in. butterfly catheter (Unolok, Hindustan syringe, Medical device LTD Faridabad, India) and secured with adhesive tape around the neck. Response to treatment {#Sec5} --------------------- After 3 days of treatment, bloat reoccurred. Treatment was initiated a second time by giving antibiotic pen strep (Pen & Strep^@^~,~ 100 ml, York Vet, USA): 5 ml, (IM), every 24 h (q24hrs) for 2 days while the calf had been fastening. Additional supportive therapy of 40% glucose (100 ml/kg/day IV), isotonic saline solution (10 ml/kg/hrs IV) and a multivitamin (Multivitamin injection 100 ml, Norbrook Laboratories Limited, Ireland), was administered 10 ml IM once at a time (Stat.) during the time that the milk was withheld. After 2 days of treatment, the calf was allowed to suckle milk from the dam; however, the calf exhibited bloat again 5 h after milk consumption. Rumenostomy {#Sec6} ----------- Ruminal fistulation (rumenostomy) was conducted to prevent recurrence according to a procedure described by Turner and Mcilwraith \[[@CR13]\]. Before the surgical procedure milk was withheld from the calf overnight while IV fluids and glucose were administered at the dose rate explained above. The left paralumbar fossa was prepared by shaving the hair and washing skin aseptically using 7.5% povidone-iodine surgical scrub (Povidone-iodine cleansing solution, Wockhard LTD, Mumbai, India) while the calf was standing. A circular area of 6 cm (cm) in diameter just below the transverse process of the lumbar vertebrae was marked and infiltrated with local anesthetic, 2% Lidocaine (Zoetis Canada, Kirkland,Quebec), at the concentration of 20 mg per milliliter (mg/ml). Approximately a 2 cm diameter circular incision was made to remove the skin. After skin removal the abdominal muscles were dissected bluntly to expose the rumen. The rumen was grasped using sponge forceps and pulled to the exterior. The rumen wall was then tacked to the edge of the skin by four horizontal mattress sutures at "quarter hour" positions (12, 3, 6 and 9 o'clock). These sutures acted as stay sutures using a non-absorbable suture (Sofsilk™ 6--0 Black, Medtronic, USA). The rumen wall was incised carefully at one half centimeter from the wound margin/apposing skin. As the contents of the rumen came out during the procedure, we observed a high amount of milk that had entered into the rumen (Fig. [2](#Fig2){ref-type="fig"}).Fig. 2A photo showing a calf with fistulated rumen (taken right after surgery) Post Rumenostomy management {#Sec7} --------------------------- The calf was separated from the herd for 10 days to maintain close observation. Since rumenostomy is considered a clean contaminated surgery, we had to give parenteral antibiotic, penstrep, (Penstrep-400, Metaalweg, 85,804 CG Venray, Netherlands) 5 ml for 4 days, q24hrs IM to reduce the risks of peritonitis. A dexamethasone injection at 2 mg/ml (Sparhawk laboratories Inc., Lemexa, KS66215, USA) was given every eight hours (q8hrs) IM, and also served as an anti-inflammatory agent. Moreover, Deltamethrin 1% (w/v) pour-on ready-for-use formulation (Appropriate Applications Ltd., USA) at a dose rate of 10 ml per 100 kg body weight was used to prevent insect infestation and miyiasis. The surgical wound was examined and monitored every day until closure for any complications such as wound dehiscence or rumen attachment to the skin. Rumen contents leaking out onto the flank area and outer surgical site were cleaned by using antiseptic solution (Chlorhexidine) and clean towels. The rumen was repeatedly flushed through the fistula with 0.5--1 l of warm tap water adjusted to the calf's body temperature. This flushing helped to prevent desiccation and was used for buffering purpose. The calf was allowed to suckle milk from its dam twice a day during the follow-up period. After 10 days post-operation, the calf was provided with some hay and fresh grasses to stimulate rumen function. Bloat resolved by the time the calf started solid feeds and the wound was closed surgically just after a week of feeding grasses and hay. Response to Rumenostomy {#Sec8} ----------------------- The calf was followed for 6 months after the procedure. Shortly after wound closure, the amount of milk that the calf was getting was reduced to encourage the intake of hay and grasses. During this time, bloat did not occur as it had been observed prior to surgery. Although ruminal contents spilled onto the flank post-operatively, this did not appear to upset the calf, and its general condition improved gradually. We recommended that the owner reintroduce the calf with the existing herd 6 months after the surgery, and advised the owner to inform us of any observable complications. We promised the owner that we would visit the calf at one year; however, the owner had sold the calf at 9 months of age to a beef farmer in another area of the country. Discussion and conclusion {#Sec9} ========================= The occurrence of bloat in calves that have started on a hay/grass/concentrate or a calf starter is not a new phenomenon, however, in newborn calves that are only suckling milk, it is unusual \[[@CR14]\]. According to some studies, \[[@CR3], [@CR12]\] ruminal bloat is a secondary consequence to esophageal groove dysfunction in calves at this age. Esophageal groove dysfunction is the major cause of ruminal bloat in newborn calves that are directly suckling its dam, or could be the result of overfeeding concentrate feed \[[@CR15]\]. The later cause of ruminal bloat doesn't seem to be appearing a factor for this case because the calf wasn't turned onto concentrate feed at the moment, and was suckling only the milk from its dam. We were confronted with a paradox justification because in some studies \[[@CR3], [@CR4], [@CR8], [@CR12]\] esophageal groove dysfunction in calves directly suckling their dam is not common, but failure of the esophageal groove can occur when calves drink cold milk, are tube fed, or fed from a bucket \[[@CR16]\]. In normally functioning esophageal grooves, milk should bypass the rumen and reticulum \[[@CR17]\]. The presence of milk in the rumen was confirmed during surgery, when the ruminal contents were observed and a high quantity of fermented milk and milk clots were noted. Abomasal bloat can be a differential diagnosis for this case, however, there were certain things that set ruminal bloat apart from abomasal bloat. For instance, in ruminal bloat abdominal distention is higher on the left side \[[@CR3]\], which was seen in this case. In addition, during ruminal bloat rumen contents can easily be released out with the help of a stomach tube whereas, in abdominal bloat it is difficult to introduce a stomach tube into the abomasum and flush its contents out while the animal is in a standing position \[[@CR2]\]. When we manipulated the stomach tube while the calf was standing, we were confident that the content was coming out from the forestomach. In the case of choke, bloat must accompany drooling of saliva \[[@CR3]\] and recurrences should not have occurred after checking the patency of esophagus using the stomach tube. The largest problem that we have failed to demonstrate was the underlying cause or factor for esophageal groove failure. Gentile \[[@CR12]\] reported that pathological conditions (diarrhea, otitis, phlebitis, vagus nerve problem, etc.), inadequate feeding technique (irregular feeding time, bucket feeding, very cold milk feeding etc) and stress are some of the causes of esophageal groove dysfunction. In our investigation, the calf wasn't exhibiting diarrhea or any other gross pathological conditions except abdominal distention. Furthermore, we also investigated the feeding technique of the calf and realized that suckling was the only feeding technique and it was regular, twice everyday (12 h interval of milk feed per day). Bloat in older animals is associated with grazing legumes in legume-dominant pastures, feeding high-grain diets, and impaired eructation processes \[[@CR18]\]. Despite the primary cause of bloat being multifactorial, it's clear that the esophageal groove is not functional in those animals as it regresses when they start solid feeds \[[@CR19]\]. Hence, "esophageal groove dysfunction" cannot be an ideal term to use to describe bloat in older animals. Apart from other treatment protocols, several scholars \[[@CR20]--[@CR22]\] suggest that rumenostomy is a therapeutic option for animals with recurrent or non-resolving bloat in young or older animals. Amanda et al. 2015 \[[@CR23]\] mentioned that of 42 rumenostomy treated cases, 20 cases were indicated for bloat. According to the authors, half of the calves were followed for long periods in the herd and they had better health conditions until they were culled. While the primary associated factor of esophageal groove failure is unclear, the presence of milk clot and fermented fluid in high amounts in the rumen at an early age suggests a malfunction in the normal physiology of the esophageal groove. To the best of our knowledge, case like this has never been reported so far in naturally suckling calves. As treatment intervention, withholding milk whilst giving IV fluid would give temporary relief. However, considering the fact that milk is a natural feed that should not be taken away from every calf at this age, we rather encourage calves to consume milk as it would not results in bloat as far as rumen fistula is being created. Furthermore, encouraging calves to consume starter feed (fresh grasses and hay) earlier than usual recommended period whilst decreasing milk intake would hasten the rumen function in those calves. Therefore we concluded that esophageal groove dysfunction should be suspected when severe and recurrent bloat occurs in calves that consume only milk by suckling. Nevertheless, since we did not investigate the underlying cause, detailed study on the primary causes of esophageal groove dysfunction in young calves should be encouraged. We also found that rumenostomy is a better management option over conservative approaches in similar clinical cases. Despite rumenostomy considered a better option, it degrades the appearance and the value of the animal, and we suggest additional studies on alternative treatment methods. Im : Intramuscular Iu : International unit Iv : Intravenous q24hrs : every 24 hours q8hrs : every 8 hours Stat. : Once at a time W/V : weight per volume We would like to thank Slagel Clare, C. for devoting her time in editing language and technical aspects. T Kassa has recorded all the information of case history, diagnosis of the case and treatment intervention, BA involved post operative management and follow up of the case; T Kaba compiled data for write up of the manuscript. All authors read and approved the final manuscript. Ethics approval and consent to participate {#FPar1} ========================================== Not applicable. Consent for publication {#FPar2} ======================= The owner gave informed consent for publication. Competing interests {#FPar3} =================== The authors declare that they have no competing interests. Publisher's Note {#FPar4} ================ Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.	{ "pile_set_name": "PubMed Central" }
Introduction {#Sec1} ============ Radish (Raphanus sativus L., 2n = 2× = 18), one of the most important worldwide root vegetable crops, is an annual or biennial herb belonging to Brassicaceae family. The fleshy taproot is the significant edible part in radish plants, which contains abundant nutrient substances such as carbohydrates, crude fiber, vitamin C, protein, and other secondary metabolites including glucosinolate, and it determines the final yield and quality^[@CR1]^. In addition, it is of significant value on diet and medicine^[@CR2]--[@CR5]^. Therefore, it is urgent to clarify the molecular mechanism underlying taproot thickening in radish plants. In the past years, the morphological, physiological, and anatomical characterization of taproot thickening have been extensively studied on radish^[@CR6],[@CR7]^. The morphogenetic process displays dynamic changes in the period of taproot thickening, which is determined by the interactions of genetic, environmental, and physiological factors^[@CR8],[@CR9]^. Essentially, fleshy taproot development is the result of related genes programmed expression^[@CR8],[@CR9]^. In recent years, with the rapidly developed "omics" technology, the draft genome sequences and transcriptome studies of R. sativus have been reported, which provide a valuable database for identification of the critical genes and genetic manipulation in radish^[@CR10]--[@CR13]^. Using RNA-Seq technique, characterization of transcriptome and miRNA may dissect the molecular mechanism underlying taproot thickening, with several miRNAs and differentially expressed genes (DEGs) identified during three different stages of taproot thickening (pre-cortex splitting stage, cortex splitting stage, and expanding stage)^[@CR8],[@CR14]^. However, the molecular mechanism underlying taproot thickening in radish has not been comprehensively uncovered at the proteome level. Proteomics studies provide a powerful tool for exploring related proteins in specific tissues with different stages in the post-genomic era, therefore comparative proteomic analysis for identifying proteins involved in radish taproot thickening would be an indispensable way for complementing genomics analysis to further explore molecular mechanism governing radish taproot thickening. Currently, the two-dimensional gel electrophoresis (2-DE) in combination with mass spectrometry (MS) has been employed for proteomic studies providing useful tools for protein separation and quantification^[@CR15]--[@CR18]^. However, it has some limitations in protein abundance identification. More recently, isobaric tags for relative and absolute quantitation (iTRAQ), a new and powerful technology, was used in quantitative proteomics by isotope tagging and high-performance liquid chromatography (HPLC)^[@CR19],[@CR20]^. High sensitivity (less than 1 ppm), low detection limit (proteins of less than 10 kD or greater than 200 kD) in identification and quantitation of proteins aspects make iTRAQ a valuable approach for proteomics studies in some plant species including cassava^[@CR21]^, potato^[@CR22],[@CR23]^, and Brassica napus^[@CR24]^. Previous studies reported that cortex splitting stage is a sign of transition from primary growth to secondary growth, and it is of significance to investigate the expression changes of proteins during the pre-cortex splitting stage, cortex splitting stage, and expanding stage of taproot thickening in radish^[@CR9],[@CR11]^. In this study, comparative proteomic analysis of monitoring differentially expressed proteins (DEPs) from three libraries of pre-cortex splitting stage (S1, 10 days after sowing (DAS)), cortex splitting stage (S2, 20 DAS), and expanding stage (S3, 40 DAS) were conducted by iTRAQ-coupled LC-MS/MS. Furthermore, association analysis between DEGs and DEPs in three development stages of taproot thickening was performed for further screening of critical genes involved in taproot formation in radish. The outcomes of this study could be beneficial for further dissection of the molecular mechanism governing radish taproot thickening, and provide a fundamental basis for genetic improvement of taproot formation in root vegetable crops. Materials and methods {#Sec2} ===================== Plant materials {#Sec3} --------------- Seeds of an advanced inbred line of radish (R. sativus L.) "NAU-YH" were germinated on moist filter paper in dark at 25 °C for 3 days, and then cultured in plastic pots with 16 h light (25 °C) and 8 h dark (18 °C) for the rest of the experimental period. At pre-cortex splitting stage (S1, 10 DAS), cortex splitting stage (S2, 20 DAS), and expanding stage (S3, 40 DAS), taproots were sampled, respectively, and three biological replicates were collected for each stage. Equal amounts of taproot samples from three independent biological replicates of each stage were pooled and immediately frozen in liquid nitrogen, then stored at −80 °C for proteomic analysis. Protein preparation {#Sec4} ------------------- The taproot samples were ground in liquid nitrogen to a fine powder and extracted in lysis buffer, then 1 mM phenylmethyl sulfonyl fluoride (PMSF), 2 mM ethylene diamine tetraacetic acid (EDTA), and 10 mM dithiothreitol (DTT) were added. The samples were sonicated for 15 min and centrifuged at 25,000×g for 20 min. The supernatant was mixed well with 5× volume of chilled acetone and incubated at −20 °C for 2 h. After centrifugation at 16,000×g for 20 min, the supernatant was discarded and then repeated previous processes once. The protein concentrations were measured with the Bradford assay^[@CR25]^, and 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was used for separating proteins and checking the proteins. iTRAQ labeling and SCX fractionation {#Sec5} ------------------------------------ The extracted total protein from each sample solution (100 μg) was digested with Trypsin Gold with a ratio of protein:trypsin = 20:1 (w/w) at 37 °C for 4 h, and then Trypsin Gold (protein:trypsin = 20:1) added once more at 37 °C for 8 h. After trypsin digestion, peptides were dried by vacuum centrifugation and reconstituted in 0.5 M tetraethyl-ammonium bromide (TEAB) with the next steps following the manufacturer's procedure of 8-plex iTRAQ reagent (Applied Biosystems). Briefly, 1 U of iTRAQ reagent was thawed and reconstituted in 70 μL isopropanol. Samples from pre-cortex splitting stage S1, cortex splitting stage S2, and expanding stage S3 were labeled with isobaric tags 116, 117, and 118, respectively, and incubated at room temperature for 2 h, and then pooled and dried by vacuum centrifugation^[@CR26]--[@CR28]^. Then the labeled peptides were separated by SCX chromatography using separation column of 4.6 × 250 mm (Ultremex SCX column) with LC-20AB liquid system. In detail, the mixed labeled peptides were reconstituted in 4 mL buffer A (25 mM NaH~2~PO~4~ in 25% ACN, pH 2.7), and gradient-eluted in buffer B at a flow rate of 1 mL/min in column using 5% buffer B (25 mM NaH~2~PO~4~, 1 M KCl in 25% ACN, pH 2.7), 7 min; 5--60% in 20 min; 60--100% and maintenance for 1 min; 5% maintenance for 10 min^[@CR29]^. The eluted process was supervised under 214 nm absorption photometry. The columns were cleaned to eliminate salts using StrataX. The extracted liquid was lyophilized and stored at −80 °C. LC-ESI-MS/MS analysis {#Sec6} --------------------- The peptide was dissolved in buffer C (5% ACN, 0.1% FA) and centrifuged at 20,000×g for 10 min. The supernatant (approximately 10 μL) was loaded on a trap column and then the peptides were eluted onto an analytical column by LC-20AD nanoHPLC (Shimadzu, Kyoto, Japan) according to the manufacturer's instruction. In detail, the peptides were loaded at 8 μL/min for 4 min, then the gradient program was performed as follows: it was started from 2 to 35% in buffer D (95% ACN, 0.1% FA) at 300 μL/min, followed by a 5 min linear gradient to 60%, then a 2 min linear gradient to 80% and maintenance for 4 min, and finally return to 5% in 1 min^[@CR29]^. Data were acquired by a TripleTOF 5600 System. The MS was operated with more than 30,000 resolutions. A sweeping collision energy was set for 35 ± 5 eV with a dynamic exclusion setting of 1/2 of peak width (15 s), ensuring that the same precursor ion was fragmented not more than twice. Protein identification {#Sec7} ---------------------- For peptide data analysis, raw data was employed with the Mascot search engine (Matrix Science, London, UK; version 2.3.02) against the Raphanus_sativus (224,406 sequences) protein sequence database (ftp://ftp.kazusa.or.jp/pub/radish/). For protein identification, a series of standard parameters were set as follows: 0.1 Da fragmented mass tolerance, 0.05 Da peptide mass tolerances and one max missed cleavage. To reduce the probability of false peptide identification, only peptides at the 95% probability were used for protein identification by a Mascot probability analysis, and each confident protein contains at least one unique peptide^[@CR28]^. For protein quantitation, protein has at least two unique spectra, and the parameter "median" was considered as the standard of quantitative protein ratios^[@CR26],[@CR30]^. The identification results of significant DEPs were filtered with P-value \< 0.05 and fold change (FC) \> 1.2^[@CR31]^. Bioinformatics analysis {#Sec8} ----------------------- A hierarchical cluster analysis of the DEPs was carried out using Cluster 3.0 software. The similarity of proteins was calculated with Euclidean distance, and the average-linkage method was selected for clustering^[@CR32]^. Functional annotations of the DEPs were employed by KOBAS 2.0 program search against the non-redundant (nr) protein database deposited in NCBI. The database of Kyoto Encyclopedia of Genes and Genomes (KEGG), Gene Ontology (GO), and Cluster of Orthologous Groups (COG) were performed to categorize and group the DEPs^[@CR31],[@CR33]^. RT-qPCR analysis {#Sec9} ---------------- Quantitative real-time PCR (RT-qPCR) was employed to validate the quality of iTRAQ results. Total RNAs from taproot samples including S1, S2, and S3 stages was extracted and reverse transcribed to cDNA following the manufacturer's instructions (Tiangen Biotech Co., Ltd., China). Each reaction was carried out using 10 μL 2× SYBR green reaction mix, 2.0 μL diluted cDNA, and 0.2 μM of each primer in a total volume of 20 μL system. RT-qPCR amplification reactions were conducted on a Real-Time PCR Detection System (Bio-Rad iQ5, USA) following the reported protocol^[@CR34]^. The specific primers used for RT-qPCR were designed with Beacon Designer 7.0 software (Premier Bio-soft International, USA), which are listed in Supplementary Table [S1](#MOESM1){ref-type="media"}. Three replicates for each gene assay were performed and mRNA expression levels were normalized by Rs-Actin gene. The relative gene expression value was calculated with the 2^−∆∆CT^ method. Results {#Sec10} ======= Protein identification {#Sec11} ---------------------- Based on LC-ESI-MS/MS analysis, a total of 411,811 spectra were generated in this study, which included 95,124 matched spectra and 46,912 unique spectra. Totally 27,080 peptides containing 16,722 unique peptides as well as 6342 proteins were identified. Moreover, protein mass was dominantly enriched in 20--60 kDa, and the peptide number and peptide length were mainly distributed at the range of 1--10 and 6--29 (in amino acids), respectively (Supplemental Figure [S1](#MOESM9){ref-type="media"}). Functional annotation of proteins {#Sec12} --------------------------------- GO function classification showed that the proteins were mainly annotated to the terms of cellular process (14.77%, in biological process ontology), cell part (22.82%, in cell component ontology), and binding (43.40%, in molecular function ontology) (Fig. [1](#Fig1){ref-type="fig"}a--c).Fig. 1GO and KEGG pathway functional classification of the identified proteins.a The cell component category of GO classification. b The molecular function category of GO classification. c The biological function category of GO classification. d KEGG pathway functional classification and annotation KEGG pathway analysis revealed two dominant pathways: "Metabolic pathways" (ko01100) and "Biosynthesis of secondary metabolites" (ko01110). In addition, "Starch and sucrose metabolism" (ko00500) pathway was the seventh top of all, which was one of the most important pathways involved in taproot thickening (Fig. [1](#Fig1){ref-type="fig"}d)^[@CR13]^. The COG database was employed for protein orthologous classification, and a total of 5260 proteins were aligned to 24 COG terms. These results showed that the category of "general functions prediction only" (17.60%) was the largest, followed by "Posttranslational modification, protein turnover, chaperones" (12.36%), "Translation, ribosomal structure and biogenesis" (9.13%), "Carbohydrate transport and metabolism" (7.89%), and "Energy production and conversion" (7.13%). In contrast, the categories of "Cell motility" and "Nuclear structure" were less than five proteins with high homology (Supplemental Figure [S2](#MOESM10){ref-type="media"}). Screening of differentially expressed proteins (DEPs) {#Sec13} ----------------------------------------------------- Base on the expression level of proteins, a FC value \>1.2 or less than 0.83 with P-value \< 0.05 were used as thresholds to judge DEPs. Using these standards, a total of 1222, 1046, and 940 DEPs were detected from S1 vs. S2, S1 vs. S3, and S2 vs. S3 pairs, respectively. Among these, totally 528, 459, and 515 proteins were up-regulated, and 694, 587, and 425 proteins were down-regulated in S1 vs. S2, S1 vs. S3, and S2 vs. S3 pairs, respectively (Fig. [2a, b](#Fig2){ref-type="fig"}; Supplementary Table [S2](#MOESM2){ref-type="media"}). A total of 282 DEPs including 26 up-regulated proteins were presented among three comparison pairs (Fig. [2b](#Fig2){ref-type="fig"}), while 270, 107, and 391 DEPs were specifically expressed in up-regulated pairs of S1 vs. S2, S1 vs. S3, and S2 vs. S3, respectively (Fig. [2c](#Fig2){ref-type="fig"}). These results suggested that the specifically expressed proteins corresponding to S1 vs. S2, S1 vs. S3, and S2 vs. S3 pairs played certain roles on cortex splitting, primary expanding and secondary expanding in radish.Fig. 2The characteristic of DEPs distribution in three libraries.a The number of DEPs in any two different stages. b Venn diagrams of DEPs mutual relationship among three libraries. c Venn diagrams of up-regulated proteins mutual relationship among three libraries To further identify the DEPs involved in taproot thickening in radish, two comparison pairs were overlapped. For instance, we overlapped between S1 vs. S2 and S1 vs. S3 pairs for identifying proteins related to taproot thickening initiation, and overlapped between S1 vs. S2 and S2 vs. S3 pairs for identifying proteins related to taproot thickening process. More interestingly, totally 256 up-regulated proteins including 77 unknown proteins displayed overlapped accumulation in S1 vs. S2 and S1 vs. S3 pairs, whereas 122 up-regulated proteins including 43 unknown proteins displayed overlapped accumulation in S1 vs. S3 and S2 vs. S3 pairs (Supplementary Table [S3](#MOESM3){ref-type="media"}). Our findings suggest that these proteins may play critical roles in taproot thickening initiation and taproot formation in radish. GO and pathway enrichment of DEPs {#Sec14} --------------------------------- We used GO and pathway enrichment analysis of DEPs to determine significantly enriched DEPs compared with the background of all DEPs at P-value \< 0.05 as the threshold value, and thereby explored the main biological functions, biochemistry metabolism, and signal transduction pathways. In the S1 vs. S2 comparison, GO enrichment analysis showed that 1154, 1005, and 1090 DEPs were assigned to 271, 582, and 1708 GO terms of cell component, molecular function, and biology process, respectively. Among these, 83, 110, and 377 GO terms were significantly enriched in cell component, molecular function, and biology process, respectively (Supplementary Table [S4](#MOESM4){ref-type="media"}). In the S1 vs. S3 comparison, 993, 864, and 944 DEPs were annotated to 254, 549, and 1689 GO terms of cell component, molecular function, and biology process, respectively. Of these, 76, 125, and 408 GO terms were significantly enriched in cell component, molecular function, and biology process, respectively (Supplementary Table [S4](#MOESM4){ref-type="media"}). In the S2 vs. S3 comparison, 882, 742, and 833 DEPs were assigned to 250, 513, and 1601 GO terms of cell component, molecular function, and biology process, respectively. Of all, 65, 76, and 219 GO terms were significantly enriched in cell component, molecular function, and biology process, respectively (Supplementary Table [S4](#MOESM4){ref-type="media"}). In addition, the shared significantly enriched GO terms of three libraries (Cluster frequency ≥50%) were calculated, and cytoplasm, cytoplasmic part, catalytic activity, metabolic process, cellular metabolic process, response to stimulus, and single-organism metabolic process were obtained. Pathway enrichment analysis showed that 26, 24, and 14 pathways were significantly enriched in the S1 vs. S2, S1 vs. S3, and S2 vs. S3 comparison, respectively (Supplementary Table [S5](#MOESM5){ref-type="media"}). The significantly enriched pathways of three libraries were "Metabolic pathways" (ko01100), "Glycolysis/Gluconeogenesis" (ko00010), "Biosynthesis of secondary metabolites" (ko01110), "Cysteine and methionine metabolism" (ko00270), "Glutathione metabolism" (ko00480), "Phenylalanine metabolism" (ko00360), and "Ribosome" (ko03010) (Table [1](#Tab1){ref-type="table"}). In addition, several DEPs were identified to be involved in "Plant hormone signal transduction" (ko04075) and "Starch and sucrose metabolism" (ko00500) (Supplementary Table [S6](#MOESM6){ref-type="media"}).Table 1The significant enrichment pathways for DEPs in three taproot thickening stages in radishTermsP-ValuePathway IDS1 vs. S2S1 vs. S3S2 vs. S3Specifically enrichment in S1 vs. S2 pair Tyrosine metabolism0.002338993----ko00350 Pyruvate metabolism0.003318677----ko00640 Isoquinoline alkaloid biosynthesis0.005839398----ko00950 Tropane, piperidine, and pyridine alkaloid biosynthesis0.02102198----ko00960 Phenylalanine, tyrosine, and tryptophan biosynthesis0.02465798----ko00400 Fatty acid metabolism0.02649325----ko00071 beta-Alanine metabolism0.03322849----ko00410 Phagosome0.04160965----ko04145Specifically enrichment in S1 vs. S3 pair Photosynthesis--2.23E−06--ko00195 Phenylpropanoid biosynthesis--0.001141693--ko00940 Arachidonic acid metabolism--0.00592627--ko00590 Linoleic acid metabolism--0.01070429--ko00591 Oxidative phosphorylation--0.01803258--ko00190 Citrate cycle (TCA cycle)--0.04408532--ko00020Specifically enrichment in S2 vs. S3 pair Amino sugar and nucleotide sugar metabolism----0.007055244ko00520 Glycosphingolipid biosynthesis----0.02370965ko00603 Ether lipid metabolism----0.03840109ko00565S1 vs. S2 and S1 vs. S3 pair share Tryptophan metabolism0.0031749370.003847518--ko00380 Fructose and mannose metabolism0.0035460470.009507623--ko00051 Ubiquinone and other terpenoid--quinone biosynthesis0.0087227740.002321026--ko00130 Nitrogen metabolism0.01111430.02068339--ko00910 Glyoxylate and dicarboxylate metabolism0.013152270.00145352--ko00630 Pentose phosphate pathway0.02362230.04274795--ko00030 Glucosinolate biosynthesis0.030623380.03704522--ko00966 Carbon fixation in photosynthetic organisms0.031350753.75E−05--ko00710 Arginine and proline metabolism0.036175070.01826023--ko00330S1 vs. S2 and S2 vs. S3 pair share Propanoate metabolism0.003318677--0.0138437ko00640 Alanine, aspartate, and glutamate metabolism0.02117078--0.003866935ko00250S1 vs. S3 and S2 vs. S3 pair share Cyanoamino acid metabolism--1.69E−050.03116905ko00460 alpha-Linolenic acid metabolism--0.0006147910.009060137ko00592S1 vs. S2, S1 vs. S3, and S2 vs. S3 share Metabolic pathways6.47E−066.29E−140.000324501ko01100 Glycolysis/gluconeogenesis6.56E−060.0009681210.006929212ko00010 Biosynthesis of secondary metabolites9.98E−063.34E−060.001973937ko01110 Cysteine and methionine metabolism0.0001919970.0026534020.000150698ko00270 Ribosome0.0074525150.034169240.03002672ko03010 Glutathione metabolism0.0088764450.00010460.003305723ko00480 Phenylalanine metabolism0.036175070.0044211690.03331485ko00360 Associated analysis of mRNA and proteins during taproot thickening in radish {#Sec15} ---------------------------------------------------------------------------- Proteome and transcriptome analysis showed two difference levels, which reflect genes expression. In order to mutual corroborate the data reliability, association analysis of DEPs data with previous transcriptome data was performed in this study (Supplementary Table [S7](#MOESM7){ref-type="media"})^[@CR14]^. The result showed that 187, 181, and 96 DEPs were successfully matched with DEGs in the pairs of S1 vs. S2, S1 vs. S3, and S2 vs. S3, respectively (Fig. [3a--c](#Fig3){ref-type="fig"}; Supplementary Table [S8](#MOESM8){ref-type="media"}), and the corresponding Spearman correlation coefficient for proteome and transcriptome (R, Spearman) were 0.3050, 0.4009, and 0.1371, respectively (Fig. [3d--f](#Fig3){ref-type="fig"}). These DEPs/DEGs results could be categorized into four groups: (i) the expression patterns of DEPs and DEGs with both up-regulated (up-DEPs & up-DEGs; 47, S1 vs. S2; 43, S1 vs. S3; 35, S2 vs. S3); (ii) the expression patterns of DEPs and DEGs with both down-regulated (down-DEPs & down-DEGs; 74, S1 vs. S2; 78, S1 vs. S3; 23, S2 vs. S3); (iii) the expression patterns of DEPs and DEGs with the opposite, i.e., either up-regulated DEPs and down-regulated DEGs (up-DEPs & down-DEGs; 19, S1 vs. S2; 28, S1 vs. S3; 26, S2 vs. S3), or down-regulated DEPs and up-regulated DEGs (down-DEPs & up-DEGs; 47, S1 vs. S2; 32, S1 vs. S3; 12, S2 vs. S3). Interestingly, although the Spearman correlation coefficient between DEPs and DEGs was weak (0.3050, S1 vs. S2; 0.4009, S1 vs. S3; 0.1371, S2 vs. S3), the same expression trends of DEPs and DEGs were relatively strong among the association results (0.7269, S1 vs. S2; 0.7710, S1 vs. S3; 0.7916, S2 vs. S3) (Fig. [3g--i](#Fig3){ref-type="fig"}). The result may suggest that gene regulation mechanisms were different in mRNA and protein level, and the associated proteins/genes were critical for exploring the molecular mechanism of taproot thickening in radish. Furthermore, the cluster analysis was performed to identify the characteristic of correlation between transcriptome and proteome (Fig. [4](#Fig4){ref-type="fig"}).Fig. 3Association results of proteome and transcriptome in three libraries.a--c Venn diagram of associated DEPs/DEGs. d--f R value of DEPs and DEGs association. g--i R value of the same trends DEPs/DEGs. a, d, g S1 vs. S2 pair. b, e, h S1 vs. S3 pair. c, f, i S2 vs. S3 pair. R is the Pearson correlation coefficientFig. 4Cluster diagram for the concordance between proteome and transcriptome based on pairwise comparison among the three libraries.a S1 vs. S2 pair. b S1 vs. S3 pair. c S2 vs. S3 pair RT-qPCR validation {#Sec16} ------------------ To evaluate the validity of iTRAQ results, 17, 11, and 9 proteins from three pairwise comparisons (S1 vs. S2, S1 vs. S3, and S2 vs. S3) were randomly selected and detected with RT-qPCR analysis. The related transcript of these proteins include signaling transduction (MAPK3, MAPK5, SUR1, AUR, JA, SNX1, LOX1), cell division and expanding (CDC5, EXPB1), sucrose biosynthesis and metabolism (PDC2, BG1, PPC, INV, SEX4, PGDH), amino acid biosynthesis and metabolism (GLN, ASP, GAD2, GLS1, ASP), HSP (HSC70), FP6 and CYP79F1 (Supplementary Table [S1](#MOESM1){ref-type="media"}). As shown in Fig. [5](#Fig5){ref-type="fig"}, the transcript levels of BG1, PPC, INV, CYP79F1, LOX1, GLS1, GLN, PGDH, and EXPB1 were higher in the pre-cortex splitting stage than in the cortex splitting stage and expanding stage, whereas SNX1 and PDC2 were higher in the cortex splitting stage and expanding stage than in the pre-cortex splitting stage at mRNA levels. Interestingly, the transcript levels of FP6, SEX4, HSC70, and GLN were higher in the expanding stage than in the pre-cortex splitting stage and cortex splitting stage, while the transcript of CDC5 was higher in cortex splitting stage than in the expanding stage. Furthermore, the mRNA levels of signaling transduction proteins including MAPK3, MAPK5, and AUR were higher in the pre-cortex splitting stage than in the cortex splitting stage. Overall, it was found that there was a good concordance between iTRAQ and mRNA expression levels (Fig. [5](#Fig5){ref-type="fig"}), indicating the reliability of iTRAQ-based quantitative proteomic data analysis.Fig. 5Validation of the mRNA expression levels of proteins from iTRAQ results by RT-qPCR.a S1 vs. S2 pair. b S1 vs. S3 pair. c S2 vs. S3 pair Discussion {#Sec17} ========== As an important storage organ, radish taproot thickening determines its final yield and quality of the product. miRNA and mRNA of radish taproot thickening at pre-cortex splitting stage (S1), cortex splitting stage (S2), and expanding stage (S3) have been studied, which provided novel insights into the genetic regulatory network during taproot thickening^[@CR9],[@CR14]^. However, the molecular mechanism underlying taproot thickening in radish is far from being fully clarified. In this study, iTRAQ-based proteomic approach was employed to monitor DEPs from three different stages of radish taproot thickening (S1, S2, and S3). Furthermore, based on the integrative analysis of transcriptomic and proteomic data, an overview of proteome changes involved in radish taproot thickening at three developmental stages was put forward (Fig. [6](#Fig6){ref-type="fig"}). To the best of our knowledge, this is the first comprehensive investigation to characterize the potential functional proteins involved in taproot thickening in radish.Fig. 6An overview of proteome changes involved in radish taproot thickening at three developmental stages.The blue font represents overlap proteins between DEPs and DEGs, and the normal font represents unique proteins in this study. MAPK4 mitogen-activated protein kinase 4, CAM putative calmodulin, SRP54 signal recognition particle subunit, PP2C58 probable protein phosphatase 2C 58, SPP1 sucrose-phosphatase 1, PFK3 6-phosphofructokinase 3, PFK5 6-phosphofructokinase 5, bglL beta-glucosidase L, UGD1 UDP-glucose 1-dehydrogenase, SUS1 sucrose synthase 1, PME51 phosphoglucomutase 1, bZIP59 bZIP transcription factor 59, EF1BB elongation factor 1B alpha-subunit 2, PP2C59 protein phosphatase 2C 59, Gns7 glucan endo-1,3-beta-glucosidase 7, MAPK6 mitogen-activated protein kinase 6, BSK8 brassionsteroid-signaling kinase 8, BSL1 serine/threonine-protein phosphatase, EXPL2 expansin-like A2, EXPB1 expansion B1, XTH24/XTH32 xyloglucan endotransglucosylase 24/32, FRUCT5 beta-fructofuranosidase, bglH beta-glucosidase H, PGM1 phosphoglucomutase 1, APL3 glucose-1-phosphate adenylyltransferase, bglI beta-glucosidase I, CDC5 cell division cycle 5-like protein Proteomic studies could be a supplement to transcriptome analysis {#Sec18} ----------------------------------------------------------------- In this study, iTRAQ-based proteomic approach was employed to identify DEPs from three different stages of radish taproot thickening (S1, S2, and S3), as a follow-up to previous study^[@CR9],[@CR14]^, which could be beneficial to further dissecting the genetic molecular mechanism underlying radish taproot thickening. Furthermore, for further screening critical genes involved in taproot formation in radish, association analysis between DEGs and DEPs in three development stages of taproot thickening was also performed in this study. The relatively weak correlation of genes at mRNA and protein level indicates that an alteration in a proportion of transcripts may not be translated into changes in protein abundance. Low congruency of genes in mRNA and protein level was similar with other previous studies^[@CR20],[@CR35]^. As shown in Fig. [6](#Fig6){ref-type="fig"}, the blue font indicated the proteins with the same expression patterns in mRNA and protein level, whereas the normal font represented unique proteins. For example, proteins involved in cell wall loosening and reconstruction including XTH24 and XTH32 and proteins involved in signal transduction such as PP2C58, BSK8, and BSL1 were unique proteins that were only identified in this study but not gathered in transcriptome analysis. The results showed that the transcriptome and proteome levels were various under taproot thickening in radish, and it could be presumed that proteomic studies was used as an effective supplement to transcriptome analysis. Phytohormone is a basic regulatory factor for taproot thickening {#Sec19} ---------------------------------------------------------------- Phytohormone is critical for the growth modulation of the root system^[@CR36]^. Five main hormones including auxin, cytokinin, ethylene, abscisic acid (ABA), and gibberellic acid (GA) were extensively investigated in plant root system. Auxin-to-cytokinins ratio regulates root and shoot meristem. In general, the high ratio is favorable for the root meristem and control of root architecture, while the lower ratio is conducive to shoot meristem. The auxin influx carriers including AUX1, PGP, LAX, and auxin efflux transporters such as ABC transporters affect the distribution of auxin and regulate root morphogenesis^[@CR37]^. Previous studies showed that changing the level of auxin influences cytokinin concentration, which functioned by the mechanism of relevant proteins controlling hormone accumulation^[@CR38]^. Several other hormones including ethylene, ABA, and GA are also responsible for root development. For example, ethylene inhibits adventitious root formation in Arabidopsis^[@CR39]^, ABA could modulate main root elongation in response to drought, and GA could enhance cell elongation of root. In radish plants, cytokinin plays critical role in controlling developmental process of taproot thickening^[@CR7]^. In this study, "plant hormone signal transduction" is one of the most enriched pathways, and several DEPs including Gns7 (glucan endo-1,3-beta-glucosidase 7, Rsa1.0_00048.1_g00023.1), AHP2 (histidine-containing phosphotransfer protein 2, Rsa1.0_00568.1_g00001.1), PYR1 (abscisic acid receptor, CL14734.Contig1_CKA, CL3244.Contig2_NAU-LB), and PP2C58 (protein phosphatase 2C 58, CL6992.Contig1_NAU-LB) were up-regulated in this process, suggesting that these proteins would be likely involved in the regulation of radish taproot thickening (Supplementary Table [S6](#MOESM6){ref-type="media"}). However, a system network of hormone interactions in radish taproot thickening needs to be further investigated. Sucrose metabolism provides substance basis for taproot thickening {#Sec20} ------------------------------------------------------------------ Sucrose is not only a product of plants photosynthesis, but also a carbon source involved in other important metabolite compounds synthesis such as starch, cellulose and proteins. In this study, several candidate proteins including cell wall-related proteins (e.g., CSLD2, PME17, PME18, PME34, and PME51) and energy metabolism proteins (e.g., BXL, FRK, PGM, TPS7, PHS, BXL, CSLD2, MA1, SUS1, BFRUCT, BAM, APL, FRUCT5, BGLU, UGD1, HXK1, PGIC, and AGPS1) were identified to be differently expressed in three libraries (Supplementary Table [S6](#MOESM6){ref-type="media"}). Interestingly, PME51 (pectinesterase/pectinesterase inhibitor 51, Rsa1.0_01105.1_g00010.1) and FRK5 (fructokinase 5, CL4819.Contig2_CKA) were up-regulated in S1 vs. S2 pair, FRK1 (fructokinase 1, Rsa1.0_03943.1_g00005.1) was up-regulated both in S1 vs. S2 and S1 vs. S3 pairs, and PHS2 (alpha-glucan phosphorylase 2, CL13665.Contig1_NAU-YH), BAM3 (Beta- amylase 3, Rsa1.0_00524.1_g00011.1), APL2 (Glucose-1-phosphate adenylyltransferase, Rsa1.0_00583.1_g00001.1) and FRUCT5 (Beta-fructofuranosidase, Unigene25061_NAU-YH) were up-regulated in S2 vs. S3 pair, suggesting that these DEPs might be involved in cortex splitting, primary growth, and secondary growth of radish taproot thickening, respectively (Supplementary Table [S6](#MOESM6){ref-type="media"}). In the present study, sucrose synthase 1 (SUS1) was down-regulated in S3 vs. S1 comparison pair, which had the same trend with transcriptomic data of radish with small size taproot (radish type: cherry radish)^[@CR14]^ but exhibited a conflict trend with result from long and thick taproot radish^[@CR11]^. It was inferred that the different expression patterns of SUS1 gene between large- and small-sized radish might lead to sucrose unloading capacity difference, which partially contribute to the generation of different taproot sizes and production. Interestingly, the basic formation mechanism of the root and tuber crops (RTCs) could be interlinked, but the genes and proteins involved in these processes may be diverse. For example, sporamin including sporamin A and sporamin B were specifically identified to be up-regulated from storage roots in sweet potato^[@CR17]^, and patatin proteins (patatin protein 13 and patatin protein 15) were specifically identified from potato^[@CR16]^. The sporamin and patatin protein were not identified from taproot thickening in radish, whereas several proteins including SUS1, EXPL2, XTH32, and CDC5 were specifically identified to be significantly differentially expressed in radish, suggesting that there is certain specificity in different types of RTCs (Fig. [6](#Fig6){ref-type="fig"}; Supplementary Table [S2](#MOESM2){ref-type="media"}). However, carbohydrate metabolism was one of the most important pathways in the development of RTCs, and the proteins involved in this process including fructokinase, phosphoglycerate kinase, NADH dehydrogenase, glyceraldehyde-3-phosphate dehydrogenase, fructose-bisphosphate aldolase, ADP-glucose pyrophosphorylase, glucose-1-phosphate adenylyltransferase, and alcohol dehydrogenase were identified to be differentially expressed both in potato^[@CR16]^ and radish (Supplementary Table [S2](#MOESM2){ref-type="media"}). Meanwhile, several proteins including protein disulfide isomerase and anionic peroxidase were identified to be significantly expressed both in storage root of sweet potato^[@CR17]^ and radish taproot thickening (Supplementary Table [S2](#MOESM2){ref-type="media"}), indicating that the protein expression level in the energy supply was relatively conservative. Therefore, we speculated that the possible reason is the difference in proteins to be involved resulting in diverse accumulation of substances in potato, sweet potato, and radish. Cell division and cell expansion determining taproot thickening {#Sec21} --------------------------------------------------------------- Taproot thickening is a process characterized by changes in substances and energy. During this period, the basic metabolism of the taproot was reduced, and it would be act as a storage bank. A series of proteins are involved in the biosynthesis and metabolism of sugar, starch, and protein, which play critical roles in the initial step of taproot thickening in radish^[@CR13],[@CR40],[@CR41]^. In the present study, cell division cycle 5-like protein (CDC5), a DNA binding protein belonging to a member of MYB3R- and R2R3-type, was up-regulated in S1 vs. S2 pair, consistent with its action in Arabidopsis^[@CR42]--[@CR45]^. Interestingly, CDC5 gene was also identified to be up-regulated in S1 vs. S2 pair^[@CR14]^. Meanwhile, cell division and cell enlargement tend to be along with the cell wall loosening and reconstruction, and expansin (EXP) and xyloglucan endotransglucosylase/hydrolase (XTH) proteins played critical roles in these processes. Previous studies showed that down-regulation of EXPB1 gene enhanced storage root development in sweet potato^[@CR46]^, and XTH24 gene involved in cell wall elongation in Arabidopsis^[@CR47]^. In this study, the expansin proteins including EXPB1 (expansin B1) and EXLA2 (expansin-like A2) were identified to be differentially expressed, among them EXLA2 was all up-regulated in S1 vs. S2 pair and S1 vs. S3 pair, whereas EXPB1 was down-regulated in S1 vs. S2, all in line with gene expression pattern at mRNA level (Fig. [6](#Fig6){ref-type="fig"}; Table [2](#Tab2){ref-type="table"})^[@CR14]^. Moreover, XTH32 was up-regulated in S2 vs. S3 pair, while XTH24 that involved in GA-mediated signaling pathway, cell wall loosening, organization or biogenesis, and xyloglucan metabolic process was down-regulated in S1 vs. S2 pair and S1 vs. S3 pair (Fig. [6](#Fig6){ref-type="fig"}; Table [2](#Tab2){ref-type="table"}). The results suggested that CDC5, EXPB1, and XTH24 might play critical roles in taproot thickening process in radish.Table 2Identification of critical DEPs involved in taproot thickening in radish![](41438_2018_57_Tab1_HTML.gif){#d29e1667}![](41438_2018_57_Tab2_HTML.gif){#d29e1674} In addition, previous studies illustrated that transcription factors (TFs) including MADS-box, ABF/AREB, and homeobox were also responsible for the formation of modified stems or storage roots^[@CR38]^. In the present study, MYB and bZIP TFs were both up-regulated in S1 vs. S2 pair, suggesting that MYB and bZIP might play critical roles in regulating the development of taproot thickening in radish, which were consistent with their action of rhizome formation in lotus root^[@CR14],[@CR42]^. In conclusion, the comparative proteome changes among three developmental stages of taproot thickening were systematically investigated in radish. A total of 1862 DEPs were identified during radish taproot thickening. GO and pathway enrichment analysis showed that several DEPs were mainly involved in "plant hormone signal transduction", "starch and sucrose metabolism" and "biosynthesis of secondary metabolites". Furthermore, the integrative analysis of DEGs and DEPs data enhanced our understanding of taproot thickening molecular mechanisms in radish. Overall, it was concluded that taproot thickening initiation was triggered by phytohormone, and several functional proteins including CDC5, EXPB1, and XTH24 could contribute to cell division and expansion during radish taproot thickening. Together, these findings would provide fundamental insights into comprehensive clarification of molecular regulatory network of taproot thickening in radish. Electronic supplementary material ================================= {#Sec23} Table S1 Table S2 Table S3 Table S4 Table S5 Table S6 Table S7 Table S8 Figure S1 Figure S2 Electronic supplementary material ================================= Supplementary Information accompanies this paper at (10.1038/s41438-018-0057-7). This work was in part supported by grants from National Key Technology R&D Program of China (2017YFD0101803, 2017YFD0101806), Key Technology R&D Program of Jiangsu Province (BE2016379), and the Jiangsu Agricultural Science and Technology Innovation Fund \[CX (16)1012\], the Fundamental Research Funds for the Central Universities (KJQN201656, KJQN201734), and Project of Key Laboratory of Biology and Genetic Improvement of Horticultural Crops, MOA, China (IVF201706). Conflict of interest {#FPar1} ==================== The authors declare that they have no conflict of interest.	{ "pile_set_name": "PubMed Central" }
1 Background ============ Obesity is increasing to epidemic proportions \[[@B1]\],\[[@B2]\], starting from young ages \[[@B3]\],\[[@B4]\], and is associated with an increase in the prevalence of type-2 diabetes mellitus (T2DM) throughout the world \[[@B2]\],\[[@B5]\]. In recent years, active participation of immune cells in obesity and T2DM has become evident \[[@B6]\],\[[@B7]\]. This chronic activation of the innate immune system can lead to insulin resistance (IR), impaired tolerance to glucose and, eventually, T2DM \[[@B8]\],\[[@B9]\]. Adipose tissue produces a vast array of adipocyte-derived factors ("adipokines") that regulate the metabolism, inflammation and body mass \[[@B10]\],\[[@B11]\]. One of these adipokines, resistin, belongs to a family of cysteine-rich proteins shown to be involved in inflammation and altered insulin sensitivity in rodents \[[@B12]\],\[[@B13]\]. Resistin is produced from adipose tissue in rodents whereas, in humans, it is secreted by the mononuclear lymphocytes and stromal cells within adipose tissue \[[@B14]\],\[[@B15]\]. Studies in humanised resistin mice suggest that, even though the site of resistin production differs between species, human resistin exacerbates inflammation in white adipose tissue and contributes to IR, thereby impairing its normal effects \[[@B16]\],\[[@B17]\]. IR is defined as the genetic or learned inability of target tissues to respond normally to the action of circulating hormones. It has been described in skeletal muscle but also in liver and adipose tissues \[[@B18]\]. Some studies have shown that resistin affects glucose transport and insulin-stimulated oxidation of glucose in L6 skeletal muscle cells \[[@B19]\]-\[[@B22]\]. It also decreases the uptake and oxidation of long-chain fatty acids \[[@B23]\] and glycogen synthase kinase-3-β, as well as insulin-stimulated insulin receptor substrate-1 (IRS-1) tyrosine phosphorylation \[[@B19]\],\[[@B22]\] in the same cell line. Furthermore, high levels of resistin in rats leads to IR involving impaired insulin signalling in skeletal muscle, liver and adipose tissues, resulting in glucose intolerance, hyperinsulinemia and hypertriglyceridemia \[[@B24]\]. These observations link resistin in muscle-cell metabolism to IR. Changes in the metabolome (i.e., the whole set of metabolites) have been described as the "ultimate" response of an organism to various events, such as genetic alterations as well as disease-based or environmental influences \[[@B25]\]. However, the metabolome is dynamic and metabolites are transformed continuously in the cell. Comprehensive characterisation of the metabolic networks and their functional states requires quantitative knowledge of intracellular metabolic fluxes. These intracellular fluxes can be quantified by analysing incorporation of labelled substrates in metabolic products using appropriate bioinformatic tools \[[@B26]\]. Skeletal muscle is considered to be the main tissue involved in the maintenance of glucose homeostasis because its contribution to glucose uptake is ≈ 75% of the total contribution of peripheral tissues \[[@B27]\], and it is the main tissue responsible for insulin-dependent glucose use. Also, pyruvate homeostasis-related fluxes in skeletal muscle play a critical part in glucose homeostasis \[[@B28]\]-\[[@B32]\]. A better understanding of how these fluxes work together in muscle cells is needed to understand this process. The aim of the present study was to examine the effect of resistin on the central carbon metabolic network of rat skeletal muscle cells, and to ascertain how resistin alters the response of muscle cells to insulin at the metabolic level. We used \[1,2-^13^C~2~\]-glucose as a tracer, analysed isotopomer distributions by gas cromatography (GC) coupled with mass spectrometry (MS), and quantified metabolic fluxes using Isodyn software \[[@B26]\],\[[@B33]\]-\[[@B36]\]. In the model of the glucose metabolic network in skeletal muscle we included the glycolytic pathway and pentose phosphate pathway (PPP). Pyruvate dehydrogenase complex (PDC)-, phosphoenolpyruvate carboxykinase (PEPCK)- and pyruvate carboxylase (PC)-catalysed reactions were also included to account for pyruvate homeostasis-related fluxes. Resistin-induced alterations in the profile of metabolic fluxes of rat skeletal muscle cells were identified, and were complemented by analyses of the expression of certain genes. The study presented here offers a wider vision than previously published of the metabolic reorganisation that resistin induces in rat skeletal muscle cells and in the metabolic response of these cells to insulin. Our observations provide new data for a more precise understanding of the effect of resistin on metabolic-network adaptations of skeletal muscle cells, and help to explain the role of resistin in the IR described in myocytes. Our work reveals the importance of tackling the study of complex biological systems from a systems-biology point of view to fully understand multifactorial diseases and identify new therapeutic targets. 2 Results ========= For analyses of the effect of resistin on glucose metabolism and on the metabolic response of muscle cells to insulin, L6E9 myotubes were pretreated for 8 h with or without resistin (100 nM). After preincubation, myotubes were incubated for a further 6 h with glucose (10 mM) that was 50% enriched in \[1,2-^13^C~2~\]-glucose in the absence or presence of 100 nM insulin and in the continued presence or absence of resistin. After 6 h of incubation, glucose consumption, lactate production, intracellular levels of glycogen and glucose-6-phosphate (G6P) were determined (Table [1](#T1){ref-type="table"}), as were the mass-isotopomer distributions in glucose and lactate from the incubation medium, as well as glycogen glucose and ribose isolated from RNA in cell pellets (Table [2](#T2){ref-type="table"}). Analyses of the obtained data by Isodyn determined the distribution of the metabolic flux profile of L6E9 myotubes under various incubation conditions (Table [3](#T3){ref-type="table"}). Figure [1](#F1){ref-type="fig"} summarises the qualitative changes in metabolites and fluxes. ###### Biochemical parameters in L6E9 myotubes after different incubations with resistin and/or insulin Incubation condition ---------------------------------------- -------------------------- ----------------- ----------------- ---------------- Glucose consumption (mM) 0.50 ± 0.04 0.61 ± 0.06\* 0.78 ± 0.03\\ 0.70 ± 0.09 Lactate production (mM) 0.48 ± 0.01 0.56 ± 0.02\\ 0.75 ± 0.06\\ 0.79 ± 0.02^†^ Glycogen (μg glucose/mg prot) 9.88 ± 1.99 9.97 ± 0.98 11.34 ± 0.59 11.62 ± 1.21 Glucose-6-phosphate (nmol/mg prot) 0.92 ± 0.14 1.44 ± 0.28 2.30 ± 0.29\\ 2.10 ± 0.19^†^ L6E9 myotubes were treated (Res+) or not treated (Res--) with 100 nM resistin for 8 h and then incubated for 6 h in the absence (Ins--) or presence (Ins+) of 100 nM insulin. Glucose consumption, lactate production, and intracellular contents of glycogen and glucose-6-phosphate were determined as described in the Methods section. Results are the mean ± standard deviation (n = 3). Units of glucose consumption and lactate production refer to the volume of the incubation medium. (\) indicates p \< 0.05 and (\\) p \< 0.01 compared with the condition of neither resistin nor insulin. (†) indicates p \< 0.05 compared with the L6E9 myotubes treated only with resistin. ###### Mass isotopomer distribution of lactate, glycogen and ribose after incubations with resistin and/or insulin Mass isotopomer distributions (%)* ---------------- --------------------------------------- --------------- ---------------- -------------- *Lactate* m0 88.56 ± 0.98 88.74 ± 0.25 87.68 ± 0.22 87.63 ± 0.17 m1 0.70 ± 0.34 0.48 ± 0.11 0.63 ± 0.06 0.52 ± 0.14 m2 10.53 ± 0.30 10.85 ± 0.15 11.69 ± 0.05\* 11.78 ± 0.11 m3 0.21 ± 0.63 0.07 ± 0.18 0.00 ± 0.11 0.07 ± 0.13 *Glycogen* m0 67.28 ± 0.65 69.20 ± 1.41 65.14 ± 1.08 64.48 ± 0.45 m1 0.29 ± 0.03 0.36 ± 0.14 0.96 ± 0.10 0.84 ± 0.10 m2 31.78 ± 0.74 29.91 ± 1.02 33.20 ± 1.18 33.78 ± 0.76 m3 0.53 ± 0.19 0.43 ± 0.28 0.43 ± 0.07 0.60 ± 0.20 m4 0.00 ± 0.02 0.00 ± 0.06 0.17 ± 0.05 0.15 ± 0.02 *Ribose* m0 96.13 ± 0.55 94.28 ± 0.76 94.47 ± 0.48 92.83 ± 1.18 m1 1.34 ± 0.23 1.45 ± 0.11 2.56 ± 0.14 2.97 ± 0.15 m2 0.68 ± 0.19 1.14 ± 0.16\* 1.18 ± 0.09\* 1.40 ± 0.23 m3 1.17 ± 0.63 1.56 ± 0.58 0.87 ± 0.17 1.14 ± 0.32 m4 0.50 ± 0.44 1.17 ± 0.22\* 0.61 ± 0.14 1.17 ± 0.41 L6E9 myotubes were treated (Res+) or not treated (Res--) with 100 nM resistin for 8 h and then incubated for 6 h with 10 mM glucose, 50%-enriched in \[1,2-^13^C~2~\]-glucose in the absence (Ins--) or presence (Ins+) of 100 nM insulin. Mass isotopomer distributions at the end of incubations were determined, as described in the Methods section. Results are the mean ± standard deviation (n = 3). (\) indicates p \< 0.05 compared with the condition of neither resistin nor insulin. ###### Metabolic fluxes adjusted by Isodyn for different incubation conditions Metabolic fluxes (nmol · mL^-1^ · min^-1^)* -------------------------------------------- -------------------------------- ----------------------------------------------------------------- ------------------ ----------- ------------------ ----------- ------------------ ----------- ------------------ 0 Glucose phosphorylation 1.500 \[1.500--1.500\] 1.700\\ \[1.700--1.700\] 2.300\\ \[2.300--2.300\] 1.975^††^ \[1.975--1.975\] **Glycolytic and PPP fluxes:** 1 Phosphofructokinase 1.674 \[1.442--3.004\] 2.018 \[1.679--2.586\] 2.664\* \[2.537--2.670\] 1.907 \[1.874--1.919\] 2 Oxidative branch of PPP 0.199 \[0.196--0.199\] 0.200 \[0.198--0.200\] 0.248\\ \[0.246--0.298\] 0.198 \[0.192--0.199\] 3 Pyruvate kinase 3.078 \[3.026--3.262\] 3.582 \[2.644--4.126\] 4.671\\ \[4.573--4.873\] 4.337 \[4.087--4.468\] **Glycogen-related fluxes: 4 Glycogen phosphorylase 0.038 \[0.034--0.052\] 0.034 \[0.024--0.066\] 0.043 \[0.037--0.046\] 0.032 \[0.029--0.035\] 5 Glycogen synthase 0.040 \[0.036--0.053\] 0.035 \[0.024--0.067\] 0.050 \[0.043--0.058\] 0.038 \[0.034--0.040\] Pyruvate homeostasis-related fluxes:** 6 Pyruvate dehydrogenase complex 1.511 \[1.421--1.843\] 2.354\\ \[2.118--2.363\] 2.479\* \[1.785--2.987\] 2.125 \[1.575--2.494\] 7 Pyruvate carboxylase 0.200 \[0.113--0.291\] 0.117 \[0.027--0.201\] 0.197 \[0.155--0.254\] 0.031 \[0.028--0.055\] 8 Lactate dehydrogenase 1.366 \[1.292--1.451\] 1.511 \[1.446--1.654\] 2.033\\ \[1.620--2.738\] 2.192^††^ \[1.722--2.739\] 9 Pyruvate cycling 0.228 \[0.200--0.377\] 0.474 \[0.222--0.884\] 0.296 \[0.264--0.389\] 0.538 \[0.303--0.594\] **Tricarboxylic acid cycle fluxes:** 10 Citrate synthase 0.022 \[0.019--0.386\] 0.012 \[0.001--0.023\] 0.099 \[0.022--0.172\] 0.009 \[0.002--0.037\] 11 Citrate - \> Malate 0.035 \[0.032--0.462\] 0.072 \[0.050--0.400\] 0.185 \[0.110--0.298\] 0.370 \[0.082--0.395\] 12 Malate - \> OAA 0.640 \[0.481--2.088\] 1.152 \[0.453--3.104\] 1.509 \[0.780--2.413\] 2.078 \[0.477--3.095\] 13 OAA - \> Malate 0.605 \[0.446--1.776\] 1.080 \[0.294--2.800\] 1.324 \[0.668--2.115\] 1.708 \[0.336--2.773\] 14 Acetyl-CoA output 1.489 \[1.313--1.625\] 2.344\\ \[2.114--2.355\] 2.380\\ \[1.760--2.851\] 2.116 \[1.565--2.485\] Principal fluxes of central carbon metabolism calculated in L6E9 myotubes treated (+Res) or not treated (--Res) with 100 nM resistin for 8 h and then incubated for 6 h with 10 mM glucose, 50%-enriched in \[1,2-^13^C~2~\]-glucose in the absence (--Ins) or presence (+Ins) of 100 nM insulin. Glucose phosphorylation flux was fixed according to the glucose consumption measured (Table [1](#T1){ref-type="table"}). Remaining fluxes are expressed by median values as well as minimum and maximum values from the 20 best flux sets. Fluxes are expressed in nmol · mL^--1^ · min^--1^, where volume is the volume of the incubation medium. χ^2^ values for the averages were: Ins-- Res--: 11.35; Ins-- Res+: 5.42; Ins + Res--: 22.50; Ins + Res+: 25.56. (\\) indicates that the flux is different to the control condition with a 99% confidence interval (CI), (\) indicates that the flux is different to the control condition with a 95% CI, and (††) indicates that the flux differs from the condition in which cells were treated only with resistin with a 99% CI. PPP, pentose phosphate pathway; OAA, oxaloacetate. ![Metabolic fluxes in L6E9 myotubes not treated and treated with resistin and/or insulin.* L6E9 myotubes were treated (resistin+) or not treated (resistin--) with 100 nM resistin for 8 h and then incubated for 6 h with 10 mM glucose, 50%-enriched in \[1,2-^13^C~2~\]-D-glucose in the absence or in the presence of 100 nM insulin. Fluxes were estimated using the software Isodyn. Arrow sizes indicate net fluxes in L6E9 myotubes not treated or treated with resistin that were incubated in the absence of insulin. Colours indicate flux fold-changes in response to insulin. Fluxes plotted are the median values from the 20 best flux sets (Table [3](#T3){ref-type="table"}). Grey-coloured fluxes have not been measured. FBP, fructose-1,6-bisphosphate; Hexose-P, glucose-6-phosphate and its isomers; Mal, malate; OAA, oxaloacetate; Pentose-P, ribose-5-phosphate and its isomers; PEP, phosphoenolpyruvate; PYR, pyruvate; Triose-P, glyceraldehyde-3-phosphate and dihydroxyacetone phosphate.](s12918-014-0109-z-1){#F1} 2.1 Basal metabolic flux profile of L6E9 ---------------------------------------- Isodyn quantified the metabolic-flux distribution in L6E9 myotubes incubated in the presence of glucose (10 mM) and in the absence of insulin or resistin (Table [3](#T3){ref-type="table"}, first column). Under this incubation condition, cells had a glucose phosphorylation of 1.500 nmol · mL^--1^ · min^--1^. This flux was maintained throughout the subsequent steps of "upper glycolysis" (of which flux-1 is representative) until the split of fructose-1,6-bisphosphate into two triose-phosphates by aldolase. From this reaction onwards, the glycolytic flux (hereafter referred to as "lower glycolysis") increased twofold (as shown in flux-3). Additionally, the flux across the oxidative branch of the PPP was around eight-times lower than the upper glycolytic flux. With regard to glycogen metabolism, glycogen phosphorylase (GP) (flux-4) and glycogen synthase (GS) (flux-5) were active, and glycogen recycling was estimated to be ≈ 2% of the total glucose phosphorylation flux. However, the most important contribution of Isodyn for interpretation of experimental data was the possibility of flux quantification around the final steps of lower glycolysis and the tricarboxylic acid (TCA) cycle. In that regard, Isodyn revealed that ≈ 49% of phosphoenolpyruvate (PEP) and pyruvate entered the TCA cycle through the reactions catalysed by PDC (flux-6) or PC (flux-7), with a relative PDC:PC ratio of 8:1. Around 44% of the PEP--pyruvate flux was diverted to the synthesis and output of lactate towards the cell culture medium (flux-8), and pyruvate cycling flux (flux 9). This finding accounted for the fact that feedback of molecules from the TCA cycle back towards the PEP--pyruvate pool was around ten-times higher than the flux of the TCA cycle towards citrate synthesis (flux-10). 2.2 Effect of resistin on the metabolic flux profile of L6E9 ------------------------------------------------------------ L6E9 myotubes treated with 100 nM resistin showed 13% higher flux of glucose phosphorylation (1.700 nmol · mL^--1^ · min^--1^). Fitting of the experimental data from this incubation condition using Isodyn (Table [3](#T3){ref-type="table"}, second column) suggested that resistin increased the glycolytic flux, with a significant increase in the glucose phosphorylation flux, and a proportional increase of glycolytic fluxes. Resistin induced changes in the non-oxidative branch of the PPP (data not shown), but the relative flux through the oxidative branch (flux-2) was unchanged. Glycogen recycling-related fluxes (fluxes 4 and 5) remained unchanged, as indicated by the fact that the net ^13^C incorporation in glycogen was very similar to the control condition. Moreover, resistin increased lactate production (Table [1](#T1){ref-type="table"}), which translated to an increase in the lactate dehydrogenase flux (flux-8), though it was not significant. The most significant metabolic change after Isodyn fitting was observed in the PDC-related flux (flux-6), which was increased significantly (99% confidence). If we consider the relative ratios of the two main pyruvate incorporation fluxes (PDC and PC, fluxes 6 and 7, respectively), there was an increase of the PDC prevalence to 20:1. This finding was also supported by the level of pyruvate dehydrogenase kinase isozyme 4 (Pdk4) mRNA (Figure [2](#F2){ref-type="fig"}E), and was observed as a trend in the level of pyruvate dehydrogenase Pdha1 mRNA (Figure [2](#F2){ref-type="fig"}C). Despite the large change in this ratio, the effect on fluxes in the TCA cycle was almost negligible even though there was an increase in the acetyl-CoA output flux (flux-14). The pyruvate cycling flux was unchanged by resistin treatment alone, which was supported by the level of Pepck mRNA (Figure [2](#F2){ref-type="fig"}H). ![Gene expression of enzymes related to pyruvate homeostasis. L6E9 myotubes were treated (Res+) or not treated (Res--) with 100 nM resistin for 8 h and then incubated for 6 h with 10 mM glucose in the absence (Ins--) or presence (Ins+) of 100 nM insulin. RT-qPCR monitoring of mRNA levels of pyruvate carboxylase (Pc) (A), pyruvate kinase muscle isozyme (Pkm2) (B), pyruvate dehydrogenase E1 alpha 1 (Pdha1) (C), pyruvate dehydrogenase kinase isozyme 2 (Pdk2) (D), pyruvate dehydrogenase kinase isozyme 4 (Pdk4) (E), cytosolic NADP-isocitrate dehydrogenase (Idh1) (F), NADP-dependent malic enzyme (Me1) (G) and cytosolic phosphoenolpyruvate carboxykinase 1 (Pck1) (H) are shown. Results from three experiments are expressed as the mean ± S.E.M. (\) indicates p \< 0.05 and (\\) p \< 0.01 compared with the condition of neither resistin nor insulin. (\#) indicates p \< 0.05 and (\#\#) p \< 0.01 compared with L6E9 myotubes treated only with resistin.](s12918-014-0109-z-2){#F2} 2.3 Effect of resistin on the metabolic response of insulin in L6E9 myotubes ---------------------------------------------------------------------------- Having investigated the alterations caused by resistin on the metabolic profile of L6E9 myotubes, we examined its effect on the metabolic response of myotubes to insulin. First we ascertained if resistin affected stimulation of insulin-induced glucose transport. Thus, induction of 2-deoxy-D-\[2,6-^3^H\]-glucose (2-DG) transport was assayed 30 min after incubation with 100 nM insulin in L6E9 myotubes treated or not treated with 100 nM resistin. Insulin induced an increase in 2-DG transport in L6E9 myotubes of 79.27 ± 0.04%. However, it induced an increase in glucose transport in cells that had been pre-treated with 100 nM resistin for 8 h of 48.00 ± 0.05%. Subsequently, we analysed changes in the metabolic flux profile induced by insulin in L6E9 myotubes treated or not treated with 100 nM resistin. In the absence of resistin, insulin stimulated an increase in glucose consumption and lactate production in L6E9 myotubes. There were no changes in glycogen content, but a 150% increase in intracellular G6P content (Table [1](#T1){ref-type="table"}). Fitting of experimental data using Isodyn (Table [3](#T3){ref-type="table"}, third column) revealed that the flux of lactate production (flux-8) was 48% higher than that in cells not treated with insulin. Insulin also increased the flux of glucose phosphorylation (determined to be 2.300 nmol · mL^--1^ · min^--1^) and glycolytic fluxes (fluxes 1 and 3) by ≈ 59%, and induced a 25% increase in the oxidative PPP flux (flux-2). Moreover, fluxes related to glycogen metabolism (fluxes 4 and 5) did not change significantly after insulin treatment (albeit the median values were apparently higher than in the control condition). Insulin increased the flux through PDC (flux-6) by around 1.5-fold, but did not affect the remainder of TCA cycle-related fluxes significantly, as confirmed by the reduction in levels of Pdk4* mRNA and the increasing tendency of levels of Pdha1 mRNA (Figure [2](#F2){ref-type="fig"}E and C, respectively). The relative ratio of PDC:PC fluxes increased from 8:1 to 13:1 after insulin treatment. Insulin also led to a 59% increase in the acetyl-CoA output fluxes. In the presence of resistin, insulin increased the flux of glucose phosphorylation by 16% more than when cells were treated with resistin alone. Glycolysis fluxes did not increase significantly compared with the resistin condition, but the lactate dehydrogenase flux increased by 45% (flux-8). The increase induced by resistin treatment in the oxidative PPP pathway was not significantly altered if cells were also treated with insulin. Similarly, the GS and GP fluxes did not change significantly after resistin treatment. Interestingly, there was no change in the PDC flux (flux-6) or the acetyl-CoA production flux (flux-14), and no change was observed in the remainder of TCA cycle-related fluxes. If the ratio between PDC and PC fluxes is taken into account, it shifted from 20:1 towards 70:1 after resistin was combined with insulin. The final metabolic state of insulin-stimulated L6E9 did not match the one of cells treated with resistin. Some fluxes were equal in resistin-treated and -untreated cells: oxidative PPP flux, GS flux, GP flux, and TCA cycle-related fluxes. Most of the quantified fluxes were lower in resistin, but significant changes were found only in the lactate dehydrogenase flux, and the PDC:PC ratio. 2.4 Effect of resistin on gene expression of the enzymes involved in the PDC, PC and pyruvate cycling fluxes in L6E9 -------------------------------------------------------------------------------------------------------------------- To investigate further if the observed effects in metabolic-flux distribution were prompted by a change in the gene expression of any of the enzymes involved in their activities, we analysed the gene expression of several enzymes involved in pyruvate homeostasis at the end of the four incubation conditions tested. Neither insulin, nor resistin, nor combination of the two hormones had a significant effect on levels of pyruvate carboxylase (Pc) mRNA or pyruvate kinase muscle isozyme (Pkm2) mRNA (Figure [2](#F2){ref-type="fig"}A and B). However, insulin (but not resistin) increased levels of pyruvate dehydrogenase E1 (Pdha1) and decreased pyruvate dehydrogenase kinase isozyme 2 (Pdk2) mRNA (Figure [2](#F2){ref-type="fig"}C and D). Presence of resistin did not alter the decrease in levels of Pdk2 mRNA caused by insulin, but prevented the increase in levels of Pdha1 mRNA. mRNA expression of pyruvate dehydrogenase kinase isozyme 4 (Pdk4) decreased after treatment with both hormones (Figure [2](#F2){ref-type="fig"}E), with an additive effect if both were combined. These results suggest that resistin prevents the insulin-induced increase in Pdha1 expression, and does not prevent the increase in PDC activity caused by a reduction of expression of Pdk2 and Pdk4 genes. Moreover, mRNA levels of cytosolic NADP-isocitrate dehydrogenase (Idh1) and NADP-dependent malic enzyme (Me1), both causes of pyruvate cycling increasing NADPH production, were not affected by any of the hormones tested under these conditions (Figure [2](#F2){ref-type="fig"}F and G). However, insulin significantly decreased (by 50%) levels of Pck1 mRNA in the absence of resistin, but not in the presence of resistin (Figure [2](#F2){ref-type="fig"}H). These results suggest that resistin prevents inhibition of expression of the Pck1 gene. Hence, cytosolic flux in PEPCK may be active in the presence of resistin independent of the presence of insulin. 3 Discussion ============ After the discovery of several adipocyte-secreted molecules (including lipid metabolites and adipokines), adipocytes were recognized as part of endocrine tissue \[[@B9]\],\[[@B11]\],\[[@B37]\]. These molecules enable adipocytes to communicate with other tissues and organs, and to regulate: the metabolism of lipids and glucose; energy balance; insulin action; cell proliferation. Resistin was identified as an adipokine, and it was suggested that it might link obesity and IR \[[@B12]\],\[[@B13]\]. However, the role of resistin in the pathophysiology of IR in humans and animals, and how it acts in muscle, liver and fat, is controversial \[[@B15]\],\[[@B16]\]. Here, we investigated the effect of resistin on basal and insulin-stimulated glucose metabolism in L6E9 rat myotubes. We used tracer-based metabolomics and our in-house Isodyn software to analyse quantitatively metabolic flux distribution in this cell line under different incubation conditions. The response to insulin and other hormone stimuli in L6E9 cells is similar to that observed in skeletal muscle in vivo\[[@B38]\]. Hence, L6E9 cells are considered to be suitable models for analyses of the effect of resistin on glucose metabolism and on the metabolic response of muscle cells to insulin. Analyses with Isodyn data provided an overview of the central carbon metabolism of L6E9 skeletal muscle cells. Analyses were undertaken assuming a metabolic steady state (though not assuming an isotopic steady state), and rapid mixing of isotopic isomers of the same metabolite throughout different intracellular compartments. Incubation with a large excess of glucose ensured the required glucose uptake throughout the incubation. The other conditions of incubation did not change, so the assumption of a metabolic steady state seemed reasonable. Rapid mixing of isotopic isomers of the same species throughout the cell is accepted for analyses of data of stable isotopes \[[@B39]\]. Fitting of experimental data in the framework of these assumptions suggested a set of metabolic fluxes underlying the measured distributions of isotopic isomers. 3.1 The effect of insulin on L6E9 myotubes increases glycolysis, PPP, and incorporation of pyruvate to the TCA cycle through PDC and fatty-acid synthesis --------------------------------------------------------------------------------------------------------------------------------------------------------- Our results suggested that L6E9 myotubes were highly glycolytic, with active synthesis and degradation of glycogen, and showed TCA-cycle activity and acetyl-CoA output. Oxidative PPP was weakly active in this cell type. Furthermore, use of the same tools revealed that insulin activated the uptake and phosphorylation of glucose, glycolysis and lactate production, the oxidative branch of the PPP, pyruvate incorporation towards the TCA cycle and increased acetyl-CoA output (which is used for fatty-acid synthesis). We observed that insulin altered the ratio between PDC and PC fluxes by increasing the PDC flux through an increase of levels of Pdha1 and decrease of Pdk2 and Pdk4 mRNA. The effects of insulin on the metabolic flux distribution of L6E9 myotubes reported in the present study correlate with the known activation of PDC by profound suppression of Pdk4 expression in skeletal muscle \[[@B40]\], inhibition of expression of the Pck1 gene, and the increase in glucose uptake and lactate production observed using other approaches \[[@B38]\],\[[@B41]\]. 3.2 The effect of resistin on L6E9 myotubes increases glycolysis and fatty-acid synthesis and alters the ratio between PDC and PC, tilting it towards direct incorporation of pyruvate through PDC -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Our results revealed that, in the absence of insulin, resistin activated glucose phosphorylation and the glycolytic pathway. Also, resistin did not affect the glycogen content in skeletal muscle. Even though enrichment of the glycogen label (reflected in the m2 values of the glycogen molecule) was not significantly different to the control condition, the tendency was for glycogen-related fluxes to be lower, a result that was in accordance with other works \[[@B22]\]. Furthermore, our approach showed that, even though many of the TCA-cycle fluxes remained unchanged, the ratio between the PDC and PC fluxes was significantly altered, effectively increasing the incorporation of pyruvate through PDC into mitochondria (as suggested by the reduction observed in levels of Pdk4 mRNA). PC activity is finely regulated by mechanisms such as allosteric activation by acetyl carnitine, which was not tested in this study. The acetyl-CoA output for fatty-acid synthesis was also increased after resistin treatment, which would explain the higher flux towards the TCA cycle. 3.3 The effect of resistin impairs normal insulin metabolic actions ------------------------------------------------------------------- We showed that resistin affects the normal metabolic response of L6E9 myotubes to insulin. Thus, the increase in glucose uptake and glycolytic flux induced by insulin was lower than that observed in resistin-treated myotubes. These results are in accordance with those published by other authors for L6 muscle cells \[[@B19]\],\[[@B21]\],\[[@B22]\]. Resistin treatment, however, did not affect the insulin-stimulated content and recycling of glycogen. Moreover, in resistin-treated cells, insulin did not increase the PDC flux. After resistin treatment, levels of Pdk2 mRNA and Pdk4 mRNA were lower, accompanied with a reduction in levels of Pdha1 mRNA. The reduction in expression of the main enzyme and key regulators of its activity resulted in a null net effect on the final flux. The ratio between the PDC and PC fluxes was also altered, showing reduction in the PC anaplerotic entrance into the cycle. Fluxes through the TCA cycle and acetyl-CoA output fluxes were not increased, but instead maintained or reduced, by insulin in resistin-treated cells. This lack of metabolic activation in response to insulin can be explained by the slight decrease observed in glucose uptake and the increase of lactate dehydrogenase flux directed to lactate production, which impaired the increase in the levels of glycolytic intermediates that could be metabolized through other metabolic pathways. Our results provide mechanistic explanations for the observations of Palanivel and co-workers \[[@B22]\], who reported that the insulin-stimulated oxidation of glucose via the TCA cycle was reduced by resistin, and that this reduction would compromise oxidative respiration in the cell. 4 Conclusions ============= Several studies have identified positive correlations between resistin levels and IR in vivo\[[@B24]\] and in vitro\[[@B19]\]-\[[@B22]\]. Overall, our results showed that resistin significantly alters glucose metabolism in L6E9 myotubes and their response to insulin. Of the whole set of metabolic changes induced by resistin in L6E9 myotubes, the imbalance in PDC and PC fluxes as well as the higher acetyl-CoA output flux for fatty-acid synthesis and other metabolic processes should be emphasized. Resistin impairs the normal metabolism of insulin, and the metabolism is directed to try to maintain the imbalance between the PDC and PC fluxes. In this way, L6E9 reduces TCA-cycle fluxes and acetyl-CoA output fluxes, which leads to an important metabolic imbalance in the central carbon metabolic pathways of L6E9 myotubes. It has been reported that IR in skeletal muscle, liver and adipose tissue is usually accompanied by glucose intolerance and hypertriglyceridemia \[[@B24]\]. In this way, our results showed the importance of the balance of PDC and PC in glucose metabolism. Our results support the idea that metabolic disorders such as T2DM are caused by complex multi-molecular interactions that cannot be explained readily by an alteration in expression of a single gene or gene product, or even alteration of a single enzymatic cascade. The robustness of results obtained on the effects of insulin on L6E9 metabolism found by combining tracer-based metabolomic data and flux analyses using Isodyn showed that this approach is a suitable tool to study the effects of hormones on the central carbon metabolism network in myocytes. 5 Methods ========= 5.1 Cells and reagents ---------------------- Myoblastic cell line L6E9 was kindly provided by Dr. Nadal-Ginard (Harvard University, Boston, MD, USA). Rat recombinant resistin was produced in a HEK293 cell line. The purity of resistin was controlled by sodium dodecyl sulfate--polyacrylamide gel electrophoresis (SDS-PAGE) and estimated to be \>99%. Biological activity was validated as described \[[@B42]\]. Resistin was tested for endotoxin using a ToxinSensor Chromogenic LAL Endotoxin Assay kit (bioNova científica, Madrid, Spain). This assay indicated low levels of endotoxins (0.0015 EU/mL) at the final concentration used, which is below current USA Food and Drug Administion (FDA) limits that require eluates from medical devices to be \<0.5 EU/mL. \[1,2-^13^C~2~\]-D-glucose and \[U-^13^C-^2^H~7~\]-D-glucose were purchased from Isotec (Miamisburg OH, USA) and 2-deoxy-D-\[2,6-^3^H\]-glucose (2-DG) from GE Healthcare (Barcelona, Spain). 5.2 Conditions for cell culture and incubation ---------------------------------------------- Myoblastic cell line L6E9 was grown as described \[[@B38]\]. For biochemical and mass isotopomer assays, L6E9 myoblasts were seeded at 10^4^ cells/cm^2^ in 60-mm diameter dishes with 4 mL Dulbecco's modified Eagle medium (DMEM) without glucose or glutamine (Biological Industries, Kibbutz Beit Haemek, Israel) supplemented with 10% heat-inactivated foetal bovine serum (FBS) (Life Technologies, Madrid, Spain), 10 mM glucose, 4 mM asparagine and 5% penicillin/streptomycin (Sigma--Aldrich, Saint Louis MO, USA). Cell differentiation was induced by lowering FBS to a final concentration of 2% when myoblasts were pre-confluent \[[@B38]\]. Cells at day-4 of differentiation were washed four times in phosphate-buffered saline (PBS) and pre-treated or not with 100 nM resistin dissolved in DMEM without FBS and supplemented with 10 mM glucose, 4 mM asparagine, 2% HEPES (Sigma--Aldrich) and 1% penicillin/streptomycin. After 8 h, the medium was replaced by another with the same composition, but now with 10 mM glucose 50%-enriched in \[1,2-^13^C~2~\]-glucose in the presence or absence of 100 nM insulin (Sigma--Aldrich). At the end of 6-h incubation, media were collected and dishes washed twice with PBS before freezing in liquid nitrogen. Incubation media and dishes were stored at --80°C until biochemical and mass isotopomer analyses. 5.3 Glucose transport assay --------------------------- To carry out the glucose transport assay, 10^5^ L6E9 cells/dish were cultured on six-well plates. At day-4 of differentiation, cells were pre-treated or not with 100 nM resistin for 8 h, as described above. Next, uptake of 100 μCi 2-DG by myotubes incubated with or without 100 nM insulin for 30 min was measured as described \[[@B38]\]. 5.4 Biochemical determinations of glucose, lactate and G6P ---------------------------------------------------------- Concentrations of glucose and lactate in incubation media at the beginning and end of 6-h incubations were measured as described previously by spectrophotometric methods coupled to a Cobas Mira Plus Chemical Analyzer (HORIBA ABX, Montpellier, France). Intracellular G6P content was determined by enzymatic means as described \[[@B43]\]. 5.5 Mass isotopomer analyses of lactate, medium and glycogen glucose, and RNA ribose, and determination of glycogen content by GC/MS ------------------------------------------------------------------------------------------------------------------------------------ Mass spectral data were obtained on a QP2010 Shimadzu Mass Selective Detector connected to a GC-2010 Gas Chromatograph (Shimadzu, Kyoto, Japan) using helium as the carrier gas and isobutane 0.0016 Pa as reagent gas in chemical ionization analyses. The settings were: GC inlet, 250°C for glucose, ribose and deoxyribose, and 200°C for lactate; transfer line, 250°C; MS source, 200°C. A Varian VF-5 Capillary Column (30-m length, 250-μm diameter, 0.25-μm film thickness) was used to analyze all compounds. Glucose from the frozen media was purified using a tandem set of Dowex-1X8/Dowex-50WX8 (Sigma--Aldrich) ion-exchange columns. The glucose from media was converted to its glucose aldonitrile pentaacetate derivative as described \[[@B43]\]. Ion clusters around m/z 328 (carbons 1--6 of glucose, chemical ionization) were monitored. Lactate from the media was extracted with ethyl acetate after acidification with hydrochloric acid (HCl) and derivatized to its propylamide-heptafluorobutyric form. Clusters around m/z 328 (carbons 1--3 of lactate, chemical ionization) were monitored \[[@B43]\]. RNA ribose was isolated by acid hydrolysis of cellular RNA after Trizol purification of cell extracts. Ribose isolated from RNA was derivatized to its aldonitrile-acetate form using hydroxylamine in pyridine and acetic anhydride. We monitored ion clusters around m/z 256 (carbons 1--5 of ribose, chemical ionization) to find the molar enrichment and positional distribution of ^13^C labels in ribose as described \[[@B44]\]. Glycogen in frozen cell monolayers from different analytical conditions was extracted and quantified as described \[[@B44]\]. Extraction was done by direct digestion of sonicated extracts with amyloglucosidase. Mass isotopomer analyses of glycogen glucose was undertaken as described above for medium glucose. Measurement of glycogen content was carried out using the isotopomer \[U-^13^C-^2^H~7~\]-glucose as recovery standard and internal standard quantification procedures. The ion cluster for the \[U-^13^C-^2^H~7~\]-glucose of the glucose aldonitrile-pentaacetate derivative was monitored from m/z 339 to m/z 341. Spectral data were corrected using regression analysis to extract natural ^13^C enrichment from results \[[@B45]\]. Measurement of ^13^C label distribution determined the different relative distribution percentages of the mass isotopomers, m0 (without any ^13^C labels), m1 (with one ^13^C), m2 (with two ^13^C), etc., which were reported as molar fractions. 5.6 Analyses and modelling of data using Isodyn ----------------------------------------------- Transfer of ^13^C from \[1,2-^13^C~2~\]-glucose medium into intracellular metabolites was simulated by Isodyn, a software program written in C++ and designed for analyses of data from stable isotopic tracers \[[@B26]\],\[[@B33]\],\[[@B34]\]. Latest modifications of this software have been described in detail \[[@B35]\],\[[@B36]\]. This software automatically constructs and solves the large system of ordinary differential equations describing the evolution of isotopomer concentrations of metabolites produced in glycolysis, TCA cycle, and PPP (Figure [1](#F1){ref-type="fig"}). All the reactions of the non-oxidative branch catalyzed by transketolase and transaldolase are accounted for, assuming pentose-5-phosphates to be a single pool. The reactions of the oxidative branch are passed to one reaction leading from G6P to pentose-phosphates. At this initial moment, all metabolites (except added labeled substrate) are assumed to be non-labeled, and the latter is set in accordance with the known isotopomer composition of the used labeled substrate. Initial total concentrations of intracellular metabolites are calculated as functions of model parameters assuming a steady state at the initial moment. Functions designed especially for each type of reaction (e.g., carboxylation, decarboxylation, transketolase-type) simulate transformation of the carbon skeleton (specific transitions of labeled carbons), and the production and consumption rate for each isotopomer of the considered system. These transformations redistribute ^13^C isotopes in all metabolites. Individual rates are computed for each isotopomer, and these rates determine the values of the derivatives for the isotopomers. This system is solved using a method of numerical integration that can be chosen arbitrarily (Runge-Kutta, BDF, Dassl). Starting from the initial values corresponding to the experimental conditions of incubation, Isodyn simulates a real-time course of label propagation. Experimental and computed data for corresponding time points are compared. In this way, reaching an isotopic steady state is not required. 5.7 RNA extraction and real-time RT-qPCR gene-expression analyses ----------------------------------------------------------------- Total RNA was isolated from L6E9 cells using an RNeasy Mini kit and a QIAcube Automat (Qiagen, Venlo, the Netherlands). RNA integrity was checked by capillary electrophoresis (Experion; Bio-Rad, Hercules, CA, USA). Concentrations were checked with Ribogreen (Molecular Probes, Eugene, OR, USA) and a Victor-X5 Multilabel Reader (PerkinElmer, Waltham, MA, USA). Labelled cDNA was synthesized from 600 ng of total RNA using SuperScript-II, reverse transcriptase and random hexamers (Life Technologies, Madrid, Spain). Oligonucleotide sequences for the reverse transcriptase reaction, synthesized by Eurogentec (Angers, France), were: pyruvate carboxylase (PC), Pc, rPC-F, TTCCGTGTCCGAGGTGTAAAG, and rPC-R, CGCTAGGAACTGCTGGTTGTT; pyruvate kinase muscle isozyme, Pkm2, rPKm-784F, GAGCAGGACGTGGACATGGT, and rPKm-864R, CTCTCCCAGGACCTTCCTAACC; pyruvate dehydrogenase E1 alpha 1, Pdha1, rNm_001004072.2F, TTTGTCTTCTGTGCTGGGAGACTG and rNm_001004072.2R, GTAGATGGGTGGCTTCAAGTTTGC; pyruvate dehydrogenase kinase isozyme 2, Pdk2, rNm_030872.1F, CATGGCTAAGCTCCTGTGTGAC, and rNm_030872.1R, GGACGTAGACCATGTGAATGGG; pyruvate dehydrogenase kinase isozyme 4, Pdk4, rNm_053551.1F, TGCTCATGAACCAGCACATCCTC and rNm_053551.1R, TCCCAATGTGGCTTGGGTTTCC; cytosolic NADP-isocitrate dehydrogenase, Idh1, rNm_031510.1F, AGACGTCCACCAATCCCATTGC, and rNm_031510.1R, TCAAGCTTTGCTCTGTGGGCTAAC; NADP-dependent malic enzyme, Me1, rNm_012600.2F, TGCAAGACCATGGTTCCCAGAC, and rNm_012600.2R, AAGCTAAGCCCAGGGACATTAGGG; cytosolic phosphoenolpyruvate carboxykinase (PEPCK) 1, Pck1, rPEPCK1-2228F, TGGTTCCACTTCGAGGTCACT, and rPEPCK1-2310R, CAAGTATGTTTTCTGTGCACTTTAGCT. Real-time PCR (qPCR) was conducted with a Sybrgreen Master Mix Reagent using Sureprime Core kit (MP Biomedical, III Kirch, France) in an iCycler (Bio-Rad, Marnes-la-Coquette, France) apparatus. The standard curve method was used for relative quantification of PCR products, and gene expressions normalized to 18S RNA quantification, which has been found to be a reliable internal control gene in our hands and by other scholars \[[@B46]\]-\[[@B48]\]. 5.8 Statistical analyses ------------------------ Experiments relating to glucose transport were carried out using three cultures each time for each treatment regimen and then repeated thrice. Metabolic characterization assays were carried out in triplicate and repeated twice. RNAs for RT-qPCR assays were isolated from three independent extracts. Statistical analyses of glucose transport as well as biochemical and metabolite mass isotopomer distributions were calculated by two-way ANOVA between groups. Data for real-time PCR were analyzed by the one-way ANOVA between groups. Confidence intervals at a confidence level of 95% or 99% (p \< 0.05 and p \< 0.01, respectively) were taken as indicating significant differences in each parameter analyzed. Differences from the mean were considered significant for p \< 0.05. In the case of the metabolic fluxes estimated using Isodyn, the difference between calculated and experimental data (χ^2^) was minimized using a simulated annealing algorithm. The goodness of best fit (corresponding to the minimal deviation of calculated isotopomer fractions from the experimental ones) was checked based on the value of χ^2^ and number of degrees of freedom \[[@B36]\]. In the case that, according to this checking, the fit was deemed acceptable, the metabolic fluxes corresponding to the best fit were accepted as consistent with the measured isotopomer distribution. The confidence intervals for the fluxes were determined as the upper and lower limits of fluxes that produced χ^2^ that was less than the fixed threshold \[[@B36]\]. Abbreviations ============= 2-DG: 2-deoxy-D-\[2,6-^3^H\]-glucose G6P: Glucose-6-phosphate GP: Glycogen phosphorylase GS: Glycogen synthase IR: Insulin resistance PC: Pyruvate carboxylase PDC: Pyruvate dehydrogenase complex PEP: Phosphoenolpyruvate PEPCK: Phosphoenolpyruvate carboxykinase PPP: Pentose phosphate pathway T2DM: Type-2 diabetes mellitus TCA: Tricarboxylic acid Competing interests =================== The authors declare that they have no competing interests. Authors' contributions ====================== SG, SM, FS, AZ, PR and MC conceived and designed the research. SG, SM, AM, RH, FS, AT and YD undertook the experiments. SG, SM, VS, AM, RH, FS, AT, YD and MC analysed data. SG, SM, AM, VS, JC, AZ, PR, and MC interpreted the results of the experiments. SG, SM, JC, AM, RH, FS, AT, YD and PR prepared the figures. SG, SM, AM, VS, PR and MC drafted the manuscript. SG, SM, AM, VS, JC, AZ, PR and MC edited and revised the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ This study was supported by European Union-funded project ETHERPATHS (FP7-KBBE-222639) and from Agència de Gestió d\'Ajuts Universitaris i de Recerca (AGAUR)-Generalitat de Catalunya (2014SGR1017 and 2006ITT-10007). MC aknowledges the support received through the prize \"ICREA Academia\" for excellence in research, funded by ICREA foundation-Generalitat de Catalunya. We also thank Region Midi-Pyrenees for financial support of this work.	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ All-trans retinoic acid (tRA) is the primary bioactive form of endogenous retinoids derived from dietary vitamin A and plays important roles in regulating a myriad of physiological events [@pone.0045725-Napoli1]. One of the major mechanisms through which tRA exerts its biological activity is by binding and activating its cognate nuclear receptors, the retinoic acid receptors (RARs) α, β, γ, and retinoid X receptors (RXRs) α, β, γ, which heterodimerize to act as transcription factors, thereby modulating gene transcription [@pone.0045725-RochetteEgly1]. Direct target genes of tRA are often characterized by the presence of one or more retinoic acid response elements (RAREs) in the gene regulatory region, which serve as anchorage points for the RXR-RAR heterodimers [@pone.0045725-Balmer1], [@pone.0045725-Lalevee1]. A RARE typically consists of two direct repeats (DR) of the hexameric motif PuG(G/T)TCA, separated by 1, 2, or 5 nucleotides, referred to as DR1, DR2, and DR5, respectively [@pone.0045725-RochetteEgly1]. While DR5 RAREs represents the most potent classical RARE for transcriptional activity regulation, other forms of non-classical RAREs, e.g., hexameric motifs spaced by more than 5 nucleotides, and imperfect hexameric motifs have been described [@pone.0045725-Balmer1]. Other than the canonical tRA/RAR/RARE signaling, various non-canonical signaling events of retinoids, such as RAR-independent signaling and the involvement of ligands other than tRA, have been described [@pone.0045725-Amann1]. In addition to the multiplicity of retinoid signaling, the presence of endogenous retinoids is also intricately controlled by multiple synthesizing and metabolizing enzymes, including medium-chain and short-chain dehydrogenases, retinaldehyde dehydrogenases (Raldhs), and the cytochrome P450 family 26 [@pone.0045725-Theodosiou1]. It is thus not surprising that retinoid signaling is highly complex, reflected in its diverse and seemingly paradoxical effects, depending on different cell types and different settings. It is well established that the endogenous tRA and RARs are indispensable for embryonic kidney development [@pone.0045725-Xu1]. Even mild gestational vitamin A deficiency leads to a deficit in nephron number, which may predispose the kidney to abnormal development or function when associated with other morbidities [@pone.0045725-LelievrePegorier1]. Using RARE-hsp68-lacZ mice as a reporter model, we recently described the presence of a RARE reporter signal indicative of RAR-dependent tRA activity in the collecting ducts (CDs) of young and adult mouse kidneys [@pone.0045725-Wong1]. Our observation of this RARE reporter signal in the CDs is similar to the finding of Rosselot et al., who described the presence of a RARE reporter signal in the ureteric buds (UBs) of embryonic kidney, which are the embryonic precursors of CDs [@pone.0045725-Rosselot1]. Thus, RARE activity in healthy kidneys appears to be confined to the UB/CD cell lineage. The observation of RARE reporter activity in the UB/CD cell lineage has important implications. Initially Batourina et al proposed that kidney stromal mesenchyme is the primary site of action for tRA/RAR signaling in embryonic kidneys; the signal being delivered to the UB to regulate Ret expression, which stimulates ureteric bud branching and initiates kidney development [@pone.0045725-Batourina1]. The same group recently revised this model and proposed that tRA/RAR signaling was actually initiated in the UB cells rather than in the stromal mesenchyme [@pone.0045725-Rosselot1]. The revised model is in good concordance with the presence of a RARE reporter signal in UB cells. Nevertheless, tRA/RAR target genes in UB cells remain largely unknown, as are the tRA/RAR target genes in UB-derived CDs. Because RARs are nuclear transcription factors known to regulate gene transcription, and in view of the continuity of tRA/RAR signaling in the UB/CD cell lineage in embryonic and postnatal kidneys [@pone.0045725-Wong1], [@pone.0045725-Rosselot1], it is relevant to explore what genes are under the control of the endogenous tRA/RARs in CD cells. To explore target genes of endogenous tRA/RARs in the UB/CD cell lineage, we performed microarray experiments using mIMCD-3 as an in vitro cell model. mIMCD-3 is a well-characterized inner medullary CD cell line [@pone.0045725-Rauchman1], which has been widely used in a variety of studies to explore the functions and activities of UB/CD cells: ion channel signaling [@pone.0045725-Goel1], [@pone.0045725-Sproul1], signaling during osmotic and hypertonic stress [@pone.0045725-Cai1], urea signaling [@pone.0045725-Cohen1], and branching morphogenesis [@pone.0045725-Lee1]. We used two different chemicals to disrupt endogenous tRA/RAR signaling at different stages: (i) AGN193109 [@pone.0045725-Johnson1], a pan-antagonist of RARs, which competes with endogenous tRA for RAR binding, and (ii) 4-(diethylamino)benzaldehyde (DEAB) [@pone.0045725-Russo1], an inhibitor of Raldhs, which inhibits endogenous tRA biosynthesis. To confirm the specificity of these chemical reagents, exogenous tRA was added simultaneously to determine if their effects could be abolished ([Figure 1](#pone-0045725-g001){ref-type="fig"}). ![Strategy employed to examine endogenous tRA/RAR signaling and its target genes.\ In cells, all-trans retinol (tRol), i.e., vitamin A, is oxidized reversibly to all-trans retinaldehyde (tRal), catalyzed by the enzymes retinol dehydrogenases (Rdhs) or alcohol dehydrogenases (Adhs). tRal is then oxidized irreversibly to all-trans retinoic acid (tRA), catalyzed by retinaldehyde dehydrogenases (Raldhs). tRA translocates into cell nucleus to bind and activate retinoic acid receptors (RARs), modulating gene transcription. AGN193109 (purple) and 4-(diethylamino)benzaldehyde (DEAB, orange) were used to compete with tRA for RAR binding and to inhibit Raldhs in the conversion of tRal into tRA, respectively. Specificity of inhibition was confirmed by adding exogenous tRA (red) simultaneously to examine if the effects of AGN193109 and DEAB could be at least partially abolished.](pone.0045725.g001){#pone-0045725-g001} Results {#s2} ======= The mIMCD-3 Cell Line Simulates CD Cells in vivo, Expressing Epithelial and CD Principal Cell Markers and Demonstrating Constitutive tRA/RAR Activity {#s2a} ------------------------------------------------------------------------------------------------------------------------------------------------------- mIMCD-3 cells expressed E-cadherin, a marker of epithelial cells ([Figure 2A](#pone-0045725-g002){ref-type="fig"}i), and aquaporin 2, a marker of CD principal cells ([Figure 2A](#pone-0045725-g002){ref-type="fig"}ii), confirming its epithelial and CD-like phenotype. After confirming the identity of mIMCD-3 cells, we examined the presence of RARE-reporter activity by transfecting the cells with pGL3-RARE-luciferase reporter plasmid. By treating mIMCD-3 cells with AGN193109, basal RARE-luciferase activity was reduced to about 50% of that of the vehicle control group; when exogenous tRA was added with AGN193109, the reduction of RARE-luciferase activity was abolished in a dose-dependent manner ([Figure 2B](#pone-0045725-g002){ref-type="fig"}i). Similarly, when mIMCD-3 cells were treated with DEAB, a reduction of basal RARE-luciferase activity was observed, which was reversed when exogenous tRA was added ([Figure 2B](#pone-0045725-g002){ref-type="fig"}ii). Note that the RARE-luciferase activity was reversed to a level similar to the vehicle control group at 0.1 nM tRA; the RARE-luciferase activity was saturated by 1 nM tRA, which was about 1.75-fold of that of the vehicle control group and about 3.5-fold of that of the DEAB-treated group ([Figure 2B](#pone-0045725-g002){ref-type="fig"}ii). Under culture conditions, when RARE-luciferase transfected mIMCD-3 cells were treated with exogenous tRA alone, the induction of RARE-luciferase activity was weak and did not reach statistical significance ([Figure 2B](#pone-0045725-g002){ref-type="fig"}iii), suggesting that the constitutive RARE activity induced by endogenous tRA/RARs was close to saturation for activating the reporter construct. ![mIMCD-3 cells as an in vitro model to examine target genes of endogenous tRA/RARs.\ mIMCD-3 cells expressed E-cadherin (E-cad) (Ai) and aquaporin 2 (AQP2) (Aii) proteins. Left panels showed positive staining of specific antibodies, whereas right panels showed negative controls of cells stained with the respective non-immune IgGs. Original magnification was 200×. Bi. Treating cells with 1 µM AGN193109 for 24 h resulted in a decrease of RARE-luciferase activity to about 50% of that of the vehicle control group; the reduction was at least partially abolished with simultaneous addition of exogenous tRA at 0.01--1 µM, in a dose-dependent manner. \\ and \\\: p\<0.01 and p\<0.001, respectively. Bii.* Treating cells with 25 µM DEAB for 72 h resulted in a suppression of RARE-luciferase activity to about 50% of that of vehicle control group; when exogenous tRA was added during the last 24 h, the suppression of RARE-luciferase activity was reversed in a dose-dependent manner, saturated at 1 nM tRA. \# and \#\#: p\<0.05 and 0\<0.01 vs vehicle control group, respectively; \\ and \\\: p\<0.01 and p\<0.001 vs DEAB-only group, respectively. Biii.* When cells were treated with exogenous tRA alone at 0.001--0.1 µM for 24 h, a slight trend of dose-dependent increase of RARE-luciferase activity was noted but the difference was not statistically significant. Each dot represents mean value of triplicates or quadruplicates from a single biological experiment.](pone.0045725.g002){#pone-0045725-g002} Candidate Target Genes of Endogenous tRA/RARs {#s2b} --------------------------------------------- Two sets of independent biological experiments (N = 3 each) were performed: (i) cells treated with AGN193109, with and without tRA, to identify candidate genes regulated by RARs, and (ii) cells treated with DEAB, with and without tRA, to identify candidate genes regulated by endogenous tRA. Samples taken from these studies were subjected to microarray analysis, in which candidate target genes of endogenous tRA/RARs were defined as those genes regulated in the same direction by both AGN193109 and DEAB, with the regulation partially or completely abolished by exogenous tRA at concentrations used in the earlier studies shown in [Figure 2](#pone-0045725-g002){ref-type="fig"}. Partial abolishment included a statistical difference between the "inhibitor-only" and "inhibitor with tRA" groups, or no statistical difference between the "inhibitor with tRA" and "vehicle control" groups. Prior to microarray experiments, a small group of pilot reverse-transcription-quantitative polymerase chain reaction (RT-qPCR) studies was performed to examine the expression of a few genes of interest to verify the effectiveness of our strategy, as well as to guide the process of shortlisting target genes in the ensuing microarray studies. For this purpose, we selected Bmp7, Foxa1, Pax2, and Wnt7b, which have been implicated in kidney development, water transport, and defense/protection mechanism against injury [@pone.0045725-Cai1], [@pone.0045725-Zeisberg1], [@pone.0045725-Yu1], [@pone.0045725-Behr1], [@pone.0045725-Piscione1]. These genes have been reported in other systems to be regulated by tRA with or without known functional RARE [@pone.0045725-Paralkar1], [@pone.0045725-Balmer2], [@pone.0045725-Jacobs1], [@pone.0045725-Liu1]. As shown in [Figure 3](#pone-0045725-g003){ref-type="fig"}, Bmp7 and Foxa1 mRNAs were suppressed by both AGN193109 and DEAB; the suppression was at least partially abolished when exogenous tRA was added simultaneously. On the other hand, Wnt7b and Pax2 mRNAs were not significantly regulated by either AGN193109 or by DEAB (data not shown). ![Regulation of Bmp7 and Foxa1 by AGN193109 and 4-(diethylamino)benzaldehyde (DEAB) in pilot study.\ mRNA expression of Bmp7 (Ai) and Foxa1 (Aii) was suppressed by AGN193109; the suppression was at least partially abolished in the presence of 0.2 µM tRA. Bi. Bmp7 mRNA was suppressed by DEAB; the suppression was reversed to a level slightly higher than basal level in the presence of 0.01 µM tRA. Bii. Expression of Foxa1 mRNA was suppressed by DEAB; the suppression was partially abolished in the presence of 0.01 µM tRA. Each dot represents mean value of three technical replicates from a single biological experiment. \, \\, and \\\: p\<0.05, p\<0.01, and p\<0.001, respectively.](pone.0045725.g003){#pone-0045725-g003} Given that the magnitude of regulation for Foxa1 was lower than that of Bmp7, and the effects of exogenous tRA did not reach significance, Foxa1 was used as a cut-off threshold for shortlisting candidate target genes (details described in the Materials and Methods section): 403 and 439 unique genes were generated from AGN193109 experiments and DEAB experiments, respectively. As shown in [Figure 4A](#pone-0045725-g004){ref-type="fig"}, there were 133 overlapping genes; of these, 125 were similarly regulated by AGN193109 and by DEAB, and the regulation was at least partially abolished in the simultaneous presence of exogenous tRA. These 125 genes were categorized as the group 1 genes that represent specific target genes of endogenous tRA/RARs. In addition to the group 1 genes, there were 213 genes designated as group 2 genes that were regulated only by AGN193109, but not DEAB. These were candidate genes regulated by RARs, but not endogenous tRA. There were 266 additional genes designated as group 3 genes that were regulated only by DEAB, but not AGN193109. They were considered as candidate genes regulated by endogenous tRA, but independent of RARs ([Figure 4A](#pone-0045725-g004){ref-type="fig"}). When the genes were ranked by fold-changes, 16 out of the top 20 and 9 out of the top 10 most highly regulated genes overlapped ([Figure 4B and 4C](#pone-0045725-g004){ref-type="fig"}), suggesting that the most regulated genes were dependent on both tRA and RARs. Thus, further analysis was focused on the tRA/RAR-dependent genes (group 1). ![Candidate target genes short-listed from microarray experiments. A.\ A total of 403 and 439 unique genes were short-listed from AGN193109 experiment and DEAB experiment, respectively. 133 genes were regulated by AGN193109 and DEAB; among them, 125 genes (designated as group 1 genes), were regulated by AGN193109 and DEAB at similar directions, which the regulation was at least partially abolished in the simultaneous presence of exogenous tRA. 270 genes were regulated only by AGN193109, of which regulation of 213 genes (designated as group 2 genes) was at least partially abolished in the simultaneous presence of exogenous tRA; 306 genes were regulated only by DEAB, of which regulation of 266 genes (designated as group 3 genes) was at least partially abolished in the simultaneous presence of exogenous tRA. Note that the fold-changes of intersected genes were higher than those of non-intersected genes. When ranked based on fold-changes, 80% of the top 20 most regulated genes (B) and 90% of the top 10 most regulated genes were overlapped (C).](pone.0045725.g004){#pone-0045725-g004} The number and fold-changes of all group 1 genes are summarized in [Figure 5](#pone-0045725-g005){ref-type="fig"}. The total number of genes suppressed by both AGN193109 and DEAB were approximately 3-fold more than those induced by both AGN193109 and DEAB. Of note, there were 19 and 12 genes suppressed by AGN193109 and DEAB by 2-fold or more, respectively, but none was induced by 2-fold or more, indicating that tRA/RARs play a more important role in maintaining, rather than suppressing, gene expression. The top 20 most down- and up-regulated genes within group 1 are listed in [Table 1](#pone-0045725-t001){ref-type="table"} and [Table 2](#pone-0045725-t002){ref-type="table"}, respectively. The complete lists of genes regulated by AGN193109 and by DEAB, genes regulated only by AGN193109 and genes regulated only by DEAB, are compiled in [Tables S1](#pone.0045725.s004){ref-type="supplementary-material"}, [S2](#pone.0045725.s005){ref-type="supplementary-material"}, [S3](#pone.0045725.s006){ref-type="supplementary-material"}. Original raw data of microarray experiments have been deposited in NCBI's Gene Expression Omnibus (GEO) and are accessible through GEO Series accession number GSE33955 (<http://www.ncbi.nlm.nih.gov/geo/query/acc>. cgi?acc = GSE33955). 10.1371/journal.pone.0045725.t001 ###### Top 20 most down-regulated genes. ![](pone.0045725.t001){#pone-0045725-t001-1} Gene Symbol AGN193109 AGN193109+ tRA DEAB DEAB + tRA Entrez Genes ------ --------------- ----------- ---------------- ------- ------------ -------------- 1\. Ppbp −9.49 −1.53 −6.15 +1.85 57349 2\. Dhrs3 −8.59 −3.78 −5.02 +2.03 20148 3\. Sprr1a −6.85 −1.38 −4.62 +2.53 20753 4\. Cpm −5.15 −1.69 −2.30 +1.33 70574 5\. 9930023K05Rik −4.36 −1.59 −2.35 +1.60 226245 6\. Tns1 −3.22 −1.82 −2.29 +1.27 21961 7\. Itga2 −3.22 −1.16 −2.56 +1.25 16398 8\. Klhdc7a −3.10 −1.87 −2.11 +1.03 242721 9\. Csn3 −2.91 −1.91 −2.18 +1.18 12994 10\. Sorcs2 −2.66 −1.46 −2.05 +1.18 81840 11\. Ebf1 −2.66 −1.43 −2.43 +1.33 13591 12\. Lcn2 −2.48 −1.28 −1.78 +1.16 16819 13\. 2310007B03Rik −2.31 −1.58 −1.42 1.00 71874 14\. Galns −2.27 −1.55 −1.69 +1.07 50917 15\. Npr3 −2.21 −1.03 −1.99 +1.52 18162 16\. Muc20 −2.09 −1.84 −2.18 +1.08 224116 17\. Clca4 −2.07 +1.31 −1.78 +1.94 229927 18\. Upk3b −2.05 −1.50 −1.47 +1.07 100647 19\. Slc37a1 −2.00 −1.33 −1.53 +1.16 224674 20\. Hrsp12 −1.98 −1.27 −1.65 +1.06 15473 Shown here are mean fold-changes of gene expression compared to vehicle control from three experimental groups. Genes were sorted by fold-changes of AGN193109 group compared to vehicle group. Minus and plus numbers indicate folds of suppression and induction, respectively, in comparison to the vehicle control group, which was normalized as 1. 10.1371/journal.pone.0045725.t002 ###### Top 20 most up-regulated genes. ![](pone.0045725.t002){#pone-0045725-t002-2} Gene Symbol AGN193109 AGN193109+tRA DEAB DEAB+tRA Entrez Gene ------ -------------------------- ----------- --------------- ------- ---------- ------------------- 1\. Anxa8 +1.73 −1.10 +1.43 −1.54 11752 2\. Ptgs2 +1.67 +1.48 +1.24 −1.23 19225 3\. Gsdma +1.61 −1.01 +1.39 −1.33 57911 4\. Cpeb2 +1.57 +1.18 +1.24 −1.15 231207 5\. Ahnak2 +1.57 −1.11 +1.42 −1.42 100041194 6\. Casp14 +1.56 −1.11 +1.16 −1.31 12365 7\. Peg10 +1.52 +1.21 +1.35 −1.16 170676 8\. Ndrg1 +1.45 −1.02 +1.32 −1.24 17988 9\. Rnf39 +1.45 +1.11 +1.18 −1.19 386454 10\. Pkp1 +1.40 +1.11 +1.25 −1.20 18772 11\. Alcam +1.39 1.00 +1.36 −1.05 11658 12\. Timp3 +1.39 −1.03 +1.12 −1.15 21859 13\. Adora1 +1.38 +1.15 +1.27 −1.08 11539 14\. Car5b +1.36 +1.03 +1.13 −1.28 56078 15\. Atp6v0a4///D630045J12Rik +1.34 +1.11 +1.21 −1.23 140494///330286 16\. Egr1 +1.33 +1.24 +1.07 −1.02 13653 17\. Ly6c1///Ly6c2 +1.31 −1.01 +1.14 −1.20 17067///100041546 18\. Cav1 +1.30 +1.02 +1.20 −1.11 12389 19\. Lama3 +1.28 −1.02 +1.04 −1.21 16774 20\. Lmna +1.26 −1.02 +1.17 −1.17 16905 Shown here are mean fold-changes of gene expression compared to vehicle control from three experimental groups. Genes were sorted by fold-changes of AGN193109 group compared to vehicle group. Minus and plus numbers indicate folds of suppression and induction, respectively, in comparison to the vehicle control group, which was normalized as 1. ![Candidate target genes of endogenous tRA/RARs: number of induced and suppressed genes.\ A total of 31 and 94 genes were up-(red) and down-(green) regulated, respectively, by AGN193109 and by DEAB; regulation of these genes were at least partially abolished in the presence of tRA. Of the 31 up-regulated genes, 24 were up-regulated by AGN193109 by less than 1.5-fold, 7 by more than 1.5-fold but less than 2-fold, and none by 2-fold and more; none of the genes were up-regulated by DEAB by 1.5-fold and more. Of the 94 down-regulated genes, 38 were down-regulated by AGN193109 by less than 1.5-fold, 37 by more than 1.5-fold but less than 2-fold, and 19 by 2-fold and more; 68 were down-regulated by DEAB by less than 1.5-fold, 14 by more than 1.5-fold but less than 2-fold, and 12 by 2-fold and more.](pone.0045725.g005){#pone-0045725-g005} Validation of Microarray Results {#s2c} -------------------------------- The top 20 most highly regulated genes shortlisted from the microarray experiments were all down-regulated by AGN193109 and by DEAB. Regulation of these genes was validated by RT-qPCR. As shown in [Table 3](#pone-0045725-t003){ref-type="table"}, among the 20 genes examined, 19 were significantly down-regulated by AGN193109; suppression of all these genes, except Muc20, was at least partially abolished in the presence of exogenous tRA. The same 19 genes were also significantly down-regulated by DEAB and suppression was abolished in the presence of exogenous tRA. Thus, while Csn3, which was not amplified, and Muc20 need further investigation, the remaining 18 genes, along with Bmp7 and Foxa1 examined in the pilot study, were confirmed as specific target genes of endogenous tRA/RARs. 10.1371/journal.pone.0045725.t003 ###### Validation of top 20 most down-regulated genes. ![](pone.0045725.t003){#pone-0045725-t003-3} Gene Symbol AGN193109 AGN193109+tRA DEAB DEAB+tRA ------ --------------- -------------------------------------------------- ----------------------------------------------------------------------------------- -------------------------------------------------- ------------------------------------------------------------------------------------- 1\. Dhrs3 0.01±0.0009[\\\](#nt106){ref-type="table-fn"} 0.14±0.0495[∧∧∧](#nt109){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 0.04±0.0052[\\\](#nt106){ref-type="table-fn"} 2.88±0.4050[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 2\. Sprr1a 0.05±0.0126[\\\](#nt106){ref-type="table-fn"} 0.60±0.0575[∧∧∧](#nt109){ref-type="table-fn"} 0.15±0.0195[\\\](#nt106){ref-type="table-fn"} 4.11±0.3857[+++](#nt112){ref-type="table-fn"} [\\\](#nt106){ref-type="table-fn"} 3\. Ppbp 0.07±0.0056[\\\](#nt106){ref-type="table-fn"} 0.73±0.0586[∧∧∧](#nt109){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 0.14±0.0129[\\\](#nt106){ref-type="table-fn"} 2.56±0.1259[+++](#nt112){ref-type="table-fn"} [\\\](#nt106){ref-type="table-fn"} 4\. 9930023K05Rik 0.12±0.0072[\\\](#nt106){ref-type="table-fn"} 0.63±0.0326[∧∧∧](#nt109){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 0.28±0.0270[\\\](#nt106){ref-type="table-fn"} 2.09±0.1978[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 5\. Cpm 0.13±0.0217[\\\](#nt106){ref-type="table-fn"} 0.47±0.0053[∧∧](#nt108){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 0.31±0.0240[\\\](#nt106){ref-type="table-fn"} 1.48±0.0393[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 6\. Tns1 0.14±0.0282[\\](#nt105){ref-type="table-fn"} 0.37±0.0587[∧](#nt107){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 0.32±0.0272[\\\](#nt106){ref-type="table-fn"} 1.52±0.0506[+++](#nt112){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 7\. Lcn2 0.17±0.0090[\\\](#nt106){ref-type="table-fn"} 0.70±0.0604[∧∧∧](#nt109){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 0.26±0.0057[\\\](#nt106){ref-type="table-fn"} 1.53±0.1460[+++](#nt112){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 8\. Itga2 0.19±0.0360[\\](#nt105){ref-type="table-fn"} 0.90±0.1579[∧∧](#nt108){ref-type="table-fn"} 0.27±0.0064[\\\](#nt106){ref-type="table-fn"} 1.44±0.1289[+++](#nt112){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 9\. Npr3 0.21±0.0356[\\](#nt105){ref-type="table-fn"} 0.93±0.1878[∧∧](#nt108){ref-type="table-fn"} 0.35±0.0374[\\](#nt105){ref-type="table-fn"} 2.09±0.1123[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 10\. Klhdc7a 0.21±0.0110[\\](#nt105){ref-type="table-fn"} 0.43±0.0740[∧](#nt107){ref-type="table-fn"} [\](#nt104){ref-type="table-fn"} 0.34±0.0249[\\\](#nt106){ref-type="table-fn"} 1.27±0.0792[+++](#nt112){ref-type="table-fn"} 11\. Sorcs2 0.25±0.0625[\\](#nt105){ref-type="table-fn"} 0.68±0.0667[∧](#nt107){ref-type="table-fn"} 0.34±0.0026[\\\](#nt106){ref-type="table-fn"} 1.34±0.1383[+++](#nt112){ref-type="table-fn"} 12\. Clca4 0.26±0.0526[\\\](#nt106){ref-type="table-fn"} 1.70±0.3294[∧∧∧](#nt109){ref-type="table-fn"} 0.53±0.0504[\\](#nt105){ref-type="table-fn"} 2.69±0.1611[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 13\. 2310007B03Rik 0.25±0.0269[\\\](#nt106){ref-type="table-fn"} 0.55±0.0100[∧∧](#nt108){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 0.59±0.0356[\\](#nt105){ref-type="table-fn"} 1.12±0.0751[++](#nt111){ref-type="table-fn"} 14\. Muc20 0.31±0.0248[\\\](#nt106){ref-type="table-fn"} 0.41±0.0573[\\](#nt105){ref-type="table-fn"} 0.36±0.0440[\\](#nt105){ref-type="table-fn"} 1.21±0.0852[++](#nt111){ref-type="table-fn"} 15\. Upk3b 0.35±0.0061[\\\](#nt106){ref-type="table-fn"} 0.65±0.0077[∧∧∧](#nt109){ref-type="table-fn"} 0.58±0.0125[\\\](#nt106){ref-type="table-fn"} 1.16±0.0190[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 16\. Galns 0.37±0.0007[\\\](#nt106){ref-type="table-fn"} 0.69±0.0407[∧∧](#nt108){ref-type="table-fn"} 0.44±0.0507[\\](#nt105){ref-type="table-fn"} 1.16±0.1101[+++](#nt112){ref-type="table-fn"} 17\. Slc37a1 0.40±0.0781[\\](#nt105){ref-type="table-fn"} 0.76±0.0342[∧](#nt107){ref-type="table-fn"} 0.61±0.0181[\\\](#nt106){ref-type="table-fn"} 1.30±0.0573[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 18\. Ebf1 0.44±0.0681[\\](#nt105){ref-type="table-fn"} 0.80±0.0313[∧](#nt107){ref-type="table-fn"} 0.54±0.0237[\\](#nt105){ref-type="table-fn"} 1.55±0.1321[+++](#nt112){ref-type="table-fn"} [\\](#nt105){ref-type="table-fn"} 19\. Hrsp12 0.52±0.0435[\\](#nt105){ref-type="table-fn"} 0.80±0.0561[∧∧](#nt108){ref-type="table-fn"} 0.57±0.0382[\\](#nt105){ref-type="table-fn"} 1.19±0.0826[+++](#nt112){ref-type="table-fn"} 20\. Csn3 NA NA NA NA Shown here are means of gene expression levels relative to vehicle control group ± standard errors (SE); the vehicle control group was normalized as 1. Taking Dhrs3 expression as an example, 0.01 and 2.88 in the AGN193109 group and the DEAB+tRA group indicate 100-fold reduction and 2.88-fold induction relative to the vehicle group, respectively. p\<0.05 vs vehicle, p\<0.01 vs vehicle, p\<0.001 vs vehicle; p\<0.05 vs AGN193109, p\<0.01 vs AGN193109, p\<0.001 vs AGN193109; p\<0.05 vs DEAB, p\<0.01 vs DEAB, p\<0.001 vs DEAB; NA: Not amplified. Gene Ontology Analysis {#s2d} ---------------------- The 125 candidate target genes of endogenous tRA/RARs were subjected to GeneGO Metacore™ analysis to determine if they are enriched for a particular set of biological processes and diseases. As shown in [Table S4](#pone.0045725.s007){ref-type="supplementary-material"}, the top 50 biological processes in which the candidate target genes are involved can be broadly categorized into six main groups: 42% of the top 50 biological processes are under the umbrella of developmental processes or morphogenesis, 20% are in cell fate determination, 16% are in response towards stimuli, 14% are in regulation of signal transduction, 4% are in cell communication, and the remaining 4% are in other processes ([Figure 6A](#pone-0045725-g006){ref-type="fig"}). These genes were also reported to be dysregulated in some diseases, including fibrosis, wounds healing, immune system diseases, and renal inflammation ([Figure 6A](#pone-0045725-g006){ref-type="fig"}). ![Gene ontologies, expression in native tissues, and presence of retinoic acid response element (RARE). A.\ Enrichment analysis on the 125 candidate target genes showed that they are associated with many biological processes (green box) and diseases (red box), some of which are highly relevant to renal physiology and pathology. Those marked with an asterisk were reported to be associated with tRA/RAR signaling. The full list of gene ontologies for the 125 candidate target genes are in [Table S4](#pone.0045725.s007){ref-type="supplementary-material"}. B.* (i) The 20 validated target genes of endogenous tRA/RARs were reported to be involved in retinol metabolism (A), cell-cell, cell-substrate interaction (B), ureteric bud branching (C), immune/inflammatory processes (D), oxidative stress (E), repair/regeneration (F), ion/solute/water transport and metabolism (G), and gene transcription/translation (H); functions of 9930023K05Rik, Klhdc7a, Sorcs2, and 2310007B03Rik are not yet known. Pink: GeneGO Metacore™, Yellow: DAVID [@pone.0045725-Huang1], Green: GeneGO Metacore™ and DAVID, Blue: additional literature review [@pone.0045725-Zeisberg1], [@pone.0045725-Behr1], [@pone.0045725-Levashova1], [@pone.0045725-Starkey1], [@pone.0045725-Pradervand1], [@pone.0045725-Bonilla1], [@pone.0045725-Zhang1]. The full list of gene ontologies for the 20 validate target genes are in [Table S5](#pone.0045725.s008){ref-type="supplementary-material"}. (ii) Expression of the validated genes in native samples, available through Gene Expression Omnibus datasets (Ω: GSE6290, θ: GSE7891), was reviewed. E11.5 UB: embryonic day 11.5 ureteric bud cells from mice (three samples), E15.5 mCD: embryonic day 15.5 medullary collecting duct cells from mice (three samples), rat IMCD: inner medullary collecting duct cells from rat (one sample from 6-week-old and two samples from 10-week-old). +: present, −: absent, ud: undetermined; ?: present in 6-week-old sample but absent from 10-week-old samples. (iii) The number of direct-repeat 5 (DR5) RARE that are present in the validated genes in whole mouse genome is summarized here, based on report from Lalevee et al. [@pone.0045725-Lalevee1]. \: At least one of the DR5 RAREs is present within 10 kb from transcriptional start site and from gene end.](pone.0045725.g006){#pone-0045725-g006} Expression of the validated genes in native tissues was determined using two GEO series deemed relevant to mIMCD-3 cells: (i) E11.5 mouse UBs and E15.5 mouse medullary CD cells (GEO accession: GSE6290), contributed by GenitoUrinary Development Molecular Anatomy Project (GUDMAP) (<http://www.gudmap.org/>), and (ii) inner medullary CD cells derived from 6- and 10-week-old rat kidney (GEO accession: GSE7891), contributed by Uawithya et al [@pone.0045725-Uawithya1]. There were three biological replicates in each of the database; a gene was considered as positively expressed if at least two out of three replicates had positive hybridization signal. As shown in [Figure 6B](#pone-0045725-g006){ref-type="fig"}, expression of all the 20 genes, except for 9930023K05Rik, 2310007B03Rik, and Itga2, was found in at least one native sample of UB/CD cells. In addition, by referring to a recently published database of DR5 RARE in mouse whole genome [@pone.0045725-Lalevee1], it was found that 13 out of the 20 validated genes contain at least one DR5 RARE. Of these 13 genes, 11 of them have at least one DR5 RARE within 10 kb from the transcription start sites and from gene ends ([Figure 6B](#pone-0045725-g006){ref-type="fig"}), which are regions deemed highly relevant for transcriptional activity modulation [@pone.0045725-Lalevee1]. Gene ontologies for each validated gene were also examined and classified since these genes showed the greatest amount of change ([Figure 6B](#pone-0045725-g006){ref-type="fig"}). These genes were reported to be involved in regulating the conversion of tRal into tRol (Dhrs3), regulating UB branching and kidney development (Ppbp and Bmp7), maintaining CD cell polarity and integrity by regulating cell-cell and cell-substrate interactions (Tns1 and Itga2), regulating cellular responses to stress and modulating tissue repair (Sprr1a, Cpm, Lcn2, and Bmp7), regulating water and solute transport mechanisms, and kidney stone formation (Npr3, Foxa1, Clca4, Slc37a1, and Galns). The known functions of these identified tRA/RAR target genes might help in understanding how tRA/RARs function at the molecular level in the UB/CD cell lineage and guide future studies. In addition, further studies of the function of some genes with unknown functions (9930023K05Rik, Klhdc7a, Sorcs2, and 2310007B03Rik) or functions not yet associated with UB/CD cells and the kidney (Ebf1, Upk3b, and Hrsp12) might not only help to better understand the functions of these genes, but also might help to discover novel functions of the UB/CD cell lineage. The complete list of gene ontologies of the validated genes was listed in [Table S5](#pone.0045725.s008){ref-type="supplementary-material"}. Discussion {#s3} ========== In this study, we have detected the presence of constitutive RAR activation following continuous stimulation by endogenous tRA in mIMCD-3 cells ([Figure 2](#pone-0045725-g002){ref-type="fig"}). Specifically, AGN193109 and DEAB treatment reduced RARE-luciferase activity, indicating the presence of basal tRA/RAR signaling in mIMCD-3 cells; this effect was abolished by exogenous tRA, in a dose-dependent manner, confirming the specificity of AGN193109 and DEAB. Interestingly, exogenous tRA treatment alone resulted in only weak induction of RARE-luciferase activity, although induction was 1.75-fold higher compared with the vehicle control group when exogenous tRA and DEAB were added simultaneously to the cells. We interpreted this to mean that there was already constitutive RAR activation by endogenous tRA in mIMCD-3 cells and that further activation was limited by low cellular uptake of exogenous tRA, availability of RARs and/or transcriptional co-regulators. On the other hand, when endogenous tRA synthesis was abolished by DEAB, enhanced cellular uptake of exogenous tRA and/or enhanced expression of RARs and co-factors might have taken place, thus potentiating RARE-luciferase induction to significantly higher than the basal level ([Figure 2](#pone-0045725-g002){ref-type="fig"} Bii). Thus, for cells with constitutively active endogenous tRA/RAR signaling such as mIMCD-3 cells, instead of treating them directly with exogenous tRA or RAR agonists to identify tRA/RAR target genes, it would be more appropriate to determine genes differentially regulated when endogenous tRA/RAR signaling was inhibited ([Figure 1](#pone-0045725-g001){ref-type="fig"}). By treating mIMCD-3 cells with AGN193109 and DEAB, with and without exogenous tRA, we were able to catalogue a list of candidate target genes specifically regulated by both endogenous tRA and RARs. Note that when all shortlisted genes were overlapped directly the intersected genes appeared to be the minority ([Figure 4A](#pone-0045725-g004){ref-type="fig"}). However, only when the top 20 and top 10 most highly regulated genes were overlapped, did the intersected genes become the majority ([Figure 4B and 4C](#pone-0045725-g004){ref-type="fig"}). This finding was also reflected by the greater fold-changes observed for the intersected genes compared with the non-intersected genes ([Figure 4A](#pone-0045725-g004){ref-type="fig"}). Thus, the data suggest that RAR-independent tRA signaling or tRA-independent RAR signaling might regulate specific gene subsets, respectively. Nonetheless, genes most dependent on retinoid signaling in mIMCD-3 cells were those regulated by the canonical tRA/RAR signaling. We also found that endogenous tRA/RAR signaling plays a primary role as "inducer" to support gene expression, rather than as "suppressor" to inhibit gene expression in mIMCD-3 cells, since genes suppressed by AGN193109 and by DEAB greatly exceeded genes induced by AGN193109 and by DEAB, in number of genes and magnitude of regulation ([Figure 5](#pone-0045725-g005){ref-type="fig"}). Among the 20 validated target genes of endogenous tRA/RARs, including Foxa1 and Bmp7 identified in the pilot study ([Figure 6B](#pone-0045725-g006){ref-type="fig"}), some were reported to be regulated by tRA and/or RARs in previous reports, i.e., Dhrs3 (up-regulated) [@pone.0045725-Cerignoli1], [@pone.0045725-Feng1], Sprr1a (down-regulated) [@pone.0045725-Fujimoto1], [@pone.0045725-Tesfaigzi1], Cpm (down-regulated) [@pone.0045725-Kreutz1], Itga2 (up-regulated) [@pone.0045725-Froeling1], Ebf1 (up-regulated) [@pone.0045725-Chen1], Lcn2 (up-regulated) [@pone.0045725-Nelson1], Bmp7 (up-regulated) [@pone.0045725-Paralkar1], and Foxa1 (up-regulated) [@pone.0045725-Balmer2]. The remaining genes, i.e., Ppbp, 9930023K05Rik, Tns1, Klhdc7a, Sorcs2, 2310007B03Rik, Galns, Npr3, Clca4, Upk3b, Slc37a1, and Hrsp12, represent novel target genes of endogenous tRA/RARs, at least in mIMCD-3 cells. Except for 9930023K05Rik, 2310007B03Rik, and Itga2, all validated genes are expressed in native samples of UB/CD cells, suggesting that they are of physiological relevance. We also noted that half of these genes contain at least one DR5 RARE; among them, almost all have at least one DR5 RARE within 10 kb from transcription start sites and from gene ends, which are the regions deemed most relevant for transcriptional activity modulation [@pone.0045725-Lalevee1]. These genes might have been regulated via their DR5 RARE(s), although other mechanisms, e.g., through other forms of RARE or though RARE-independent mechanisms, might also be involved. It is also worth noting that several well-established target genes of tRA that have more than one functional DR5 RAREs, e.g., Rarb and Cyp26a1 [@pone.0045725-Balmer2], were not shortlisted in the microarray experiments. In fact, there was no statistically significant difference in the retinoid binding proteins, metabolizing enzymes, and the retinoid nuclear receptors, between the vehicle, inhibitors alone, and inhibitors with exogenous tRA groups, as shown in the microarray data. This could be explained by the cell type-dependent activity of tRA, but treatment length might also have an impact. Additional studies at earlier and later time points should complement the existing profile of endogenous tRA/RAR target genes in mIMCD-3 cells. Among the top 50 biological processes that were significantly enriched ([Figure 6A](#pone-0045725-g006){ref-type="fig"}), 42% was within the category of development and morphogenesis, in agreement to the established role for tRA as a morphogen. Apart from kidney development and other previously reported tRA/RAR-dependent biological processes [@pone.0045725-Xu1], [@pone.0045725-Xu2], these genes are also involved in responses to external stimuli, including oxygen levels, hormones, and chemicals, consistent with the functions of CD cells to withstand the highly variable oxygen levels, particularly in the medulla, as well as in regulating water and solute transport. Besides the classical functions of CD cells, emerging publications are suggesting novel functions for these cells, e.g., in regulating inflammation, epithelial-mesenchymal transition and fibrogenesis [@pone.0045725-Butt1], [@pone.0045725-Fujiu1], as well as in defense against bacterial infection [@pone.0045725-Chassin1], etc. Moreover, post-natal vitamin A deficiency has been associated with renal anomalies, including increased incidence of pyelonephritis, kidney inflammation and fibrosis, polyuria and dysregulated urinary ion/solute content, urolithiasis, and delayed tissue repair [@pone.0045725-Xu1], [@pone.0045725-Kavukcu1], [@pone.0045725-Kancha1], [@pone.0045725-vanLeersum1], [@pone.0045725-Zile1]. Thus, the roles for the tRA/RAR target genes identified in this study in the aforementioned renal anomalies associated to post-natal vitamin A deficiency deserve further studies. Among the validated target genes, the top three genes that were most dramatically regulated by tRA/RARs could have important roles in mediating the biological processes and diseases mentioned above. For instance, Ppbp could be regulating kidney development [@pone.0045725-Levashova1], inflammation [@pone.0045725-Walz1], and defense against bacterial infection [@pone.0045725-Nguyen1]; Sprr1a is another gene highly dependent on tRA/RAR signaling in mediating tissue repair and regeneration [@pone.0045725-Starkey1] relevant to kidney injury. It is important that while some validated genes, e.g., Dhrs3, Cpm, and Tns1, are expressed as cellular components, other genes, e.g., Ppbp and Lcn2, are expressed as secreted proteins. This supports the view that tRA/RARs not only regulate CD cell function in an autocrine manner, but also modulate the function and health of the kidney or the whole body in a non-autocrine manner. The importance of tRA/RAR signaling in CD cells is also supported by the regulation of Dhrs3 by AGN193109, DEAB and exogenous tRA. Dhrs3 is involved in converting retinaldehyde back to vitamin A to reduce tRA biosynthesis [@pone.0045725-Feng1]. The pronounced regulation of Dhrs3 might represent a self-regulatory mechanism of endogenous tRA/RAR signaling in CD cells. In summary, we have detected endogenous tRA/RAR activity in mIMCD-3 cells, an in vitro* cell model of the UB/CD cell lineage. Using this model, a panel of genes regulated by both endogenous tRA and RARs has been identified. Many of these genes represent novel target genes of endogenous tRA/RARs. We propose that endogenous tRA/RARs may play crucial roles in kidney development and in maintaining normal function of CD cells and the kidney, at least in part, by regulating these tRA/RAR target genes. Given the complexity of retinoid signaling, which is highly dependent on cell type and environment, further studies are warranted to examine the regulation of these genes by tRA/RAR signaling in vivo and to explore the potential role of these endogenous tRA/RAR target genes in normal and abnormal renal function. Materials and Methods {#s4} ===================== Cell Culture {#s4a} ------------ mIMCD-3 cells (LGC Standards, Middlesex, UK) were routinely grown in DMEM-F12 (PAA Laboratories Ltd, Somerset, UK) containing penicillin (100 IU/ml), streptomycin (100 µg/ml) (PAA Laboratories Ltd) and amphotericin B (2.5 µg/ml) (Life Technologies Ltd, Paisley, UK) herein referred as complete medium, supplemented with 5% fetal calf serum (FCS; Life Technologies Ltd), at 37°C and 5% CO~2~. Immunocytochemistry {#s4b} ------------------- Cells were seeded at 6×10^5^ cells/35 mm^2^ dish in 2 ml complete medium, supplemented with 5% FBS. After an overnight culture until confluent, cells were fixed with 5% ice-cold formalin for 15 min on ice, followed by washing with three changes of PBS, 5 min each wash. Cells were then permeabilized with 0.1% Triton X-100 for 3 min and washed with three changes of PBS, 5 min each wash. After fixation and permeabilization, cells were incubated with 1% BSA for 2 h at room temperature to reduce non-specific antibody binding, followed by 1 h incubation of rabbit anti-AQP2 IgG (dilution 1∶150, Millipore, Watford, UK), or mouse anti-E-cad IgG2a (dilution 1∶50, BD Biosciences, Oxford, UK), at room temperature. Cells incubated with non-immune rabbit IgG and non-immune mouse IgG2a (Insight Biotechnology Ltd, Middlesex, UK) at the same concentrations served as negative controls for anti-AQP2 IgG and for anti-E-cad IgG2a, respectively. Cells were then washed with three changes of PBS, 5 min each wash, and were incubated with goat anti-rabbit conjugated with Alexa Fluor 488 (dilution 1∶1000, Life Technologies Ltd) for AQP2 detection or goat anti-mouse conjugated with Alexa Fluor 555 (dilution 1∶1000, Life Technologies Ltd) for E-cad detection, for 1 h at room temperature. At the end of secondary antibody incubation, cells were washed with three changes of PBS, 5 min each wash. One milliliter of fresh PBS was then pipetted onto the cell monolayer and fluorescence microscopy was performed immediately. Microscopy {#s4c} ---------- Phase-contrast and fluorescence microscopy was performed on a Nikon Eclipse TE2000-S epifluorescence microscope equipped with a standard RGB filter wheel (Nikon Instruments Europe B.V., Amstelveen, The Netherlands). Images were captured with a DXM1200F Nikon digital camera (Nikon UK Limited, Surrey, UK), then processed and merged with Adobe Photoshop (Adobe Systems Europe Ltd, Uxbridge, UK). Reagents {#s4d} -------- tRA (Sigma-Aldrich Company Ltd, Dorset, UK) was reconstituted in 100% ethanol to 10 mM; DEAB [@pone.0045725-Russo1] (Sigma-Aldrich Company Ltd) was reconstituted in 100% dimethylsulphoxide (DMSO) (Sigma-Aldrich Company Ltd) to 25 mM; AGN193109 [@pone.0045725-Johnson1], was reconstituted in 100% DMSO to 1 mM. All reagents were first diluted with their respective diluents to 1000× more concentrated than the working concentration, then diluted 1000× with culture medium to the working concentrations immediately before treatment was commenced. Control groups were treated with 0.1% ethanol and/or 0.1% DMSO. Transient Transfection {#s4e} ---------------------- Lipofectamine™ LTX, Plus™ reagents and Opti-MEM® I Reduced Serum Medium (Life Technologies Ltd) were used in transient transfection of plasmid DNA according to the manufacturer's instructions. pmaxGFP (Lonza Wokingham Ltd, Berkshire, UK) was used to assess transfection efficiency; the transfection efficiency was estimated to be around 70% ([Figure S1](#pone.0045725.s001){ref-type="supplementary-material"}). pGL3-RARE-luciferase plasmid [@pone.0045725-Hoffman1], was used to assess RAR-dependent endogenous tRA activity. pCI-β-galactosidase plasmid (Promega, Madison, USA) was used for normalization of RARE-luciferase activity. Background signal was determined from wells treated with transfection medium only without plasmid DNA. Ratio of pGL3-RARE-luciferase: pCI-β galactosidase was 5∶1 when co-transfected into mIMCD-3 cells. Cells were lysed with Reporter Lysis Buffer (Promega UK, Southampton, UK); luminescence signal was detected with Luciferase Assay System and Beta-Glo® Assay System (Promega UK) and was expressed as relative light unit (RLU). RLU of treatment group was normalized to vehicle control group and expressed as fold-change. RNA Extraction and Reverse Transcription-quantitative Polymerase Chain Reaction (RT-qPCR) {#s4f} ----------------------------------------------------------------------------------------- Total RNA was extracted using RNeasy® Mini Kit and QIAshredder spin column (Qiagen Ltd, West Sussex, UK) following manufacturer's protocol. Quantity and quality of total RNA was determined with a NanoDrop® ND-1000 machine (Labtech International Ltd, East Sussex, UK); an absorbance ratio of 260/280 and 260/230 for all total RNA samples were within the range of 1.8--2.0. Reverse transcription was performed using Oligo-d(T) primer and Omniscript® Reverse Transcription Kit (Qiagen Ltd) following manufacturer's protocol. qPCR was performed using Taqman® universal master mix and Taqman® Gene Expression Assays (Life Technologies Ltd), as listed in [Table 4](#pone-0045725-t004){ref-type="table"}. Conditions for qPCR reactions were as follows: 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s and 60°C for 1 min. Gene expression was computed using the 2−^ΔΔ^CT formula with glyceraldehyde 3-phosphate dehydrogenase (Gapdh) as an internal control; gene expression of the vehicle control group was normalized as 1. Expression of Gapdh was found to be stable across all experimental groups ([Figure S2](#pone.0045725.s002){ref-type="supplementary-material"}). 10.1371/journal.pone.0045725.t004 ###### Assay ID and amplicon length of Taqman® Gene Expression Assays used. ![](pone.0045725.t004){#pone-0045725-t004-4} Gene Assay ID Amplicon length ------ --------------- ----------------- ----------------- 1\. Bmp7 Mm00432101_m1 82 2\. Clca4 Mm00519742_m1 68 3\. Cpm Mm01250796_m1 63 4\. Csn3 Mm02581554_m1 99 5\. Dhrs3 Mm00488080_m1 84 6\. Ebf1 Mm00432948_m1 59 7\. Foxa1 Mm00484713 \_m1 68 8\. Galns Mm00489576_m1 80 9\. Gapdh Mm99999915 \_m1 107 10\. Hrsp12 Mm00476177_m1 73 11\. Itga2 Mm00434371_m1 63 12\. Klhdc7a Mm00557861_s1 117 13\. Lcn2 Mm01324470_m1 84 14\. Muc20 Mm00524818_m1 65 15\. Npr3 Mm00435329_m1 63 16\. Pax2 Mm01217939\_ m1 55 17\. Ppbp Mm00470163_m1 62 18\. Slc37a1 Mm00461949_m1 78 19\. Sorcs2 Mm00473050_m1 64 20\. Sprr1a Mm01962902_s1 94 21\. Tns1 Mm00452886_m1 91 22\. Upk3b Mm00558406_m1 89 23\. Wnt7b Mm00437357\_ m1 88 24\. 9930023K05Rik Mm00554061_m1 110 25\. 2310007B03Rik Mm00549644_m1 73 Microarray {#s4g} ---------- Samples were prepared using Ambion® WT kit (Life Technologies Ltd), Affymetrix fragmentation and labeling kit, and Affymetrix wash, stain, and hybridization kit (Affymetrix UK Ltd, High Wycombe, UK). Mouse Gene 1.0 ST array (Affymetrix UK Ltd) was used for gene expression profiling. Agilent Nano Chip and bioanalyser (Agilent Technologies UK Ltd, Edinburgh, UK) were used to assess the quality of total RNA; there was no evident of RNA degradation, as shown in [Figure S3](#pone.0045725.s003){ref-type="supplementary-material"}. Nucleotide size of sense-cDNA was also examined before and after fragmentation. An Affymetrix® Gene Control Console software was used for data normalization using the Robust Multichip Average method. Candidate target genes were short-listed using an Qlucore Omics Explorer software with multiple comparison analysis, by using Foxa1 as the cut-off threshold of statistical significance (variance = 0.098, q = 0.65 for AGN193109 experiment; variance = 0.053, q = 0.30 for DEAB experiment). The RNA samples used for microarray experiments were subsequently used in the validation experiments. Gene Ontology Analysis {#s4h} ---------------------- Gene ontology analysis was performed using Database for Annotation, Visualization and Integrated Discovery (DAVID) (<http://david.abcc.ncifcrf.gov/>) [@pone.0045725-Huang1] and GeneGO Metacore™ (GeneGO Inc., St. Joseph, Michigan); additional literature reviews were performed to complement gene ontology analysis. Experimental Protocols {#s4i} ---------------------- For reporter assays involving AGN193109 and tRA, mIMCD-3 cells were seeded into a 24-well plate at 4×10^4^ cells/well in 500 µl of DMEM-F12 medium without antibiotics and anti-mycotic, supplemented with 5% FCS. After culturing for 16--24 h to about 70% confluence, cells were co-transfected with pGL3-RARE-luciferase and pCI-β-galactosidase plasmids, and cultured for another 5 h. Medium was then changed to fresh complete medium supplemented with 1% FCS. After an overnight incubation, medium was changed to fresh complete medium supplemented with 1% FCS containing 1 µM AGN193109, with and without 0.01 µM, 0.1 µM, or 1 µM tRA, and cultured for 24 h. For reporter assays involving tRA alone, the protocol was similar as above, except that cells were treated with 0.001 µM, 0.01 µM, or 0.1 µM tRA for 24 h. For reporter assays involving DEAB and tRA, mIMCD-3 cells were seeded into a 24-well plate at 2×104 cells/well in 500 µl of DMEM-F12 medium without antibiotics and anti-mycotic, supplemented with 5% FCS. Cells were cultured for 16--24 h to about 40--50% confluence, and medium was changed to fresh DMEM-F12 without antibiotics and anti-mycotic, supplemented with 1% FCS, and contained 25 µM DEAB or vehicle. Cells were then cultured for 16--24 h until about 70% confluence, after which medium changing was repeated as described above and cells were co-transfected with pGL3-RARE-luciferase and pCI-β-galactosidase plasmids. After an overnight incubation, medium was again changed to fresh DMEM-F12 without antibiotics and anti-mycotic, supplemented with 1% FCS, and contained 25 µM DEAB, with and without 0.0001 µM, 0.001 µM, 0.01 µM, or 0.1 µM tRA. Cells were then cultured for 24 h. For RT-qPCR and microarray experiments, cells were seeded at 1.2--2×105 cells/35 mm^2^ dish in 2 ml complete medium, supplemented with 5% FCS. After an overnight culture, medium was changed to fresh complete medium supplemented with 1% FCS. After an overnight culture, medium was again changed to fresh complete medium supplemented with 1% FCS, containing 1 µM AGN193109 with and without 0.2 µM tRA, or 25 µM DEAB with and without 0.01 µM tRA, and cultured for 24 h. Control groups were treated with vehicle in all aforementioned protocols. For reporter assays, three independent experiments, with triplicate or quadruplicate wells for each group, were performed. For microarray and RT-qPCR experiments, three independent experiments were performed. Statistical Test for Reporter Assay and qPCR {#s4j} -------------------------------------------- For reporter assays and RT-qPCR experiments, One-way Analysis of Variance (ANOVA) statistical analysis with Tukey post-test was performed on log-transformed fold-change values using GraphPad Prism, Version 4.0. p\<0.05 was considered as statistically significant. Supporting Information {#s5} ====================== ###### Transfection efficiency assessed from green fluorescent protein (GFP) expression. Around 60% to 70% of the total mIMCD-3 cells transfected with pmaxGFP plasmid expressed GFP (green). No GFP expression was observed in cells where only transfection reagents were added. Original magnification was 100×. (TIFF) ###### Click here for additional data file. ###### Amplification plot of glyceraldehyde 3-phosphate dehydrogenase (Gapdh). Expression of Gapdh did not vary much between the vehicle control group and 24 h treatment of AGN193109 with and without tRA (A), and of DEAB with and without tRA (B), evident by a tight overlapping amplification plots across all the samples. Representative amplification plots derived from technical triplicates of a single biological experiment are shown here. (TIFF) ###### Click here for additional data file. ###### RNA integrity of samples for microarray experiments. Integrity of RNA samples from AGN193109 experiment (A) and DEAB experiment (B) was examined with bioanalyser. The two sharp major peaks correspond to 18s and 28s ribosomal RNA, respectively. The high RNA Integrity Number (RIN) values, within the range of 9.70--10.00, suggest a good quality of RNA samples with minimum degradation of RNA. There is no evidence of genomic DNA contamination in the RNA samples, given the presence of thin and sharp 18s and 28s RNA peaks, as well as absence of additional peaks other than the expected ribosomal RNA peaks. (TIFF) ###### Click here for additional data file. ###### 133 intersected genes. (XLS) ###### Click here for additional data file. ###### 270 genes regulated by AGN193109. (XLS) ###### Click here for additional data file. ###### 306 genes regulated by DEAB. (XLS) ###### Click here for additional data file. ###### Gene ontology for 125 candidate target genes. (XLS) ###### Click here for additional data file. ###### Gene ontology for validated genes. (XLS) ###### Click here for additional data file. We thank Dr. R. A. S. Chandraratna (Allergan, Irvine, CA) for the kind gift of AGN193109 and Dr. T. M. Underhill (University of British Columbia, Vancouver, BC, Canada) for the kind gift of pGL3-RARE-luciferase plasmid. [^1]: Competing Interests:The authors have declared that no competing interests exist. [^2]: Conceived and designed the experiments: QX YFW. Performed the experiments: YFW. Analyzed the data: YFW QX MA. Contributed reagents/materials/analysis tools: MA. Wrote the paper: YFW QX PDW RJU BMH JTN.	{ "pile_set_name": "PubMed Central" }
INTRODUCTION {#s1} ============ Malignant pleural mesothelioma (MPM) is an aggressive tumour arising from the mesothelial lining of the pleura, which remains compartmentalized for most of the disease course \[[@R1]\] \[[@R2]\]. Current therapeutic approaches include surgery, radiotherapy and chemotherapy. Cisplatin (CDDP) alone or with pemetrexed (PMTX) still represents the current standard of care \[[@R3]\] \[[@R4]\] \[[@R5]\]. Similarly to most solid tumours, the acquired resistance impedes the success of the therapeutic response. Overall patient\'s survival ranges from 8 and 18 months. Furthermore, there is not any diagnostic tool that allows an early detection of the disease \[[@R2]\] \[[@R6]\]. Given the role played from the asbestos in the etiology of the mesothelioma, this tumor is still considered an occupational disease \[[@R7]\] \[[@R8]\] \[[@R9]\]. Mesothelioma is an orphan disease with an increasing worldwide incidence, especially in the developing countries where the asbestos is still exported and not banned \[[@R9]\] \[[@R10]\]. After inhalation asbestos fibers infiltrate pleural spaces and accumulate in black spots. This leads to the induction of pleural fibrosis that might result in cancer transformation \[[@R11]\] \[[@R12]\]. Within the last five decades, the natural products have continuously contributed to drug discovery and development process \[[@R13]\] \[[@R14]\] \[[@R15]\]. Increasing evidences have shown that natural products, including extracts and isolated chemicals, are multi-targeted and can be considered as a model to approach chronic diseases such as cancer. Actually, most natural agents do not induce high level of toxicity and target simultaneously multiple signalling pathways involved in cell growth, apoptosis, invasion, angiogenesis and metastasis \[[@R16]\] \[[@R17]\]. Since cancer is the result of a deregulation of multiple signalling pathways and natural products elicit multi-targeted activities, the latter could hold a great potential for treating human tumors \[[@R18]\] \[[@R6]\]. The long period between asbestos exposure and the development of disease may be a window of opportunity for chemoprevention or dietary interventions. It has been shown that some natural agents such as coffee, resveratrol, curcumin and butein exert anticancer activities in mesothelioma cellular systems \[[@R2]\] \[[@R10]\] \[[@R19]\] \[[@R19]\] \[[@R20]\]. Observational studies have shown that high adherence to a Mediterranean diet, which encompasses a combination of characteristic foods such as fruits, vegetables fresh produce, fish and seafood, nuts, legumes/pulses and olive oil \[[@R21]\], is associated with a significant reduction in the risk of overall cancer mortality (10%), colorectal cancer (14%), prostate cancer (4%) and aero digestive cancer (56%) \[[@R22]\]. The most diffuse phytochemicals in nature are the classes of polyphenols. For long time, polyphenols have been considered only for their antioxidants properties. Recent data have revealed that polyphenols affect different cell signalling pathways. This also occurs through the modulation of microRNAs thereby affecting the expression of their mRNA targets whose encoded proteins are critical components of different pathways \[[@R23]\] \[[@R23]\]. Globe artichoke (Cynara scolymus) is a fundamental component of Mediterranean diet \[[@R24]\]. It has been reported that the edible parts (receptacles with inner and intermediate bracts) and leaves of artichokes represent a potent source of polyphenols fractions. Recent data have revealed that polyphenols affect different cell signalling pathways and exert their anticancer effects along the different steps of carcinogenesis \[[@R25]\] \[[@R26]\] \[[@R27]\]. Furthermore, the main components of Cynara scolymus are caffeoylquinic acid derivatives (cynarin and chlorogenic acid), flavonoids (luteolin and apigenin) and bitters (cynaropicrin) \[[@R28]\] \[[@R29]\] \[[@R28]\]. Several in vitro and in vivo experiments have shown that Cynara scolymus exhibits choleretic, hepatoprotective, antibacterial, antinflammatory, antithrombotic and hypocolesterolemic properties \[[@R30]\] \[[@R31]\] \[[@R32]\] \[[@R33]\]. Artichoke extracts are reported to induce apoptosis and cytotoxic effects in cancer cells \[[@R34]\] \[[@R35]\] \[[@R36]\]. In the present report we aimed to demonstrate the anti-cancer activity of the artichoke leaf extract through direct experimental tests, through the normalizing effect of the extract on the cancer metabolic alterations and through the evidence of extract antitumoral activity due to its impact on signalling pathways of oncogenic significance. We found that artichoke leaf extracts (freeze-dried extract prepared as indicated in the methods section by the ABOCA company- <http://www.aboca.com>) severely affect in vitro and in vivo mesothelioma tumorigenicity. Indeed, the artichoke leaf extract significantly reduces cell proliferation and colony formation of diverse mesothelioma cell lines. It also promotes apoptosis and restrains mesothelioma cell migration and invasion. It also impairs engraftment and reduces tumor volume of xenografted mesotheliomas. These effects are similar to those induced by pemetrexed. Protein array analyses reveal that the artichoke leaf extract activates distinct set of proteins to those of pemetrexed or cisplatin that might be critical mediators of its antitumoral activities. RESULTS {#s2} ======= The artichoke leaf extract inhibits MPM cell growth and proliferation {#s2_1} --------------------------------------------------------------------- We aimed first to test the effects of the artichoke extract on mesothelioma cell growth and proliferation. To this end, we treated MSTO-211H, MPP-89 and NCI-H28 mesothelioma cell lines and untransformed mesothelial cells, HMC (Figure [1A--1C](#F1){ref-type="fig"}) with increasing concentrations (ranging from 3 to 200 μg/ml) of artichoke leaf extract for 72 hrs. We determined the half-maximal concentration of growth inhibition (IC50) for the extract phytocomplex in MPM cells (Figure [1A--1C](#F1){ref-type="fig"}, [Supplementary Table 1](#SD1){ref-type="supplementary-material"}). We found that the artichoke extract inhibited cell viability in a dose dependent manner (Figure [1A--1C](#F1){ref-type="fig"}). Moreover, MSTO-211H cells treated with the artichoke extract showed a change in the cellular morphology both in a short-term assay that in long term as showed in [Supplementary Figure 1](#SD1){ref-type="supplementary-material"}. In contrast, HMC cells were more resistant to the growth inhibitory effect of the artichoke extract (Figure [1A--1C](#F1){ref-type="fig"}). Next, we performed a colony-forming assay to evaluate the capability of mesothelioma cells to form colonies after the removal of the extract. We found that the artichoke leaf extract inhibited the colony forming ability of MSTO-211H, MPP-89 and NCI-H28 cells (Figure [1D--1F](#F1){ref-type="fig"}). Altogether these findings indicated that the artichoke leaf extract treatment is highly effective on cell proliferation and colony forming ability of mesothelioma cell lines. ![The artichoke leaf extract strongly affects the growth of MPM cells in a dose dependent manner\ A, B, C. Viability of MPM cells (MSTO-211H, MPP89, NCI-H28) and normal untransformed mesothelial cell line (HMC) treated for 72 hrs with Cynara scolymus leaf extracts. The IC-50 value was indicated by a dotted line in each panel. Data are represented as mean +/− SD. Statistics (t-test): p \< 0.05. D, E, F. Colony forming assay. Histograms showing average colony counts from duplicate experiments. Bars indicate the average of three independent experiments. Statistics (t-test): p \< 0.05.](oncotarget-06-18134-g001){#F1} The artichoke leaf extract induces apoptosis of MPM cell lines {#s2_2} -------------------------------------------------------------- The induction of apoptosis is a pivotal event for successful cancer treatment by natural agents. To this end, we assessed whether the artichoke leaf extract induced apoptosis of mesothelioma cell lines using different approaches. Cytofluorimetric analysis revealed that the artichoke extract induced the appearance of a subG1 peak in MSTO-211H and MPP-89 cells. This effect was dose-dependent (Figure [2A--2D](#F2){ref-type="fig"}). Furthermore, cells treated with the artichoke leaf extract for 24 hrs became Annexin V-positive in a dose-dependent manner (Figure [2E--2F](#F2){ref-type="fig"}). We also found that the extract treatment led to increased cleavage of caspase 3, caspase7 and Parp (Figure [2G](#F2){ref-type="fig"}). Comet assay performed in MSTO-211H and in HMC cells revealed that Cisplatin (7, 5 μg/ml, for 20 h) treatment induced DNA damage of both cell lines ([Supplementary Figure 2A, 2B](#SD1){ref-type="supplementary-material"}). Interestingly, the exposure to the artichoke leaf extract, used at not apoptotic concentration, 3 μg or 6 μg/ml, did not induce DNA damage ([Supplementary Figure 2A, 2B](#SD1){ref-type="supplementary-material"}) and reduced that induced by CDDP in HMC cells ([Supplementary Figure 2A](#SD1){ref-type="supplementary-material"}). Overall, these results clearly demonstrate that the artichoke leaf extract affects cell viability of MPM cell lines by inducing apoptosis. ![The artichoke leaf extract induces apoptosis of MPM cells\ A, B, C, D. Percentage of subG1 cell population of MPM cells (MSTO-211H and MPP-89) treated with the indicated doses of the artichoke extract for 48 or 72 hrs. Error bars represent mean +/− SD. Statistics (t-test): p \< 0.05. E, F. Histograms showing the percentage of Annexin V+/PI− over vehicle. MPM cells (MSTO-211H, MPP-89) were treated at the indicated doses of the artichoke extract for 24. Error bars represent mean +/− SD. Statistics (t-test): p \< 0.05. G. Representative protein gel blot of whole cell lysates obtained from MSTO-211H and MPP-89 cell lines treated for 24 hrs with 100 μg/ml of artichoke leaf extract and stained with anti-cleaved parp, anti-caspase-3 and anti-caspase-7 antibodies. Actin staining was used as loading control.](oncotarget-06-18134-g002){#F2} The artichoke leaf extract severely impairs migration and invasion of mesothelioma cell lines {#s2_3} --------------------------------------------------------------------------------------------- It has been previously reported that dietary phenolic acids, monophenols and polyphenols possess inhibitory properties against the invasive and metastatic behaviours of different cancer cells lines. To this end, we investigated whether the artichoke leaf extract impaired migration and invasiveness of MPM cell lines. First, we performed scratch wound closure assay (Figure [3A, 3B](#F3){ref-type="fig"}) in MPM cell lines treated with different, not apoptotic, concentrations of the extract. We found that the artichoke extract inhibited, in a time dependent manner, migration of MSTO-211H and MPP89 as wound closure occurred slower than that of the control cells (Figure [3A--3B](#F3){ref-type="fig"}). As matter of fact, the artichoke leaf extract at the concentration of 6 μg/ml and 12 μg/ml inhibited MSTO-211H cells migration of 50% and 35% respectively as well as in MPP89 cell line (Figure [3A--3B](#F3){ref-type="fig"}). Second, we performed invasion assay in MSTO-211H cells using a 24-well chamber with a non-coated 8-mm pore size filter in the presence of different concentrations of extract (6 μg and 12 μg/ml). As shown in Figure [3C](#F3){ref-type="fig"}, the artichoke leaf extract markedly reduced the invasion of MSTO-211H cells. Altogether these results indicated that the artichoke extract inhibited migration and invasion of mesothelioma cell lines. ![The artichoke leaf extract impairs the migration and invasion of MPM cells\ A, B. Left side: representative micrographs of wound healing closure assays from MSTO-211H and MPP-89 cells treated for 36 hrs with 12, 5 μg/ml of artichoke leaf extract. Right side: histogram showing the healing closure efficiency of the cells treated with vehicle or the artichoke extract (6 μg and 12 μg) after different times of treatment (0, 12, 24 and 36 hrs). Bars indicate the average of three independent experiments. Statistics (t-test): p \< 0.05. C. Percentage of invading cells over vehicle. MSTO-211H cells were treated at the indicated doses of extract. Error bars represent mean +/− SD. Statistics (t-test): p \< 0.05.](oncotarget-06-18134-g003){#F3} The artichoke leaf extract affects in vivo tumor growth {#s2_4} --------------------------------------------------------- We aimed to assess in vivo the antitumoral effects of the extract. First, we evaluated whether the artichoke leaf extract could affect the engraftment of MSTO-211H cells. To this end, MSTO-211H cells were treated for 24 hrs with vehicle or artichoke extract at 50 μg/ml. Cell suspensions were subcutaneously injected into CD1 mice. As shown in Figure [4A](#F4){ref-type="fig"}, cells pre-treated with the extract engrafted less efficiently than vehicle-treated cells when inoculated into CD1 mice. We next evaluated the efficacy of the extract to inhibit the growth of xenografted mesotheliomas subcutaneously implanted in CD1 nude mice. To this end, CD1 mice were subcutaneously transplanted with MSTO-211H (2 × 10^6^). At the evidence of tumor progression, mice were randomly divided in five different groups (n = 6) and were beverage with vehicle or artichoke leaf extract at different concentrations (25, 50 and 75 mg/ml) in drinking water (Table [2](#T2){ref-type="table"}). Dietary feeding of the artichoke leaf extract for 3 weeks reduced significantly and dose dependent the growth of xenografted mesothelioma tumors (Figure [4B](#F4){ref-type="fig"}). This effect was similar to that induced by PMTX, a drug commonly employed in the treatment of mesothelioma (Figure [4B](#F4){ref-type="fig"}). To evaluate the proliferation index xenografted tumors were stained for Ki67 expression. Strikingly, we found that xenografted tumors derived from the extract treated mice exhibited significantly lower Ki67 expression than those treated with vehicle (Figure [4C, 4D](#F4){ref-type="fig"}). We also found that artichoke leaf extracts sensitized MSTO-211H and NCI-H28 mesothelioma cell lines to pemetrexed-induced cell killing ([Supplementary Figure 3A-B](#SD1){ref-type="supplementary-material"}). Similarly, artichoke leaf extract at 75 mg/ml potentiated the antitumoral effect of pemetrexed on tumor volume of xenografted tumors ([Supplementary Figure 3C](#SD1){ref-type="supplementary-material"}). Altogether these findings indicate that the artichoke leaf extract can affect in vivo the growth of mesothelioma cell lines. ![The artichoke leaf extract inhibits in vivo mesothelioma tumor growth\ A. Artichoke extract inhibits engraftment of MSTO-211H cells. Suspensions of 2 × 10^6^ MSTO-211H cells were pre-treated with either vehicle or the artichoke leaf extract (50 μg/ml) for 24 hrs and subcutaneously injected into CD1 mice as described in Material and Methods. No further treatment was given to mice. Horizontal bars represent the average tumor volume of the vehicle (n = 6) and the artichoke extract (n = 6) treated engrafted tumors. Tumors were collected 24 days after MSTO-211H cells injection. Statistics (t-test): p \< 0.05. B. The artichoke leaf extract beverage inhibits in vivo mesothelioma tumor progression. Tumour volumes of mice (n = 6) treated with either vehicle, pemetrexed or the artichoke extract in drinking water are reported. Statistics (t-test): p \< 0.05. C. Representative micrographs of the excised tumours stained with anti-Ki-67 antibody. Scale bar, 100 μm. D. Histograms showing the percentage of Ki-67 positive nuclei scored in seven fields randomly chosen for each tumours. Statistics (t-test): p \< 0.05.](oncotarget-06-18134-g004){#F4} ###### Artichoke leaf extract components --------------------------------------------------------------------------------------------------------------------- Classes of Compounds Levels found in Artichoke freeze-dried extract ----------------------------------------------------- ------------------------------------------------- ------------- PolyPhenols (%) total\ 34.251\ of which lignins total\ 1.98\ of which tannins total\ 5.96\ of which Caffeoylquinic acid derivatives total\ 10.33\ of which Chlorogenic acid\ 4.07\ of which phenypropanoid derivatives total\ 7.01\ of which phenols and phenolic acids total\ 0.703\ of which flavonoids total\ 8.27\ of which salicilates total 0.001 Terpenes (%) total\ 1.5\ of which Cynaropicrin 1.5 Organic Acids (%) total 6.3 Protein (%) total\ 5.6\ of which water soluble proteins 5.6 Amino Acids fre (%) total 0.53 Polysaccharides (%) total\ 1.7\ of which soluble dietary fiber\ 0.8\ of which Insoluble dietary fiber 0.9 Saccharides (%) total\ 24.69\ of which monosaccharides\ \ Fructose\ 2.58\ Glucose\ 8.02\ of which disaccharides\ \ Sucrose\ 8.89\ of which fructo-oligosaccharides 5.2 Fats (%) total 0.087 Phytic acid (%) total 0.45 Minerals (Oligo, micro and macroelements) (%) total 9.5 --------------------------------------------------------------------------------------------------------------------- Different classes of compounds present in the artichoke whole phytocomplex are reported. The phytocomplex is derived from artichokes (Cynara scolymus). ###### Scheme of treatment ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Group n° of animals Cell line n° of cells via Volume Treat.\ Treat. A Treat.\ Treat.\ Treat. B Treat.\ Treat.\ starting admin schedule admin. schedule --------- --------------- ----------- ------------- ----- ------------------- ------------------------ ------------------------ --------- -------------------- ---------------------- --------- ----------------- Group 1 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) \- Group 2 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance \- Artichoke (20 ug/ml) OS drinkable water Group 3 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance \- Artichoke (50 ug/ml) OS drinkable water Group 4 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance \- Artichoke (75 ug/ml) OS drinkable water Group 5 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance Pemetrexed (100 mg/kg) IP 5 consecutive days Group 6 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance Pemetrexed (100 mg/kg) IP 5 consecutive days Artichoke (20 ug/ml) OS drinkable water Group 7 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance Pemetrexed (100 mg/kg) IP 5 consecutive days Artichoke (50 ug/ml) OS drinkable water Group 8 6 MSTO 2\10\^6 SC 0.2 ml (Matrigel) After tumor appearance Pemetrexed (100 mg/kg) IP 5 consecutive days Artichoke (75 ug/ml) OS drinkable water ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- CD1 mice were subcutaneously transplanted with MSTO-211H (2 × 10^6^). At the evidence of tumor progression (when tumor volume reached 60 mm^3^) mice were randomly allocated in placebo and in the artichoke leaf extract-treated groups as described in the table. A control group of mice was intraperitoneally injected with Pemetrexed at the dose of 100 mg/kg for five consecutive days. Artichoke leaf extract induces changes in metabolic profiles of MSTO-211H cells {#s2_5} ------------------------------------------------------------------------------- Since metabolic alterations are among the hallmarks of a cancer cell we aim to assess whether the artichoke leaf extract could affect the metabolism of MSTO-211H cells. ^1^H-NMR metabolomics profiles of MSTO-211H cell culture media displayed significant differences between artichoke-treated and vehicle-treated cells. Initially, we explored the NMR data through an unsupervised approach such as Principal Component Analysis (PCA) carried out on a dataset of treated and untreated medium samples. The results reported in Figure [5A](#F5){ref-type="fig"} highlighted significant differences between the two groups on PC1 (p = 0.013). Subsequently, to identify the variables (metabolites) that have a major contribution to the discrimination between samples, we analyzed the loadings values from O-PLS-DA with a threshold of 0.8 (Figure [5B, 5C](#F5){ref-type="fig"}). LV1 included the following variables with the highest correlation levels: 3-methyl-2-oxovalerate, acetate, N-acetyl groups, glutamax and succinate with positive loadings and arginine, pyruvate, tyrosine and phenylalanine with negative loadings. Therefore, considering the net balance of these metabolites, LV1 indicated that artichoke extract induced higher consumption of arginine, pyruvate, tyrosine and phenylalanine as well as a lower consumption of glutamax and succinate. Moreover, artichoke extract-treated cells exhibited higher production of 3-methyl-2-oxovalerate, acetate and N-acetyl groups than vehicle-tread ones. This suggests that exposure to artichoke leaf extract mainly affected the citrate cycle (TCA cycle), arginine and proline metabolism, glutamine and glutamate metabolism, alanine, aspartate tyrosine metabolism and phenylalanine metabolism (Table [3](#T3){ref-type="table"}). ![Metabolic response of MSTO-211H cell lines to the artichoke leaf treatment\ PCA A. and O-PLS-DA B. models built on the 1H-NMR dataset of media samples from the artichoke extract-treated and vehicle-treated MSTO-211H cell cultures. The score plots show the metabolic differences between the two cell groups. Panel C shows the fold changes relative to vehicle-treated samples (means +/− SD for three independent experiments; (t-test): p \< 0.05) of the most discriminant metabolites between the two groups from the O-PLS-DA model. 3-MOV, 3-Methyl-2-oxovalerate (p); Arg, arginine (c); Ac, acetate; N-Acetyl, group of N-acetyls; Glutamax, glutamax (c); Pyr, pyruvate (c); Suc, succinate; Tyr, tyrosine (c); Phe, phenylalanine (c). "c" and "p" for each metabolite indicate consumption or production, respectively.](oncotarget-06-18134-g005){#F5} ###### Metabolic pathways perturbed by the artichoke leaf extract in MSTO-211H cells as determined by analysis of OPLS-DA loadings Metabolite Artichoke leaf extract vs. vehicle Kegg Related Pathways[\](#tfn-001){ref-type="table-fn"} --------------------------------------------- ------------------------------------ ---------------------------------------------------------- 3-Methyl-2-oxovalerate higher production Valine, leucine and isoleucine degradation Arginine higher consumption Arginine and Proline metabolism Alanine, aspartate and glutamate metabolism Glycine, serine and threonine metabolism Citrate cycle (TCA cycle) Group of N-acetyls higher production Arginine and proline metabolism Amino sugar and nucleotide sugar metabolism Alanine, aspartate and glutamate metabolism Acetate higher production Glycolysis/Gluconeogenesis Pyruvate metabolism Fatty acid metabolism Glutamax lower consumption D-Glutamine and D-glutamate metabolism Purine metabolism Pyrimidine metabolism Alanine, aspartate and glutamate metabolism Arginine and Proline metabolism Pyruvate higher consumption Pyruvate metabolism Citrate cycle (TCA cycle) Glycolysis Pentose phosphate pathway Succinate lower consumption Citrate cycle (TCA cycle) Oxidative phosphorylation Alanine, aspartate and glutamate metabolism Tyrosine metabolism Phenylalanine metabolism Carbon metabolism Tyrosine higher consumption Tyrosine metabolism Phenylalanine higher consumption Phenylalanine metabolism The metabolites are mapped to their respective biochemical pathways as delineated in the Kyoto Encyclopedia of Genes and Genomes (Release 69.0, January 1, 2014; KEGG, <http://www.genome.jp/kegg>). The artichoke leaf extract antitumoral activity impinges on different signalling pathways {#s2_6} ----------------------------------------------------------------------------------------- Since the artichoke extract appears to exert broad anti-tumoral effects on mesothelioma cell lines, we aimed to assess which signalling pathways might be involved. To this end, we performed a phospho-antibody array containing 1318 antibodies representative of key proteins involved in over than 30 signalling pathways altered in cancer among which proliferation, apoptosis, invasion, migration, metabolism and angiogenesis. MSTO-211H cells were treated for 24 hrs with the artichoke extract (50 μg/ml) or vehicle and the derived protein lysates were used to probe the antibody array. The heatmap shown in Figure [6A](#F6){ref-type="fig"} represented the signal intensities of those proteins whose levels of expression or phosphorylation were changed upon the artichoke leaf extract treatment. Among them, we found a group of proteins, which included p70S6k (phospho-Ser 418), NMDAR1 (Ab-897), CAMK1-a (Ab-177), HDAC6 (phospho-Ser22), ACK1 (phosphoTyr284), p38 MAPK (phosphor Tyr-182) that were upregulated and phosphorylated upon the extract treatment compared to vehicle (Figure [6C](#F6){ref-type="fig"}). Strikingly, the artichoke leaf extract treatment activated key proteins such as p53, BAX p38, and Caspase 3 leading to apoptosis and downregulated survival and pro-tumorigenic factor among which c-Abl, STAT1, AKT, and VEGFR2 (Figure [6C](#F6){ref-type="fig"}) \[[@R37]\] \[[@R38]\]. By comparing the pattern of modulated proteins in MSTO-211H cells treated with Pemetrexed or Cisplatin we found that the artichoke extract modulated distinct and common sets of proteins when compared to the two drugs respectively (Figure [6B](#F6){ref-type="fig"} and Table [4](#T4){ref-type="table"}). Among the proteins upregulated by the artichoke treatment, GSK 3a-β (Ab-216/279) appear to be pivotal components of multiple signalling pathways such PI3K/PTEN/AKT/mTORC1, Ras/Raf/MEK/ERK, Hedgehog, Notch and WNT. Since aberrant activation of the WNT signalling pathway plays a role in mesothelioma tumorigenesis we investigated whether the treatment with the artichoke extract of MPM cell lines modulated the expression of diverse components of this pathway. We found that the artichoke leaf extract reduced β-catenin nuclear staining in two mesothelioma cell lines (Figure [7A--7B](#F7){ref-type="fig"}). This paired with the reduced expression of well-known β-catenin target genes such as LEF1, FGF8, COX2, MMP2 and VEGF transcripts (Figure [7C](#F7){ref-type="fig"}) \[[@R39]\]. Altogether these findings highlight the possibility that the broad antitumoral activities of the artichoke leaf extract on mesothelioma cell lines might occur through the modulation of diverse signalling pathways commonly altered in human cancers. ![Phosphoprotein analysis of MSTO-211H cells treated with the artichoke leaf extract\ A.* Heatmap overall intensities of protein and phospho-protein level of spotted 1318 proteins, after 24 hrs of treatment with vehicle or the artichoke extract (50 μg/ml). Red and blue shadings indicate high and low or undetectable protein levels respectively. B. Venn diagram showing the number of proteins and phospho-proteins whose result deregulated after 24 hrs of treatment with the artichoke extract (50 μg/ml), cisplatin (CDDP, 8 μM) or pemetrexed (PMTX, 30 μM). C. List of all proteins and phospho-proteins that result deregulated after 24 hrs of artichoke extract treatment. Red and blue colours indicate up and down regulation respectively.](oncotarget-06-18134-g006){#F6} ###### List of deregulated or phosphorylated proteins upon CDDP, PMTX or artichoke leaf extract treatment ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Pemetrexed Artichoke leaf extract Cisplatin ---------------------------------------- -------------------------- ------------------------------------------ -------------------------------------------- ---------------------------------------- -------------------------------------- -PKC theta (Ab-538)\ -Ezrin (Phospho-Tyr478)\ -HDAC6 (Phospho-Ser22)\ -Histone H3.1 (Phospho-Ser10)\ -VEGFR2 (Phospho-Tyr1054)\ -Smad1 (Ab-465)\ -NFkB-p105/p50 (Phospho-Ser337)\ -EPB41 (Ab-418/660)\ -ACK1 (Phospho-Tyr284)\ -HSL (Phospho-Ser552/563)\ -HER2 (Ab-877)\ -Cytokeratin 18 (Ab-52)\ -LYN (Ab-507)\ -Dok-1 (Phospho-Tyr362) -P38 MAPK (Phospho-Tyr182)\ -Hsp90 co-chaperone Cdc37 (Phospho-Ser13)\ -P38 MAPK (Ab-180)\ -HER2 (Phospho-Tyr877)\ -MAPKAPK2 (Ab-334)\ -P70S6K (Phospho-Ser418)\ -AKT (Phospho-Ser473)\ -Ras-GRF1 (Phospho-Ser916)\ -EGFR (Phospho-Tyr1092)\ -FGFR1 (Ab-654)\ -GSK3a-b (Ab-216/279)\ -PKA CAT (Phospho-Thr197)\ -Smad2 (Ab-245)\ -G3BP-1 (Phospho-Ser232)\ -Abl1 (Ab-204)\ -CaMK1-a (Ab-177)\ -DAB1 (Phospho-Tyr220)\ -RapGEF1 (Phospho-Tyr504)\ -MEK1 (Phospho-Ser298)\ -DAXX (Ab-668)\ -NMDAR1 (Ab-897)\ -VEGFR2 (Phospho-Tyr1059)\ -FLT3 (Phospho-Tyr969)\ -Smad3 (Ab-213)\ -IKK-a/b (Ab-176/177)\ -Trk A (Phospho-Tyr791)\ -AKT1 (Ab-308)\ -BAX (Ab-167)\ -Tau (Phospho-Ser356)\ -RAD52 (Ab-104)\ -eNOS (Phospho-Ser1177)\ -EPHB1/2 (Ab-594/604)\ -P38 MAPK (Ab-322)\ -Tau (Phospho-Thr205)\ -Merlin (Ab-10)\ -BAX (Ab-167)\ -Opioid Receptor (Ab-375)\ -Caspase-3 (Ab-150)\ -Androgen Receptor (Phospho-Ser213)\ -Rb (Phospho-Ser780)\ -P38 MAPK (Ab-322)\ -Amyloid beta A4 (Phospho-Thr743/668)\ -CD19 (Ab-531)\ -NMDAR1 (Phospho-Ser897)\ -MAPKAPK2 (Phospho-Thr334)\ -Caspase-3 (Ab-150)\ -c-Abl (Phospho-Tyr412)\ -Mst1/Mst2 (Ab-183)\ -PLCG1 (Phospho-Tyr1253)\ -Ephrin B1/B2/B3 (Phospho-Tyr324)\ -CD19 (Ab-531)\ -COT (Phospho-Thr290)\ -p53 (Phospho-Ser9)\ -Rb (Phospho-Ser807)\ -HCK (Ab-410)\ -Mst1/Mst2 (Ab-183)\ -KSR (Phospho-Ser392)\ -SREBP-1 (Phospho-Ser439)\ -Synaptotagmin (Phospho-Ser309)\ -DAXX (Phospho-Ser668)\ -p53 (Phospho-Ser9)\ -Zap-70 (Phospho-Tyr493)\ -MAP3K7/TAK1 (Ab-439)\ -p21Cip1 (Ab-145)\ -AKT1 (Ab-129)\ -SREBP-1 (Phospho-Ser439)\ -STAT1 (Phospho-Ser727)\ -PFKFB2 (Phospho-Ser483)\ -Smad1 (Phospho-Ser465)\ -CD32 (FcgammaRIIb) (Ab-292)\ -MAP3K7/TAK1 (Ab-439)\ -ACC1 (Phospho-Ser79)\ -CASP1 (Ab-376)\ -VAV2 (Ab-142)\ -NFkB-p65 (Phospho-Ser468)\ -PFKFB2 (Phospho-Ser483)\ -CDC25B (Phospho-Ser353)\ -GRB10 (Ab-67)\ -Dok-1 (Phospho-Tyr362)\ -NFkB-p100 (Phospho-Ser872)\ -CASP1 (Ab-376)\ -BTK (Phospho-Tyr550)\ -Tyrosine Hydroxylase (Phospho-Ser31)\ -ATP-Citrate Lyase (Phospho-Ser454) -RapGEF1 (Phospho-Tyr504)\ -GRB10 (Ab-67)\ -NFkB-p65 (Phospho-Ser468)\ -IkB-alpha (Phospho-Ser32/36)\ -FLT3 (Phospho-Tyr969)\ -\'Tyrosine Hydroxylase (Phospho-Ser31)\ -Dok-1 (Phospho-Tyr362)\ -PAK2 (Ab-192) -BAX (Ab-167)\ -IkB-alpha (Phospho-Ser32/36)\ -ATP-Citrate Lyase (Phospho-Ser454) -HDAC1 (Ab-421)\ -PAK2 (Ab-192) -Caspase-3 (Ab-150)\ -Mst1/Mst2 (Ab-183)\ -p53 (Phospho-Ser9)\ -SREBP-1 (Phospho-Ser439)\ -PFKFB2 (Phospho-Ser483)\ -CASP1 (Ab-376)\ -GRB10 (Ab-67)\ -Tyrosine Hydroxylase (Phospho-Ser31)\ -PAK2 (Ab-192) ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- The list of proteins whose result induced or downregulated by Cisplatin, Pemetrexed or the artichoke extract treatment compared to the vehicle is shown. In italics are reported the proteins which were modulated in more than one treatment. ![The artichoke leaf extract downregulates Wnt/β-catenin signalling\ A-B. Histograms showing the percentage of β-catenin positive nuclei/total nuclei of MSTO-211H A and MPP-89 cells B treated for 24 hrs with either vehicle or the artichoke extract (50 μg/ml). Bars indicate the average of three independent experiments. Statistics (t-test): p \< 0.05. C. Q-PCR for the expression of the indicated Wnt/β-catenin target genes from either vehicle or artichoke leaf extract-treated MSTO 211H cells (50 μg/ml, 24 hrs). Bars indicate the average of three independent experiments. Statistics (t-test): p \< 0.05.](oncotarget-06-18134-g007){#F7} DISCUSSION {#s3} ========== Malignant pleural mesothelioma is an extremely rare, highly lethal tumor: mesothelioma is highly resistant to current chemotherapeutic agents, thus the disease-free-survival of the treated patients is very poor. Here we investigated whether, a phytocomplex derived from Artichoke leaves extracts, affects malignancy of pleural mesothelioma. It has been previously reported that artichoke phenol extracts can promote apoptosis and impair invasiveness of different cancer cell lines \[[@R34]--[@R36]\]. We found that the artichoke extract exerts in vitro and in vivo a broad anti-tumoral effect by strongly affecting MPM growth, migration and invasion. Next-generation sequence technologies have tremendously contributed to decipher tumor landscape and to dissect tumor heterogeneity. This has further established that cancer is the consequence of a broad array of genetic and epigenetic alterations. Modern drugs in the form of single-chemical entities have been successful in the treatment of acute conditions such as infectious diseases. Drug action is fast and predictable, leading to a favourable resolution of a critical situation in a short period of time. The situation is markedly different when treating complex chronic conditions such as cancer, type 2 diabetes and cardiovascular diseases. In particular considering cancer development, carcinogenesis process is multistep (that is, it results from accumulated genetic and epigenetic alterations), multipath (that is, multiple functional pathways are involved, such as self-sufficiency in growth signals, insensitivity to anti-growth signals, apoptosis evasion, limitless replicative potential, tissue invasion and metastasis and sustained angiogenesis), and multifocal (both multiclonal - e.g., field cancerization - and clonal expansion leading to intraepithelial spread). More than 250 population-based studies, including case--control and cohort studies, indicate that people who eat about five servings of fruit and vege day have approximately half the risk of developing cancer, particularly cancers of the digestive and respiratory tracts, of those who eat fewer than two servings. The NCI has identified about 35 plant-based foods that possess cancer-preventive properties. These include garlic, soybeans, ginger, onion, turmeric, tomatoes and cruciferous vegetables (for example, broccoli, cabbage, cauliflower and Brussels sprouts). Thus, the plants contain numerous substances, some of them pharmacologically active other inert but still potentially interacting with the active ones to potentiate the effectiveness as anticancer agents. The effectiveness as anticancer agent of a given plant could be endowed in its complexity (number of composing substances) coupled with multi-targeting activity. In aggregate this might allow concomitant untangling of diverse cancer pathways. Congruently, both metabolomics and antibody array analysis of MPM cells revealed that the artichoke leaf extract affects concomitantly different signalling pathways. We found that the extract affects β-catenin nuclear expression which pairs with the downregulation of the expression of its target genes such as LEF-1, FGF18, COX-2, MMP2 and VEGF. It was previously shown that direct targeting of β-catenin sensitizes mesothelioma cell lines to the treatment with cytotoxic drugs \[[@R40]\]. WNT is together with RAS, PI3K, Hippo and BCL among those pathways that are mostly altered in mesotheliomas \[[@R41]\]. We also found that the artichoke extract induces phosphorylation of the tumor suppressor protein p53 on Ser9. Saito et al., have shown that phosphorylation of p53 on Ser 46, as well as on Ser 9 was dependent from ATM protein kinases in response to ionizing radiation \[[@R42]\]. This led to p53 stabilization, transcriptional activation and induction of apoptosis. ATM-mediated phosphorylation on Ser 15 and at nearby residues including Ser 6 and Ser 9 enhanced p53 apoptotic activities. Piccolo\'s group has shown that p53 phosphorylation on Ser 9 was instrumental for the activation of the TGF-β cytostatic program \[[@R43]\]. AKT/PKB is frequently activated in tumors and is an important player for signalling pathways that regulate growth and survival. Fully active AKT/PKB mediates numerous cellular functions including metabolism, angiogenesis, growth, proliferation, survival, protein synthesis, transcription, and apoptosis. We found that artichoke leaf extract reduced both protein levels of AKT1 and of its phosphorylation at Ser473 that enables fully activation of the signalling axis involving insulin receptor, IRS and PI3 kinase. This pairs with our metabolomics data indicating that artichoke leaf extract affects the citrate cycle (TCA cycle), arginine and proline metabolism, glutamine and glutamate metabolism, alanine, aspartate and glutamate metabolism, tyrosine metabolism and phenylalanine metabolism (Table [3](#T3){ref-type="table"}). Thus, artichoke leaf extract might impact on crucial metabolic pathways that are aberrantly activated in human cancers. Altogether these evidences contribute to categorize artichoke leaf extract as a natural product with broad antitumoral activity on malignant pleural mesothelioma. Due to the evidence that the signalling pathways altered in mesothelioma and affected by artichoke leaf extracts are common to other human malignancies we might expect that cynara scolymus could exert antitumoral effects on other types of human cancers ([Supplementary Figure 5](#SD1){ref-type="supplementary-material"}). Data shown in Figure [1A-F](#F1){ref-type="fig"} and [Supplementary Figures 2A-B](#SD1){ref-type="supplementary-material"} and [3A-C](#SD1){ref-type="supplementary-material"} provide in vitro and in vivo evidence for a co-treatment of artichoke leaf extracts with either cisplatin or pemetrexed. Indeed, artichoke leaf extract, unlike cisplatin that provoked DNA damage to both mesothelial (HMC) and mesothelioma cell lines, exerted a more pronounced apoptotic effects on MPM cell lines than HMC. Interestingly artichoke leaf extracts reduced DNA damage of HMC cells upon cisplatin treatment, thereby suggesting that its co-treatment with either cisplatin or pemetrexed not only potentiates their antitumoral effects but might also confers selectivity toward tumor cells. Natural products are also the leading compounds for chemopreventive strategies. This is mainly based on the evidence that are well tolerated and have no or very minimal side effects on the subjects enrolled in chemoprevention trials. MPM is an occupational disease with a latency up to 40--50 years that represents an extraordinary time window for a chemopreventive approach. Indeed, a Phase II clinical trial (NCT02076672) aiming to investigate the anti-cancer activity of artichoke leaf extract in an asbestos-exposed population is currently enrolling participants in Ontario. MATERIALS AND METHODS {#s4} ===================== Cell lines {#s4_1} ---------- The human MPM cell lines MSTO-211H, NCI-H28, MPP89 were purchased from the ATCC (Rockville, MD). HMC, human mesothelial cells, were purchased from Tebu-Bio (Le Perray en- Yvelines, France). All MPM cell lines, were cultured as monolayers at 37°C and 5% CO2 in DMEM/F12 + GLUTAMAX (InVitrogen, Carlsbad, CA) supplemented with 10% non-heat inactivated FBS (Gibco, Life Technologies, USA) and 0, 5 Unit/ml insulin (Humulin, Eli Lilly and Company, Indiana, USA), while HMC were cultured in Mesothelial cell basal medium (Zen-Bio, Research Triangle Park, NC). Extraction procedure and phytochemical characterization {#s4_2} ------------------------------------------------------- Frozen artichoke leaf samples were extracted (DER from 5 to 6:1) by hydro-alcoholic procedure (ETOH 50%). The phytocomplex consists of a partially purified mixture of polyphenolic compounds and terpenes extracted from artichoke (Cynara scolymus L.) enriched in caffeoylquinic acids, chlorogenic acid and cynaropicrine. It is a pending Aboca\'s patent (RM2014A000685). It was submitted to a complete characterization of the composition by means of metabolomic analysis (LC-ESI/MS) (as previously described \[[@R28]\] and by quantitative analysis of caffeoylquinic acids (SFM), chlorogenic acid (HPLC-UV) and cynaropicrine (HPLC-UV). The complete chemical classes of compounds (phenols, terpenes, fats, proteins, amino acids, minerals, polysaccharides, etc.) addictionally analysed and presents in the phytocomplex are listed in Table [1](#T1){ref-type="table"}. Chemicals and antibodies {#s4_3} ------------------------ Artichoke capsules (Aboca, Sansepolcro, Italy); Pemetrexed (ALIMTA, Eli Lilly and Company, Indiana, USA) and Cisplatin (Pfizer Pharmaceuticals Group, New York, USA) were dissolved according to the manufacturer\'s instructions. The following primary antibodies were used: anti-cleaved PARP (Asp214) (Cell Signaling, \# 9541); anti-caspase 7 (Cell Signaling, \#9492); anti-caspase 3 (Enzo life Science, \#31A1067); anti-beta actin (A-2228, SIGMA); β-catenin antibody (clone CAT-5H10, \# 18--0226. Zymed). Secondary horseradish peroxidase-conjugated was purchased from Santa Cruz; secondary antibody for immunofluorescence Alexa Fluor 594 (mouse) conjugated was obtained from Molecular Probes (Inc, Eugene, OR, USA). ECL reagent (Amersham, GE Healthcare, Piscataway, NJ, USA) was employed for the chemo-luminescence detection. Annexin V FITC (0, 2 μg/ml) (Abcam, ab-63556); DAPI staining (Sigma) was used for nuclear detection. RNA processing and qRT-PCR {#s4_4} -------------------------- Total RNA from mesothelioma cell lines was extracted by using Trizol Reagent following manufacturer\'s instructions (InVitrogen). The first-strand cDNA was synthesized according to the manufacturer\'s instructions (M-MLV RT kit, Invitrogen). Gene expression was measured by real-time PCR using the FastStart SYBR Green Master Mix (Applied Biosytems) on a 7900HT instrument (Applied Biosystems). Sequences of qPCR primers are LEF-1 Fw: 5′-AGCGAATGTCGTTGCTGAGTGTA-3′, Rv 5′-CT CTTGCAGACCAGCCTGGATAA-3′. FGF18 Fw: 5′-CT CTACAGCCGGACCAGTG-3′, Rv: 5′-CCGAAGGTGTC TGTCTCCAC-3′. ACTIN Fw: 5′-GGCATGGGTCAGA AGGATT-3′, Rv: 5′-CACACGCAGCTCATT GTAGA AG-3′. C-MYC Fw: 5′-CTCCTGGCAAAAGG TCAGA G-3′, Rv: 5-TCGGTTGTTGCTGATCTGTC-3′. COX2 Fw: 5′-GAATGTTCCACCCGCAGTACA-3′, Rv: 5′-GCATAAAGCGTTTGCGGTAC-3′. VEGF Fw 5′-CGA GGGCCTGGAGTGTGT-3′, Rv: 5′-CGCATAATCTGCAT GGTGATG-3′. CyclinD1 Fw: 5′-GCCCTCGGTGTCC TACTTC-3′, Rv: 5′-AGGAAGCGGTCCAGGTAGTT-3′. MMP2 Fw: 5′-AAGTCTGGAGCGATGTGACC-3′, Rw: 5′-GAGTCCGTCCTTACCGTCAA-3.′ Cell viability assay {#s4_5} -------------------- Cell viability of treated cells was assessed using ATPlite assay (Perkin Elmer, Massachusset, USA) accordingly to the manufacturer\'s instructions. Cell viability tests were carried out to determine the efficacy of Cynara scolymus concentrations using different plant fractions and lot numbers. Cells (8 × 10^2^ cells) were seeded in 96 well-plates and cultured for 24 hrs and treated for 72 hrs with Cynara scolymus leaf extracts. Each plate was evaluated immediately on a microplate reader (Expire Technology, Perkin Elmer). Clonogenic assays {#s4_6} ----------------- MPM cell lines were grown at 70% confluence and treated with Cynara scolymus leaf extracts or with vehicle. Sixteen hrs later, cells were detached and seeded at 600 cells per 6 well into six-well dishes (Corning-Costar, Tewksbury, MA, USA) in drug-free media. Fresh media (25%) was added every three days. After 15--21 days, colonies were stained with crystal violet and colonies counted. Apoptosis detection {#s4_7} ------------------- To determine the effect of artichoke leaf extracts on the cell cycle FACS analysis was carried out. For propidium iodide (PI) staining, cells were seeded in 6-well plates at a density of 10^4^ cells/ml. After 24 hrs cells were treated with indicated plant extract concentrations for different time intervals. Floating and attached cells were harvested, washed in PBS, fixed in ice-cold ethanol (70% v/v) and stored at −20°C. For the analysis, cells were washed in PBS and incubated with RNase A (1 mg/ml) and PI (40 μg/ml) was added. For PI/Annexin V double staining treated cells were harvested and suspended in binding buffer (HEPES pH 7.4, CaCl2 2.5 mM, NaCl 140 mM). Aliquots of cells were incubated for 15 min with Annexin V FITC (0, 2 μg/ml) (Abcam, ab-63556) and PI (5 mg/ml) (Invitrogen). For each FACS analysis, 3 × 10^3^ events for each sample were analyzed. Flow cytometry analyses were carried out with Easycyte 8HT (Guava, Millipore) followed by analysis using InCyte software (Millipore). Transwell invasion assay {#s4_8} ------------------------ Migration assay was performed using a 24-well Boyden chamber with a non-coated 8-mm pore size filter in the insert chamber (BD Falcon, Franklin Lakes, NJ, USA). Cells were suspended in 0.5 ml DMEM/F12 media without containing FBS and seeded into the insert chamber. Cells were allowed to migrate for 24 hrs into the bottom chamber containing 0.5 ml of DMEM/F12 media containing 10% FBS in a humidified incubator at 37°C in 5% CO2. Migrated cells that attached to the outside of the filter were visualized by staining with DAPI and counted. The average number of cells per field was expressed as percentage of the control after normalizing for cell number. Wound--healing migration assay {#s4_9} ------------------------------ MPM cells were grown to 80% of confluence in 6-well tissue culture plates and wounded with a sterile 10-mL pipet tip to remove cells. PBS washing was used to remove loosely attached cells. The progression of migration was at photographed at different times under a light microscope. The number of cells migrated into the scratched area was calculated. Comet assays {#s4_10} ------------ Cells were pre-treated with or without 3 μg or 6 μg/ul of Cynara scolymus for 16 hrs. After that, cells were treated or not with CDDP, 7.5 μg/ml, for 20 hrs. After treatment, cells were detached with trypsin and embedded in 1% low melting agarose (Sigma) in phosphate-buffered saline and spread onto microscopy slides coated previously with 1% agarose (Bio-Rad). Cells were lysed in the lysis solution (2.5 M NaCl, 100mM ethylenediaminetetraacetic acid, 10mM Tris base, 8 g/l NaOH, 1% Triton X-100, 10% dimethyl sulfoxide) for 1 h at room temperature and then run in running solution (300mM NaOH, 1mM ethylenediaminetetraacetic acid, pH 13.0) for 30 min at 25V and 250 mA. DNA was equilibrated with 0.4M Tris (pH 8.0) and slides were dried with methanol. DNA was stained with propidium iodide (Sigma) and pictures were taken using 63 × magnification at an Axiovert 200 M microscope and Axiovision acquisition program (Zeiss). At least 300 cells were scored for each slide. Western blot analysis {#s4_11} --------------------- Cell lysis was performed on ice for 30 min in NP40 lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% NP-40, 1 mM EGTA, 1 mM EDTA) supplemented with protease and phosphatase inhibitors (5 mM PMSF, 3 mM NaF, 1 mM DTT, 1 mM NaVO4). Equal amounts of total proteins extracts (30 μg) were resolved by 8% denaturing SDS polyacrylamide gel electrophoresis (SDS-PAGE), and transferred for 2 hrs to polyvinylidene difluoride membrane. Membranes were blocked in 5% milk-TBS-0.05% Tween 20 for 1 hour and incubated overnight with the specific primary antibodies (see chemicals and antibodies section). Immunofluorescence microscopy {#s4_12} ----------------------------- Briefly, MSTO-211H cells were seeded into eight-chamber culture slides (BD Falcon). The next day, cells were rinsed with ice-cold PBS buffer and fixed with 4% paraformaldehyde for 10′ at room temperature and then permeabilized with 1% Triton X-100. The cells were incubated overnight with the indicated antibody. The day after, cells were washed with cold PBS three times for 3 min each and stained for 2 hrs with a secondary antibody Alexa 488-conjugated goat anti-mouse IgG (Molecular Probes Cells) and counterstained with DAPI (40, 6-diamidino-2-phenylindole, dihydrochloride). Cells were examined under a Zeiss LSM 510 laser scanning fluorescence confocal microscope (Zeiss, Wetzlar, Germany). Sample preparation for NMR spectroscopy {#s4_13} --------------------------------------- Each medium sample (2 ml) was lyophilized, then dissolved in 700 μl of 1 mM TSP \[sodium salt of 3-(trimethylsilyl) propionic-2, 2, 3, 3-d~4~ acid\], 10 mM sodium azide D~2~O phosphate buffer solution (pH = 7.4) and finally homogenized by vortex mixing for 1 min. After centrifugation (10 min, 10.000 RCF at 22°C), 600 μl of each resulting supernatant was transferred to a 5-mm NMR tube and used for the NMR analysis. ^1^H-NMR spectroscopy {#s4_14} --------------------- 2D ^1^H J-resolved (JRES) NMR spectra were acquired on a 500 MHz Varian/Agilent spectrometer (Agilent, Santa Clara, CA) using a double spin echo sequence with 4 transients per increment for a total of 32 increments. These were collected into 16 k data points using spectral widths of 6 kHz in F2 and 40 Hz in F1. There was a 2.0 s relaxation delay. Each FID was Fourier transformed after a multiplication with sine-bell/exponential function in the F2 dimension and a sine-bell function in the F1 dimension. JRES spectra were tilted by 45°, symmetrised about F1, referenced to TSP at d~H~ = 0.0 ppm and the proton-decoupled skyline projections (p-JRES) exported using Agilent VNMRJ 3.2 software. Metabolites responsible for the separation between treated and untreated samples were identified using an in-house NMR database and Chenomx NMR suite v. 7.7 (Chenomx Inc., Alberta, Canada). NMR spectra pre-processing treatment {#s4_15} ------------------------------------ The 1D skyline projections exported were aligned and then reduced into spectral bins with ranging from 0.01 to 0.02 ppm by using the ACD intelligent bucketing method (1D NMR Manager software (ACD/Labs, Toronto, Canada). To compare the spectra, the integrals derived from the binning procedure were normalized to the total integral region, following exclusion of bins representing the residual water peak (4.33--5.17 ppm) and the TSP peak (0.5--0.5 ppm). The resulting data was used as input for multivariate analysis: Principal Component Analysis (PCA and Orthogonal projections to latent structures discriminant analysis (OPLS-DA) were performed using SIMCA-P + version 12 (Umetrics, Umea, Sweden). Phospho-protein profiling by the phospho explorer antibody microarray {#s4_16} --------------------------------------------------------------------- The Phospho Explorer antibody microarray, which was designed and manufactured by Full Moon Biosystems, Inc. (Sunnyvale, CA), contains 1318 antibodies \[[@R44]\]. Each of the antibodies has two replicates that are printed on a coated glass microscope slide, along with multiple positive and negative controls ([Supplementary Figure 4](#SD1){ref-type="supplementary-material"}). The antibody array experiment was performed using Full Moon Biosystems, according to their established protocol. In brief, cell lysates obtained from MSTO-211H treated with the artichoke extract at 50 μg/ml or with Cisplatin at 8 μM or Pemetrexed at 30 μM of concentrations for 24 hrs, were biotinylated with the antibody array assay kit (Full Moon Biosystems, Inc.). The antibody microarray slides were first blocked with a blocking solution (Full Moon Biosystems, Inc.) for 30 min at room temperature, rinsed with Milli-Q grade water for 3--5 min. The slides were then incubated with the biotin-labeled cell lysates in coupling solution (Full Moon Biosystems, Inc.) at room temperature for 2 hrs. The array slides were washed 4 to 5 times with 1x Wash Solution (Full Moon Biosystems, Inc.) and rinsed extensively with Milli-Q grade water before detection of bound biotinylated proteins using Cy3-conjugated streptavidin. Each slide (containing six replicates) hybridized and Cy3 fluorescence acquired by microarray scanner with a scan resolution of 10 mm (Agilent Technologies). The images were quantified using Agilent Feature Extraction (AFE) software (Agilent Technologies). The fluorescence signal of each antibody was obtained from the fluorescence intensity of this antibody spot after subtraction of the blank signal (spot in the absence of antibody). Statistical analysis {#s4_17} -------------------- Bionformatic analysis was performed with Matlab (The MathWorks Inc.). Z score transformation was used to express the background corrected spot intensity values as unit of a standard deviation from the normalized mean of zero \[[@R45]\]. Features were selected basing on Z ratios calculated by taking the difference between the averages of the observed protein Z scores and dividing by the standard deviation of all the differences for that particular comparison. A Z-ratio that was higher than 1.96 was inferred as significant. Unsupervised Hierarchical Clustering was used to investigate clusters of samples. Pathway analysis was performed by DAVID program \[[@R46]\] \[[@R47]\]. EnSpire^®^ cellular label-free platform {#s4_18} --------------------------------------- MSTO-211H cells were seeded in specially designed 384-well plate with highly precise optical sensors able to measure changes in light refraction resulting from dynamic mass redistribution (DMR) within the cell\'s monolayer. Change in the light refraction was indicated by a shift in wavelength. Experimental animals and ethics statement {#s4_19} ----------------------------------------- ### Xenograft transplantation {#s4_19_1} MSTO-211H cells were pre-treated with the artichoke extract (50 μg/ml) for 24 hrs. Suspensions of 2 × 10^6^ MSTO-211H cells × mouse (n = 6) were subcutaneously injected in PBS 1x/Matrigel (BD Biosciences San Jose, CA, USA) into 6-weeks-old female CD1 mice (Charles River, Milan). Tumor volume was calculated by using the formula: V 1/2 × length × width^2^ (by electronic caliper). ### Animal studies {#s4_19_2} CD1 mice were subcutaneously transplanted with MSTO-211H (2 × 10^6^). At the evidence of tumor progression (when tumor volume reached 60 mm^3^) animals were randomly divided in five groups. Drinking water and a complete pellet diet (GLP 4RF21, Mucedola) were supplied ad libitum. Three groups of mice (n = 6) were given the artichoke extract in drinking water at the following concentrations: 25, 50 and 75 mg/ml, whereas mice of the control group (n = 6) were given tap. In the last group, pemetrexed was injected intraperitoneally at the dose of 100 mg/kg for five consecutive days. Body weight and clinical signs of the mice were checked every 3 days. All tumorigenicity assays were performed according to the guidelines set by the internal ethical committee. At the end of the experiment tumor masses were collected and fixed in 10% buffered formalin. ### Immunoistochemical analysis {#s4_19_3} Formalin-fixed and paraffin-embedded 5 μm sections from mice tumor sections were stained with haematoxylin and eosin or stained with anti-ki67 antibody (ab15580, Abcam). Seven fields chosen randomly from each sample were scored. SUPPLEMENTARY FIGURES AND TABLE {#s5} ===============================	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec0005} =============== Myositis ossificans (MO), otherwise known as heterotopic ossification, is a non-neoplastic, localized tumor-like lesion of new true bone formation that affects the muscles, ligaments, and fascia. Most cases of MO occur as a result of trauma, and thus the main demographics are adolescents and young adults [@bib0005], [@bib0010]. The anterior muscle groups of the thighs and arms are more frequently affected as these sites are subjected to high-risk injuries [@bib0005], [@bib0015]. Despite its being a clinically and histologically distinct entity, diagnosis may be very difficult, especially when it is presented in an uncommon location. MO can be confused with malignant lesions, such as osteosarcoma and soft- tissue sarcoma. Appropriate imaging is crucial for excluding infections or malignancies. Computed tomography generally is the imaging tool of choice in difficult cases and to planning for surgical resection. In this article, we present a rare case of a MO lesion located in the dorsal forefoot soft tissue. In addition, we review the imaging and histopathological features of the MO lesion that are useful for differential diagnosis. 2. Case report {#sec0010} ============== A 50-year-old woman was referred to our hospital because of a 6-month history of a mass in the right forefoot region. The patient described a minor trauma with hard object to the mass region from the medial side and subsequently, she noticed a painful and enlarging lesion on the medial aspect of her dorsal forefoot, which became firm and gradually less painful. The patient's significant medical history was negative and she specifically denied any weight loss, malaise, anorexia, fever, or chills. On examination she had a firm 2-cm lump that was well demarcated and palpable. Minimal local tenderness were noted. Her great toe movements were slightly restricted with pain. Her laboratory findings, including white blood cell count, erythrocyte sedimentation rate and C-reactive protein level, were normal. Radiographs of the foot was unremarkable and there was no suspicion of a bone lesion or periosteal reaction. Contrast-enhanced computed tomography (CT) of the foot ([Fig. 1](#fig0005){ref-type="fig"}) demonstrated dystrophic appearing subcutaneous calcifications scattered throughout the outlines of a well defined 18 × 9 mm non enhancing soft tissue mass-like lesion just anterior to the head of the first metatarsal bone and metatarsal phalangeal joint without osseous involvement. A primary diagnosis of MO was made based on the clinical history and imaging findings and after a discussion with the patient and her family, the patient was referred to surgical excision for a definitive diagnosis. After 6 days, surgical excision was performed as planned and under general anaesthesia, the lesion was exposed after a careful dissection at the dorsal surface of medial side of forefoot and proximal phalynx of the great toe and was excised with 1 cm of tumor-free margins after careful seperation from the extensor hallucis longus tendon. The resected specimen was well circumscribed, totally seperated from the extensor hallucis longus tendon sheath and showed a distinct zonal pattern of myositis ossificans ([Fig. 2](#fig0010){ref-type="fig"}a). The histopathological study reflected morphologic changes compatibles with MO. No malignant signs were observed in the sample. The definitive histopathological analysis described a central zone of immature fibroblastic spindle cells intersecting a myxochondroid stroma and the periphery comprised calcification and mature lamellar bone compatible with MO ([Fig. 2](#fig0010){ref-type="fig"}b). The edges of the sample were tumor free. The surgical wound healed in 2 weeks without complications and the patient recovered without incident. At the 2-months follow-up examination, the patient was asymptomatic and exhibited normal function and a full range of motion of all digits of the foot, particularly great toe flexion and extension. 3. Discussion {#sec0015} ============= Myositis ossificans (MO) is a localized, self-limiting, extraskeletal formation of heterotopic bone and cartilage in soft tissues that usually occur after trauma in 60% to 75% of all cases (as in our case) and therefore called in sometimes myositis ossificans traumatica (MOT) [@bib0020]. The most agreed etiologic mechanism includes an osteoblast stimulation as a consequence of a bone or soft tissue damage causing a formation of new bone, dystrophic calcifications or calcified chondroid matrix. However, in approximately 25% of cases, there is no apparent history of preceding trauma, and in some of these cases, an infectious process has been implicated to be a possible cause or the initiating factor. Other provocative causes include burns, neuromuscular disorders and hemophilia. MOT usually appears in adolescents or young adults. The majority of the patients are male. The lesions are predominately involves the high-risk sites of injury in about 80% of cases, such as thigh, elbow and buttocks [@bib0025]. The most commonly affected sites in the lower extremity are the quadriceps femoris and gluteus muscles, and those in the upper extremity are the brachialis muscle [@bib0020]. However, involvement of the foot is rare, and only a few cases have been reported to current date. Allard et al. [@bib0030] reported a rare case of plantar forefoot MO while De Maeseneer et al. [@bib0035] reported a rare case of MO located inferior to the second and third metatarsal bones in 37 years old woman. Early in the disease, the lesion is soft and painful, and within a few weeks a firm and often painful mass develops in the affected muscles. This lesion matures over 6--12 months, and eventually ossifies and becomes painless [@bib0020]. The lesion may cause limitation in the range of movements according to its site and size. In this case, toe movements were slightly restricted especially the flexion and extension while other fingers show reserved full range of movements. The microscopic histological findings vary according to the age of the lesion and are mirrored by radiographic findings. Early in the disease course, the lesion is mostly cellular with fibroblastic tissue resembling a granulation tissue, and radiographs are often negative. As the area of ossification expands, radiographs demonstrate flocculent radiodensities or calcifications. As the lesion matures, it completely ossifies. The most pivotal diagnostic feature of MO is that the bone maturity occurs from the periphery in a zonal manner with a fibroblastic center, whereas the central part presents with loose spindle cells with no cytological atypia [@bib0040], [@bib0045]. However, in sometimes as in our case, the radiograph cannot detect the mineralization of MO. Ultrasonography has limited role in early diagnosis of MO, mature lesions have the echogenicity and dense shadowing of cortical bone. However, ultrasonography is an operated-dependent examination, and no data are available on the value of ultrasonography in the diagnosis of MO in the foot [@bib0050]. Magnetic resonance imaging (MRI) is not routinely used for the evaluation of MO. Typical MRI findings include a low-signal-intensity rim and a heterogeneous, high-signal-intensity and tumor-like enlargement of affected tissues. Intravenous gadolinium administration results in early, intense, heterogeneous enhancement of the lesions [@bib0050]. CT is the preferred imaging modality to demonstrate the zonal pattern in posttraumatic MO [@bib0055]. It optimally identifies the typical patterns of this disease, including the separation of the mass from the adjacent cortex and the decreased attenuation of the center of the mass. The term myositis ossificans is inconsistent because the inflammation is absent and, if present, it is usually minimal and the muscle may not be involved. Hence, the term heterotopic ossification is more agreeable. When MO occurs in its common locations and/or if a clear clinical history and findings are obtained, the diagnosis is straightforward. However, the differential diagnosis may be problematic in patients with a lesion that lacks the characteristic zoning phenomenon and grows in an irregular multifocal or multilobulated form [@bib0020]. In our patient, the rare location of the lesion necessitated histopathological assessment to distinguish the lesion from extraskeletal osteosarcoma. There are many differential diagnosis for skeletal and soft tissue tumors that originate in the foot and share similar presentation with MO including periosteal and parosteal osteosarcoma, synovial sarcoma, periosteal chondroma, osteomyelitis, giant cell tumors of tendon sheath and vascular tumors. Other non-neoplastic soft tissue processes with bone formation such as pseudomalignant osseous tumor of soft tissues, florid reactive periostitis, and bizarre parosteal osteochondromatous proliferation occur more commonly in the foot than myositis ossificans. Presenting features of malignant tumors that arise in the foot are similar to those of benign lesions, including palpable mass, swelling, increased warmth, limp and pain; and many of patients reports history of trauma [@bib0060]. In contrast to MO, osteosarcoma presents a more disorganized growth of hyperchromatic and pleomorphic cells with osteoid formation. Additionally, the greater degree of cellular atypism and infiltration of adjacent tissues in a destructive fashion are highly indicative of this lesion. Although mitotic figures are present in immature MO and osteosarcoma lesions, clearly atypical or tripolar forms guide toward malignancy [@bib0020]. Furthermore, unless adequate biopsy material is obtained, including the central and peripheral components of the lesion, histological differentiation may also be difficult. The histological features of our case included a lobulated, well demarcated and partially encapsulated lesion with peripheral calcification and mature cartilaginous tissue. There was a central zone made of a cellular proliferation with mesenchymal features. Nor cytological atypia neither mitosis were seen in the lesion. Early in the disease course, rest, ice, compression, and elevation are universally recommended [@bib0005]. MO is a reactive self limiting condition, can spontaneously resolve and there is no compelling evidence that malignant degeneration ever occurs [@bib0010]. Surgery may be necessary in cases of considerable decreases in the range of motion, muscle atrophy, unremitting pain, and deterioration of function after 6--12 months of unsuccessful conservative care [@bib0065]. Excision is only indicated if the lesion is completely ossified because removal of immature bone may cause extensive local recurrence. Some studies suggest that using prophylactic indomethacin and etidronate can be beneficial in reducing postsurgical ectopic calcification [@bib0005]. In our patient, we decided to perform surgery as the lesion was located in an unusual site. Intra-operatively, we found that the MO occupied a limited space just above the extensor hallucis longus tendon. The patient's post-operative course was nevertheless uneventful, and no functional deficit was apparent. In summary we report a rare case of a 50-year-old woman with MO in the dorsal forefoot. The history of antecedent trauma and computed tomography findings allowed preoperative accurate diagnosis. Because of the rare location of the lesion, surgical excision and a histopathological examination was necessary to establish the diagnosis. Pathology reports confirmed the MO diagnosis. The patient recovered without complications. Increasing awareness on the unusual sites for MO occurance is necessary for differentiating a MO lesion from a malignant soft- tissue tumors and avoiding diagnostic pitfalls and unnecessary investigations, which can have major consequences and complications for patients. Conflict of interest {#sec0020} ==================== The authors declare that they have no conflict of interest. Funding {#sec0025} ======= None. Ethical approval {#sec0030} ================ Not requested as the study not involved the patient. Consent {#sec0035} ======= Informed consent was obtained from the patient for publication of this case report and any accompanying images. Author contribution {#sec0040} =================== As a 1st author and corresponding author, Qays A. Hassan contribute to case report writing, data collection and discussion writing. Mohammed Al-edani performed surgery. Both authors reviewed critically the manuscript for important intellectual content. Guarantor {#sec0045} ========= The Guarantor is Qays A. Hassan. ![Axial contrast-enhanced CT scan images of foot show dystrophic appearing subcutaneous calcifications scattered throughout the outlines of a well defined non-enhancing soft tissue mass-like lesion (arrows) just anterior to the head of the first metatarsal bone and metatarsal phalangeal joint with normal underlying bony structures.](gr1){#fig0005} ![a) Macroscopic image of the surgical specimen that consisted of one tissue fragment measuring 2.3 × 1.6 × 1.5 cm. b) Histopathological examination showing a central zone of immature fibroblastic spindle cells intersecting a myxochondroid stroma and the periphery comprised calcification and mature lamellar bone.](gr2){#fig0010}	{ "pile_set_name": "PubMed Central" }
Introduction {#s1} ============ Infectious pancreatic necrosis virus (IPNV) is an aquatic virus that causes acute contagious diseases in freshwater and marine fish, which can result in heavy losses to the aquaculture industry. IPNV is a member of the Birnaviridae family [@pone.0016740-Dobos1]. Birnaviruses contain two genome segments (A and B) of double-stranded RNA contained within an unenveloped, medium-sized, icosahedral capsid [@pone.0016740-Dobos2]. The birnavirus genome encodes three to five structural proteins that are generated through various posttranslational cleavages. VP1 is a viral polymerase that is encoded by the smaller segment, B [@pone.0016740-Duncan1]. The larger segment, A, encodes a polyprotein that is processed into the capsid proteins VP2 and VP3 as well as the viral protease VP4 [@pone.0016740-Duncan2]. Another, smaller open reading frame (ORF) on segment A encodes one 17-kDa non-structural protein, VP5 [@pone.0016740-Magyar1], which is a viral Bcl-2 (B-cell CLL/lymphoma 2) family member that can regulate Mcl-1 and viral protein expression to inhibit apoptosis of infected cells [@pone.0016740-Hong1], [@pone.0016740-Hong2]. Two main types of cell death can be easily distinguished: apoptosis and necrosis [@pone.0016740-Jeurissen1], [@pone.0016740-Majno1], [@pone.0016740-Wyllie1]. Apoptotic cell death is a physiological event that is important during the development and maintenance of tissues. Apoptosis is an active and energy-conserving form of cell death that eradicates aged or diseased cells and poses little threat to the organism. Indeed, it does not lead to activation of the immune system but rather results in the quick clearance of the dying cells by phagocytes without the concomitant induction of an inflammatory response. In contrast, cell death induced by other means, such as injury, leads to necrosis, a form of non-programmed and destructive cell death. Necrosis is characterized by the disturbance of energy metabolism, disruption of cellular membranes, and release of cytoplasmic and nuclear components into the extracellular environment. However, it has become clear that necrotic cell death is as tightly controlled as caspase-dependent apoptosis, and it may be an important mode of cell death that is both pathologically and physiologically relevant [@pone.0016740-Berghe1], [@pone.0016740-Festjens1]. TNFα (tumor necrosis factor alpha) is a pro-inflammatory cytokine that plays important roles in diverse host responses, including cell proliferation, differentiation, necrosis, apoptosis, and the induction of other cytokines. TNFα can induce either NF-κB-mediated survival or apoptosis depending on the cellular context [@pone.0016740-Rahman1]. TNFα mediates powerful anti-microbial responses, including the induction of apoptosis, the killing of infected cells, the inhibition of intracellular pathogen replication, and the up-regulation of diverse host response genes. Many viruses have evolved strategies to neutralize TNF by direct binding and inhibition of the ligand or its receptor or modulation of various downstream signaling events [@pone.0016740-Benedict1]. Furthermore, TNF receptor-1 (TNFR1) has been shown to initiate necrotic cell death [@pone.0016740-Morgan1], and TNFα and other cytokines that bind to receptors of different classes have been reported to lead to the generation of ROS (reactive oxygen species) that function as second messengers in the necrotic cell death pathway [@pone.0016740-Shen1]. In a recent study investigating the molecular mechanisms regulating necrosis, Sato et al. identified the gene expression profile induced in mouse mammary FM3A tumors, which required gene expression to trigger necrosis following treatment with an anticancer agent, 5-fluoro-2′-deoxyuridine [@pone.0016740-Sato1]. TNFα activates the RIP1 kinase-mediated signaling cascade that is necessary for the induction of downstream genes influencing necrosis or apoptosis [@pone.0016740-Hitomi1], [@pone.0016740-Holler1]. Previous studies have shown that IPNV infection induces both apoptosis and secondary necrosis both in a fish cell line [@pone.0016740-Hong3], [@pone.0016740-Hong4] and in vivo [@pone.0016740-Santi1]. IPNV infection can trigger the tyrosine kinase-mediated death pathway to induce the pro-apoptotic protein Bad [@pone.0016740-Hong5], which may act via NF-κB [@pone.0016740-Hong6]. Then, IPNV can down-regulate the survival factor Mcl-1 [@pone.0016740-Hong2] and induce MMP (mitochondrial membrane permeabilization), which is blocked by the ANT (adenine nucleotide translocator) inhibitor BKA [@pone.0016740-Chen1]. Furthermore, IPNV infection can also cause the activation of caspase-9 and -3 [@pone.0016740-Hong7]. Finally, the submajor capsid protein VP3 can trigger cell death in fish cells [@pone.0016740-Chiu1]. In this study, we examine how IPNV-induced apoptotic cell death is linked to secondary necrosis in the zebrafish cell line ZF4. We used zebrafish oligo-microarray and real-time RT-PCR assays to screen the IPNV-induced cell death-associated gene expression profiles in zebrafish embryonic cells. Early in replication in IPNV-infected cells, the pro-inflammatory cytokine TNFα was up-regulated up to six-fold relative to the negative control. Furthermore, we demonstrate that the IPNV-mediated up-regulation of TNFα regulates both the Bad/Bid-mediated apoptotic pathway and the RIP1 (receptor-interacting protein-1)/ROS-mediated secondary necrosis pathway. Results {#s2} ======= IPNV-induced gene expression profiles in zebrafish embryonic cells {#s2a} ------------------------------------------------------------------ We used the zebrafish embryonic cell line ZF4 as a model system to screen IPNV-induced transcriptomes. We first determined if IPNV (multiplicity of infection (MOI) = 5) can infect ZF4 cells. The IPNV replication stages can be divided into early (6 h post-infection (p.i.)), middle (12 h p.i.), and late replication stages (24 h p.i.) in the ZF4 cell system. As seen in [Figure S1A](#pone.0016740.s001){ref-type="supplementary-material"}, the viral protein VP2 could be detected in IPNV-infected ZF4 cells at 6, 9, and 12 h p.i. Next, ZF4 cells were infected with different viral doses, and cell death was monitored using a viability assay ([Figure S1B](#pone.0016740.s001){ref-type="supplementary-material"}). We then used the ZF4 cells to analyze the gene expression profile of IPNV-infected cells. The cells were infected with IPNV (MOI = 5), and total RNA was isolated from infected and uninfected control cells at 0, 6, 12, and 24 h p.i. The zebrafish 14K oligo microarray we used comprised 1800 zebrafish gene sequences from the NCBI and a database of 12,768 putative open reading frames derived from NCBI zebrafish EST (expressed sequence tag) sequence information. Overall, the gene expression pattern seen in cells 12 h p.i. was similar to the expression pattern observed at 24 h p.i. ([Figure 1A](#pone-0016740-g001){ref-type="fig"}). Furthermore, the expression of genes that were differentially expressed at 6 h p.i. was significantly different from the expression at both 12 h p.i. and 24 h p.i. ![Transcriptional profile of IPNV-infected ZF4 cells 6, 12, and 24 h p.i.\ (A) Hierarchical clustering of the mRNA expression pattern, analyzed using DNA microarrays, compared with the 0 h timepoint. Red indicates up-regulation of genes, green indicates down-regulation of genes, and black is used to represent no change in expression. (B) Venn diagram detailing the number of genes up-regulated at 6, 12, and 24 h p.i. in each of three independent experiments examining IPNV-infected ZF4 cells. (C) Venn diagram detailing the number of genes down-regulated at 6, 12, and 24 h p.i. in each of three independent experiments examining IPNV-infected ZF4 cells.](pone.0016740.g001){#pone-0016740-g001} Student\'s t-test was used to identify the genes with significant changes in expression relative to the control. We identified 299 transcripts \[211 up-regulated ([Figure 1B](#pone-0016740-g001){ref-type="fig"}) and 88 down-regulated ([Figure 1C](#pone-0016740-g001){ref-type="fig"})\] that demonstrated at least a two-fold change in expression at 6 h p.i. Furthermore, using the same two-fold threshold, 258 (132 up-regulated and 126 down-regulated) and 295 (163 up-regulated and 132 down-regulated) transcripts were differentially regulated at 12 and 24 h p.i. ([Figure 1B--C](#pone-0016740-g001){ref-type="fig"}). [Table S1](#pone.0016740.s004){ref-type="supplementary-material"} lists the number of significantly (p\<0.05) differentially expressed genes (greater than two-fold change with respect to the control cells). These transcripts that were significantly modulated following IPNV infection were divided into twelve functional categories: immune response, apoptosis, transcription, signal transduction, lipid and cholesterol metabolism, carbohydrate metabolism, oxidative phosphorylation, cell cycle, protein degradation, protein folding and stress response, protein synthesis, nucleoside metabolism and synthesis. Quantitative real-time RT-PCR was used to confirm the transcriptional changes in select genes. Five up-regulated genes (mmp9, isgf3g, bcl-xl, cebpb, and tnfa) and three down-regulated genes (lpl, jun, and hsp47) were analyzed, using the expression of ef1a as an internal control ([Table S2](#pone.0016740.s005){ref-type="supplementary-material"}). The real-time RT-PCR data confirmed the same relative transcriptional regulation of the selected genes. The DNA array data were confirmed using RT-PCR. The expression of the pro-apoptotic genes bad, bmf1, bmf2, noxa, and bax ([Figure S2A](#pone.0016740.s002){ref-type="supplementary-material"}) was up-regulated at 6 h p.i. At 12 h p.i., the up-regulation of pro-apoptotic genes bid, puma, bok, and bok2 ([Figure S2B](#pone.0016740.s002){ref-type="supplementary-material"}) was analyzed. Blockade of TNFα-mediated death signals enhances host cell viability {#s2b} -------------------------------------------------------------------- We used Pathway Studio 6.0 to search for genes that showed a two-fold or greater expression change in the cDNA microarray and quantitative RT-PCR experiments to see whether TNFα may directly regulate some of the genes ([Figure S3](#pone.0016740.s003){ref-type="supplementary-material"}). We hypothesized that TNFα plays a crucial role in regulating either the apoptotic or necrotic cell death pathway at different replication stages. TNFα production was specifically inhibited using tyrphostin AG-126, a compound that inhibits the activity of the tyrosine kinases necessary for TNFα production [@pone.0016740-Levitzki1]. Following treatment with 50 µM AG-126, the expression of tnfa was reduced six-fold (at 6 h p.i., [Figure 2A](#pone-0016740-g002){ref-type="fig"}, lane 5), eight-fold (12 h, lane 6) and four-fold (24 h, lane 7) when compared with the untreated IPNV-infected cell (lanes 2--6; 6, 12, and 24 h p.i., respectively). The western blot results were confirmed using real-time RT-PCR, and similar results were obtained. Following treatment with either 50 µM or 100 µM AG-126, the tnfa expression level was reduced approximately 10-fold at the 6, 12, and 24 h p.i. timepoints ([Figure 2B](#pone-0016740-g002){ref-type="fig"}). We also used RNA interference to investigate whether knocking down TNFα would affect IPNV pathogenesis. The transcriptional expression of tnfα was reduced to 25.6% after TNFα-specific siRNA treatment in IPNV-infected cell ([Figure 2C](#pone-0016740-g002){ref-type="fig"}). The expression level of TNFα protein was also significantly decreased following siRNA treatment in IPNV-infected cell ([Figure 2D](#pone-0016740-g002){ref-type="fig"}). ![Treatment with TNFα-specific siRNA or AG-126 blocks TNFα expression by IPNV-infected ZF4 cells.\ (A) TNFα protein expression level in IPNV-infected ZF4 cells (MOI = 1) 0, 6, 12, and 24 h p.i. The protein was detected using western blot with a polyclonal antibody specific for TNFα. Lanes 1--4: ZF4 cells were pretreated with 50 µM AG-126 and infected with IPNV for 0 (lane 1), 6 (lane 2), 12 (lane 3), or 24 (lane 4). Lanes 5--7: Untreated ZF4 cells were infected with IPNV for 6 (lane 5), 12 (lane 6), or 24 h (lane 7). The expression of actin was used as an internal control. (B) TNFα mRNA expression in IPNV-infected ZF4 cells was quantified using RT-PCR. The ZF4 cells were pre-treated with 50 µM or 100 µM AG-126 for 2 hours, infected with IPNV (MOI = 1), and incubated for 0, 3, 6, 12, or 24 h. The expression of ef1a (elongation factor 1-alpha) was used as an internal control. (C) The tnfa expression was inhibited by TNFα-specific siRNA in IPNV-infected cells. TNFα expression was efficiently inhibited by TNFα-specific siRNA after IPNV infection. Sample 1: ZF4 cells infected by IPNV. Sample 2: ZF4 cells pretreated with scrambled siRNA and then infected by IPNV. Sample 3: ZF4 cells pretreated with TNFα-specific siRNA and then infected by IPNV. The quantification of gene expression in normal versus siRNA-treated cells was calculated relative to ef1a. (D) Detection of TNFα in untreated or TNFα-specific siRNA-treated ZF4 cells after IPNV infection by western blotting. Lane 1: untreated ZF4 cells; lane 2: ZF4 cells treated with control siRNA; lane 3: ZF4 cells treated with TNFα-specific siRNA. The expression of actin was used as an internal control. (E) Cell viability of IPNV-infected ZF4 cells pre-treated with TNFα-specific siRNA or AG-126 at 0, 6, 12, 24, 36 and 48 h p.i. The viability of each sample was determined in three individual experiments. Data shown are the mean ± SD. Student\'s t tests indicate significant differences compared to IPNV infection only or untreated control: \, p\<0.05; \\, p\<0.01.](pone.0016740.g002){#pone-0016740-g002} In addition, the treatment of IPNV-infected cells (MOI = 5) with 50 µM AG-126 increased their viability up to 21.8% (at 12 h p.i.), 33.4% (24 h), and 37.8% (48 h) relative to untreated, IPNV-infected cells. Treatment of IPNV-infected cells (MOI = 5) with TNFα-specific siRNA increased their viability up to 33.8% (at 12 h p.i.), 49.5% (24 h), and 55.3% (48 h) relative to untreated, IPNV-infected cells. Both mock-infected cells and AG-126-treated, uninfected cells were used as negative controls ([Figure 2E](#pone-0016740-g002){ref-type="fig"}). Recognition of a TNFα-mediated death signal may regulate the expression of the pro-apoptotic genes bad* and bid {#s2c} ------------------------------------------------------------------------------------------------------------------ As seen in [Figure 3](#pone-0016740-g003){ref-type="fig"}, inhibition of the IPNV infection-induced TNFα up-regulation reduced the expression of the pro-apoptotic gene Bad by TNFα-specific siRNA at 6 h p.i. (six-fold, [Figure 3A](#pone-0016740-g003){ref-type="fig"}, lane 5), 12 h p.i. (eight-fold, lane 6), and 24 h p.i. (four-fold, lane 7) when compared with untreated, IPNV-infected cells ([Figure 3A](#pone-0016740-g003){ref-type="fig"}, lanes 1--4). These results were confirmed using quantitative RT-PCR ([Figure 3B](#pone-0016740-g003){ref-type="fig"}). Expression of Bid and formation of truncated-Bid were suppressed by TNFα-specific siRNA at 6 h ([Figure 3C](#pone-0016740-g003){ref-type="fig"}, lane 2) and 12 h p.i. (lane 3) when compared with the untreated IPNV-infected group ([Figure 3C](#pone-0016740-g003){ref-type="fig"}, lanes 4 and 5) and the negative control (0 h, lane 1). The transcriptional expression of bid was suppressed after pre-treated with TNFα-specific siRNA or AG-126 in the IPNV-infected ZF4 cells ([Figure 3D](#pone-0016740-g003){ref-type="fig"}). ![A TNFα-mediated death signal regulates the expression of pro-apoptotic Bad and Bid in IPNV-infected ZF4 cells during the early-middle stage of replication.\ (A) The expression level of the pro-apoptotic proteins Bad and Bid in IPNV-infected ZF4 cells (MOI = 1) at 0, 6, 12, and 24 h p.i. was determined. The proteins were detected using western blot with a polyclonal antibody specific for mouse Bad. Lanes 1--4: ZF4 cells were pretreated with TNFα-specific siRNA and infected with IPNV for 0 (lane 1), 6 (lane 2), 12 (lane 3), or 24 h (lane 4). Lanes 5--7: untreated ZF4 cells were infected with IPNV for 6 (lane 5), 12 (lane 6), or 24 h (lane 7). The expression of actin was used as an internal control. Results are expressed as the ratio of Bad/actin. The mRNA expression of bad (B) and bid (D) in IPNV-infected ZF4 cells was quantified using quantitative RT-PCR. ZF4 cells were pre-treated with TNFα-specific siRNA or AG-126 and infected with IPNV (MOI = 1) for 0, 6, 12, or 24 h. The expression of ef1a was used as an internal control. Data shown are mean ± SD. Student\'s t tests indicate significant differences compared to untreated control: \\, p\<0.01. (C) The expression level of the pro-apoptotic proteins Bid and t-Bid in IPNV-infected ZF4 cells (MOI = 1) at 0, 6 and 12 h p.i. was determined. The proteins were detected using western blot with a polyclonal antibody specific for Bid. Lanes 2--3: ZF4 cells were pretreated with TNFα-specific siRNA and infected with IPNV for 6 (lane 2) or 12 h (lane 3). Untreated ZF4 cells were infected with IPNV for 6 (lane 4) or 12 h (lane 5). Untreated ZF4 cells were infected with IPNV for 0 h (Lane 1). The expression of actin was used as an internal control. Results are expressed as the ratio of Bid/actin or t-Bid/actin.](pone.0016740.g003){#pone-0016740-g003} A TNFα-mediated death signal may induce apoptosis and activate caspases {#s2d} ----------------------------------------------------------------------- Treatment with either AG-126 (50 µM) or TNFα-specific siRNA (20 nM) can reduce the number of apoptotic, annexin V-positive cells by up to 5.2% and 25% ([Figure 4A](#pone-0016740-g004){ref-type="fig"}) at 6 h and 12 h p.i. when compared with the IPNV-infected group and mock-infected group. IPNV infection induced dramatic caspase-9 activation at 6 h and 12 h p.i. ([Figure 4B](#pone-0016740-g004){ref-type="fig"}), which was only partially blocked by inhibiting TNFα expression. Treatment with TNFα-specific siRNA or AG-126 resulted in a two-fold and three-fold decrease in the number of apoptotic cells at 6 h and 12 h p.i., respectively. Caspase-8 was also activated at 6 h p.i. (5.4-fold) and 12 h (4.7-fold), and this activation was blocked by TNFα-specific siRNA or AG-126 treatment (only a twofold activation was observed) ([Figure 4C](#pone-0016740-g004){ref-type="fig"}). Caspase-3 was also activated at 6 and 12 h p.i., and this activation was reduced two- and three-fold, respectively, following treatment with TNFα-specific siRNA or AG-126 ([Figure 4D](#pone-0016740-g004){ref-type="fig"}). ![Inhibition of TNFα production can reduce caspase activation and the number of annexin V-positive IPNV-infected cells.\ (A) Detection of annexin V-positive cells following infection with IPNV. ZF4 cells were pre-treated with TNFα-specific siRNA or AG-126, infected with IPNV (MOI = 1), and incubated for 0, 6, 12, 18 and 24 h. Three individual experiments were performed for each sample. (B--D) Caspase-9, -8, and -3 activities were analyzed. ZF4 cells were pre-treated with TNFα-specific siRNA or AG-126 for 2 hours, infected with IPNV (MOI = 1), and incubated for 0, 6, 12, or 24 h. Luminogenic substrate assays were performed in triplicate. Data shown are the mean ± SD. Student\'s t tests indicate significant differences compared to IPNV infection only: \, p\<0.05.](pone.0016740.g004){#pone-0016740-g004} A TNFα-mediated death signal may regulate the apoptotic and secondary necrotic death pathway during the middle stage of replication {#s2e} ----------------------------------------------------------------------------------------------------------------------------------- IPNV infection (MOI = 5) caused a 2.1-fold, 3.0-fold, and 4.1-fold increase in ROS production in ZF4 cells 12, 18, and 24 h p.i., respectively. TNFα-specific siRNA (20 nM)- or AG-126 (50 µM)-treated cells displayed a reduction in ROS production at 12 h (1.5-fold), 18 h (1.8-fold) and 24 h p.i. (2.5-fold) ([Figure 5A](#pone-0016740-g005){ref-type="fig"}). RNA interference was used to investigate whether knocking down RIP1 would affect IPNV-induced necrosis. The expression level of RIP1 protein was decreased following duplex siRNA treatment ([Figure 5B](#pone-0016740-g005){ref-type="fig"}). Following infection with IPNV, the percentages of annexin V-positive, apoptotic cells, were 12.9%, 38.2%, and 9.2% at 6, 12, and 18 h p.i., respectively. Treatment with RIP1-specific siRNA reduced the percentage of PI-positive cells to 7.1%, 24.2%, and 8.4% at 6, 12, and 18 h p.i., respectively ([Figure 5C](#pone-0016740-g005){ref-type="fig"}); similar results were seen following treatment with RIP1 inhibitor necrostatin-1 (Nec-1) [@pone.0016740-Degterev1], [@pone.0016740-Vandenabeele1] ([Figure 5D](#pone-0016740-g005){ref-type="fig"}). Nox1 is a NADPH oxidase that interacts with Rac1 to promote ROS formation [@pone.0016740-Herman1], [@pone.0016740-Lee1]. TNF-induced necroptosis is delayed by inhibition of Nox1 activity [@pone.0016740-Lee1]. BHA (butylated hydroxyanisole) is a type of anti-oxidant that efficiently inhibits TNF-induced necrotic cell death [@pone.0016740-Chen1]. The percentages of annexin V-positive cells were reduced after BHA or DPI (diphenylene iodonium, an inhibitor of Nox1) treatment in IPNV-infected cells ([Figure 5E--F](#pone-0016740-g005){ref-type="fig"}). ![Inhibition of TNFR1 necrotic signaling complex formation inhibits apoptosis and ROS formation in IPNV-infected cells.\ (A) Detection of ROS production in TNFα-specific siRNA or AG-126 pre-treated cells after IPNV infection at 0, 6, 12, 18 or 24 h p.i. Fluorescence assays were performed in triplicate. Determination of the percentage of PI-positive cells after IPNV infection. (B) Detection of RIP1 in untreated or TNFα-specific siRNA-treated ZF4 cells by western blotting. Lane 1: untreated IPNV-infected ZF4 cells; lane 2: IPNV-infected ZF4 cells treated with scrambled siRNA; lane 3: IPNV-infected ZF4 cells treated with RIP1-specific siRNA. The expression of actin was used as an internal control. Detection of annexin V-positive cells following infection with IPNV. ZF4 cells were pre-treated with RIP1-specific siRNA (C), Nec-1 (D), DPI (E) or BHA (F), infected with IPNV (MOI = 1), and incubated for 0, 6, 12, 18 and 24 h. Three individual experiments were performed for each sample. Data shown are the mean ± SD. Student\'s t* tests indicate significant differences compared to IPNV infection only: \, p\<0.05.](pone.0016740.g005){#pone-0016740-g005} Following infection with IPNV, the percentages of PI-positive, necrotic cells, detected using a necrotic death assay, were 9.2%, 21.2%, and 36.5% at 12, 18, and 24 h p.i., respectively. Treatment with TNFα-specific siRNA reduced the percentage of PI-positive cells to 3.8%, 7.0%, and 11.7% at 12, 18, and 24 h p.i., respectively ([Figure 6A](#pone-0016740-g006){ref-type="fig"}); similar results were seen following treatment with 50 µM AG-126 ([Figure 6B](#pone-0016740-g006){ref-type="fig"}). Recent research has suggested that RIP1 is required for triggering the ROS-mediated necrotic death pathway [@pone.0016740-Ting1]. Treatment with RIP1-specific siRNA reduced the percentage of PI-positive cells to 3.3%, 5.8% and 9.0% at 12, 18 and 24 h p.i., respectively ([Figure 6C](#pone-0016740-g006){ref-type="fig"}). After treating with RIP1 inhibitor Nec-1, the percentage of PI-positive cells was reduced to 3.3% (12 h p.i.), 6.0% (18 h p.i.) and 9.5% (24 h p.i.) ([Figure 6D](#pone-0016740-g006){ref-type="fig"}). Treatment with DPI resulted in a reduction in the percentage of PI-positive cells to 7.2%, 14.5% and 23.3% at 12, 18 and 24 h p.i., respectively ([Figure 6E](#pone-0016740-g006){ref-type="fig"}). The percentage of PI-positive cells was reduced to 5.3% (12 h p.i.), 9.2% (18 h p.i.) and 14.0% (24 h p.i.) after BHA treatment in IPNV-infected ZF4 cells ([Figure 6F](#pone-0016740-g006){ref-type="fig"}). Cells treated with caspase inhibitor (zVAD-fmk) also showed a reduction in the ratio of PI-positive cells after IPNV infection ([Figure 6G](#pone-0016740-g006){ref-type="fig"}). ![Inhibition of TNF-α or TNFR1 necrotic signaling complex formation inhibits secondary necrosis in IPNV-infected cells.\ Detection of PI-positive cells following infection with IPNV. ZF4 cells were pre-treated with TNFα-specific siRNA (A), AG-126 (B), RIP1-specific siRNA (C), Nec-1 (D), DPI (E), BHA (F) and z-VAD (G) then infected with IPNV (MOI = 5), and incubated for 12, 18 and 24 h. (H) Detection of ROS production in RIP1-specific siRNA-, Nec-1- or DPI-pre-treated cells after IPNV infection at 0, 6, 12, 18 or 24 h p.i. Fluorescence assays were performed in triplicate. Data shown are the mean ± SD. Student\'s t* tests indicate significant differences compared to IPNV infection only or untreated control: \, p\<0.05.](pone.0016740.g006){#pone-0016740-g006} To determine if ROS production is activated through the formation of TNFR necrotic signaling complex, cells were pre-treated with RIP1-specific siRNA, Nec-1 or DPI and then infected by IPNV. ROS production was increased in IPNV-infected cells at 12 (2.1-fold), 18 (3.1-fold), and 24 h p.i. (4.3-fold) ([Figure 6H](#pone-0016740-g006){ref-type="fig"}). RIP1-specific siRNA (20 nM) or Nec-1 (20 µM) treated cells displayed a reduction in ROS production at 12 h (1.4-fold), 18 h (1.9-fold) and 24 h p.i. (2.7-fold). Pre-treating ZF4 cells with DPI caused a 1.6-fold, 2.3-fold, and 3.1-fold increase in ROS production 12, 18, and 24 h p.i., respectively. Discussion {#s3} ========== IPNV causes acute contagious diseases in aquaculture. Therefore, designing effective control or preventive measures against this virus is important. However, an understanding of the mechanisms underlying infection and immunity is essential. In this study, using a zebrafish cell line system, we demonstrated that IPNV regulates the apoptotic and necrotic death pathways through the up-regulation of TNFα. Thus, this study provides new insights into IPNV-induced molecular pathogenesis. The zebrafish as a model animal system to examine pathogen-induced transcriptomes {#s3a} --------------------------------------------------------------------------------- Zebrafish have been recognized and established as a model animal of infectious disease and are considered to have great potential for studying the development and function of the vertebrate immune system. Zebrafish have been shown to be susceptible to infection with and to allow the subsequent replication of various bacterial and viral pathogens. Zebrafish have the advantages of real-time visualization and genetic screens compared to other animal models of infection [@pone.0016740-Jenner1], [@pone.0016740-vanderSar1]. IPNV is able to persist and possibly replicate in adult zebrafish [@pone.0016740-LaPatra1]. IPNV also activates caspases and promotes host cell apoptosis in a zebrafish cell line [@pone.0016740-Hong7]. Therefore, zebrafish could be an efficient disease model of IPNV. In our system, we first established the DNA array screening system in ZF4 cells to better understand the virus-host interaction. Our results provide information on the virus as well as alterations in the expression of host genes related to immunity, apoptosis, transcription regulation, unfolded protein response, protein degradation, and the metabolism of cholesterol and carbohydrates following infection with IPNV ([Figure S2](#pone.0016740.s002){ref-type="supplementary-material"} and [Table S1](#pone.0016740.s004){ref-type="supplementary-material"}). TNFα-mediated apoptosis and necrosis pathway {#s3b} -------------------------------------------- TNFα is a crucial regulator in the innate and adaptive immune response against microbial infection via regulation of cell death and survival [@pone.0016740-Fiers1]. TNF is a pro-inflammatory cytokine that plays an important role in diverse host responses such as cell proliferation, differentiation, necrosis, apoptosis, and the induction of other cytokines. Recently, it has been shown that TNF can induce either an NF-κB-mediated survival or apoptotic pathway depending on the cellular context [@pone.0016740-Ting1]. Many viruses have strategies to neutralize TNF by direct binding and inhibition of the ligand or its receptor or modulation of the various downstream signaling events [@pone.0016740-Benedict1], [@pone.0016740-Benedict2]. The death receptors, including TNFR1, Fas, death receptor 3, DR4, DR5, and the TRAIL receptors, contain an intracellular "death domain" that activates downstream signaling pathways by means of homotypic interactions with adaptor proteins, such as FADD, TRADD, and RIP1 [@pone.0016740-Zheng1]. These death receptors induce apoptosis in many cell types through the activation of caspase-8. Activated caspase-8 may act indirectly to induce apoptosis through cleavage of Bid. Truncated Bid acts on the mitochondria to cause the release of cytochrome c, which further activates caspase-9. TNFR1 has been shown to initiate necrotic cell death [@pone.0016740-Morgan1]. TNFα and other cytokines that bind to receptors of different classes have been reported to induce the formation of ROS that function as second messengers in the necrotic cell death pathway [@pone.0016740-Meurer1], [@pone.0016740-Shakibaei1]. RIP1 is an intracellular adaptor molecule with kinase activity [@pone.0016740-Festjens2]. RIP1 [@pone.0016740-Holler1] and RIP3 [@pone.0016740-He1] appear to be crucial in the signaling through cell death receptors that do use caspases to induce death. RIP1 is also necessary for the generation of ROS by TNFα [@pone.0016740-Meurer1], [@pone.0016740-Shakibaei1]. RIP1 is required for apoptotic death induced by TNFα [@pone.0016740-Wang1]. IPNV infection of fish cells results in an initial wave of apoptosis that is followed by a second wave of necrosis [@pone.0016740-Hong3]. In this study, we propose that IPNV induced TNFα up-regulation early in replication when it plays an important role in controlling apoptotic and necrotic cell death [@pone.0016740-Hitomi1]. IPNV infection can up-regulate pro-apoptotic genes during the early-middle replication stage, as seen in [Figure S2](#pone.0016740.s002){ref-type="supplementary-material"}, and this up-regulation can be suppressed by blocking production of TNFα ([Figure 3](#pone-0016740-g003){ref-type="fig"}). Furthermore, blockade of TNF production can reduce the apoptotic ratio ([Figure 4A](#pone-0016740-g004){ref-type="fig"}) and caspase-3, -8, and -9 activities ([Figure 4B--D](#pone-0016740-g004){ref-type="fig"}). During the middle-late replication stage, the TNFα-mediated death signal triggers the necrotic death pathway via the subsequent ROS production ([Figure 5A](#pone-0016740-g005){ref-type="fig"}), which is a novel death signal pathway in the zebrafish cell system. The formation of TNFR1 signaling complex including RIP1, TRADD, Nox1, NOXA1 and Rac1 (small GTPase) is induced by TNF [@pone.0016740-Kim1], [@pone.0016740-VandenBerghe1]. The kinase activity of RIP1 is essential only for signaling to necrosis [@pone.0016740-Degterev1]. RIP1 is also necessary for the ROS generation by TNFα [@pone.0016740-Festjens2], [@pone.0016740-Lin1]. Both superoxide generation and cell death in response to TNFα are prevented by knockdown of Nox1 [@pone.0016740-Kim1]. The activation of Nox1 is downstream of the mitochondrial ROS production [@pone.0016740-Lee1]. In our study, inhibition of TNFα activation suppressed ROS production in the pathogenesis of IPNV. Inhibition of TNFα, RIP1, Nox1 or generation of ROS could decrease the percentage of annenix V-positive and PI-positive cells after IPNV infection ([Figure 5](#pone-0016740-g005){ref-type="fig"}--[6](#pone-0016740-g006){ref-type="fig"}). Inhibition of activities of the TNFR1 necrotic signaling complex also suppressed ROS production after IPNV infection. IPNV-induced necrosis might require the formation of TNFR1 necrotic signaling complex. Inhibition of caspase activation was also affected by necrosis induced by IPNV ([Figure 6G](#pone-0016740-g006){ref-type="fig"}). In summary, as shown in [Figure 7](#pone-0016740-g007){ref-type="fig"}, we demonstrate that IPNV-induced caspase-mediated, apoptosis and RIP1/ROS-mediated, secondary necrosis requires a TNFα-triggered death signal. Our study may provide new insight into RNA viral pathogenesis. ![IPNV induces apoptotic and necrotic death cascades via TNFα induction.\ A hypothesis of how pro-inflammatory TNFα is up-regulated by IPNV infection and how it regulates the apoptotic and necrotic death pathways. When a cell is infected with IPNV (E1-S), the virus binds to the cellular receptor, penetrates the cell, uncoats (the entry stage), and up-regulates the expression of TNFα (a) during the early replication stage (0--6 h p.i.). This up-regulation of TNFα regulates the next wave of gene expression, including the pro-apoptotic genes bad* and bid, during the early-middle stage of replication (6--12 h p.i.; b), at which point the TNFR necrotic signaling complex and reactive oxygen species (ROS) are produced. During this stage, phosphatidylserine (PS) is externalized and endonuclease is released from the mitochondria, resulting in DNA restructuring and cleavage. Furthermore, the cell finally enters the post-apoptotic, necrotic stage during the middle-late stage of replication (12--18 h p.i.; c). During the late replication stage (18--24 h p.i.; d), the cells are broken down. The TNFα-mediated death signal is halted by treatment with a specific inhibitor of TNFα production. (e) TNFα-specific siRNA or AG-126, which block both the pro-apoptotic Bad/Bid-mediated death pathway and the TNFR/ROS-mediated secondary necrotic death pathway. (f) DPI, Nec-1 or RIP-1-specific siRNA, which block the activity of RIP-1 or Nox1 and block the formation of the TNFR necrotic signaling complex. (g) BHA blocks the formation of ROS.](pone.0016740.g007){#pone-0016740-g007} Materials and Methods {#s4} ===================== Cells, viruses and reagents {#s4a} --------------------------- The zebrafish ZF4 cell line, which was originally derived from 24-hpf (24 hours post-fertilization) zebrafish embryos, was purchased from the American Type Culture Collection (CRL-2050) and cultured in RPMI 1640 medium supplemented with 10% (v/v) fetal bovine serum and penicillin/streptomycin. The isolated virus, E1-S, a member of the Ab strain of IPNV, was isolated from Japanese eels in Taiwan. The E1-S virus was propagated on a ZF4 cell monolayer at a multiplicity of infection (MOI) of 0.01. Infected cultures were incubated at 18°C until extensive cytopathogenic effects were observed [@pone.0016740-Dobos1]. The virus plaque assays [@pone.0016740-Nicholson1] and TCID~50~ were performed on a confluent monolayer of ZF4 cells. Benzyloxycarbonyl-Val-Ala-Asp (Ome) fluoromethylketone (zVAD-fmk; Promega, USA) was used at 10 µM. 5-(Indol-3-ylmethyl)-(2-thio-3-methyl) hydantoin (Nec-1), BHA, DPI chloride and AG-126 (Sigma Aldrich, USA) were used at 20, 100, 25 and 50 µM, respectively. Cell viability assays {#s4b} --------------------- For the cell viability assays, the cells were infected with IPNV or pre-treated with AG-126 (50 µM) or TNFα-specific siRNA (20nM) and divided into virus- or mock-infected groups. Cell viability was measured using a colorimetric assay based on the cleavage of the tetrazolium salt WST-1 by mitochondrial dehydrogenase (Cell Proliferation Reagent WST-1; Roche, USA). Western blot {#s4c} ------------ Approximately 10^5^ ZF4 cells/ml were seeded in a 60-mm Petri dish (Nunc, Denmark) and cultured for more than 24 h. These cells were then infected with IPNV at an MOI of 5 and incubated for 0, 6, 12, or 24 h. At the completion of each incubation period, the culture medium was aspirated, and the cells were washed with PBS and lysed in 0.5 ml lysis buffer (10 mM Tris base, 20% glycerol, 10 mM SDS, 2% β-ME, pH 6.8). Proteins present in the cell lysate were separated using SDS-PAGE, electro-blotted and subjected to immunodetection as described by Kain et al. [@pone.0016740-Kain1]. The blots were incubated with a 1∶1500 dilution of monoclonal antibody specific for IPNV VP2 (Microtek, Canada), Bad (BD Biosciences, USA), Bid (Millipore, USA), RIP1 (Abcam, UK) or Actin (Millipore, USA) and a 1∶50,000 dilution of a peroxidase-conjugated goat anti-mouse or anti-rabbit antibody (Sigma Aldrich). Alternatively, the blots were incubated with a 1∶3000 dilution of a polyclonal anti-TNFα antibody (AnaSpec, USA) and a 1∶50,000 dilution of a peroxidase-conjugated goat anti-rabbit antibody (Sigma Aldrich). Chemiluminescence detection was performed according to the instructions provided with the Western Exposure Chemiluminescence Kit (GE Healthcare, USA). Resulting western blots were scanned with an imaging densitometer (LAS-3000; Fujifilm, Japan), and optical densities of specific proteins were analyzed with Image Gauge software (Fujifilm). Annexin V-FITC labeling {#s4d} ----------------------- An analysis of phosphatidylserine on the outer leaflet of apoptotic cell membranes was performed using the Annexin-V-FLUOS staining kit (Roche), which contains annexin V-fluorescein and propidium iodide (PI) to differentiate apoptotic cells from necrotic cells. At the end of the various incubation times (0, 6, 12, 18 and 24 h), each sample was removed from the medium and washed with PBS. The cells were incubated with staining solution for 10--15 min. Apoptosis was detected using fluorescence microscopy (Olympus IX70, Japan) with 488-nm excitation and a 525-nm filter for detection [@pone.0016740-Hong3]. Necrosis was detected using fluorescence microscopy (Olympus IX70) with 535-nm excitation and a 620-nm filter for detection. Each sample group was counted three times, with at least 300 cells counted each time. The mean of the three counts for each different group was used to calculate the apoptotic and necrotic cell indices and their respective standard error. RNA preparation {#s4e} --------------- Approximately 10^5^ ZF4 cells/ml were seeded in a 100-mm Petri dish (Nunc) and cultured for more than 24 h. These cells were then infected with IPNV at an MOI of 5 and incubated for 0, 6, 12, or 24 h. At the completion of each incubation period, the culture medium was aspirated, and the cells were washed with PBS. Total RNA was extracted using TRIzol (Invitrogen) and was further purified using an on-column RNase-free DNase digestion (QIAGEN, Germany) to remove possible genomic DNA contamination. The RIN value of the RNA samples before being applied to the microarray was measured using an Agilent 2100 Bioanalyzer (USA) and was 10.0. Microarray preparation {#s4f} ---------------------- The zebrafish 14K oligo microarray comprising 14,067 zebrafish oligonucleotides was designed and synthesized by MWG Genomic Company (Germany) based on 1800 zebrafish gene sequences from the NCBI and a database of 12,768 putative ORFs using NCBI zebrafish EST sequence information [@pone.0016740-Schultz1]. The zebrafish 14K 50-mer oligos were printed on an UltraGAPS Coated Slide (Corning, USA) using an OmniGrid 100 microarrayer (Genomic Solutions, Ann Arbor, USA) according to the manufacturer\'s instructions. After printing, the slides were baked at 80°C for 6 h, incubated in a glass chamber for 45 min at 42°C in pre-warmed block solution (4× SSC, 0.5% SDS, 1% BSA), quickly washed with distilled water at room temperature, and dipped in room temperature isopropanol. The slides were dried by brief centrifugation. Microarray hybridization {#s4g} ------------------------ Amino-allay dye coupling was carried out using the SuperScript Plus Indirect cDNA Labeling System (Invitrogen) according to the manufacturer\'s instructions. We optimized the reverse transcription labeling protocol to use 40 µg total RNA, 5 µg anchored oligo (dT) primer ((dT)20VN), and SuperScript III reverse transcriptase. After a 3-h incubation at 46°C, the reaction was stopped by incubating the reaction at 70°C for 15 min in the presence of 1 N NaOH; the solution was then neutralized by adding 1 N HCl. The reaction mixture was brought to a final volume of 100 µl with nuclease-free water. The amine-modified DNA was purified using a MinElute PCR Purification Kit (QIAGEN), following the instructions in the kit, except that the washes were performed twice instead of once and the probe was eluted with 0.1 M NaHCO~3~. One aliquot of Alexa Fluor Dye was then resuspended in 20 µl of aa-cDNA labeled probe and incubated at RT in the dark overnight. The MinElute PCR Purification was then repeated using EB buffer for the elution. Yeast tRNA (10 µg) was added to the sample, which was then dried in a speedvac at 45°C and redissolved in 70 µl of formamide-based hybridization buffer (MWG, Germany). The mixture was then denatured at 95°C for 2 min. The solution was collected by a brief centrifugation and applied onto the oligo area on the microarray slides. Coverslips (60×22 mm) were applied, and the slides were then placed in a chamber and immersed in a water bath for hybridization overnight at 42°C. The arrays were sequentially washed with 2× SSC/0.1% SDS, 1× SSC/0.1% SDS, 0.5× SSC, and 0.1× SSC at room temperature for 5 min per wash. The slides were subsequently dried by brief centrifugation. The arrays were scanned using an Axon GenePix 4000B scanner, and the median spot intensity was determined using an Axon GenePix Pro5.1 (Molecular Devices, Sunnyvale, USA). Microarray data analyses {#s4h} ------------------------ The data files were imported into GeneSpring GX 7.3 (Agilent Technologies, Foster City, USA) for further analysis. "LOWESS Normalization" was applied for data normalization in GeneSpring. The expression data sets must have passed the following quality control categories before they were used for cluster analysis: 1) the hybridization results were not flagged as bad; 2) the net intensity of both channels was equal to or greater than 500; and 3) the statistical analyses were applied to the triplicate data for each spot and repeated three times. For expression data, ratios equal to or greater than 2 were considered up- or down-regulated. DNA microarray data sets were deposited at NCBI\'s Gene Omnibus Express under the accession number GSE21077. All of the DNA microarray data is MIAME compliant. Quantitative real-time PCR {#s4i} -------------------------- The primers used for quantitative PCR were designed using Primer Express 2.0 software (Applied Biosystems, USA) and are listed in [Table S3](#pone.0016740.s006){ref-type="supplementary-material"}. For real-time quantitative PCR, first-strand cDNA from ZF4 cells was synthesized using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) with random primers. Quantitative PCR was performed using the Power SYBR Green PCR Master Mix (Applied Biosystems) and an ABI Prism 7000 Sequence Detection System. Knockdown of TNFα and RIP1 by RNA interference {#s4j} ---------------------------------------------- Duplex small interfering RNA (siRNA) that specifically targeted the mRNA encoding TNFα (BC124141) or RIP1 (BC163762) and scrambled siRNA were commercially synthesized (Sigma, Singapore). The sequences of scrambled siRNA and duplex siRNA that specifically targeted TNFα or RIP1 are listed in [Table 1](#pone-0016740-t001){ref-type="table"}. Duplex siRNAs (20 nM) were transfected into ZF4 cells that were cultured in 60 mm-diameter plastic tissue culture plates (Nunc) using the GeneMute siRNA transfection reagent (SignaGen Laboratories, USA). After a 6-h incubation period, 1 ml RPMI 1640 culture medium containing 10% FBS (Invitrogen) was added to each well without removing the transfection reagent. The cells were then infected with IPNV at an MOI of 1 in 10% FBS RPMI 1640 at 18°C. 10.1371/journal.pone.0016740.t001 ###### Primer sequences for duplex siRNA that specifically targeted the mRNA encoding TNFα or RIP1 and scrambled siRNA. ![](pone.0016740.t001){#pone-0016740-t001-1} Oligo Name Sequence 5′→3′ Oligo Name Sequence 5′→3′ ----------------------- --------------------- --------------------------- --------------------- tnfa-siRNA-sense CUCUGAAUCAAAGACCUUA tnfa-siRNA- antisense UAAGGUCUUUGAUUCAGAG rip1-siRNA-sense CAAAUGCACUCAUUCCUGA rip1-siRNA-antisense UCAGGAAUGAGUGCAUUUG Scrambled-siRNA-sense GAUCAUACGUGCGAUCAGA Scrambled-siRNA-antisense UCUGAUCGCACGUAUGAUC Caspase activity assays {#s4k} ----------------------- Approximately 10^5^ ZF4 cells/mL were seeded in a 60-mm Petri dish (Nunc) and cultured for 24 h. The cells were infected with IPNV at a MOI of 5 and incubated for 6, 12, or 24 h at 18°C. Caspase-3, -8, or -9 activation assays [@pone.0016740-Hong7] were performed using 10^6^ cells per timepoint. The cleavage of the Z-DEVD, Z-LETD, or Z-LEHD synthetic caspase-3, -8 and -9 substrates, respectively, was used to determine caspase activation using the Caspase-Glo Assay Kit (Promega). The assays were performed in 96-well plates and analyzed using a luminometer (VICTOR X2, PerkinElmer, USA). The amount of luminescence detected is directly proportional to the amount of caspase activation. All of the luminogenic substrate assay experiments were performed at the same time. Both the mock- and IPNV-infected caspase activation profiles were the same in all experiments and are included in each figure to facilitate comparisons. The results of all experiments are reported as the mean ± SEM. Intracellular ROS detection {#s4l} --------------------------- Following infection with IPNV for the indicated time period (6, 12, 18 or 24 h), ZF4 cells were collected, washed twice with PBS, and incubated in warm HBSS/Ca/Mg solution containing 25 µM carboxy-H~2~DCFDA (Invitrogen) for 30 min at 37°C to detect ROS in live cells. ROS production was quantified using a fluorescent plate reader (VICTOR X2, PerkinElmer) with 485-nm emission and 525-nm absorption. Supporting Information {#s5} ====================== ###### IPNV infection of zebrafish embryonic cells (ZF4) induces host cell death. (A) Detection of the viral protein expression profile in ZF4 cells following infection with IPNV (MOI=1) using western blot. The blot was probed using a polyclonal VP2-specific antibody, and lanes 1--4 correspond to 0, 6, 9, and 12 h p.i., respectively. The blot was probed with an actin-specific antibody as an internal control. (B) Viability of ZF4 cells infected with IPNV at an MOI of 1, 5, or 10 after 6, 12, 24, and 48 h. The viability for each sample was determined in three individual experiments. Data shown are the mean ± SD. (DOC) ###### Click here for additional data file. ###### Determination of gene expression levels at 0, 6, 12 and 24 h p.i. using quantitative real-time RT-PCR. The expression profile of pro-apoptotic genes (A--B) were detected using quantitative real-time RT-PCR. The fold-change values of IPNV-infected cells compared to uninfected cells for genes representative of each of these groups is shown. The quantification of gene expression in IPNV-infected cells compared to uninfected control cells was calculated relative to the expression of ef1a as an internal control. Student\'s t tests indicate significant differences compared to 0 h: \, p\<0.05; \\, p\<0.01. (DOC) ###### Click here for additional data file. ###### TNFα has the highest connectivity among the altered genes in microarray and quantitative RT-PCR experiments.* All of the altered genes were analyzed by Pathway Studio 6.0. The software is available from Ariadne Genomic Inc. (DOC) ###### Click here for additional data file. ###### List of differentially expressed zebrafish mRNAs during IPNV infection identified using microarray analyses. ^a^ Minus, decreased gene expression; no minus, increased gene expression; boldface, \>2-fold-increased or decreased gene expression. (DOC) ###### Click here for additional data file. ###### Comparison of the fold changes determined between microarray analysis and real-time PCR. a) Quantitative real-time RT-PCR validation of oligo microarray data for genes that were up- or down-regulated in IPNV-infected versus uninfected control host cells at various timepoints post-infection. The genes validated using RT-PCR are also listed in Table S3. The quantification of gene expression in IPNV-infected versus uninfected control cells was done relative to the ef1α gene. b) Significant change in gene expression between IPNV-infected cells and uninfected cells as determined using microarray, p\<0.05. c) Significant change in gene expression between IPNV-infected cells and uninfected cells as determined using quantitative real time RT-PCR, p\<0.01. (DOC) ###### Click here for additional data file. ###### Primer sequences for quantitative RT-PCR. (DOC) ###### Click here for additional data file. We thank Dr. Jiann-Ruey Hong, who reviewed and provided useful comments regarding this manuscript. Competing Interests:The authors have declared that no competing interests exist. Funding:This study was funded by grants (NSC95-2313-B-001-004, NSC96-2313-B-001-001) awarded to Dr. Jen-Leih Wu from the National Science Council, Taiwan, Republic of China. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. [^1]: Conceived and designed the experiments: W-LW H-YG J-LW. Performed the experiments: W-LW G-HL WL. Analyzed the data: W-LW G-HL C-CL. Contributed reagents/materials/analysis tools: W-LW J-RH M-WL. Wrote the paper: W-LW J-RH J-LW.	{ "pile_set_name": "PubMed Central" }
Background ========== Left ventricular ejection fraction (LVEF) is the single most used non-invasive measure of cardiac function in clinical practice and is an important prognostic factor for survival after myocardial infarction (MI), in stable coronary artery disease (CAD), and in heart failure \[[@B1]-[@B3]\]. Health-related quality of life (HRQoL) has also been identified as a predictor of survival in patients with CAD and heart failure \[[@B4]-[@B7]\]. However, the relationship between LVEF and HRQoL has not been settled and whether LVEF can predict HRQoL is still controversial. Some studies have observed an association \[[@B8]-[@B11]\] while others have not \[[@B12]-[@B17]\]. Most of these studies reported on highly selected heart patients enrolled in clinical trials \[[@B9],[@B13],[@B14]\], patients with known reduced LVEF \[[@B13],[@B14]\], and patients with chronic heart failure \[[@B11],[@B17]\]. Only two previous studies included unselected MI patients, of which one observed an association between LVEF measured at the time of the MI and later HRQoL \[[@B10],[@B16]\]. Better understanding of the relationship between cardiac function and quality of life might contribute to tailor treatment that would maintain or improve the patients\' daily functioning. Evidently, more research on this subject is needed. Accordingly, our aim was to assess the relationship between LVEF measured during hospitalization for acute MI and HRQoL 2.5 years later in an unselected MI patient population, using both a generic and a disease-specific HRQoL measure. Methods ======= Study design and sample ----------------------- We established a cohort of hospitalized patients with a discharge diagnosis of acute MI, defined as codes I21 and I22 in ICD-10 (The International Statistical Classification of Diseases and Related Health Problems, tenth revision) \[[@B18]\]. We wanted our cohort to be representative of Norwegian patients with MI, and hence recruited them from teaching and non-teaching hospitals in different regions of Norway. We included a total of 754 consecutive patients who were discharged alive from 15 hospitals during a 3-month period between August 1, 1999, and January 31, 2000. Before discharge, LVEF was measured in 406 (54%) patients. The patient population is described in Figure [1](#F1){ref-type="fig"}. ![Flow chart of study population. Study population stratified on patients with and without their left ventricular ejection fraction measured](1471-2261-8-28-1){#F1} In 2002, we mailed a questionnaire to patients who were still alive according to hospital information systems and the National Population Register of Statistics Norway. The questionnaires were mailed from the hospital of discharge, along with a cover letter signed by the head of the hospital\'s cardiology unit. After 4 weeks, we sent a reminder to non-respondents. At the time of the survey, 191 of the 754 patients had died, 8 had an unknown address, and 7 were excluded for miscellaneous reasons. Hence, we mailed the questionnaire to the remaining 548 patients of whom 408 (74%) returned completed questionnaires. A total of 256 of the 408 respondents (63%) had data on LVEF (Figure [1](#F1){ref-type="fig"}). The mean time from the index MI to questionnaire response was 2.5 (SD = 0.2) years, range 2.1--3.1 years. Review of medical records ------------------------- Baseline characteristics were abstracted from the patients\' medical records and included previous medical history, presenting features, in-hospital treatment, and medication at discharge. Cardiovascular morbidities from before the index MI were categorized as previous MI, hypertension, angina pectoris, heart failure, peripheral vascular disease, or stroke. The diagnoses were based on either previous ICD-codes or explicit statements in the medical record. The major indications for each cardiovascular drug prescribed at discharge were classified as secondary prevention, hypertension, angina pectoris, heart failure, or other. More details about sampling and collection of data from the patients\' medical records are available elsewhere \[[@B18]\]. Measurement of left ventricular ejection fraction ------------------------------------------------- We measured LVEF by one of two methods: Multiple gated-acquisition radionuclide ventriculography (MUGA) in 129 (51%) patients and echocardiography in 127 (49%) patients. MUGA is a reliable reference method, and the widely used echocardiographic method correlates fairly well with radionuclide imaging \[[@B19],[@B20]\]. LVEF \>50% is considered to indicate normal left ventricular function, while LVEF \<40% indicates reduced function. According to this classification we categorized LVEF as normal (LVEF \>50%), intermediate (40--50%) or reduced (\<40%). Questionnaire ------------- The questionnaire focused on HRQoL, using the EQ-5D Index, the EQ-5D Visual Analogue Scale (EQ-VAS), and the Kansas City Cardiomyopathy Questionnaire (KCCQ). In addition, we asked about subsequent cardiac events and revascularization procedures after hospital discharge for the index MI. The KCCQ is a self-administered 23-item questionnaire designed for measuring HRQoL in patients with chronic heart failure. It comprises six scales: Symptoms, Symptom stability, Physical limitation, Social limitation, Self-efficacy and Quality of life \[[@B21]\]. Four of the scales, Symptoms, Physical limitation, Social limitations and Quality of life, are aggregated to an overall score, the KCCQ clinical summary score \[[@B21]\]. Each item is scored on a 5 to 7-point Likert scale \[[@B22]\]. Each scale score is calculated as the mean of its item scores and transformed to a 0--100 scale, with higher score indicating higher level of functioning. The KCCQ has been translated into Norwegian, and its psychometric properties have been documented in post MI patients \[[@B23]\]. Change in KCCQ score of 5, 10, and 15 points correspond with small, moderate, and large clinical change respectively \[[@B24]\]. The EQ-5D is a self-administered HRQoL instrument with 5 items: Mobility, Self-care, Usual activities, Pain/discomfort, and Anxiety/depression. Each item is scored on a 3-point Likert scale: no problems (score of 1), moderate problems (2), and extreme problems (3). Responses to these items can be converted to a utility score, the EQ-5D Index, by applying an algorithm derived from time trade-off valuations of health status obtained from the general population \[[@B25]\]. A score of 1.0 represents perfect health and 0 represents dead. For the EQ-5D Index negative utilities are possible, representing states perceived to be worse than dead. We used a UK time trade-off tariff \[[@B26]\]. The EQ-5D Index has been used and documented in post MI patients \[[@B27]-[@B29]\]. In addition to the QE-5D Index the EQ-5D questionnaire include the EQ-VAS, a visual analogue scale ranging current overall health by one single number on a scale from 0 (worst imaginable health state) to 100 (best imaginable health state) \[[@B28]\]. The EQ-VAS has documented acceptable reliability and validity in patients with CAD \[[@B29]-[@B31]\]. Statistical analysis -------------------- We present descriptive statistics with means and SDs, or proportions. For group comparisons, we used the t-test, analysis of variance, Kruskal-Wallis test for three independent groups, Wilcoxon rank-sum test, or chi-square test where appropriate. We used multiple linear regression analysis to identify determinants for KCCQ clinical summary score, EQ-5D Index, and EQ-VAS at a mean of 2.5 years after MI. To reduce problems with multicollinearity, we checked pairwise Pearson correlations (r) between independent variables. However, none of the pairwise correlations had r\> 0.70. Variables with p\< 0.25 in bivariable linear regression analysis were included in multivariable modeling. In the multivariable models, we first included age at admission, sex, length of education in years, and LVEF in the models, and then added the other potential independent determinant variables in a forward stepwise fashion. In addition to age, sex, education, and LVEF, we retained all independent variables with p \< 0.05 in multivavariable linear regression in the final model. The final models were checked for interactions. The coefficient of determination (R-square) is a measure of explained variance in regression analysis. To determine the marginal contribution of LVEF to R-square, in KCCQ clinical summary score, EQ-5D Index score, and EQ-VAS, we compared the R-square in the final models adjusted for degrees of freedom, with R-square in the final models without LVEF. We used a 5% significance level with two-sided tests. Standard statistical software was used for all analyses (SPSS version 12.0, SPSS, Chicago, IL). The Regional Committee for Medical Research Ethics and the Norwegian Data Inspectorate approved the study. Demographics ------------ Respondents (n= 408) were younger than non-respondents (n= 140), comprised a higher proportion of males, had less cardiovascular morbidity at admission, and fewer cardiovascular indications for medication at discharge. Respondents and non-respondents did not differ in the proportion of patients with ST-segment elevation at admission, Q-wave infarction, or localization of the index MI. Respondents who had their LVEF measured (n= 256) had higher LVEF than non-respondents who had their LVEF measured (n= 71) (normal, intermediate, reduced: 65%, 22%, 13% vs. 51%, 27%, 23%, p= 0.02). A more extensive comparison of respondents and non-respondents has been presented elsewhere \[[@B23]\]. Results ======= Respondents who had their LVEF measured (n= 256) during hospitalization for the index MI were younger than respondents who had not (n= 152), had less cardiovascular morbidity at admission, and a higher proportion were smokers. Further, a larger proportion of respondents who had their LVEF measured had ST-segment elevation MI, underwent acute revascularisation, and developed Q-wave MI, and a lower proportion had medication for angina pectoris at discharge (Table [1](#T1){ref-type="table"}). ###### Possible determinant variables All respondents LVEF value ------------------------------------------ ------------------- -------------- --------- ----------- ----------- ----------- --------- n 256 152 167 56 33 Time since index MI, years 2.5 (0.2) 2.6 (0.2) \<0.001 2.5 (0.2) 2.6 (0.2) 2.5 (0.2) 0.046 Sex (% women) 29 29 1.0 29 30 27 1.0 Age, years 64 (12) 68 (12) 0.002 62 (12) 68 (11) 68 (13) 0.001 Education, years, range 7--21 10 (3) 10 (4) 0.9 11 (3) 10 (3) 10 (3) 0.3 Smoker at admission 49^a^ 39^b^ 0.07 51^c^ 48^d^ 45^e^ 0.8 Comorbidities before the index MI Diabetes mellitus 10 6 0.3 8 11 15 0.5 Chronic obstructive pulmonary disease 8 7 0.7 6 13 6 0.2 Hypertension 29 35 0.2 30 29 24 0.8 Previous MI 17 24 0.08 11 24 36 0.001 Angina pectoris 22 30 0.08 20 22 33 0.2 Peripheral vascular disease 4 9 0.08 2 7 9 0.1 Stroke 5 6 0.5 3 7 12 0.07 Heart failure 2 8 0.009 1 2 9 0.02 Index MI, characteristics ST-segment elevation 59 35 \<0.001 58 70 46 0.07 Localization 0.1 \<0.001 Anterior wall 44 30 34 57 70 Inferior wall 41 38 51 27 12 Unknown 16 24 15 16 18 Q-wave 48 33 0.004 47 48 52 0.9 In-hospital treatment Revascularization 49 22 \<0.001 48 63 33 0.02 Indication for medication at discharge Secondary prevention 99 98 0.7 99 100 97 0.4 Hypertension 11 12 0.8 14 5 6 0.1 Angina pectoris 18 35 \<0.001 19 18 12 0.6 Heart failure 19 22 0.4 7 29 61 \<0.001 Subsequent events New MI 5 7 0.5 5 5 3 0.8 Percutaneous coronary intervention 15 11 0.3 20 5 6 0.008 Coronary artery bypass graft surgery 11 13 0.7 12 7 15 0.5 Background variables, characteristics of the index myocardial infarction, treatment and subsequent events according to whether left ventricular ejection fraction (LVEF) was measured or not and in patients with normal (\>50%), intermediate (40--50%), and reduced (\<40%) LVEF. Numbers represent mean (SD) or percent. ^a^n= 227; ^b^n= 137; ^c^n= 150; ^d^n= 48; ^e^n= 29; MI, Myocardial Infarction Among respondents who had their LVEF measured, those with reduced LVEF were older and had increased prevalence of heart failure at admission. There was an increased prevalence of previous MI, anterior wall infarction, and use of medication on the indication heart failure with falling LVEF. Further a smaller proportion of respondents with reduced LVEF underwent acute revascularization or had subsequent percutaneous coronary intervention (Table [1](#T1){ref-type="table"}). Health-related quality of life ------------------------------ The mean score for all patients with LVEF measured was 80 for the KCCQ clinical summary scale, 0.80 for the EQ-5D Index, and 69 for the EQ-VAS. HRQoL scores differed between patients with different levels of LVEF both for the KCCQ clinical summary score (p\< 0.001) the EQ-5D Index (p= 0.005), and the EQ-VAS (p= 0.001) (Table [2](#T2){ref-type="table"}). ###### Health-related quality of life scores Left ventricular ejection fraction ----------------------------- -------------------------------------- ------------- ------------- n 167 54 33 KCCQ clinical summary score 85 (18) 75 (22) 68 (21) n 160 53 30 EQ-5D Index 0.83 (0.18) 0.72 (0.27) 0.76 (0.14) n 148 46 26 EQ-VAS 72 (18) 65 (21) 57 (20) KCCQ clinical summary score, EQ-5D Index score, and EQ-VAS score according to left ventricular ejection fraction, mean (SD) KCCQ, Kansas City Cardiomyopathy Questionnaire; EQ-VAS, EuroQol Visual Analogue Scale In multivariable linear regression analysis, age ≥ 70 years, a history of chronic obstructive pulmonary disease (COPD), subsequent MI, and intermediate or reduced LVEF measured during hospitalization were all independent determinants of lower KCCQ clinical summary score 2.5 years after the index MI (Table [3](#T3){ref-type="table"}). Female sex, medication for angina pectoris at discharge, and intermediate LVEF were independent determinants a lower EQ-5D Index score, while a history of peripheral vascular disease was associated with a higher EQ-5D Index score (Table [3](#T3){ref-type="table"}). Age 70 ≥ years, a history of COPD, and reduced LVEF were independent determinants of lower EQ-VAS score (Table [3](#T3){ref-type="table"}). ###### Multivariable linear regression analyses KCCQ clinical summary score EQ-5D Index EQ-VAS --------------------------------------------- ------------------------------- ----------------- ---------- ------- ------------------ ------- ------- ----------------- ------- n 253 242 219 Sex (0 = male, 1 = female) -3.4 (-8.8 to 2.1) 0.2 -0.10 (-0.16 to -0.04) 0.001 -1.7 (-7.7 to 4,2) 0.6 Age \< 70 years = 0, ≥ 70 years = 1 -6.9 (-12.1 to -1.6) 0.01 -0.03 (-0.09 to 0.03) 0.3 -6.4 (-12.0 to -0,7) 0.03 Education, years (range 7--21) 0.5 (-0.3 to 1.2) 0.2 -0.00 (-0.01 to 0.01) 0.5 -0.09 (-0.9 to 0.7) 0.8 Left ventricular ejection fraction \>50% 40--50% -6.9 (-13.0 to -0.9) 0.02 -0.10 (-0.17 to -0.04) 0.001 -4.3 (-10.7 to 2.1) 0.2 \<40% -15.4 (-22.6 to -8.3) \<0.001 -0.07 (-0.14 to -0.01) 0.08 -13.5 (-21.4 to -5.6) 0.001 Peripheral vascular disease 0.16 (0.04 to 0.28) 0.01 Chronic obstructive pulmonary disease -12.5 (-21.5 to -3.5) 0.006 -12.1 (-21.5 to -2.8) 0.01 Medication for angina pectoris at discharge -0.11 (-0.17 to -0.04) 0.001 Subsequent myocardial infarction -11.2 (-21.9 to -0.6) 0.04 Total adjusted R-square 0.16 0.16 0.10 Independent determinants of health-related quality of life 2.5 years after myocardial infarction in patients with left ventricular ejection fraction measured during index hospitalization. In the final multivariable models, R-square was 0.16 for both KCCQ clinical summary score and EQ-5D Index, and 0.10 for the EQ-VAS (Table [3](#T3){ref-type="table"}). When excluding LVEF in the final models, R-square was 0.09, 0.12, and 0.06 respectively, indicating that LVEF accounted for 25% to 44% of the variation explained by the final multivariable models. One hospital measured LVEF routinely in all MI patients, using MUGA (n= 101). When applying our final multivariable models in this subset of patients, the relationship between LVEF and HRQoL was essentially the same as in all patients with measured LVEF. In this subset, using KCCQ clinical summary score as the dependent variable, the unstandardized regression coefficient (B) and 95% confidence interval (CI) for patients with intermediate LVEF and reduced LVEF were -10.4 (-21.0 to 0.3) (p= 0.06) and -13.6 (-26.9 to -0.4) (p= 0.04) respectively. With EQ-5D Index as dependent variable Band 95% CI were -0.13 (-0.25 to -0.02) (p= 0.02) and -0.02 (-0.16 to 0.12) (p= 0.7). Discussion ========== In this study LVEF measured during hospitalization for the index MI was a determinant of HRQoL 2.5 years later, with poorer HRQoL in patients with reduced LVEF. This relationship persisted after adjusting for comorbidities, sociodemographic variables, and variables related to the index MI. The summary score of the condition-specific KCCQ questionnaire, which has been validated in an earlier study \[[@B23]\], and the EQ-VAS showed a trend of falling scores with falling LVEF, while the EQ-5D Index only captured a difference between patients with normal and intermediate LVEF. A difference in score for the KCCQ clinical summary score between patients with normal LVEF and intermediate or reduced LVEF of 10 and17 points respectively, indicates a moderate to large clinical difference between the groups of patients with different LVEF \[[@B24]\]. The observed falling HRQoL scores with falling LVEF both for KCCQ clinical summary score and EQ-VAS indicate that the level of systolic heart function is a determinant for HRQoL. However, some influence from the mere fact that an MI has occurred cannot be ruled out since also patients with normal LVEF tend to have reduced EQ-5D scores compared to US norms \[[@B32]\]. If so, it is possible that other cardiac pathophysiological mechanisms or psychological mechanisms are involved. An understanding of the way LVEF may influence HRQoL can be provided from a conceptual model outlined by Wilson and Cleary \[[@B33]\]. In their model biological and physiological variables influence symptoms, that is, patients with reduced LVEF may experience symptoms such as fatigue, dyspnoea, and sleep disturbances. These symptoms can in turn affect the patients\' functional status, general health, and overall quality of life. The impact on individual patients is further modified by psychological and socioeconomic variables \[[@B33]\]. In our study the level of LVEF immediately after the index MI had a statistically significant impact on later HRQoL and in accordance with the usual interpretation of the KCCQ the clinical importance was moderate to large \[[@B24]\]. However, the observed absolute change in R-square was small indicating that the amount of variation in HRQoL scores explained by LVEF was moderate. A moderate amount of variation in HRQoL score explained by a single clinical variable can be expected if Wilson and Cleary\'s model is valid, as other interrelated variables intervene between pathophysiology of the heart and quality of life \[[@B33]\]. Three previous studies have observed an association between LVEF and HRQoL in patients with CAD \[[@B8]-[@B10]\]. However, two of these studies reported on patients not comparable to those in the present study. One reported on patients admitted to hospital with acute chest pain, thus including patients with acute MI, unstable angina, and chest pain of other reasons \[[@B8]\]. The second study presented results from patients enrolled in a clinical trial of thrombolysis, thus representing a highly selected group of MI patients \[[@B9]\]. In the third study Ecochard et al. showed that LVEF \< 46% measured by ventricular angiography within a month after the index MI was associated with reduced function on the Physical mobility dimension of the Nottingham Health Profile (NHP) assessed at one year \[[@B10]\]. However, they did not observe any association between LVEF and the five other dimensions of the NHP (Energy level, Sleep, Pain, Emotional reactions, and Social isolation). By contrast, we found a relationship between LVEF and composite HRQoL scales covering physical, emotional, and social aspects of the HRQoL concept. Furthermore, we observed falling HRQoL scale scores with falling LVEF for the KCCQ clinical summary score and the EQ-VAS. Ecochard et al. did not find a similar fall by use of NHP, neither did we by use of the EQ-5D Index \[[@B10]\]. The dissimilarities might be due to the fact that both NHP and EQ-5D Index are generic instruments and thus supposed to be less sensitive to smaller changes in health status and disease severity than the disease-specific KCCQ \[[@B34]\]. Another reason why reduced LVEF was not independently associated with lower EQ-5D Index score in our study was the reduced statistical power due to low number of respondents with LVEF \< 40%. Studies that did not find an association between LVEF and later HRQoL also varied with regard to inclusion criteria and most of them reported on selective groups of patients and not on unselected acute MI patients as we did \[[@B12]-[@B16]\]. One reported on patients 65 years of age or older with CAD \[[@B12]\], two on clinical trials of post MI patients with reduced LVEF \[[@B13],[@B14]\], and one on patients with their first MI receiving thrombolysis \[[@B15]\]. McBurney et al. did not detect any difference in HRQoL between patients with LVEF \< 40% and patients with normal and sub-normal LVEF in a patient sample comparable to ours \[[@B16]\]. However, they used a generic questionnaire, the short form 12 (SF-12) which probably is less sensitive to changes in health status and disease severity \[[@B34]\]. In our study on MI patients, age ≥ 70 years was associated with lower KCCQ clinical summary score. This contrasts with previous results on heart failure patients in which increased age independently correlated with higher KCCQ Quality of life scale score \[[@B35]\]. One possible explanation for this discrepancy is that the KCCQ clinical summery score, which we used as an HRQoL measure in our study, is an overall scale including, in addition to the KCCQ Quality of life scale, scales on symptoms, physical limitations, and social limitations. Another possible explanation is that in our sample of patients with previous MI, increasing age is associated with more advanced CAD and age might act as a surrogate for disease severity \[[@B36]\]. This might not be the case in a sample of patients with heart failure as the underlying cause in younger patients with heart failure more commonly is dilated cardiomyopathy or other cardiomyopathies rather than CAD \[[@B35]\]. We did not observe an association of age with EQ-5D Index scores even though EQ-5D Index US norms report a small decrease in scores with increasing age \[[@B32]\]. We identified sex as an independent determinant of EQ-5D Index score, but not of KCCQ clinical summary score. A difference in score between men and women who are otherwise comparable might be less likely when applying a disease specific instrument than a generic instrument which has a broader perspective. Thus, the EQ-5D Index US norms have reported a small difference between men and women with poorer score in women \[[@B32]\]. Our observations are in agreement with these observations. Our study showed that the presence of COPD was associated with reduced HRQoL after MI. COPD and CAD share some causal risk factors, and might to some extent present with similar symptoms, for example dyspnoea. Therefore, it is not surprising that patients with COPD who suffer an MI are at increased risk of impaired HRQoL. Similarly, medication for angina pectoris was associated with reduced EQ-5D Index score, and presumably, the reason for this is the fact that such medication reflects enhanced disease severity. In our study the presence of peripheral vascular disease was associated with higher EQ-5D Index score. However, this finding is based on only 11 patients with peripheral vascular disease, and therefore should be interpreted with caution. Due to the rather long time from the MI to measuring of HRQoL in our study, the association between LVEF and HRQoL could have been distorted. Although the patients reported on intervening major cardiac events, such as new MI or coronary revascularization procedures, other major life events, worsening or improvement of illness not reported in our study, might have had an effect on current HRQoL. We addressed this issue by entering the time between the index MI and HRQoL assessment as a variable in the multivariable analyses. This factor was not, however, independently associated with HRQoL. In the context of study limitations the representativeness of the sample should be discussed. The eligible patients were representative of survivors of an unselected MI population. Three quarters of the patients responded to the survey, and LVEF was measured in two thirds of the respondents, thus we completed our analysis on approximately half of the eligible patients. This might have skewed the representativeness. However, results from a sub-group analysis of patients in one hospital in which all MI patients had their LVEF measured, were largely in line with the overall results, indicating that the main results pertains to unselected MI patients. Another limitation is the uncertainty of whether the clinicians categorizing the indications for drugs prescribed at discharge did assign the most important indication for each drug, as the protocol instructed. Cardiovascular drugs may have more than one indication for use, and we did not assess the reliability and validity of this classification. However, the classification was undertaken by experienced physicians, most of them cardiologists ensuring the best possible assessment. Conclusion ========== LVEF measured during hospitalization for acute MI is an independent determinant for later HRQoL also after taking sociodemographic and clinical variables into account. The magnitude of the difference in HRQoL score between patients with normal, intermediate, and reduced LVEF was of clinical importance. As expected, in accordance with theoretical models, only a moderate amount of the observed variation in HRQoL score was explained by the level of LVEF. Restoration of myocardial function after an MI is, in addition to being important for live expectancy, also crucial for long-term daily functioning after the event. Abbreviations ============= CAD: coronary artery disease; COPD: chronic obstructive pulmonary disease; EQ-VAS: EuroQol Visual Analogue Scale; HRQoL: health-related quality of life; KCCQ: Kansas City Cardiomyopathy Questionnaire; LVEF: left ventricular ejection fraction; MI: myocardial infarction; MUGA: multiple gated-acquisition radionuclide ventriculography; NHP: Nottingham Health Profile. Competing interests =================== The authors declare that they have no competing interests. The work was performed at the Norwegian Knowledge Centre for the Health Services and at the Department of Pharmacotherapeutics, University of Oslo, Oslo, Norway. The project had no external funding or grants. Authors\' contributions ======================= KIP: Had the idea to the study, organized the data collection and data preparation, performed statistical analysis, drafted the first version of the manuscript, revised the manuscript and approved the final version. EK: Participated in data collection and data preparation, participated in discussion of study findings, revision of the manuscript and approved the final version. AR: Participated in discussion of study findings, revision of the manuscript and approved the final version. KS: Participated in study design, statistical analysis, manuscript preparation and revision, and approved the final version. AR: Participated in data collection, discussion of study findings, manuscript preparation and revision, and approved the final version. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2261/8/28/prepub> Acknowledgements ================ We thank the following physicians for participating in the study: E. Anker, T. Dahl, H.P. Dørum, T. Grønvold, J. Hærem, T. Indrebø, K. Knutsen, K.T. Lappegård, A. Mangschau, C. Platou, J. Vegsundvåg, A. von der Lippe, K. Waage, and A. Zalmai.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1-ijerph-16-00683} =============== Medicine has been revolutionized by antibiotics throughout the 20th century, due to their ability to prevent and treat bacterial infections. However, the overuse and misuse of antibiotics prompted bacteria to develop resistance. Our capacity for treating infectious diseases is challenged by the increase in antibiotic-resistant bacteria (ARB) and antibiotic resistance genes (ARGs) \[[@B1-ijerph-16-00683]\]. Indeed, the World Health Organization classifies antimicrobial resistance as a "serious threat" to global public health \[[@B2-ijerph-16-00683]\]. This problem is especially prevalent in China because China is one of the largest producers and consumers of antibiotics in the world. More than 90 million tons of antibiotics are consumed annually in China, with approximately 48% procured for human use and 52% for agriculture \[[@B3-ijerph-16-00683]\]. The prescription of antibiotics accounts for about 10% of all drugs prescribed by hospitals in developed countries, but it exceeds 50% in China \[[@B4-ijerph-16-00683]\]. Cephalexin, amoxicillin, tetracycline, ofloxacin, and norfloxacin were the most frequently used antibiotics \[[@B3-ijerph-16-00683]\]. Antibiotics are usually used to control the disease and stimulate growth of animals in modern agriculture \[[@B5-ijerph-16-00683]\]. Amoxicillin, florfenicol, penicillin, lincomycin, and enrofloxacin were the five most frequently used antibiotics in the livestock industry, and the annual usage in China has exceeded 4000 tons \[[@B3-ijerph-16-00683],[@B6-ijerph-16-00683]\]. Generally, antibiotics are incompletely absorbed in mammalian bodies and, hence, are excreted and dispersed into the environment. This aggravates the dissemination of ARB and ARGs in the environment. Therefore, hospital and animal husbandry are the primary contributors of antibiotic and ARG dissemination into receiving environments through waste disposal processes, municipal wastewater treatment, and application of manure to land. Numerous studies have addressed the detection of antibiotics and ARGs in different environmental conditions, including air, water, sediment, sludge, and soil \[[@B7-ijerph-16-00683],[@B8-ijerph-16-00683],[@B9-ijerph-16-00683],[@B10-ijerph-16-00683],[@B11-ijerph-16-00683],[@B12-ijerph-16-00683],[@B13-ijerph-16-00683]\]. High-throughput quantitative PCR (HT-qPCR) combined with Illumina sequencing has been developed to provide a comprehensive insight into the profile of an extensive number of ARGs in different environmental samples \[[@B14-ijerph-16-00683],[@B15-ijerph-16-00683],[@B16-ijerph-16-00683],[@B17-ijerph-16-00683]\]. Su et al. \[[@B18-ijerph-16-00683]\] used this method to characterize the geographical and seasonal distributions of ARGs and their potential hosts from 116 urban sewage samples. A total of 381 ARGs were determined and widely shared in all samples. The results demonstrated that human activity was the main driver of antibiotic resistance and its distribution. In recent times, humans spend the majority of time (almost 90%) indoors. There has been a growing concern over the health effects associated with indoor air quality since airborne diseases, including influenza \[[@B19-ijerph-16-00683]\], tuberculosis \[[@B20-ijerph-16-00683]\], the common cold \[[@B21-ijerph-16-00683]\], and legionellosis \[[@B22-ijerph-16-00683]\], exist in indoor air. Several studies have documented that humans who are exposed daily to these airborne particles can induce allergic and chronic inflammatory responses \[[@B23-ijerph-16-00683]\]. While current research in indoor air quality has mainly focused on pathogenic microorganisms, the occurrence and abundance of antibiotics and ARGs in indoor environments is still unclear. Air-conditioning (AC) aims to achieve a more comfortable interior environment, particularly for humans and animals. However, AC systems allow microbial organisms to proliferate and infectious contaminants to accumulate through the collection of moisture and air exchange \[[@B24-ijerph-16-00683]\]. Extensive epidemiologic research has identified relationships between AC systems and indoor air quality \[[@B25-ijerph-16-00683]\]. In this study, 12 samples were collected from hospitals, farms, city residences, and village residences. HT-qPCR and Illumina sequencing of bacterial 16S rRNA gene were used (1) to comprehensively characterize the ARG profiles in air-conditioning filters from different sites; (2) to compare ARG profiles in different samples and locations, i.e., hospitals and farms; and (3) to reveal the correlations among the ARGs and bacteria in situ. 2. Materials and Methods {#sec2-ijerph-16-00683} ======================== 2.1. Site Selection and Sample Collection {#sec2dot1-ijerph-16-00683} ----------------------------------------- A total of 12 samples were collected from four types of environment in Ningbo city (China) in October 2017. These environments included (1) hospitals (general hospital, medical hospital, surgical hospital); (2) farms (chicken farm, duck farm, swine farm); (3) city residences (3 samples); and (4) village residences (3 samples). Air-conditioning units had been running for the summer, and the air-conditioning filter had not been cleaned for 4 months. The dust in the air-conditioning filter was collected, sieved through a 2.0 mm mesh, and then stored at −20 °C for DNA extraction. 2.2. DNA Extraction and Sequencing {#sec2dot2-ijerph-16-00683} ---------------------------------- The biomass from dust samples was too low to obtain sufficient DNA for HT-qPCR and Illumina sequencing analysis using commercial kits. Therefore, DNA was extracted from 2 g dust samples by previously described freeze-grinding and SDS lysis-based methods \[[@B26-ijerph-16-00683]\]. DNA was purified by electrophoresis on a low melting agarose gel, followed by phenol extraction according to Zhou et al. \[[@B27-ijerph-16-00683],[@B28-ijerph-16-00683]\]. A NanoDrop spectrophotometer (ND-1000, Thermo Fisher Scientific, Waltham, MA, USA) was applied to assess DNA concentration and quality. All DNA samples were stored at −20 °C for later use. In order to determine the diversity of microbial communities, the hypervariable V4 region of the 16S rRNA gene was amplified using the primer pair 515F (5'-GTGCCAGCMGCCGCGG-3') and the reverse primer 907R (5'-CCGTCAATTCMTTTRAGTTT-3') \[[@B29-ijerph-16-00683]\]. Each PCR amplification, using a 3 × 50 μL reaction system, was set up under the following program: 95 °C for 5 min; 30 cycles of 30 s at 95 °C, 30 s at 58 °C, 30 s at 72 °C; and a final extension step of 10 min at 72 °C. The PCR products were purified using the Universal DNA Purification kit (Tiangen, Beijing, China) and then quantified using the NanoDrop spectrophotometer (ND-1000, Thermo Fisher Scientific, Waltham, MA, USA). Pre-mixed samples were sent for sequencing at Novogene (Beijing, China) on an Illumina MiSeq PE300 platform. 2.3. Phylotype Analysis {#sec2dot3-ijerph-16-00683} ----------------------- All Illumina sequencing data were processed using the Quantitative Insights into Microbial Ecology (QIIME 1.9.1) platform as previously described \[[@B30-ijerph-16-00683]\]. Sequences were sorted and assigned by matching the unique barcode for each sample exactly. Low-quality reads, such as primer mismatches, ambiguous bases, and errors in barcodes were filtered out. Chimeric and noisy sequences were also removed. Qualified sequence reads were clustered into operational taxonomic units (OTUs) at 97% identity \[[@B31-ijerph-16-00683]\]. The most abundant OTU was conducted using a Ribosomal Database Project (RDP) Classifier at an 80% confidence threshold \[[@B32-ijerph-16-00683]\]. Rarefaction analysis was applied to determine Chao1 diversity, phylogenetic diversity, and observed species metrics at different sequencing depth. The reads were deposited into the NCBI short-reads archive database (accession number: SRP158950). 2.4. ARGs Determination {#sec2dot4-ijerph-16-00683} ----------------------- The WaferGen SmartChip Real-Time PCR system (WaferGen Inc., Fremont, CA, USA) with 296 validated primer sets was applied ([Table S1](#app1-ijerph-16-00683){ref-type="app"}). To determine the relative abundance of ARGs (copies/16S rRNA gene) as previously described \[[@B33-ijerph-16-00683]\], we targeted 285 ARGs from all major classes of antibiotics. Except for the 285 ARGs, the smart-chip system also included eight transposases, one class 1 integron-integrase gene (intI1), one clinical class 1 integron-integrase gene (cintI1), and one 16S rRNA gene. qPCR reactions were performed according to a previously published study by An et al. \[[@B34-ijerph-16-00683]\] with a 100 nL reaction system. Melting curve analysis was performed after amplification. 2.5. Analysis of ARGs {#sec2dot5-ijerph-16-00683} --------------------- SmartChip qPCR software (V 2.7.0.1, WaferGen Inc., Mountain View, CA, USA) was applied to analyze HT-qPCR data. Multipeak reactions and reactions with amplification efficiency out of the acceptable range (0.9--1.1) were removed \[[@B33-ijerph-16-00683],[@B35-ijerph-16-00683],[@B36-ijerph-16-00683]\]. In this study, a cycle threshold (Ct) value \<31 was regarded as positive. Moreover, samples without three positive technical replicates were discarded. The relative abundance (copies/16S rRNA gene) of ARGs were calculated according to Looft et al. \[[@B37-ijerph-16-00683]\] (see Equation (1)). 2.6. Statistical and Network Analysis {#sec2dot6-ijerph-16-00683} ------------------------------------- Excel 2010 (Microsoft Office 2010, Microsoft, Redmond, WA, USA) was applied to determine the averages and standard deviations. One-way analysis of variation (ANOVA) and principal component analysis (PCA) were performed using SPSS version 18.0 (IBM, Armonk, NY, USA). For testing of statistically significant differences, a significance level of 0.05 was used. Heatmap plotting was performed in R 3.5.0 with pheatmap package. To investigate co-occurrence patterns of microbial communities and ARGs, correlation matrices were constructed by calculating all possible pairwise Spearman's rank correlations between ARG subtypes and bacterial genera in R environment using the "psych" package. A correlation between any two items was considered to be a robust correlation if the Spearman's correlation coefficient (ρ) was \>0.7 and the p value was \<0.01. Network visualization was conducted using the interactive platform Gephi 0.9.1 software (<https://gephi.org/>). 3. Results {#sec3-ijerph-16-00683} ========== 3.1. Occurrence and Abundance of ARGs {#sec3dot1-ijerph-16-00683} ------------------------------------- In total, 177 ARGs were detected in all samples. Hospital and farm locations were major sources of antibiotics, which contained 146 and 154 ARGs, respectively. However, in city and village samples, ARGs were also detected, numbering 103 and 85, respectively. These were significantly lower than in hospital and farm samples. The composition of ARG types in the four treatments were similar; multidrug, beta-lactam, aminoglycoside, tetracycline, and macrolide--lincosamide--streptogramin B resistance (MLSB) were the five most dominant resistance gene types, comprising 78.7% to 81.3% of the total detected ARGs ([Figure 1](#ijerph-16-00683-f001){ref-type="fig"}). The relative abundances of ARGs (copies/16S rRNA) in hospital, farm, city, and village locations were 8.76 × 10^−2^, 3.29 × 10^−2^, 3.64 × 10^−6^, and 3.29 × 10^−3^ copies/16S rRNA, respectively. The relative abundance of ARGs in city samples was the lowest, and significantly lower than in the other three locations ([Figure 2](#ijerph-16-00683-f002){ref-type="fig"}). The top 79 most abundant ARGs were extracted and are shown in [Figure 3](#ijerph-16-00683-f003){ref-type="fig"}. MLSB, beta-lactam, multidrug, integrase, and tetracycline resistance genes were the most abundant ARG types. All of these ARGs were detected in farm samples except for the vanxd gene, while ermb and CIntI were the two most abundant subtypes. Resistance gene bl3_cpha was the most abundant gene identified in hospital and village samples (6.03 × 10^−4^ and 8.79 × 10^−2^ copies/16S rRNA, respectively), while the number and relative abundance of ARGs was higher in hospital samples compared with village samples. The similarity of ARG compositions in all 12 samples was determined by PCA ([Figure 4](#ijerph-16-00683-f004){ref-type="fig"}). The ARG profiles from farms significantly separated along PC1, indicating that the farms had different resistome profiles. No distinct difference was observed between hospital, city, and village samples. Factor loadings of individual resistance genes on PC1 confirmed that ant3ia (score −0.639), qacEdelta1 (score −0.402), and Tn25 (score −0.338) were relatively abundant in farm samples. The most frequently prescribed antibiotics in hospitals and farms were beta-lactam and tetracycline, respectively \[[@B38-ijerph-16-00683]\]. Therefore, the beta-lactam and tetracycline resistance genes were analyzed in both hospital and farm samples. Six beta-lactam resistance genes (bl1_ampC, bl2_ges, bl2b_tem1, bl2d_oxa10, bl3_cpha, fox5) and eight tetracycline resistance genes (tetg, teth, tetk, tetl, tetm, teto, tetq, tetx) were detected in both hospital and farm samples. Compared to farm samples, hospitals had a higher level of beta-lactam resistance genes and a lower level of tetracycline resistance genes ([Figure 5](#ijerph-16-00683-f005){ref-type="fig"}a,b). The resistance gene bl3_cpha was the most abundant subtype in hospital samples with an abundance of 2.01 × 10^−4^ copies/16S rRNA, while in farm samples, the bl3_cpha gene abundance was only 5.08 × 10^−12^ copies/16S rRNA. The resistance genes tetq and tetg were the most abundant subtypes in farm samples with abundances of 1.14 × 10^−3^ and 1.36 × 10^−3^ copies/16S rRNA, respectively, while tetq and tetg gene abundances were only 1.84 × 10^−8^ and 2.49 × 10^−9^ copies/16S rRNA in hospital samples, respectively. Moreover, the relative abundance of multidrug and MLSB resistance genes were also higher in farm samples than in hospital samples ([Figure 5](#ijerph-16-00683-f005){ref-type="fig"}c,d). 3.2. Characterization of Bacterial Community {#sec3dot2-ijerph-16-00683} -------------------------------------------- In total, 661,301 high quality sequences (ranging from 27,303 to 88,263 sequences per sample, median = 53,302) were obtained from 12 samples, which were clustered into 25,311 OTUs at 97% similarity cutoff. There was no significant difference in bacterial diversity (chao1 index) among the four treatments ([Figure 6](#ijerph-16-00683-f006){ref-type="fig"}). Proteobacteria, Chloroflexi, Acidobacteria, Actinobacteria, Nitrospirae, and Cyanobacteria were the six dominant phyla, accounting for 83.8% to 86.2% of the total abundance within the 12 samples ([Figure 7](#ijerph-16-00683-f007){ref-type="fig"}). Proteobacteria were the most abundant phylum, and the relative abundances were 43.6% to 53.4% in the four treatments. Some pathogenic microorganisms, such as Staphylococcus, Enterobacteriaceae, Pseudomonas, and Corynebacterium were also detected. 3.3. Correlation of ARGs with Bacterial Community {#sec3dot3-ijerph-16-00683} ------------------------------------------------- Co-occurrence patterns between ARGs and bacterial taxa (at the genus level) were performed by network analysis. The correlation analysis showed strong (ρ \> 0.7) and significant (p \< 0.01) correlations between ARGs and bacterial assemblages. Proteobacteria, Actinobacteria, and Chloroflexi were the potential hosts of ARGs. Proteobacteria had non-random connections with beta-lactam, MLSB, aminoglycoside, multidrug, sulfonamide, and tetracycline resistance genes. Actinobacteria was a potential host of beta-lactam, aminoglycoside, multidrug, and tetracycline resistance genes. The multidrug ARGs were mainly associated with Proteobacteria and Actinobacteria ([Figure 8](#ijerph-16-00683-f008){ref-type="fig"}). 4. Discussion {#sec4-ijerph-16-00683} ============= In this study, we provide detailed information on ARG profiles in air-conditioning filters from hospitals, farms, and residences. Although the relative abundances of ARGs in AC filters were much lower (less than 1--5 orders of magnitude) than those of soil, sludge, and wastewater \[[@B18-ijerph-16-00683],[@B34-ijerph-16-00683],[@B39-ijerph-16-00683],[@B40-ijerph-16-00683]\], 10 resistance gene types---including 177 subtypes---were determined, including somemultidrug resistance genes. Environmental ARGs can be transferred to both pathogenic and nonpathogenic microorganisms through the lateral transfer of genes \[[@B41-ijerph-16-00683],[@B42-ijerph-16-00683]\], leading to drug resistance by inhalation of ARGs located within pathogens for susceptible populations \[[@B43-ijerph-16-00683]\]. These results suggest that the ARGs in AC units might be a possible public health threat through ventilation and respiration. Indoor air quality is of critical importance as humans now spend the majority of their time indoors. While current research in this field has mostly focused on the indoor microbiome, numerous studies have characterized the community composition, abundance, and specific sources of indoor microorganisms \[[@B44-ijerph-16-00683],[@B45-ijerph-16-00683]\]. However, less is understood concerning the ARGs in indoor environments. Previous studies have shown that outdoor air could contribute to the variation in composition or abundance of microbes in air circulating indoors \[[@B46-ijerph-16-00683]\]. The diversity and abundance of ARGs in the particulate matter of severe smog were determined \[[@B47-ijerph-16-00683]\]. Similar to our results, a total of 205 airborne ARG subtypes were detected. Among the detectable ARGs, tetracycline, beta-lactam, aminoglycoside, chloramphenicol, MLSB, and multidrug resistance genes exhibited the highest abundance. However, integrase resistance genes were also abundant in farm samples in our study. Hu and colleagues found the levels of ARGs to be similar among airborne particulate matter, soil, and activated sludge, and these were all higher than observed in our study \[[@B47-ijerph-16-00683]\]. Shen et al. \[[@B48-ijerph-16-00683]\] suggested that automobile exhaust emissions, dust, and industrial gas could provide more attachment points for airborne microorganisms in smog. Moreover, particulate matter tends to be suspended in air, rather than deposited onto land \[[@B48-ijerph-16-00683]\]. Our results showed that the ARG abundances in farm and hospital locations were significantly higher than those in city and village residences. However, PCA revealed that ARG profiles from the farm samples were different compared to those of the hospital samples. With the exception of beta-lactam, the relative abundance of resistance genes for tetracycline, multidrug, integrase, and MLSB were significantly higher in the farm samples than in the hospital samples. Many types of antibiotics (especially tetracycline) are used in livestock industries, and intensive animal farming is considered to be a primary contributor to the increased environmental burden of ARGs. High-capacity qPCR arrays were used to study the drivers and abundant ARGs in three Chinese swine farms. A total of 149 unique ARGs were detected in all samples, which potentially conferred resistance to aminoglycosides, beta-lactam, tetracyclines, and MLSB \[[@B49-ijerph-16-00683]\]. Hong et al. \[[@B50-ijerph-16-00683]\] demonstrated that tetracycline resistance genes (TetB, TetH, TetZ, TetO, TetQ, and TetW) existed in high abundance, ranging from 9.55 × 10^2^ to 1.69 × 10^6^ copies per ng DNA extracted from the bioaerosols sampled from both pig and poultry farms. The relative abundances of the integron-integrase genes (CIntI and IntI) were considerably high in one farm sample in our study ([Figure 3](#ijerph-16-00683-f003){ref-type="fig"}). IntI is frequently linked to ARGs and has penetrated into diverse pathogenic bacteria. Moreover, IntI is easily transmissible between diverse commensal and pathogenic bacteria associated with humans and animals, and could increase the risk of horizontal gene transfer \[[@B51-ijerph-16-00683]\]. Continued selective pressure by different antibiotics has resulted in multidrug resistance organisms and multidrug resistance genes \[[@B52-ijerph-16-00683]\]. A high abundance of multidrug resistance genes in farm samples were found in this study, which suggested that the microorganisms of farms could be resistant to multiple antibiotics. Beta-lactam is the most widely used antibiotic (occupies 65% of the total injectable antibiotics in hospital) due to its low toxicity and wide applicability \[[@B53-ijerph-16-00683],[@B54-ijerph-16-00683]\]. Therefore, numerous studies have demonstrated that beta-lactam ARGs and beta-lactam-resistant bacteria were detected in sputum, urine, blood, wastewater, and air samples obtained from hospitals \[[@B55-ijerph-16-00683],[@B56-ijerph-16-00683]\]. Rowe et al. \[[@B57-ijerph-16-00683]\] compared the abundances of beta-lactam ARGs in hospital and farm effluents. They found that two beta-lactam resistance genes were overexpressed in hospital effluent samples, but this overexpression was not observed in farm effluents. The high abundances of ARGs in effluents were mainly related to the heavy use of beta-lactam in hospitals. Gao et al. \[[@B58-ijerph-16-00683]\] characterized the microbial community in five nosocomial bioaerosols. Similar to our results, they found that Proteobacteria represented the most prevalent microbes at almost 45% representation. Staphylococcus saprophyticus, Corynebacterium minutissimum, Streptococcus pneumoniae, Escherichia coli, Arcobacter butzleri, Aeromonas veronii, Pseudomonas aeruginosa, and Bacillus cereus were the most abundant pathogenic microorganisms. The non-random co-occurrence patterns between ARGs and the microbial community could indicate potential host information relating to ARGs, since specific microbial taxa appear to carry specific ARGs \[[@B8-ijerph-16-00683]\]. Therefore, network analysis is widely used to provide us with new insights into the ARGs and their role as potential hosts in complex environmental conditions \[[@B59-ijerph-16-00683]\]. Previous studies have identified some ARG hosts, such as Lactobacillus, which was considered to be a host to a number of ARGs, including gentamicin, tetracycline, and kanamycin \[[@B47-ijerph-16-00683]\]. Proteobacteria, Cyanobacteria, Bacteroidetes, and Verrucomicrobia were regarded as potential hosts of tetracycline resistance genes in coastal aquaculture \[[@B60-ijerph-16-00683]\]. In the present study, network analysis revealed that Proteobacteria and Actinobacteria were the possible hosts of resistance genes for beta-lactam, aminoglycoside, multidrug, and tetracycline. In our study, Proteobacteria and Actinobacteria were the first and fourth most abundant phyla, and this cannot be ruled out as a reason for the close association between the two phyla and ARGs. 5. Conclusions {#sec5-ijerph-16-00683} ============== The ARG profiles in air-conditioning filters from different environments were comprehensively characterized using HT-qPCR and Illumina sequencing. The detectable number and the relative abundances of ARGs in hospital and farm samples were significantly higher than in city and village samples, indicating that hospitals and farms were major sources of ARGs. With the exception of beta-lactam resistance genes, the relative abundance of resistance genes for tetracycline, multidrug, integrase, and MLSB were significantly higher in farm samples than in hospital samples. Our results suggested that the presence of ARGs was primarily linked to the heavy usage of antibiotics. Network analysis revealed that Proteobacteria and Actinobacteria were possible hosts of beta-lactam, MLSB, aminoglycoside, multidrug, sulfonamide, and tetracycline resistance genes. This study may help to better understand the general distribution patterns of ARGs in indoor environments, and more efforts should be devoted to exploring the relationship between ARGs and pathogenic microorganisms. The following is available online at <https://www.mdpi.com/1660-4601/16/5/683/s1>. Table S1: Primer list of antibiotic resistance genes in this study. ###### Click here for additional data file. Conceptualization, Y.L. and H.Y.; methodology, Y.L.; software, Y.L. and H.L.; validation, Y.L., H.Y. and H.L.; formal analysis, Y.L.; investigation, Y.L.; resources, Y.L.; data curation, H.L.; writing---original draft preparation, Y.L.; writing---review and editing, Y.L.; visualization, Y.L.; supervision, H.Y.; project administration, Y.L.; funding acquisition, Y.L. and H.Y. All authors interpreted findings from the analysis and drafted the manuscript. This research was funded by the National Key R & D Program of China (2016YFC0502704), the National Natural Science Foundation of China (41525002), and the Knowledge Innovation Program of the Chinese Academy of Sciences (IUEMS201403). The authors declare no conflict of interest. ![The number of detected antibiotic resistance gene (ARG) types in air-conditioning filters. H: hospital; F: farm; C: city residence; V: village residence. MLSB: macrolide--lincosamide--streptogramin B resistance. Error bars indicate standard errors of three replicates, different lowercase letters indicate a significant difference (p \< 0.05).](ijerph-16-00683-g001){#ijerph-16-00683-f001} ![The relative copy number (copies/16S rRNA, relative abundance) of ARGs. H: hospital; F: farm; C: city residence; V: village residence. Error bars indicate standard errors of three replicates, different lowercase letters indicate a significant difference (p \< 0.05).](ijerph-16-00683-g002){#ijerph-16-00683-f002} ![The heat map of detected ARGs. H: hospital; F: farm; C: city residence; V: village residence. MLSB: macrolide--lincosamide--streptogramin B resistance.](ijerph-16-00683-g003){#ijerph-16-00683-f003} ![The ARG composition by principle component analysis (PCA). H: hospital; F: farm; C: city residence; V: village residence.](ijerph-16-00683-g004){#ijerph-16-00683-f004} ![The relative abundance of (a) beta-lactam resistance genes, (b) tetracycline resistance genes, (c) multidrug resistance genes, and (d) MLSB resistance genes. H: hospital; F: farm. MLSB: macrolide--lincosamide--streptogramin B resistance. Error bars indicate standard errors of three replicates, different lowercase letters indicate a significant difference (p \< 0.05).](ijerph-16-00683-g005){#ijerph-16-00683-f005} ![Rarefied Chao 1 index. H: hospital; F: farm; C: city residence; V: village residence. Error bars indicate standard errors of three replicates, different lowercase letters indicate a significant difference (p \< 0.05).](ijerph-16-00683-g006){#ijerph-16-00683-f006} ![The abundance of dominant bacteria at the phylum level. H: hospital; F: farm; C: city residence; V: village residence. Error bars indicate standard errors of three replicates, different lowercase letters indicate a significant difference (p \< 0.05).](ijerph-16-00683-g007){#ijerph-16-00683-f007} ![The co-occurrence patterns among ARGs and microbial taxa. The nodes were colored according to ARG types and genus. The size of each node is proportional to the number of connections, and the thickness of each connection between two nodes is proportional to the value of Spearman's rank correlation coefficients ρ \> 0.7. MLSB: macrolide--lincosamide--streptogramin B resistance.](ijerph-16-00683-g008){#ijerph-16-00683-f008}	{ "pile_set_name": "PubMed Central" }
Introduction ============ The major common endpoint in most human disease is thrombosis and the major causes of arterial thrombosis (such as diabetes) have been established for decades. The risk factors for venous thromboembolism (VTE) (such as obesity and orthopedic surgery) are also now becoming well established and the penetrance of these factors in the pathophysiology of thrombosis is so strong that anticoagulation is often mandatory. Indeed, VTE is a frequent complication among hospital in-patients and contributes to longer hospital stays, morbidity, and mortality.[@b1-vhrm-5-693]--[@b4-vhrm-5-693] Some VTEs may be subclinical and patients presenting with sudden pulmonary embolus (PE) or deep vein thrombosis (DVT) are common in acute medicine. Using ultrasonic Doppler and venographic techniques, DVT of the lower limb has been documented in half of all major lower limb orthopedic operations performed without antithrombotic prophylaxis, a quarter of patients with acute myocardial infarction (MI) and more than half of patients with acute ischemic stroke. Indeed, some risk factors are more likely to evoke a VTE than others ([Table 1](#t1-vhrm-5-693){ref-type="table"}). Accordingly, anticoagulation is demanded. However, the perfect anticoagulant (rapid mode of action and cessation, oral availability, minimal side effects, etc) does not exist. Instead, we have a crude poison with a long time to action and long half-life with a wide range of effect in different patients leading to difficulties in management (that is, warfarin), and an injectable agent with poor pharmacokinetics such that regular blood testing is required, and a risk of thrombocytopenia (that is, heparin).[@b5-vhrm-5-693] More recently, some of the disadvantages of heparin have been reduced with the introduction of an improved preparation: low-molecular-weight heparin (LMWH).[@b6-vhrm-5-693],[@b7-vhrm-5-693] However, there are several different species of LMWH available for the prevention of VTE, and all have slightly different properties and licenses for different risk situations. Fortunately, new anticoagulants are coming on to the market and it is conceivable that they will slowly take over from both warfarin and the heparins over the coming decade. The objective of this communication is to review the role of the different LMWHs in the prevention of VTE. In order to do this the position of (unfractionated) heparin will be first be reviewed, following by an in-depth examination of LMWH. It will conclude with an examination of new anticoagulants. Heparin ======= The modern era of heparin can be traced to 1880 with the description of anticoagulant preparations that evolved into heparin[@b8-vhrm-5-693]--[@b11-vhrm-5-693] that ultimately became clinically useful.[@b12-vhrm-5-693]--[@b14-vhrm-5-693] Brinkhous and colleagues[@b15-vhrm-5-693] subsequently demonstrated the requirement of heparin for another substance in order for anticoagulation to be effected, a factor that we now recognize as antithrombin (formerly antithrombin III). Commercial heparin was thus ready for industrial/pharmacological scale production. By 1979, its value in prophylaxis for venous thrombosis, PE, and other possible problems were established.[@b5-vhrm-5-693] The laboratory potential of a more efficacious fraction of whole, unfractionated heparin, ie, LMWH, was recognized in 1979[@b16-vhrm-5-693],[@b17-vhrm-5-693] and clinical note was made shortly afterwards.[@b18-vhrm-5-693],[@b19-vhrm-5-693] Mode of action -------------- Heparin is a natural product present on the cell membrane, but when provided as a therapeutic drug, it is a mixture of glycosaminoglycans and polysaccharides composed of long chains of repeated disaccharide units (hexosamine and glucuronic or iduronic acid), although the composition of different macromolecules vary markedly. In its whole, unfractionated form, with species of molecular weights ranging from 3 to 30 kDa (although most is in the range 12--15 kDa), perhaps only one-third of a standard heparin preparation has anticoagulant activity.[@b5-vhrm-5-693] Heparin by itself is not an anticoagulant: it is a co-factor in the activity of antithrombin, a 58 kDa single chain polypeptide synthesized in the liver.[@b20-vhrm-5-693] Heparin binding to specific sites on antithrombin induces a conformational change in the latter, exposing a site that binds serine proteases factor Xa, thrombin (IIa) (in approximately equal proportions) and, with less affinity, factors IXa, XIa, XIIa, kallikrein, plasmin, and C~1~-esterase, although almost all of its effects are against thrombin and factor Xa.[@b5-vhrm-5-693] However, heparin fractions less than 18 monosaccharide units (3--4 kDa) do not enhance the inhibition of thrombin by antithrombin and probably act via factor Xa inhibition. The major inhibitor of coagulation, the effects of antithrombin are accelerated some 1000-fold in the presence of heparin. It follows, therefore, that special measures are required in subjects deficient in antithrombin (whether by genetics or excess consumption). In high doses, heparin also prolongs the bleeding time by inhibiting platelet aggregation in vitro[@b21-vhrm-5-693] and may exert some effect on the endothelium,[@b22-vhrm-5-693] although it is unclear (and possibly academic) whether or not these influence clinical efficacy. Clinical effectiveness ---------------------- Due to variability of effect in different individuals, mostly due to complex pharmacokinetics (such as degree of binding to numerous proteins including, platelet factor 4, vitronectin, fibronectin and von Willebrand factor),[@b23-vhrm-5-693],[@b24-vhrm-5-693] laboratory monitoring of the effects of heparin is necessary and can be followed with the relatively straightforward activated partial thromboplastin time (APTT), thrombin time, and to lesser extent, the prothrombin time. The APTT derives from the ability of citrated plasma to clot an artificial 'platelet' substitute of phospholipids and other substances to which calcium has to be added: quality control is crucial as the quality of reagents can vary markedly.[@b25-vhrm-5-693] In practice, most clinicians aim for a degree of anticoagulation that prolongs the APTT by anything from 1.5 to 3 times that of normal, uncoagulated blood, ie, an APTT patient/control ratio of 1.5:3.0.[@b4-vhrm-5-693],[@b26-vhrm-5-693] The primary use of heparin has been, and continues to be, in the treatment and prophylaxis of VTE.[@b5-vhrm-5-693],[@b26-vhrm-5-693],[@b27-vhrm-5-693] As with aspirin, early crucial work was performed in the 1960's with studies looking at the effect of heparin on PE and DVT.[@b28-vhrm-5-693],[@b29-vhrm-5-693] However, heparin is not absorbed from the gastrointestinal tract and so must be given intravenously or subcutaneously, with the former providing a more rapid effect. Subsequent trials not only confirmed the efficacy of this agent, but also showed that continuous intravenous infusion was frequently preferable to subcutaneous or intermittent infusions.[@b30-vhrm-5-693],[@b31-vhrm-5-693] One 'typical' trial reported a frequency of asymptomatic extension of DVT or PE of 8.2% of patients with proximal DVT taking heparin and oral anticoagulation compared to a frequency of 39.6% in those taking oral anticoagulants alone.[@b32-vhrm-5-693] There was also a reduction in the report of symptomatic events (6.7% versus 20%), but no difference in hemorrhagic complication (3% versus 5%). In practice, until recently, hospital treatment of, and prophylaxis against, VTE, generally began with heparin, moving oral anticoagulants following discharge, for varying periods up to six months.[@b33-vhrm-5-693] Clinical side effects --------------------- As with aspirin, the major problem with heparin is in dose-dependent excess bleeding. Early commentaries reported a rate of hemorrhagic complications of up to 30%.[@b34-vhrm-5-693] However, a later review reported an incidence of 6.8% in patients given continuous infusion and 14.2% in patients given intermittent intravenous injections,[@b5-vhrm-5-693] emphasizing four important variables: dose, patient's anticoagulant response, method of administration, and other patient-related factors such as renal failure or chronic alcohol use. Gallus and colleagues[@b35-vhrm-5-693] reported an incidence of major bleeding of 11% in high-risk patients, compared to 1% in low-risk patients. Surprisingly, heparin-induced thrombocytopenia (HIT; generally platelet count \<140,000 cells/μL, and reversible) has been described for 50 years,[@b36-vhrm-5-693] with two forms (nonimmune and immune \[often an immunoglobulin G \[IgG\] targeting a heparin/platelet factor 4 complex\]) being recognized.[@b37-vhrm-5-693] When the platelet count falls, especially if it falls from normal to less than \<100,000 cells/μL, cessation is mandatory and alternate anticoagulant cover (eg, argatroban)[@b26-vhrm-5-693],[@b38-vhrm-5-693] is necessary. Curiously, 0.4% of patients with HIT suffer arterial thrombosis (that may follow platelet aggregation in vivo) or venous thrombosis (that may result from heparin resistance caused by the neutralizing effect of the heparin-induced release of platelet factor 4). These events may be related to the generation of thrombin,[@b39-vhrm-5-693] and also to platelet activation by heparin.[@b21-vhrm-5-693],[@b40-vhrm-5-693] If present, HIT is likely to be apparent 3 to 15 days after initiation (such that a platelet count is advised on days 3--5), but may also occur very rarely within hours, with the platelet count returning to normal within four days of discontinuation.[@b41-vhrm-5-693] Rates of HIT vary between \<0.1% and 3%.[@b26-vhrm-5-693],[@b37-vhrm-5-693],[@b42-vhrm-5-693] Other uncommon side effects of heparin (especially in long-term and high-dose use) include fractures and osteopenia, hyperkalemia, elevations in liver enzymes, skin necrosis at the site of administration, alopecia, hypersensitivity, priapism, and hypoaldosteronism.[@b5-vhrm-5-693],[@b26-vhrm-5-693],[@b43-vhrm-5-693]--[@b48-vhrm-5-693] Thus, despite the proven safety and effectiveness of continuous infusions of heparin, its limitations and adverse side effect profile, unpredictable pharmacokinetic response, daily laboratory monitoring with dose adjustments, and requirement for hospitalization leaves room for alternatives. One such alternative is a low-molecular-weight variant of the standard (unfractionated) heparin preparation. Low-molecular-weight heparin ============================ The laboratory demonstration of a more efficacious fraction of whole, unfractionated heparin, ie, LMWH,[@b16-vhrm-5-693]--[@b19-vhrm-5-693] has ushered in a new era of anticoagulation. LMWH can be manufactured from unfractionated heparin by diverse routes, such as benzylation with alkaline hydrolysis, nitrous acid depolymerization or digestion, heparinase digestion, and isoamyl nitrate digestion. Each provides a different LMWH with mean molecular weights varying from 4371 Da to 5866 Da, so that each has different pharmacokinetics and activities.[@b49-vhrm-5-693]--[@b51-vhrm-5-693] However, there are common class advantages over unfractionated heparin ([Table 2](#t2-vhrm-5-693){ref-type="table"}), most of which address mode of action and side effects. Broadly speaking, compared to unfractionated heparin, LMWH has a longer subcutaneous half-life and therefore has potential for outpatient use, a more predictable anticoagulant response requiring less monitoring, and has better antifactor Xa effect. Building on early trials,[@b18-vhrm-5-693],[@b19-vhrm-5-693] several large studies have unequivocally demonstrated the value of LMWH in the initial treatment of DVT.[@b52-vhrm-5-693]--[@b54-vhrm-5-693] Lensing and colleagues[@b52-vhrm-5-693] described a recurrence rate of 3.1% in patients on LWMH compared to 6.6% (p \< 0.01) in those on standard heparin, with 0.8% and 2.8% (p \< 0.005) suffering a major bleed, respectively, which was confirmed by other analyses. However, later trials[@b55-vhrm-5-693],[@b56-vhrm-5-693] failed to find a difference in efficacy between the two types of heparin with equivalence of adverse events. Interestingly, these meta-analyses also reported a lower mortality with LWMH due to causes unrelated to VTE in the patients particularly with cancer. A subsequent more formal meta-analysis of 629 cancer patients did indeed demonstrate a benefit in three-month mortality for LMWH versus unfractionated heparin.[@b57-vhrm-5-693] While this is an unsought for endpoint and therefore of questionable value, there is nevertheless good mechanistic evidence for this benefit.[@b58-vhrm-5-693] The value of LMWH compared to unfractionated heparin now extends to prevention of thrombosis in general surgery, orthopedic surgery, hip fracture, multiple trauma, and neurosurgery,[@b51-vhrm-5-693] whilst efficacy and safety issues in medical patients have been addressed.[@b59-vhrm-5-693] Apart from its clinical effectiveness and better side effect profile, a further advantage of LMWH is in home use by appropriate patients with DVT who are not severely ill.[@b60-vhrm-5-693],[@b61-vhrm-5-693] In both studies, LMWH administered subcutaneously twice-daily was compared with standard continuous infusion in hospital. Both reported no significant difference in recurrence of thromboembolism with similar rates of bleeding. Those on LMWH in the study of Levine and colleagues[@b60-vhrm-5-693] remained in hospital for a mean of 1.1 days, suffering 13 embolic events, compared to 6.5 days for those on unfractionated heparin, with 17 events, of which two were fatal. The health economics of these figures is apparent, even after adjustment for the higher costs of the LMWH. The study of Koopman and colleagues[@b61-vhrm-5-693] included quality of life data: overall, there was no difference between the LMWH and unfractionated heparin groups, but those on LMWH reported better physical activity and social functioning at weeks 1--2. Data demonstrating the superiority of LMWH compared to unfractionated heparin (eg, shorter time to effective anticoagulation and more days of effective anticoagulation) continues to be published.[@b62-vhrm-5-693] Whilst the weight of literature focuses on (unprovoked) DVT and PE, LMWH is also effective in those with conditions predisposing to VTE, such as cancer,[@b63-vhrm-5-693]--[@b65-vhrm-5-693] although more data on other groups, such as the obese and those in renal failure or with thrombophilia[@b66-vhrm-5-693]--[@b69-vhrm-5-693] is needed and these are slowly becoming available. However, despite the weight of literature on LMWH, questions and doubts remain. For example, Knight and colleagues conclude that heparins may, in some circumstances, be pro-thrombotic.[@b70-vhrm-5-693] Thus there remains scope for more efficacious, yet safe, anticoagulants.[@b71-vhrm-5-693] Which LMWHs are available? ========================== In the United Kingdom, pharmaceutical practice in NHS hospitals is dominated by the British National Formulary (BNF).[@b72-vhrm-5-693] This book, also available online, is updated each six months, and lists all government-approved pharmaceutical agents and their licensed indications. It follows that doses and guidelines may differ in other countries, and that doses in prefilled syringes may also differ. Four types of LMWH are described In issue 57 of the BNF. These are (in alphabetic order) bemiparin, dalteparin, enoxaparin, and tinzaparin. All are given subcutaneously and regimes are markedly different for prophylaxis of VTE versus treatment of VTE. Furthermore, all have numerous cautions and contraindications. Principal among these is the need to refer to renal function because several agents are cleared by this organ so that defective clearance effectively leads to higher biologically active levels and thus a risk of overanticoagulation, which may lead to hemorrhage. An additional concern is the liver, as failure of this organ to synthesize key coagulation proteins may also lead to a risk of hemorrhage. Other LMWHs are available and include nadroparin and certoparin. Bemiparin --------- This agent is licensed for the prophylaxis of DVT. For patients at moderate risk (for the definition of risk, see [Tables 4](#t4-vhrm-5-693){ref-type="table"} and [5](#t5-vhrm-5-693){ref-type="table"} and the discussion to come), the recommended dose is 2500 units two hours before or six hours after surgery, then 2500 units every 24 hours for 7--10 days. High-risk patients should receive 3500 units two hours before to six hours after surgery, then 3500 units every 24 hours for 7--10 days. Treatment of DVT (with or without PE) is likely to be 115 units/kg every 24 hours for 5--9 days and until adequate oral anticoagulation (inevitably warfarin) is established. Dalteparin ---------- This agent is licensed for additional indications. For prophylaxis of DVT in surgical patients at moderate risk, 2500 units 1--2 hours are given before surgery, then 2500 units every 24 hours for 5--7 days or longer. High-risk patients are also likely to require 2500 units 1--2 hours before surgery, but then 2500 units 8--12 hours later (or 5000 units on the evening before surgery, then 5000 units on the following evening), with 5000 units every 24 hours for 5--7 days or longer (five weeks in hip replacement). For prophylaxis of DVT in medical patients, 5000 units are given every 24 hours. Treatment of DVT and PE in adults is dose-adjusted to body weight less than 46 kg, 7500 units/day, 46--56 kg 10,000 units/day, 57--68 kg, 12,500 units/day, 69--82 kg, 15,000 units/day, and body weight over 83 kg 18,000 units/day, with oral anticoagulant until the international normalized ratio (INR) is in the therapeutic range (which may take up to seven days). Monitoring of the effect of the LMWH is not usually needed but if so is by antifactor Xa assay. There is an unlicensed indication for dalteparin for the treatment of VTE in pregnancy. This calls for dose-adjusted injections based on early pregnancy body weight: under 50 kg, 5000 units twice daily; 50--70 kg, 6000 units twice daily; 70--90 kg 8000 units twice daily, and body weight over 90 kg, 10,000 units twice daily. If called for, blood should be taken for monitoring 3--4 hours after a dose, and ideally the result will be in the range or 0.5 to 1 unit/mL of antifactor Xa activity. Monitoring is generally not required for once-daily treatments. Notably, this desired range is higher than that for prophylaxis of a VTE, which is 0.1--0.3 antifactor Xa units/mL. Daltaparin is also licensed for use in acute coronary syndromes, which are effectively unstable coronary artery disease of ST and non-ST elevation MI. The weight-adjusted dose is 120 units/kg every 12 hours (up to 10,000 units twice daily) for up to eight days. Beyond this period, such as if awaiting invasive procedures, then the dose may change. Dose also varies by sex. For women weighing less than 80 kg and men less than 70 kg, the recommended dose is 5000 units every 12 hours. Above this weight the dose is 7500 units every 12 hours. This proceeds until the day of the procedure for a maximum of 45 days. Enoxparin --------- This agent also has multiple indications. For prophylaxis of DVT in surgical patients at moderate risk, the recommended dose is 20 mg (2000 units) about two hours before surgery then 20 mg (2000 units) every 24 hours. High-risk patients (such as those undergoing orthopedic surgery) will need 40 mg (4000 units) 12 hours before surgery and repeated every 24 hours for 7--10 days. Medical patients have the same regime but for at least six days and until ambulant to a maximum of 14 days. Treatment of either forms of VTE calls for 1.5 mg/kg (150 units/kg) every 24 hours, usually for at least five days and until oral anticoagulation is effective (generally INR 2--3). In unstable angina and non-ST-segment elevation MI, 1 mg/kg (100 units/kg) every 12 hours usually for 2--8 days (minimum two days) is recommended, and there is also an unlicensed indication for the treatment of VTE in pregnancy: early pregnancy body weight under 50 kg, 40 mg (4000 units) twice daily; body weight 50--70 kg, 60 mg (6000 units) twice daily; body weight 70--90 kg, 80 mg (8000 units) twice daily; body weight over 90 kg, 100 mg (10 000 units) twice daily. Tinzaparin ---------- For the prophylaxis of DVT in general surgery, the recommended dose is 3500 units two hours before surgery, then 3500 units every 24 hours for 7--10 days. In orthopedic surgery, it may be dose-adjusted: 50 units/kg two hours before surgery, then 50 units/kg every 24 hours for 7--10 days or 4500 units 12 hours before surgery, then 4500 units every 24 hours for 7--10 days. The treatment of DVT and PE calls for 175 units/kg once daily for at least six days and until adequate oral anticoagulation, eg, with warfarin, is established. Once more, there is an unlicensed indication of VTE in pregnancy, which is 175 units/kg once daily. Choice of LMWH ============== The previous section listed the different clinical scenarios that each LMWH may be used for. Leaving aside economic considerations, the practical level issues of ease of prescribing, dose adjustment, and timing of dose seem relevant. The practitioner must also be fully aware of cautions, contraindications, and other caveats of each particular LMWH which may be relevant in a particular patient. Irritatingly, there are few (if any) good, well powered trials comparing one LMWH with another. In almost all such trials the (generally inferior) comparator is unfractionated heparin.[@b52-vhrm-5-693]--[@b59-vhrm-5-693] All other factors being equal, due to pressures on individual hospital pharmacies, many opt for a LMWH that has the broadest range of clinical applications ([Table 3](#t3-vhrm-5-693){ref-type="table"}). Use of LMWHs in practice ======================== Fortunately, there are numerous guidelines for the use of LMWHs for prophylaxis in various settings such as in obstetrics and gynecology, and after surgery, and many are available free and online.[@b73-vhrm-5-693]--[@b79-vhrm-5-693] Practice in surgery in the UK is dominated by guidelines from the National Institute for Health and Clinical Excellent (NICE).[@b79-vhrm-5-693] However, other comprehensive guidelines with more international appeal such as those of the American College of Chest Physicians[@b73-vhrm-5-693] are widely used. Unsurprisingly, there is often considerable agreement between these guidelines. The NICE document recommends that all patients about to undergo surgery should be assessed to identify their risk factors for developing VTE, which can be assessed by a scoring system. Once a particular patient's risk of VTE has been assessed ([Table 4](#t4-vhrm-5-693){ref-type="table"}), then the risk afforded by their surgery must also be assessed ([Table 5](#t5-vhrm-5-693){ref-type="table"}). Clearly, some operations are more likely to promote VTE than other, and the most risky include orthopedic surgery. Once any cautions and contraindications have been addressed (such as allergy, HIT, hepatic or renal impairment), then treatment may be given ([Table 6](#t6-vhrm-5-693){ref-type="table"}). The duration of treatment varies, bearing in mind that some patients may transfer to warfarin, especially if they have concurrent long-standing risk factors such as cancer. Notably, NICE recommends treatment for four weeks after certain orthopedic procedures.[@b79-vhrm-5-693] The section above applies to the prevention of a VTE. However, once a patient actually has a DVT or PE, then a different dosing regime is called for, which is effectively more anticoagulants ([Table 7](#t7-vhrm-5-693){ref-type="table"}). Fondaparinux ============ Technically, this agent may not be a 'heparin' or even a 'LMWH', but it certainly does inhibit the coagulation pathway in the same manner. Marketed as Arixtra^®^ by GlaxoSmithKline, fondaparinux is a novel, selective and reversible Xa-inhibitor, although based on the structure of heparin, is different from both heparin and LMWH. Accordingly, unlike heparin and LMWH, it has minimal, if any, activity against thrombin.[@b80-vhrm-5-693],[@b81-vhrm-5-693] Pharmacokinetics are characterized by a 100% bioavailability by subcutaneous route, lack of biometabolism, urinary excretion and a relatively long plasma half-life of 14--21 hours which permits once-daily injection. There is a rapid onset of action, with a peak activity reached in two hours. No interactions with aspirin, warfarin, or digitoxin have been noted. Fondaparinux, like LMWHs, does not affect the prothrombin time (PT) and has very weak effects on APPT but its activity can be determined by specific anti-Xa assays, if necessary. Thrombocytopenia (platelet count \<100 × 10^9^/L) occurs even less commonly than with LMWH. Use of fondaparinux has been approved by NICE as an alternative to LMWH in certain types of surgery, such as orthopedic.[@b79-vhrm-5-693] In this setting the prophylaxis regime recommended by the BNF[@b72-vhrm-5-693] calls for 2.5 mg six hours before surgery then 2.5 mg daily for 5--9 days (longer after hip surgery), and the same dose for thromboprophylaxis in medical patients usually for 6--19 days. For treatment of VTE, the recommended weight-adjusted dose is 5 mg (if weight under 50 kg), 7.5 mg (50--100 kg), or 10 mg (over 100 kg) every 24 hours and generally for five days and until oral anticoagulation is adequate. New anticoagulants ================== Although LMWHs are very effective agents, they still carry some problems such as a residual risk of heparin-induced thrombocytopenia and reliance of parenteral introduction. These, and other problems, are being overcome with the introduction of new agents. Some operate directly against thrombin, others target coagulation factor Xa.[@b82-vhrm-5-693]--[@b86-vhrm-5-693] Direct thrombin inhibitors -------------------------- Dabigatran etexilate (marketed as Pradaxa^®^ by Boehringer Ingelheim GmbH) is the first in a new class of anticoagulant that works by directly inhibiting thrombin. It possesses various qualities which make it potentially an attractive and promising novel oral anticoagulant with its predictable pharmacokinetics and pharmacodynamics.[@b87-vhrm-5-693] The drug has rapid absorption (within two hours) and distribution with estimated half-lives of 8--10 hours and 14--17 hours for single and multiple dose administrations, respectively. Nearly 80% of dabigatran etexilate is excreted unchanged by the kidneys with average bioavailability of 6.5%, hence high doses are required to maintain adequate plasma concentrations. Of note, the drug absorption is reduced by 20%--25% if patients are concurrently on proton pump inhibitors. Phase II and III clinical trials have been conducted assessing dosages, efficacy, and tolerability in numerous indications such as prevention of secondary VTE, DVT, stroke, and embolism due to atrial fibrillation (AF).[@b88-vhrm-5-693]--[@b93-vhrm-5-693] In the BISTRO I study,[@b88-vhrm-5-693] dabigatran etexilate demonstrated an acceptable safety therapeutic profile with doses ranging between 12.5 mg and 300 mg twice daily in patients undergoing elective total hip replacement. Historically, the latter is one type of surgery where more than 50% of patients develop VTE in the absence of thromboprophylaxis. Subsequently, the BISTRO II study[@b89-vhrm-5-693] compared four different doses of dabigatran etexilate with enoxaparin, with a significant reduction of VTE events in total hip or knee replacement patients receiving 150 mg twice daily, 300 mg once daily, and 225 mg twice daily. There is also data on stroke prevention in AF. The PETRO study[@b90-vhrm-5-693] was conducted in 502 participants randomized to receive different dozes of dabigatran alone or in combination with aspirin (81 mg or 325 mg) or open-label warfarin. This clinical study found no serious liver toxicity with only a small fraction of patients (0.9%) observed to have raised aminotransferase levels and also justified dose selection for subsequent trials. In March 2008, the regulatory authorities of the European Union approved the use of dabigatran etexilate for DVT prevention postelective total hip or knee replacement surgery. This was followed by approvals from Health Canada in June 2008. Dabigatran etexilate has been available in UK since April 2008 and was subsequently included in the DVT prevention guidance issued by NICE five months later.[@b94-vhrm-5-693],[@b95-vhrm-5-693] The BNF recommended daily doses for VTE prevention in total knee replacement surgery is 110 mg (elderly over 75 years, 75 mg) 1--4 hours after surgery, then 220 mg daily (elderly over 75, 150 mg) for nine days. For total hip replacement surgery, the same dose is recommended but it should continue for 27--34 days. As is common practice with almost all anticoagulants, patients with moderate renal impairment (glomerular filtration rate 30--50 mL/min) require reduced dosages. Factor Xa inhibitors -------------------- These agents reversibly block free factor Xa and that factor Xa that is bound to platelets within the prothrombinase complex. Two oral factor Xa inhibitors are currently under various stages of clinical development and are close to potential clinical application. Rivaroxaban (marketed as Xarelto^®^ by Bayer Schering Pharma AG) is the first selective oral direct factor Xa inhibitor advanced to phase III clinical trials. It has favorable pharmacokinetic characteristics with a bioavailability of 60%--80%. Rivaroxaban achieves peak plasma levels in three hours and has a half-life of nine hours in healthy, young subjects and about 12 hours in elderly subjects. However, the drug has been tested for clinical use only on a once-daily basis. Rivaroxaban is metabolized by the liver via CYP3A4 with up to two-thirds of the drug being eliminated by the kidneys. The use of rivaroxaban in patients with renal impairment has to be cautious because of its renal clearance. It does not significantly interact with platelet function in preclinical studies, where it demonstrated an excellent correlation between its plasma levels and achieved clotting times, while the bleeding risk was comparable to that of enoxaparin.[@b96-vhrm-5-693],[@b97-vhrm-5-693] The phase III clinical trials, the RECORD Studies, were designed to look at the effectiveness of rivaroxaban in the prevention of VTE in patients undergoing hip or knee surgery. The RECORD I and RECORD III trials revealed higher effectiveness of rivaroxaban compared to enoxaparin in VTE prevention following orthopedic surgery surgery, whilst the RECORD II study provided evidence that prolonged administration of rivaroxaban had additional clinical benefits compared to short-term therapy.[@b98-vhrm-5-693]--[@b100-vhrm-5-693] The utility of rivaroxaban for VTE treatment was assessed in two phase II studies which showed that rivaroxiban had similar efficacy when compared with enoxaparin/warfarin in the treatment of VTE. In the ODIXa-DVT study, the incidence of major and minor bleeding increased with escalating doses of rivaroxiban (range of 5.0% to 11.6%) as compared with patients treated with enoxaparin/warfarin (6.3%).[@b101-vhrm-5-693] In the EINSTEIN-DVT study, the incidence of clinically relevant bleeding occurred in 2.2%--6.0% of patients in rivaroxaban groups and in 8.8% of patients in heparin/warfarin group, with no significant dose response for bleeding.[@b102-vhrm-5-693] In these two studies, no dose-efficacy relationship with rivaroxaban treatment was observed. However, the dose-dependent increase in the risk of bleeding was reported in the ODIXa-DVT study indicating that a twice-daily regimen for rivaroxaban might increase risk of bleeding. Nonetheless, the results of these two studies had supported the safety and efficacy of rivaroxaban in the treatment of DVT. Based on these results, further phase III randomized trials of rivaroxaban for the treatment of DVT and for the prevention of stroke in AF (ROCKET-AF) with 20 mg once-daily are undergoing. Rivaroxaban was been assessed by NICE,[@b103-vhrm-5-693] who have recommended it as an option for the prevention of VTE in adults having elective total hip replacement surgery or elective total knee replacement surgery. The initial dose, subject to normal hemostasis and reflective of additional risk factors, should be 10 mg within 6--10 hours after surgery, then for two weeks in knee surgery or five weeks in major hip surgery. Apixaban is a highly selective and potent inhibitor of factor Xa at advanced clinical development. It has a small molecular weight, with a bioavailability of more than 50% and half-life of between nine and 14 hours. Apixaban has fixed twice-daily dosing and is metabolized in the liver via CYP3A4, with about 25% excreted by the kidneys and the remainder by intestinal excretion.[@b104-vhrm-5-693] It has been shown to be safe and well tolerated in initial testing in volunteers and its anticoagulant effects closely correlated to plasma concentration of the drug. In the phase II APROPOS trial apixaban prevented more cases of VTE and death after total knee replacement than enoxaparin, but the risk of bleeding correlated with drug dose.[@b105-vhrm-5-693] In a relatively small Botticelli-DVT trial (520 patients with acute symptomatic DVT), apixaban demonstrated effectiveness in the prevention of recurrent VTE similar to conventional therapy with LMWH followed by warfarin (target INR 2.0--3.0). The principal safety outcome (composite of major and clinically relevant, nonmajor bleeding) occurred in 7.3% of the apixaban-treated patients versus 7.9% of patients with reference treatment.[@b106-vhrm-5-693] Another direct thrombin inhibitor, betrixaban, is being trialed. It has a small molecular weight, with an oral bioavailability of 47% and half-life of 19 hours, and is eliminated via intestinal excretion. A randomized phase II clinical trial, the EXPERT study, investigated the safety and efficacy of betrixaban in comparison with enoxaparin for VTE prevention in patients undergoing total knee replacement surgery. More than 200 patients were randomized to either betrixaban (15 or 40 mg twice daily) or enoxaparin (30 mg twice daily). Betrixaban demonstrated antithrombotic activity and appeared well tolerated in knee replacement patients at the doses studied.[@b107-vhrm-5-693] Conclusions =========== In the UK, government initiatives have pushed anticoagulation center-stage in many aspects of medical and surgical care.[@b108-vhrm-5-693],[@b109-vhrm-5-693] The vast weight of prevention and treatment of VTE falls to the management of warfarin and LMWHs, with fondaparinux gaining use. However, the last few years has seen the development of new oral anticoagulants that may take over from currently established agents,[@b110-vhrm-5-693] and for which government bodies are releasing guidelines.[@b78-vhrm-5-693],[@b79-vhrm-5-693],[@b94-vhrm-5-693],[@b103-vhrm-5-693] It seems likely that future practice will see the increasing use of novel oral anticoagulants. Disclosures ADB has taken hospitality, speaker fees and research funds from GlaxoSmithKline, Sanofi-Aventis and Boehringer Ingleheim. The University Department of Medicine at City Hospital recruits for anticoagulant trials. ![The coagulation system simplified.](vhrm-5-693f1){#f1-vhrm-5-693} ###### Stratification of the risk factor for VTE ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Strong risk factors (increased risk \>10) Hip, pelvis or leg fracture, hip or knee replacement, major general surgery (eg, CABG), major trauma, spinal cord injury, hospital or nursing home confinement Moderate risk factors (increased risk 2--9) Arthroscopic knee surgery, central venous lines, malignancy (alone, 2--4 times, But with chemotherapy this rises to 4--6 times), congestive heart or respiratory failure, HRT, use of oral contraceptives, paralytic stroke, post-partum pregnancy, previous VTE, thrombophilia, neurological disease with extremity paresis, varicose veins at age 45, superficial vein thrombosis Weak risk factors (increased risk \<2) Bed rest \>3 days, immobility due to sitting (eg, prolonged car or air travel, wheelchair), increasing age, laparoscopic surgery (eg, cholecystectomy), obesity, ante-partum pregnancy, varicose veins at age 60 ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Abbreviations: CABG, coronary artery bypass graft; HRT, hormone replacement therapy; VTE, venous thromboembolism. ###### Differences between low-molecular-weight heparin and unfractionated heparin[@b48-vhrm-5-693]--[@b51-vhrm-5-693] Low-molecular-weight heparin Unfractioned heparin ----------------------------- ---------------------------------- ---------------------------------- Mean molecular weight (kDa) 5 15 Saccharide units 13--22 40--50 Anti-Xa/Anti-IIa activity 2:1 to 4:1 1:1 Platelet inhibition \+ + \+ + + + Inhibited by PF4 No Yes Bioavailability 92%--100% 30%--50% Half-life (hours) Intravenous 2 1 Subcutaneous 4 2 Endothelial binding No Yes Dose-dependent clearance No Yes Mode of clearance Kidney Liver/kidney Frequency of HIT Low (eg, 0%\) High (eg, 2.7%[@b42-vhrm-5-693]) Frequency of osteoporosis Low High Monitoring Anti-Xa assay Routine APTT Abbreviations:* HIT, heparin-induced thrombocytopenia; PF4, platelet factor 4; APTT, activated partial thromboplastin time. ###### Indication for the use of low-molecular-weight heparins Bemiparin Dalteparin Exonaparin Tinzaparin ------------------------------------------------ --------------- ---------------- ---------------- ---------------- Prophylaxis of VTE Yes Yes Yes Yes Treatment of VTE Yes Yes Yes Yes Treatment of VTE in pregnancy (unlicensed) No Yes Yes Yes Prophylaxis of VTE in acute coronary syndromes No Yes Yes No Notes: Data was obtained from BNF.[@b72-vhrm-5-693] Referral in practice is essential, especially in view of cautions and contraindications which may be low-molecular-weight heparin-specific. ###### Risk factors for VTE Risk factors (Score 1) Risk factors (Score 2) Risk factors (Score 3) ------------------------------ ------------------------------------------------------------------- ---------------------------- Age \>60 Estrogen-containing pill Immobile (\>72 hours) Obesity (BMI \> 30) HRT History of DVT/PE IHD, CCF or previous stroke Known thrombophilic conditions Significant COPD Malignancy Extensive varicose veins Sepsis Inflammatory bowel disease Known family history in two relatives (at least one first degree) Nephrotic syndrome Post-partum Myeloproliferative disorders Pregnancy Abbreviations: BMI, body mass index; COPD, chronic obstructive pulmonary disease; CCF, congestive cardiac failure; DVT, deep vein thrombosis; HRT, hormone replacement therapy; IHD, ischemic heart disease; PE, pulmonary embolism. ###### Additional risk factors for surgical inpatients Score Surgical procedure ----------- ---------------------------------------------------------------- 4 Major trauma; eg, lower limb fractures 4 Major joint replacement 4 Surgery for fractured neck of femur 3 Thoracotomy or abdominal surgery involving mid-line laparotomy 3 Total abdominal hysterectomy; including laparoscopic assisted 2 Intraperitoneal laparoscopic surgery lasting \>30 minutes 2 Vascular surgery (not intra abdominal) 1 Surgery lasting \>30 minutes 0 Surgery lasting \<30 minutes Note: The total risk score will give an initial guide to therapy, although low-molecular-weight heparin may not be appropriate for all patients. ###### Application of risk assessment for the use of low-molecular-weight heparin (LMWH) Low risk Score 0 or 1 Moderate risk Score 2 or 3 High risk Score ≥4 ------------------------------- ---------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------------------------------------- Early ambulationConsider GECS GECSLow dose LMWH od for surgical patientsHigh dose LMWH od for medical patients GECSHigh dose LMWH od (max. 14 days for medical in-pts)Surgical patients -- consider IPC in theatre plus high dose LMWH Note: For male patients \<57 kg and female patients \<45 kg, caution may be needed regarding the dose of LMWH prescribed. Abbreviations: IPC, intermittent pneumatic compression; GECS, graduated elastic compression stockings; od, once daily. ###### Use of low-molecular-weight heparins in prophylaxis versus treatment in an 80 kg patient Bemiparin Dalteparin Enoxaparin Tinzaparin -------------------------------------- --------------- ---------------- --------------------- ----------------------------------------- Prophylaxis in moderate-risk surgery 2,500 units 2,500 units 20 mg/2,000 units 3,500 units (general surgery) Prophylaxis in high-risk surgery 3,500 units 5,000 units 40 mg/4,000 units 4,000--4,500 units (orthopedic surgery) Treatment of VTE 9,200 units 15,000 units 120 mg/12,000 units 14,000 units Notes: Data obtained from BNF.[@b72-vhrm-5-693] Practitioners must consult this source or their own local guidelines before acting.	{ "pile_set_name": "PubMed Central" }
1. Introduction {#sec1} =============== SLE generally affects young to middle aged women, commonly presenting as a triad of fever, rash, and joint pain. However, SLE can present in a complex fashion, varying based on the degree and severity of organ involvement. Gastrointestinal symptoms are common in SLE, and more than half of the conditions are caused by adverse reactions to medications, viral or bacterial infections \[[@B1]\]. Other causes include lupus mesenteric vasculitis, which can lead to protein-losing enteropathy, intestinal pseudoobstruction, acute pancreatitis, and other rare complications, such as celiac disease and inflammatory bowel diseases. Abdominal pain in patients with SLE may also reflect underlying vasculitis and thrombosis, which can lead to life-threatening ischemia and perforation, if not promptly treated. The following case describes a young woman presenting with lupus enteritis and lupus panniculitis as the initial manifestation of SLE, the importance of early disease recognition, utilities of abdominal CT in diagnosis, and current treatment protocols for lupus enteritis. 2. Case Presentation {#sec2} ==================== A 34-year-old Hispanic female with no significant medical history arrived at the emergency department with 3 days of abdominal pain, vomiting, and painful lesions on the skin of the lower extremities. The patient described the pain as diffuse, constant, and dull in intensity, without any alleviating factors or associated symptoms. On physical examination, vitals were within normal limits. The abdomen was diffusely tender and the lower extremities revealed multiple crops, nontender 1-2 cm subcutaneous nodules, without evidence of induration or suppuration. Laboratory tests revealed microcytic anemia (hemoglobin of 9.8 g/dL), a positive antinuclear antibody titer of 1 : 180, low complement levels, negative serology for anti-DsDNA, anti-RNP, anti-Smith, anti-SSA, and anti-SSB antibodies, and undetectable cryoglobulins levels. CT of the abdomen revealed diffuse edema and thickening of the small bowel wall with a moderate amount of abdominopelvic ascites, primarily surrounding loops of small bowel, the liver, and spleen and in the pelvic cavity ([Figure 1](#fig1){ref-type="fig"}). The patient was admitted and given intravenous hydration, ciprofloxacin, and metronidazole for presumed infectious enteritis; however, the following day the symptoms did not improve. An upper endoscopy did not reveal any significant abnormalities, and biopsies along with mesenteric angiography were negative for vasculitis or ischemia. A skin biopsy of the nodular lesions was then performed which revealed lobular and septal panniculitis, with inflammation primarily composed of neutrophils without any granulomas. The patient was given systemic steroids and within a few days the abdominal pain and skin lesions began to resolve. The patient was discharged on hydroxychloroquine and a tapering dose of steroids. 3. Discussion {#sec3} ============= The clinical picture of lupus enteritis is often nonspecific, with abdominal pain, diarrhea, and vomiting being the cardinal manifestations with jejunum (80%) or with ileum (85%) involvement \[[@B2]\]. The pathogenesis is unclear but has been attributed to immune-complex deposition and complement activation, with subsequent submucosal edema \[[@B3]\]. Although considered a form of visceral or serosal vasculitis, lupus enteritis is seldom confirmed on histology, making computerized tomography the gold standard for diagnosis. There are three classic patterns suggestive of lupus enteritis: (1) bowel wall thickening greater than 3 mm, also referred to as target sign, (2) engorgement of the mesenteric vessels (coombs sign), and (3) increased attenuation of mesenteric fat \[[@B2]\] ([Figure 1](#fig1){ref-type="fig"}). Arteriography may reveal arterial narrowing and distended loops of bowel \[[@B4]\]. One should always consider mesenteric vasculitis in the differential diagnosis as overlooking the diagnosis can have grave consequences. One autopsy study found that 60--70% of SLE patients had evidence of peritonitis, whereas only around 10% of them were recognized clinically \[[@B5]\]. Signs of perforation may be subtle and masked in patients taking steroids; therefore, abdominal pain in a lupus patient must be addressed and evaluated. Lin et al. suggested that SLE should be suspected in any patient with CT findings of enteral vasculitis or ischemic enteritis, even without lupus-related symptoms or signs and complement levels \[[@B6]\]. C3/C4 levels may be helpful in the differential diagnosis. Steroids are generally considered to be first line therapy for lupus enteritis. Steroid administration may be IV or by mouth based on clinical status or other organ involvement, with preference for IV in case of severe lupus flare because of potentially reduced drug absorption from tissue edema due to enteritis \[[@B7]\]. In steroid resistant cases, oral mycophenolate may be another option \[[@B8]\]. There has been one single report of a patient who responded to the EURO lupus cyclophosphamide regimen \[[@B3]\]. Even if patients respond initially to steroids, there is a high predilection for recurrence. A predictor of risk of recurrence for lupus enteritis is bowel wall thickness greater than 9 mm and the recurrence rate of lupus enteritis correlates with a lower cumulative dosage of prednisolone and a shorter duration of treatment \[[@B3], [@B9]\]. 4. Conclusion {#sec4} ============= Lupus enteritis underlies a broad spectrum of processes which includes mesenteric arteritis, intestinal vasculitis, lupus peritonitis, and abdominal serositis. Patients who have complaints of abdominal pain should be evaluated carefully as overlooking the diagnosis and delaying treatment can result in bowel ischemia and perforation. The diagnosis of lupus enteritis is based on classical CT findings (bowel wall edema with target sign, mesenteric abnormalities, and ascites) as histopathology seldom confirms the diagnosis. Typically, lupus enteritis is steroid-responsive with an overall excellent prognosis and immunosuppressive treatment is reserved for recurrent enteritis or severe SLE cases with multiorgan involvement. This case illustrates that SLE involves various systems and can present in a multitude of ways, making it important for clinicians to consider the differential diagnosis, especially in a young woman with complex symptomatology. Consent ======= Written informed consent was not obtained from the patient for publication of this case report; no identifying information or images were used in the publication of this paper. Conflict of Interests ===================== The authors declare that they have no competing interests. ![Computed tomography of the abdomen illustrating bowel loops with edema also referred to as target sign.](CRIGM2014-962735.001){#fig1} [^1]: Academic Editor: Matteo Neri	{ "pile_set_name": "PubMed Central" }
All relevant data are within the paper and its Supporting Information files. Introduction {#sec001} ============ Crimean-Congo hemorrhagic fever virus (CCHFV) is a tick-borne viral zoonosis distributed in Africa, Asia, eastern Europe and the Balkans. Ticks belonging to the genus Hyalomma are considered the principal vectors and the broad geographic distribution of the virus correlates with that of the vector \[[@pntd.0006598.ref001],[@pntd.0006598.ref002]\]. More recently the virus has been identified as a cause of human disease in Spain \[[@pntd.0006598.ref003]\], Greece \[[@pntd.0006598.ref004]\], and India \[[@pntd.0006598.ref005]\]. There is growing concern that the virus could emerge in other southern European countries which are within the distribution range of the vector \[[@pntd.0006598.ref006],[@pntd.0006598.ref007]\]. The virus causes a disease that ranges in severity for reasons that are not clear \[[@pntd.0006598.ref001], [@pntd.0006598.ref002]\]. Fatality rates vary depending on the severity of the disease and can be as high as 30% in some countries. Current diagnosis of CCHFV is based on the detection of viral RNA using RT-PCR, isolation of the virus and/or IgM/IgG detection \[[@pntd.0006598.ref008],[@pntd.0006598.ref009]\]. Hence the detection of an antibody response against CCHFV is important for diagnosis as well as seroprevalence studies. Currently, most reagents used for development of diagnostic tools are dependent on culturing the virus within the confines of a biosafety level 4 facility. The biosafety considerations limit the number of laboratories which are able to prepare reagents \[[@pntd.0006598.ref010]\]. The current perceived risk of spread of the virus to non-endemic regions highlights the importance of increasing diagnostic capacity and serological surveillance using safe, standardized reagents \[[@pntd.0006598.ref009]\]. The viral genome consists of three RNA segments designated small (S), medium (M) and large (L). The S segment encodes the nucleocapsid, while the L segment translates into the RNA polymerase. The M segment encodes for a glycoprotein precursor that is post-translationally cleaved and generates mature G~N~ and G~C~ and a mucin-like domain \[[@pntd.0006598.ref011]\]. Serological assays have been developed based on the recombinant nucleocapsid protein \[[@pntd.0006598.ref008]\] but the use of glycoproteins has not been extensively investigated possibly due to the inherent challenges associated with preparing recombinant glycoproteins. However, peptides mimicking epitopic regions could have a potential as diagnostic tools and if the epitopes induce protective immunity they could play a role in vaccine development \[[@pntd.0006598.ref012]\]. Goedhals et al. identified two possible epitopic regions in the G~C~ of CCHFV \[[@pntd.0006598.ref013]\] which could have potential in further development of serological assays and warrant further investigation. In this study we used a microarray technique creating a high throughput and cost effective method to screen B-cell peptide epitopes \[[@pntd.0006598.ref014]--[@pntd.0006598.ref016]\] covering the complete glycoprotein precursor of CCHFV. This is the first study using microarray technology to screen clinical samples from survivors of CCHF infections in South Africa and Turkey. This data was also further examined using serum samples from vaccinated individuals that received the Bulgarian vaccine. Interestingly, we have identified several specific peptide sequences which may have application in development of serological assays and also in vaccine development. Materials and methods {#sec002} ===================== Solid Phase Peptide Synthesis (SPPS) {#sec003} ------------------------------------ Overlapping peptides representing the complete glycoprotein precursor of a strain of CCHFV including the O-glycosylated mucin-like domain and GP38 at amino acid positions 15--515, G~N~ and NS~M~ (515--1037) and G~C~ (1037--1688) (strain Turkey-Kelkit06, uniprot \#C7F6X8) were prepared by a modified automated Fmoc-SPPS (Solid-Phase Peptide Synthesis) methodology on a Syro II peptide synthesizer (MultiSynTech, Witten, Germany) as described previously \[[@pntd.0006598.ref017],[@pntd.0006598.ref018]\]. All samples were screened against a peptide library representing a CCHFV isolate from Turkey to identify conserved epitopic regions between different lineages. Printing of microarrays {#sec004} ----------------------- Peptides were selectively enriched by covalent immobilization onto amine reactive N-hydroxy succinimide activated hydrogel coated MPX16 glass slides (Schott Nexterion, SlideH) with a BioRobotics MicroGrid II spotter (Genomics Solution) using Stealth 3B Micro Spotting Pins (ArrayIt) with approximately 6 nL per spot as described previously \[[@pntd.0006598.ref017],[@pntd.0006598.ref018]\]. Printed glass slides were humidified for 30--60 min before N-hydroxy succinimide deactivation in blocking buffer (50 mM ethanolamine in 50 mM sodium borate, pH 8.5) for 30 min, then rinsed quickly in water and spun dry (VWR, Galaxy MiniArray). The blocked glass slides were fitted into superstructures; 2/16/48 well (FAST FRAME, Schleicher & Schuell (Whatman)) to make separate identical peptide libraries. To each library either 500/100/10 μL PLI-P (0.5 M NaCl, 3 mM KCl, 1.5 mM, KH~2~PO4, 6.5 mM Na~2~HPO4, pH 7.4, 3% bovine serum albumin (BSA) was added dependent on the superstructure. On-slide glycosylation of peptides {#sec005} ---------------------------------- Blocked slides were fitted with a 2-well superstructure (FAST FRAME, Schleicher & Schuell (Whatman)) to form 2 wells. The wells were filled with 500 μL of glycosylation mixture (10 μL of 100 mM UDP-GalNAc (Sigma-Aldrich), 10 μL of either 0.36 mg/mL mM GalNAc-transferase 2 \[GalNAc-T2\] (SBH Biosciences) or 0.46 mg/mL GalNAc-transferase 3 \[GalNAc-T3\] (SBH Biosciences) 10 mM and 480 μL HEPES buffer, placed in a humidification chamber, and incubated for two hours at room temperature (RT). Slides were then washed with 0.1 M AcOH (2 x 5 min, shaking) and PLI-P (5 min, shaking). Slides were again washed with phospate buffered saline (PBS) pH7.4, rinsed thoroughly with water, dried by centrifugation, and were then ready to be immediately used in a subsequent lectin binding experiment \[[@pntd.0006598.ref015],[@pntd.0006598.ref017]\]. Slides were incubated with biotinylated VVA lectin (1:500 dilution in PLI-P buffer) for 1 hour at RT, followed by incubation with streptavidin-AlexaFluor 647 (1:1000 dilution in PLI-P) for 1 hour. Immunscreening using peptide assays {#sec006} ----------------------------------- Pooled (to conserve resources) or individual CCHFV IgG antibody positive sera and CCHFV IgG negative sera or serum samples IgG positive for varicella zoster virus (VZV), Epstein-Barr virus (EBV), herpes simplex virus (HSV) and tick-borne encephalitis (TBE) were diluted (1:10, 1:20, 1:25, 1:50) in incubation buffer PLI-P (0.5 M NaCl, 3 mM KCl, 1.5 mM KH~2~PO~4~, 6.5 mM Na~2~HPO~4~, pH = 7.4, 3% BSA), added directly onto slide subarrays and incubated for minimum 1 h (up to 2 days) on a shaking plate with a slow rotation. Slides were washed three times using PBS. Positive reactors were detected using goat anti-human IgG-Cy3 (Fc specific, 10μg/mL, Sigma-Aldrich) diluted (1:1000) in PLI-P. After the final wash, slides were spun dry and scanned followed by image analysis. Slides were scanned using a ProScanArray microarray scanner (Perkin Elmer) equipped with laser for excitation at 543 nm and images were analyzed with Scan Array Express software. Spots were identified using automated spot finding with manual adjustments for irregularities in print. The final data was obtained from the mean spot Relative Fluorescence Units (RFU) from all replicate spots for each sample (3 or 5 spots). Spot intensities were determined by subtracting the median pixel intensity of the local background from the average pixel intensity within the spot. The quality control covered intra- and interchip quality analysis of replicates. For the selected peptides, serum samples with relative fluorescent values higher than two standard deviations over the mean of the control group were designated as positive. Patient sera {#sec007} ------------ The samples were routinely heated at 56°C for 1 hour prior to handling. All antibody positive samples used in the study were from convalescent or vaccinated individuals and hence would be negative for CCHFV. ### Turkey {#sec008} A total of 30 CCHF IgG positive sera from Refik Saydam National Public Health Agency, Ankara, Turkey were included in the study. All samples were confirmed CCHF IgG antibody positive using a commercial ELISA (Vektor-Best, Novosibirsk, Russia). Ethics approval was obtained for samples collected from patients in Turkey (Ethics approval number 2009--0728). ### Bulgaria {#sec009} The study included eight samples obtained from medical personnel who were voluntarily vaccinated in order to produce hyperimmune gamma-globulin. Of the ten participants, seven were female and three male, and their age ranged from 30 to 55 years (median age 48). The first group of four participants (individuals 1--4) had received several vaccinations and blood samples were taken one month after the last immunization. The second group of four participants (individuals 5--8) that had just completed a full vaccination program (consisting of three doses, plus one last dose one year after the first dose) and blood samples were taken one month after the last dose (Ethics approval number 2/2011). ### South Africa {#sec010} A total of 41 CCHF IgG positive sera from 14 laboratory confirmed patients in South Africa were included in the study. Serum samples were collected at intervals ranging from several months after infection up to 12 years after illness. All samples were confirmed CCHF IgG antibody positive using a commercial immunofluorescent antibody assay (EuroImmune). In addition, 11 CCHF IgG negative sera from 11 healthy volunteers were included as controls samples. Ethics approval was obtained for samples collected from patients in South Africa (Ethics Committee number 152/06). ### Control sera {#sec011} 37 clinically validated sera from individuals infected with other viral diseases such as TBE, (n = 9), EBV, (n = 10) \[[@pntd.0006598.ref016]\] and HSV, (n = 9) \[[@pntd.0006598.ref015],[@pntd.0006598.ref019]\] and VZV, (n = 9) \[[@pntd.0006598.ref019]\] were included in the study. ### Ethical statement {#sec012} Ethics approval was obtained for samples collected from patients in Turkey, Bulagria, South Africa (Ethics approval numbers 2009--0728 (Turkey), 2/2001 (Bulgaria)ECUFS 152/06 (South Africa). All participants (Turkey, Bulgaria and South Africa) were adults. Written informed consent was obtained from participants. All the collected samples were derived from blood donors and anonymized at the hospital/institute where the samples were originally collected, i.e. without possibility to trace samples back to the donor. Statistical analysis {#sec013} -------------------- All the data represented in this study are mean values from 3--5 replicates. The final data was obtained from the mean spot Relative Fluorescence Units (RFU) from all replicate spots for each sample (3 or 5 spots). Spot intensities were determined by subtracting the median pixel intensity of the local background from the average pixel intensity within the spot. The quality control covered intra- and interchip quality analysis of replicates. For the selected peptides, serum samples with relative fluorescent values higher than two standard deviations over the mean of the control group were designated as positive. A One-way-ANOVA (Prism Graphpad 7 software) was conducted as needed. Results {#sec014} ======= Reactivity of serum samples collected from survivors in Turkey {#sec015} -------------------------------------------------------------- A scan peptide library consisting of 168 peptides (20mer with a 10 amino acid overlap, see [S1 Table](#pntd.0006598.s001){ref-type="supplementary-material"}) representing the glycoprotein precursor which includes the mucin-like domain, GP38, and G~N~ and G~C~ region of a strain of CCHFV from Turkey, was incubated with pooled inactivated sera from survivors of CCHFV infection. Reactivity against peptides representing possible epitopic regions were identified ([Fig 1](#pntd.0006598.g001){ref-type="fig"}). Samples with a relative fluorescent value greater than two standard deviations above the mean of the control group were designated as positive. ![Reactivity of pooled Turkish sera.\ Reactivity of eight CCHFV survivor pools (blue) and six uninfected pools (black) against 168 peptides representing the complete precursor glycoprotein of CCHFV (Turkish strain, Kelkit06, Uniprot \#C7F6X8).](pntd.0006598.g001){#pntd.0006598.g001} Sera from survivors of CCHFV reacted significantly against fourteen peptides (entry 1--14, [Table 1](#pntd.0006598.t001){ref-type="table"}). Although there was minor reactivity from control sera against some peptides (entry 15--20, [Table 1](#pntd.0006598.t001){ref-type="table"}), the relative fluorescent values for these sera were significantly lower and these reactions were considered non-specific. From the results of this initial screen, 32 peptides were selected based on a combination of high reactivity of CCHFV positive sera and low reactivity of control sera ([S2 Table](#pntd.0006598.s002){ref-type="supplementary-material"}) and narrowed down the selection to the five most promising peptide epitopes (entry 6, 7, 12, 13 and 14, ([Table 1](#pntd.0006598.t001){ref-type="table"})). To further map the epitopes with respect to peptide sequence, a stepwise single-amino-acid epitope walk library of 20mers of these peptides was synthesized \[[@pntd.0006598.ref015]\]. 10.1371/journal.pntd.0006598.t001 ###### Sequences of scan peptides that showed significant reactivity with pooled Turkish sera of CCHFV survivors (1--14) and with control sera (15--20). Five possible epitopes (6, 7, 12, 13 and 14) were selected for further evaluation using a single-amino acid epitope walk analysis. ![](pntd.0006598.t001){#pntd.0006598.t001g} ID Peptide Sequence ---- --------- ---------------------------------- 1 p14 ^131^TSPSSSPSTPSTPQGIYHPA^150^ 2 p15 ^141^STPQGIYHPARSLLSVSSPK^160^ 3 p19 ^181^HSAMSRIPTPHTATRVSTEN^200^ 4 p22 ^211^SSAQQTTPSPMTSPAQSILL^230^ 5 p23 ^221^MTSPAQSILLMSAAPTAVQD^240^ 6 p24 ^231^MSA APTAVQDIHPSPTNRSK^250^ 7 p55 ^541^ETAEIHDDNYGGPGDKITIC^560^ 8 p56 ^551^GGPGDKITICNGSTIVDQRL^570^ 9 p78 ^771^RLTSDGLARHVTQCPKRKEK^790^ 10 p96 ^951^NVMLAVCKRMCFRATIEASR^970^ 11 p99 ^981^TTFVICILTLTICVVSTSAV^1000^ 12 p105 ^1041^RKPLFLDSIVKGMKNLLNST^1060^ 13 p107 ^1061^SLETSLSIEAPWGAINVQST^1080^ 14 p156 ^1551^PQSILIEHKGTIIGKQNDTC^1570^ 15 p51 ^501^SVLRQYKTEIKIGKASTGFR^520^ 16 p64 ^631^VLDACDSSCEVMIPKGTGDI^650^ 17 p74 ^731^QYRELKPQTCTICETAPVNA^750^ 18 p91 ^901^CSIGSVNGFEIESHKCYCSL^920^ 19 p164 ^1631^RRTRGLFKYRHLKDDEETGY^1650^ 20 p165 ^1641^HLKDDEETGYRRIIERLNSK^1660^ Six CCHFV IgG positive and two CCHFV IgG negative sera were analyzed individually, as shown in [Fig 2](#pntd.0006598.g002){ref-type="fig"}. Overall, serum reactivities were high (\>50%) against each peptide 20mer sequence. Two different reactivity patterns were apparent when screening the epitopes (colored in blue and orange in [Fig 2](#pntd.0006598.g002){ref-type="fig"}). Four serum samples (designated CCHFV 1--4, blue lines) reacted strongly with early epitope walk (EW) peptides within libraries p24 and p55+56. In contrast, serum samples designated CCHFV 5 and 6 (orange) reacted weakly with these peptides, or not at all, and reacted strongly with peptide 40 within library p55+56 and peptide 64 in p78. The six serum samples reacted uniformly with p78 (EW 57--61) and p96 (EW 80--86). No reactivity was detected using the serum samples that were negative for CCHF IgG antibody (uninfected, black lines). High reactivity was seen against epitope walk (EW) residues 25--32 (p55+56) for four serum samples (CCHFV 1, 2, 3 and 4), whereas the remaining samples (orange) reacted against EW peptide 40 (p55+56). All sera reacted against EW peptides 81--86 within p96. ![Epitope walk (EW) of selected five peptides (p24, p55, p56, p78 and p96).\ Figure illustrates the RFU of individual serum samples from CCHFV survivors from Turkey to selected peptides. Epitope walk sequences are found in [S3 Table](#pntd.0006598.s003){ref-type="supplementary-material"}. In general two different patterns were seen when screening the epitopes (colored in blue and orange).](pntd.0006598.g002){#pntd.0006598.g002} High reactivity was seen against epitope walk (EW) residues 25--32 (p55) for four serum samples CCHFV 1, 2, 3 and 4 (blue), whereas the remaining samples (orange) reacted against EW peptides 40 (p56). All sera reacted against EW peptides p81-86 derived from p96. Additional validation of p96 and its derived epitope walk peptides was performed with 30 sera from survivors of CCHFV from Turkey and 37 sera from individuals infected with other viral diseases such as TBE, VZV, EBV and HSV ([Fig 3](#pntd.0006598.g003){ref-type="fig"}). The results demonstrate that the CCHFV infected individuals show high specificity and sensitivity (97%) to epitope p96. In addition, the activity of 10 sera from individuals vaccinated with the Bulgarian CCHFV vaccine \[[@pntd.0006598.ref020]\] with p96 was analysed. This experiment also demonstrated that the vaccinated indivduals have significant reactivity towards this novel epitope p96 whereas the control sera (n = 37) showed no reactivity. The amini acid similarity between strains from Turkey and Bulgarian is high and the epitopic region identified has only minor differences which indicates that a similar reactivity towards this epitope could be expected. An analysis of variance showed that the reactivity of samples from Turkish survivors and from the vaccinated group relative to the control group was highly significant (F(2,74) = 68, P\<0.0001). ![Dot-plot of intensities of the selected peptide EWP 83 (p96 + 3 aa, ^954^LAVCKRMCFRATIEASRRAL^973^) after serology.\ This includes sera from Turkish CCHFV survivors, control sera from individuals infected with other viral pathogens as controls and Bulgarian individuals vaccinated against CCHFV. The dotted line represents the diagnostic cut-off value determined from mean of the control group plus two standard deviations. P\<0.0001.](pntd.0006598.g003){#pntd.0006598.g003} Lastly, since glycosylation can be important for viral envelope proteins and inducing immune responses, microarray experiments were conducted to determine whether additional epitopes could be identified using glycosylated peptides. Peptides immobilised on slides were treated with ppGalNAc transferase 2 (T2) and ppGalNAc transferase 3 (T3), as described previously \[[@pntd.0006598.ref015]\]. Glycosylation of peptides that represent parts of the mucin-domain were glycosylated as predicted with Net-O-Glyc (v3.1) algorithm. ([S1 Fig](#pntd.0006598.s004){ref-type="supplementary-material"}). We did not observe any additional reactivity of the CCHFV sera after on-chip glycosylation of the peptides. However, one should have in mind that these data are based on peptide glycosylation experiments which may not mimic the natural glycosylation pattern on the envelop proteins. Reactivity of serum samples collected from survivors in South Africa {#sec016} -------------------------------------------------------------------- To further evaluate the identified peptide epitopes, a cohort of sera from survivors of CCHFV in South Africa were screened for reactivity against the 168 20mer peptide library. A total of 41 CCHF IgG positive sera from 14 laboratory confirmed patients in South Africa were included in the study. All samples were confirmed to be CCHF IgG antibody positive using a commercial immunofluorescent antibody assay (EuroImmune). In addition, 11 CCHF IgG negative sera from 11 healthy volunteers were included as control samples. Pooled serum samples from confirmed patients reacted against peptides p55, p56, p105, p115, p119, p127 and p168 ([Fig 4](#pntd.0006598.g004){ref-type="fig"}). Reactivity against peptides 119, 127 and 168 was unique to these sera. ![Reactivity of pooled South African sera.\ Four CCHFV survivor pools (red) and one uninfected pool (black)) against 168 peptides representing the complete precursor glycoprotein of CCHFV (Turkish strain).](pntd.0006598.g004){#pntd.0006598.g004} Single serum sample experiments were performed using a 108 peptide sublibrary (see [S1 Table](#pntd.0006598.s001){ref-type="supplementary-material"}, 1--108) from the larger 168 peptide library. As seen with the Turkish cohort of sera, the majority of the South African serum samples (97,6%) reacted against the p55 (ETAEIHDDNYGGPGDKITIC) ([Fig 5](#pntd.0006598.g005){ref-type="fig"}). The specificity was even more pronounced for the South African sera. An analysis of variance showed that the reactivity of samples from Turkish survivors and SA survivors relative to the control group was highly significant (F(2,81) = 11, P\<0.0001). ![Dot-plot of intensities of the selected peptide 55 (^541^ETAEIHDDNYGGPGDKITIC^560^) after serology.\ This included sera from Turkish CCHFV survivors, control sera from individuals infected with other viral pathogens as controls and Bulgarian indivduals vaccinated against CCHFV. The dotted line represents the diagnostic cut-off value. P\<0.0001.](pntd.0006598.g005){#pntd.0006598.g005} Nine peptides were identified which included a region that reacted against either Turkish or South African samples or reacted against both samples ([Table 2](#pntd.0006598.t002){ref-type="table"}). Finally the predicted amino acid sequences for each peptide region were retrieved from UniProt and aligned to identify similarities between the geographically distinct isolates of the virus ([Table 3](#pntd.0006598.t003){ref-type="table"}). This high coverage of same peptide reactivity can be explained by high amino acid conservation along the selected epitopes ([Table 3](#pntd.0006598.t003){ref-type="table"}). In contrast the alignment confirms up to four mismatches in the predicted amino acid sequence between the Turkish peptide sequence and the South African sequence for p96, thus possibly explaining their difference in reactivity of geographically diverse samples against this peptide. 10.1371/journal.pntd.0006598.t002 ###### Reactivity of Turkish (n = 30) and South African sera (n = 41) against nine selected peptides. ![](pntd.0006598.t002){#pntd.0006598.t002g} Peptide Negative[^a^](#t002fn001){ref-type="table-fn"} Positive[^b^](#t002fn002){ref-type="table-fn"} Positive x5[^b^](#t002fn002){ref-type="table-fn"} Positive x10[^b^](#t002fn002){ref-type="table-fn"} \% Positive ------------- ------------------------------------------------ ------------------------------------------------ --------------------------------------------------- ---------------------------------------------------- ------------- ---- --- ---- --------------- --------------- 24 20 29 10 12 2 1 0 0 33 29 24 EW27 7 23 12 1 77 55 12 1 18 40 10 22 3 18 60 [98]{.ul} 56 16 17 14 24 6 5 3 2 47 59 78 27 15 3 26 1 3 0 0 10 63 79 25 16 10 6 39 95 26 15 0 0 37 96 23 33 7 8 0 0 0 0 23 20 96 EW83 1 29 0 0 [97]{.ul} 96 EW84 2 28 10 0 93 105 14 27 15 5 66 107 30 11 3 0 27 ^a^ values below 2 average of uninfected patient sera reactivity. ^b^ Positive x5/x10 indicated values 5/10 times above cut-off value. ^c^ South Africa 10.1371/journal.pntd.0006598.t003 ###### Strain similarity, comparison of amino acid sequences from eight strains of CCHFV from South Africa and one strain of CCHFV from Turkey. Sequence data was retrieved from UniProt. Amino acid differences are bolded. ![](pntd.0006598.t003){#pntd.0006598.t003g} ------------------ ---------------- ------------------------------ -------------------------------------------- -------------------------- Strain Uniprot nr p55 p56 Turkey C7F6X8 ETAEIHDDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU431/85 A0A068JCA2 ETTEIHSDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU383/87 A0A068JFM1 ENAEIHSDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU130/89 A0A068JC98 ENAEIHSDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU497/88 A0A068JD49 ENAEIHSDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU45/88 A0A068JCX3 ENTEIHSDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU187/90 A0A068JCX6 ETAEIHGDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL SA SPU48/90 A0A068JD54 ETAEIHGDNYGGPGDKITIC GGPGDKITICNGSTIVDQRL Strain Uniprot nr p78 p96 p105 Turkey C7F6X8 RLTSDGLARHYTQCPKRKEK NVMLAVCKRMCFRATIEASRRALLIR RKPLFLDSIVKGMKNLLNST SA SPU431/85 A0A068JCA2 RLTSDGLARHVMQCPKRKEK NVMLAVCKRMCFRATIEVSNRALLIR RKPLFLDSIVKGRKNLLNST SA SPU383/87 A0A068JFM1 RLTSDGLARHVIQCPKRKEK NVMLAVCKRMCFRATVEVSNKALLIR RKPLFLDSIVKGMKNLLNST SA SPU130/89 A0A068JC98 RLTSDGLARHVMQCPKRKEK NVMLAVCKRMCFRATVEVSNKALLIR RKPLSLDSIVKGMKNLLNST SA SPU497/88 A0A068JD49 RLTSDGLARHVMQCPKRKEK NVMLAVCKRMCFRATVEVSNKALLIR RKPLFLDSIVKGMRNLLNST SA SPU45/88 A0A068JCX3 RLTSDGLARHVMQCPKRKEK NVMLAVCKRMCFRATMEVSSKALLIR RKPLFLDSIVKGKKNLLNST SA SPU187/90 A0A068JCX6 RLTSDGLARHVTQCPKRKEK NVMLAVCKRMCFRATMEVSNRALFIR RRPLFLDSIVKGMKNLLNST SA SPU48/90 A0A068JD54 RLTSDGLARHVTQCPKRKEK NVMLAVCKRMCFRATMEVSDRALFIR RKPLFRDSIVKGMKNLLNST ------------------ ---------------- ------------------------------ -------------------------------------------- -------------------------- The combination of p55/p56 (ETAEIHDDNYGGPGDKITIC/GGPGDKITICNGSTIVDQRL) with p96 (NVMLAVCKRMCFRATIEASR) represents an immunodominant epitopic region that would result in a multiepitope covering the majority of the Turkish and South African strains, due to minimal differences in the amino acid sequence along the defined epitopes, defined ([Fig 6](#pntd.0006598.g006){ref-type="fig"}). An analysis of variance showed that the reactivity of samples from Turkish survivors and SA survivors and relative to the control group was highly significant (F(3,107) = 11, P\<0.0001). ![Dot-plot after combining intensities of serology with peptide p55 and p96 or EWP p83 (p96 + 3 aa).\ Control sera from healthy individuals, Turkish CCHFV survivors and South African CCHFV survivors. The dotted line represents the diagnostic cut-off value. P\<0.0001.](pntd.0006598.g006){#pntd.0006598.g006} Discussion {#sec017} ========== CCHFV is considered an emerging pathogen particularly in southern and eastern European countries. The emergence of this virus has significant public health implications. Currently there are a limited number of laboratories that can prepare reagents for diagnosis and serological surveillance as culturing the virus requires maximum containment facilities. Development of safe reagents will play a role in building capacity for diagnosis and surveillance. Development of reagents must take into consideration the global diversity of CCHFV. In this study peptides were used to identify potential epitopic regions on the glycoprotein precursor of an isolate from Turkey. The initial screen was peformed using a total of 168 peptides representing the entire GP of a Turkish strain of CCHFV. Five peptides were selected for further investigation using epitope walk analysis in an attempt to identify specific peptide sequence mimicking immunodominant linear epitopes. Despite some variability in the reactivity of the samples from survivors in Turkey, common regions were identified within p96 and p55+56. Variability in the responses was particularly noted against the peptides of p24. The differentiated response to the EW peptides of p24 is most likely caused by the fact that p24 is located in the mucin-like region of the envelope protein which is highly glycosylated. The glycosylation of the mucin-like region can differ and it is therefore expected that the serum response to the naked peptides will be diverse \[[@pntd.0006598.ref016]\]. The role of the mucin-like domain in stimulation of B cells during CCHF infections has not been well defined. By analogy with Ebola virus the mucin-like domain may block access to GP and actually inhibit immune responses \[[@pntd.0006598.ref021]\]. However as the mucin-like domain is not incorporated into viral particles its role may be very different to that proposed for Ebola \[[@pntd.0006598.ref011]\]. It would be useful to determine if the mucin-like domain is involved in immune evasion and if deletion of this region promotes an immune response to more conserved regions on the GP as has been shown for Ebola virus. This would be significant for vaccine development but less important for developing tools for detection of antibody responses. In all the experiments p55 was identified as the top candidate due to the strong response shown when incubated with serum from CCHFV infected patients. The experiments with pooled sera showed a very dominating RFU signal, thereby undermining some of the other binding when reading the scan. The reaction towards p55 was scattered with around 2/3 of the samples binding to the peptide significantly where minor to no binding was seen in other serum samples. In the second group of sera (orange pattern) the highest reaction was shifted by 10 aa residues towards p56 in comparison to the blue pattern. A small common epitope was identified in p78 (EW 57--61) with all serum samples showing similar reactivity. This epitope could be of potential interest due to the homogenous binding. The response to this epitope walk was lower than as seen with the other epitope walks except for CCHFV 5 and 6 which reacted to EW peptide 64. The epitope with the strongest homogenous response was found in p96 epitope walk representing sequences being close to the transmembrane region (aa 973--997), and the tertiary structure can be expected to be more stable and should be in reach of B-cells/antibodies/immune cells. All serum samples tested showed binding towards the peptides EW peptides 80--86. In summary, single amino acid shifts generated differentiation in reactivity patterns with no major reactive single peptide epitope for p24, p55, p56 and p78, whereas library members derived from p96 (epitope walk peptides p81-86, ^951^NVMLAVC[KRMCFRATIEAS]{.ul}RRALLIR^975^) show reactivity with all six CCHFV sera. No reactivity towards these peptides was observed with the two control sera. Identification of peptides that are cross reactive serologically against geographically distinct strains is important for development of standardised assays for detection with application on different continents. Despite some variablity in the reactivity of samples from South African patients, a commonality was identified in p55. The result of the epitope walk indicated that a strong differentiated binder could be obtained by combining the epitope ^541^ETAEIHDDNYGGPGDKITIC^560^ due to the strong signal of p55, with p96 (^951^NVMLAVCKRMCFRATIEASR^970^) for specific overall coverage. A combinational multi-epitope including these peptides could eventually secure a strong selectivity and sensitivity. The study identified several specific peptide sequences which may have application in development of serological assays and possibly vaccine development. However for vaccine development it must be taken into consideration that the immune correlates of protection for CCHFV are currently not well defined and the role of T cells and discontinuous B cell epitopes would need to be considered. Evidence exists for a role for both antibody and T cell responses. A long lived cytotoxic T cell response was recently described in survivors of infection suggesting a role for T cells in protection \[[@pntd.0006598.ref022]\]. However vaccine studies have suggested a role for both humoral responses and T cell responses. Results from a candidate vaccine employing a modified vaccinia virus Ankara poxvirus vector containing the GP in which passive and adoptive transfer of serum samples and T-lymphocytes were used in an attempt to define the role of each arm of the adaptive immune response, concluded that protective immunity likely requires both humoral and cellular involvement \[[@pntd.0006598.ref023]\]. Serum samples collected from volunteers vaccinated using the inactivated Bulgarian vaccine reacted similarly to survivors of virus infection from Turkey with regard to peptides recognised although there were differences in intensity of reactivity. Sera from vaccinated individuals reacted with lower intensity than sera from naturally infected survivors. Previous investigations of immune responses in vaccinated individuals suggested that even after several boosters the neutralising antibody response was low. Although it is not known if the epitopic regions identified in this study represent neutralising epitopes it is possible that the lower reactivity is a reflection of a less robust immune response in vaccinees compared with survivors. The use of glycoproteins has not been extensively investigated for serological assays likely due to the inherent challenges associated with expression of recombinant glycoprotein antigens. NP has been investigated as a target for diagnostic proteins due to its immunogenicty, abundance and homology between isolates. However although there is less sequence homology in the M gene compared to the S gene encoding for NP, a significant amount of the diversity is located within the mucin-like region with less than 10% amino acid diversity determined for the remainder of the GP \[[@pntd.0006598.ref024]\]. It is probable that epitopes inducing immunodominant and/or neutralising responses are more likely to be conserved between isolates and in the absence of detailed data regarding immunodominant epitopes in the GP it is appropriate to further define if there are conserved regions with possible roles in development of diagnostic tools and incorporation in vaccine design. Peptides mimicking epitopic regions could have a potential as diagnostic tools \[[@pntd.0006598.ref012]\]. Goedhals et al. identified two possible epitopic regions in the G~C~ of CCHFV and although in this study reactivity was identified in these epitopic regions using samples from South African survivors there was no significant reactivity from Turkish sera \[[@pntd.0006598.ref013]\]. This highlights the need to consider diversity when selecting peptides to mimic epitopic regions. Differences in serological reactivity between Turkish and South African samples suggested possible differences in protein sequences which were confirmed by alignment of predicted amino acid sequences. Hence the use of multiple peptides in downstream assay development will likely increase the usefulness of the assays. Lack of reactivity against serum samples collected from patients with other infectious diseases that could be considered in a differential diagnosis suggest that these peptides react more specifically against CCHFV antibodies. However the specificity and sensitivity will need to be validated in any future assay development. The cross-reactivity of these peptides with samples from a geographically distinct region where genetically diverse strains of the virus circulate enabled the identification of unique peptide epitopes from the CCHFV glycoprotein that could have application in develoment of diagnostic tools such as lateral flow or ELISA for antibody detection. Supporting information {#sec018} ====================== ###### Related to [Fig 1](#pntd.0006598.g001){ref-type="fig"} shows CCHFV GN, GC and a mucin-like domain scan peptides prepared with Fmoc-SPPS used to produce microarray data. (DOCX) ###### Click here for additional data file. ###### Sequences of 32 selected scan peptides in the first round that showed significant reactivity with pooled Turkish sera of CCHFV survivors. (DOCX) ###### Click here for additional data file. ###### Related to [Fig 2](#pntd.0006598.g002){ref-type="fig"} --sequences of peptides synthesized for further epitope mapping of scan peptides p24, p55, p56, p78 and p96. (DOCX) ###### Click here for additional data file. ###### On-slide enzymatic O-glycosylation of scan peptides (20mer with 10mer overlap) with recombinant GalNAcT2 and GalNAcT3. (DOCX) ###### Click here for additional data file. We would like to thank Prof. Tomas Bergström, Department of Infectious Diseases, Section for Clinical Virology, Institute of Biomedicine, University of Gothenburg, Gothenburg, Sweden for providing us with sera from individuals infected with tick-borne encephalitis (TBE), varicella zoster virus (VZV), Epstein-Barr virus (EBV) and herpes simplex virus (HSV). Dr. Cecilia Österholm, Karolinska Institutet is acknowledged for graphical support. [^1]: The authors have declared that no competing interests exist.	{ "pile_set_name": "PubMed Central" }
Introduction ============ Recurrent pregnancy loss (RPL) is a serious complication of pregnancy with the frequency of 1-5% among fertile couples trying to conceive ([@B1]). RPL is defined as "the occurrence of 2-3 or more consecutive pregnancy loss prior to 20th week of gestation" in reproductive age ([@B2]). RPL is a genetically heterogeneous condition resulted from both maternal and embryonic regulating factors ([@B3], [@B4]). Along with some known causes such as chromosomal abnormalities, metabolic disorders, anatomical anomalies of uterus and immunological factors, the etiology in approximately 50% of the RPL cases remains unknown (idiopathic RPL) which might be explained by immunological factors ([@B5], [@B6]). Since, fetus expresses the antigen inherited from both parents, survival of semi- allograft fetus till the term, is one of most challenging pregnancy related concepts ([@B7], [@B8]). The maternal immune system plays its critical role in successful pregnancy by controlling fertilization, implantation, progression and maintenance of pregnancy products ([@B9]-[@B11]). Therefore, investigation of immune cells, immune mediators/cytokines in the peripheral blood of mother and their coding genes would open the new window on understanding at least maternal causes of RPL. Cytokines are cell signaling proteins, mainly secreted by immune cells and mediated cell to cell communication. They are also involved in immunological, inflammatory and infectious diseases ([@B12]). They act beyond immune system in several developmental processes during embryogenesis, all steps of reproduction and then play a critical role in pregnancy outcome ([@B13]). Depending on inflammatory reactions, cytokines sub-divided into pro-inflammatory and anti-inflammatory, which are produced by T helper-1 (Th-1) and Th-2 cells, respectively. Pro-inflammatory cytokines inhibit trophoblast growth and differentiation, while anti-inflammatory cytokines promote embryonic development and placentation. On the other hand, these anti-inflammatory cytokines antagonistically control the action of Th-1 dependant cytokines ([@B14]-[@B16]). In this regard, Th-2 type immunity is believed to contribute to normal and successful pregnancy and Th-1 type immunity have been shown to be associated with pregnancy failure and may be with idiopathic recurrent miscarriage (IRM) ([@B16]). However, successful pregnancy is dependent on maintaining a fine balance between Th-1 and Th-2 immunity as well as their specific secretary cytokines ([@B17]). Interleukin (IL)-6 is a multifunctional pro- and anti-inflammatory 21-28 KDa glycoprotein, produced by different type of cells, mainly T-lymphocytes, macrophages and monocytes ([@B12], [@B18]). The gene encoded for IL-6 is mapped to the short arm of chromosome 7 (7p21) and composed of 6 exons with only 5 are coding ([@B19], [@B20]). Although, anti-inflammatory properties of this Th-2 type cytokine are well recognized, the role of IL-6 in pregnancy outcome remains unclear ([@B21]). Increased levels of this mediator have been associated with pregnancy complications such as pre-term birth ([@B22]). Several polymorphisms in IL-6 gene have been reported, some of which have been suggested to regulate its expression ([@B23]). Single nucleotide polymorphism (SNP) at position -634 C/G in the promoter region of the IL-6 gene is known to cause an altered promoter activity and thus resulting in a decreased production and secretion of IL-6 by peripheral blood mono-nuclear cells in vitro ([@B24]). In addition to identified causes of miscarriage, other factors increasing the risk of miscarriage have been reported. One of the attractive risk factors is the direct relationship between the number of live born children and pregnancy failure. Considering the potential role of cytokines in unexplained RPL, the objective of this study was analyzing the potential association between IL-6 634C/G polymorphism (rs 1800796) for the first time in Iran. Materials and methods ===================== Study population This case-control study has been performed in Arsanjan Islamic Azad University at Department of Genetics from August 2013 to October 2014. The cases comprise 121 women (17-85 years) with the history of at least two miscarriages of unknown reasons. To rule out any possible causes influencing the recurrent miscarriage, each woman underwent a diagnostic procedure includes hysteroscopy to detect uterine abnormalities, cell culture for Chlamydia and Mycoplasma infectious, maternal and paternal karyotyping, profiling hormonal status and asking for endocrine disorders. Depending on the history of previous pregnancies, women were divided to primary and secondary groups including; women with no successful pregnancy to term and those with a history of at least one pregnancy ending with live birth, respectively. 121 healthy postmenopausal women (43-76 years), were randomized to serve as a control group. No one has a history of miscarriage. All subjects in this study were from Fars province and match by geographic area. Written informed consent was obtained from all participants of the study before collecting blood samples. A questionnaire was used to elicit detailed information on demographic variables, the number of miscarriages, and number of children. Genotyping of IL-6 -634 C/G polymorphism Peripheral blood samples were collected in sterile tubes containing EDTA and genomic DNA was isolated using standard phenol- chloroform salting out method. -634C/G genotyping was done using PCR-RFLP method. The PCR primers were: 5´-GAGACGCCTTGAAGTAACTG-3´ (F) and 5´-AACCAAAGATGTTCTGAACTGA-3´ (R). PCR assay was performed for each sample in a final reaction volume of 25 µL, using 1 µL genomic DNA, 12.5 µL universal master mix (Ampliqon, Denmark), 1 µL forward primer, 1 µL reverse primer, together with 9.5 µL distilled water (DW). The PCR condition was as follows: Initial denaturation at 95^o^C for 3 min, followed by 30 cycles of: denaturation at 95^o^C for 30 Sec, annealing at 56^o^C for 30 and extension at 72^o^C for 30. Then one cycle of final extension step at 72^o^C for 5 min. All reactions were done using thermal cycler Applied Biosystems (Perkin Elmer 9600). The PCR products were digested with MbiI (BsrBI) restriction enzyme (Fermentas, \# R0102S) and put at 37^o^C for 20 hr. The products were then resolved on 2% agarose gel electrophoresis and stain with syber gold DNA safe stain dye, then visualized using a UV transilluminator. DNA molecular weight marker (Fermentas, 100 bp Ladder) was used to assess the size of the PCR-RFLP products. The amplified fragment after digestion with MbiI restriction enzymes, gave rise to 2 fragments of 120 bp and 60 bp indicating the presence of polymorphic genotype (GG), the appearance of one undigested 180 bp fragment indicates the presence of homozygous wild type (CC), while the presence of 3 fragments at 180 bp, 120 bp and 60 bp indicates the heterozygous manner (CG) ([Figure 1](#F1){ref-type="fig"}). Statistical analysis A ^2^ test was performed for determining if the case and control groups' demonstrate Hardy-Weinberg equilibrium. Statistical analysis was performed using the Statistical Package for Social Sciences (SPSS, version 16, Chicago, IL, USA). Numerical data of the scores were expressed as mean and standard deviation or median and range as appropriate. Qualitative data were expressed as frequency and percentage. Logistic regression was used to examine the relation between qualitative variables. Odds ratio (OR) with its 95% confidence interval (CI) was used for risk estimation. For quantitative data and comparison between two groups of case and control independent sample t-test was used. A p-value\< 0.05 was considered significant. Results ======= Results of PCR amplification and enzyme digestion were illustrated in [figure 1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"} respectively. 121 women with RPL (35.6±12.5 years) and 121 healthy postmenopausal women (57.6±6.2 years) were analyzed for association. Studied subjects\' characteristics are shown in [table I](#T1){ref-type="table"}. The mean number of miscarriages in patients was ranging between 2 - 10, which 23 women have 2 miscarriages and the remainders have 3 or more. The number of live births in the RPL group (1.3±2.2) was significantly lower than those in controls (4.8±2.3). This difference was significant (p\<0.001). Genotypes and allele frequencies of women with RPL and controls are given in [table II](#T2){ref-type="table"}. The frequency of polymorphic allele G in patients with RPL was 75%, while in controls was 66% (p=0.036). This allele shows an abortive effect (OR=1.5, 95%CI: 1.03-2.27). Homozygote genotype GG (OR=5.5, 95%CI: 1.05-29.3, p=0.04) and the heterozygote GC (OR=4.9, 95%CI: 0.95-24.9, p=0.05) after adjusting for age, were associated with increased risk of RPL. Under the dominant model for the allele G (GG+GC vs. CC), this allele increased the risk of disease for more than 5 times (OR=5.1, 95%CI: 1.04-25.3, p=0.04). The genotype frequencies for both patients (^2^=2.1, df= 1, p\>0.05) and controls (^2^=1.01, df= 1, p\>0.05) in the studied SNP were not significantly differ from those expected for populations in Hardy- Weinberg equilibrium. ###### Selected characteristics of the study subjects and potential risk factors for RPL Variables Cases Controls p-value ---------------------- ---------------- -------------- ------------- -- Age (mean±SD) 35.6 ± 12.5 57.6 ± 6.2 Range 17-85 43-76 Abortion (mean±SD) 3.3 ± 1.2 \- No. abortion (n, %) 2 23 (19%) \- ≥3 98 (81%) \- Abortion type (n, %) Primary 67 (55.4%) \- Secondary 54 (44.6%) \- No. of live births 1.3 ± 2.2 4.8±2.3 \<0.001 P\<0.05 were considered to indicate statistical significance. (n=121) ###### Distribution of -634 C/G genotypes and allelic frequencies in cases and controls (Gene IL-6 Genotype (n=121) Controls (n, %) Patients (n, %) [a](#TFN1){ref-type="table-fn"} p-value OR 95%CI ---------------------- --------------------- --------------------- --------------------------------------------- -------- ----------- ----------- Genotype CC 18 (14) 5 (4.1) \- \- \- CG 47 (39) 50 (41.4) 0.05 4.9 0.95-24.9 GG 56 (47) 66 (54.5) 0.04 5.5 1.05-29.3 Allele C 83 (34) 60 (25) \- \- \- G 159 (66) 182 (75) 0.036 1.5 1.03-2.27 Phenotype GG+GC vs. CC 103 (86) 116 (95.9) 0.04 5.1 1.04-25.3 CC+GC vs. GG 65 (53) 55 (45.5) 0.68 0.85 0.39-1.85 OR: Odds ratio CI: confidence interval adjusted for age. ![PCR amplification of the -634C/G polymorphism in the promoter of IL-6 gene. using designed primer set produce 180 bp amplicon in length.](ijrb-14-103-g001){#F1} ![PCR-base restriction analysis of IL-6-634C/G polymorphism was shown on 2% agarose electrophoresis. The polymorphic region was amplified by PCR resulting in digestible fragment length 120 and 60 bp in lane 1, 2 (GG homozygote), indigestible fragment length 180 bp in lane 3 and 5 (CC homozygote). Three different fragments 180, 120, 60 bp (lane 4 and 6) represent CG heterozygote. M: marker, 100 bp ladder (Fermentas, Germany). \[magnify: 640 × 480px](ijrb-14-103-g002){#F2} Discussion ========== Although extensive studies have been made on the role of several cytokines, including IL-6, IL-10, TNF-α and IFN-γ in different stage of pregnancy, the inconclusive results, make the interpretation of their effect on pregnancy difficult ([@B16], [@B25], [@B26]). One of the well-studied cytokine is IL-6, which the reduction of its protein and mRNA in the mid- secretary phase of endometrial tissue in prone women to RPL has been shown in distinct studies ([@B27]). As compared with normal women, patients with the history of RPL, have significantly lower serum concentrations of Th-2 cytokines including IL-6 and IL-10 ([@B28]). The production of pro-inflammatory and anti-inflammatory cytokines is partly under genetic control. In present research emphasis is laid on IL-6 gene polymorphism as this cytokine supposed to have crucial role on the implanting embryo. One of the most common polymorphism in the promoter region of the IL-6 gene is 174 G/C, known to alter the IL-6 transcription, which its G allele shows relation to increase serum concentrations only in one study and with decreased in others ([@B29], [@B30]). The literature review demonstrates that the vast majority of the studies assessed the role of IL-6 in pathogenesis of RPL, focused on this SNP, which none of them resulted in any association ([@B26], [@B28], [@B31]). Several studies on the association of the different cytokine gene polymorphisms and RPL were published in Iran. In one study considering the role of -174 C/G (IL-6), -197G/A (IL-17) and \[-592A/C, -819C/T, -1082A/G (IL-10)\] in the pathogenesis of the RPL, no significant differences was found between RPL cases and controls ([@B32]). In the study by Nematollahi et al the effectiveness of the -964 A\>G polymorphism in IL-27 gene in the RPL etiology was evaluated. They found no significant relationship between this polymorphism and the occurrence of the RPL in Iranian women ([@B33]). The results of the association assessment between the polymorphisms of the IL-17A (rs2275913) and the IL-17F (rs 763780) genes with the RPL in Iranian women suggested the IL-17 F as a risk factor for RPL ([@B34]). Daher et al reviewed all five available studies in Europe to explore the role of IL-6 174 polymorphism on the RPL incidence ([@B24]). No differences between RPL patients and controls were detected concerning IL-6 genotype frequencies in any of the studies, even in the Japanese population not a single polymorphic allele was detected neither in patients nor in the control group ([@B24]). Since the results influenced strongly by ethnicity, they were not comprehensive and could not apply to all populations. The Caucasian Northern Ireland group is the only one population with a lower IL-6 174 G allele frequency relative to other populations ([@B35]). As compared to the Japanese population, in the study of women from Africa, complete absence of homozygous CC genotype was seen ([@B36]). The other common regulatory polymorphism in the promoter of the gene encoding for IL-6 suggested to affect the expression of the gene is cytosine (C) to guanine (G) substitution -634 base pair downstream of the start point. The only study assessed the association between -634 C/G polymorphism and RPL was in Japan, which is resulted in the significant differences in genotype frequencies (CC vs. GG+GC) between women with recurrent pregnancy loss and the control group. Statistical analysis shows the protective role of the carriers of G allele in RPL susceptibility (OR=0.46, 95% CI:0.24-0.91) ([@B31]). To our knowledge, this is the first study of the association between IL-6 -634 C/G polymorphism and RPL in Iran, and the second one after the data published in Japan. Based on the anti-inflammatory role of the IL- 6 in reproduction and due to lower IL-6 expression, the polymorphic GG genotype might be observed in the women with RPL. This assumption confirmed by the results showing the higher frequency of GG+GC vs. CC genotypes among patients in comparison to control group (p=0.04, OR=5.1, 95%CI: 1.04-25.3). Therefore, the G allele of this polymorphic site could be considered as a risk factor for RPL among women from Iran. Whether the G allele of this polymorphic site reduces the IL-6 gene expression or not should be studied later. Our results are inconsistent with data published from the Japanese experiment, as they introduce the G allele as a protective factor against RPL (OR=0.46) ([@B31]). The number of live births among women with the history of RPL was significantly lower than healthy women (p\<0.001). This data also highlights the effectiveness and the crucial role of normal IL-6 expression in the survival of pregnancy outcome, which reduced at the presence of mutant allele G of -634 SNP. Conclusion ========== In summary, the present study suggested that polymorphism of the IL-6 gene (-634C/G) is associated with increased risk for RPL in Iranian women. Considering that the IL-6 GG genotype is associated with increased risk for RPL in Iran but not in Japanese population, most epidemiologic studies are necessary to further ascertain the relationship between IL-6 promoter polymorphism (-634C/G) and genetic predisposition to RPL ([@B31]). The authors appreciate specialist Dr. Minoo Zolghadri (Gynecologist) and her staffs for collaborating in sampling. We also thank Mohammad Moghadam for monitoring the experimental part of the project. The data provide from the thesis of the Z. Rasti, MSc student of developmental biology in Islamic Azad University, Arsanjan Branch. Conflict of interest ==================== The authors have no conflict of interest.	{ "pile_set_name": "PubMed Central" }
1. Introduction =============== Plants resist herbivorous insects through a combination of constitutive or induced defenses that are generally manifested through poor feeding, abnormal development, low fecundity or even mortality. Various molecular and biochemical approaches can be used to determine the role of constitutive or induced plant defense responses against herbivory \[[@b1-ijms-14-03921]--[@b3-ijms-14-03921]\]. These approaches are equally useful to reveal complex plant-insect interactions that may assist in identification of candidate genes involved in plant defense response \[[@b4-ijms-14-03921],[@b5-ijms-14-03921]\]. Rice is susceptible to a number of insect pests that affect its yield and quality; consequently, several modern rice varieties have so far selectively been developed with resistance to insect pests \[[@b6-ijms-14-03921]\]. Resistant varieties differ considerably in their responses to guard against pests particularly due to the presence of resistant (R) genes. For instance, rice varieties may be bred with R genes for resistance to stem borers, planthoppers or a combination of genes for resistance against multiple pests. Nevertheless, the induction of plant defense mechanisms that includes the production of nutritional and defensive proteins, phenolic compounds or protease-inhibitors and so will strongly contribute towards protecting the plants against insect damage \[[@b4-ijms-14-03921],[@b7-ijms-14-03921],[@b8-ijms-14-03921]\]. Although the presence of R genes potentiates rice defense mechanisms against herbivores, the role of other non-R gene like mechanisms and their mutual interaction with R genes during herbivory cannot be excluded \[[@b6-ijms-14-03921]--[@b9-ijms-14-03921]\]. Broadly speaking, the overall resistance to insect infestation will be a cumulative response of different cellular processes in the plant, including input of R and non-R genes that may be interacting particularly during stress to help the plant express their defense response. Elucidating the complex phenomena of rice defense is will be important to plan rice resistance strategies for existing and emerging pests. The brown planthopper (BPH), Nilaparvata lugens Stål (Hemiptera: Delphacidae), is a secondary pest of rice and causes significant economic loss to susceptible rice cultivars \[[@b10-ijms-14-03921],[@b11-ijms-14-03921]\]. Continuous feeding by BPH populations for several days on rice in the field may lead to hopperburn, a condition resulting from wilting of tillers \[[@b9-ijms-14-03921]\]. Growing resistant varieties of rice is considered the most effective and environment friendly way to control the BPH. So far, more than 20 rice genes and quantitative trait loci (QTLs) have been identified and introduced to various cultivars through breeding in order to confer BPH resistance \[[@b11-ijms-14-03921],[@b12-ijms-14-03921]\]. Rice resistance through the introduction of QTLs has been shown to be effective against BPH \[[@b13-ijms-14-03921]\]. However, due to the genetic complexity between resistant rice cultivars, it has been difficult to explain the function QTLs play in the resistance mechanisms against BPH that further hinders the performance of resistance cultivars in different environments. Expression analysis of global genes and proteins is one strategy to understand molecular responses of rice plants during BPH stress to elucidate how different genes and proteins involve and interact during defense activities and help their selection for use in breeding rice resistance against BPH. Rice defense against BPH has been well documented and the factors involved in rice resistance against BPH are usually associated with the differential regulation of genes and proteins during infestation \[[@b7-ijms-14-03921],[@b10-ijms-14-03921],[@b11-ijms-14-03921],[@b14-ijms-14-03921],[@b15-ijms-14-03921]\]. Many studies revealed physiological and metabolic changes in rice plants during BPH feeding \[[@b4-ijms-14-03921],[@b7-ijms-14-03921]--[@b11-ijms-14-03921]\]. Such alterations in rice plant with BPH infestation also accompany transcriptional activation or repression of plant genes and reorganization of the gene expression profile during stress \[[@b7-ijms-14-03921],[@b8-ijms-14-03921],[@b14-ijms-14-03921]\]. It seems that not only the genes associated with cell defense are induced by BPH, genes that are involved in plant metabolism are also altered possibly through reallocation of necessary metabolites required for growth, reproduction, and storage towards defense activities instead \[[@b11-ijms-14-03921]\]. In this process, the genes associated with abiotic stress, pathogen stress and signaling pathways are reduced, whereas photosynthesis and defense related genes are increased \[[@b7-ijms-14-03921],[@b8-ijms-14-03921],[@b14-ijms-14-03921]\]. Extensive expression analysis of genes and proteins has facilitated the identification of several distinct genes affected by BPH feeding in rice that helped to differentiate susceptible vs. resistant rice cultivars \[[@b9-ijms-14-03921],[@b11-ijms-14-03921],[@b15-ijms-14-03921]--[@b17-ijms-14-03921]\]. For example, 160 unique genes were identified that responded to BPH infestation \[[@b15-ijms-14-03921]\]. Similarly, proteomics approach differentiated a susceptible line from a resistant line carrying a resistance gene BPH15 and identified additional eight genes differentially expressed in rice with BPH infestation \[[@b9-ijms-14-03921]\]. Advances in these tools and the ability to differentiate plant reaction to BPH stress suggests for a significant role expression analysis can play in developing rice resistance to BPH. Mutational approach can play significant role in identifying proteins involved in rice response under specific physiological conditions such as abiotic and biotic stress \[[@b18-ijms-14-03921]\]. A comparative proteome analysis involving wild type rice and the mutants revealed contrasting differences in proteins induced in contrasting genotypes \[[@b19-ijms-14-03921],[@b20-ijms-14-03921]\]. Rice blast lesion mimic mutant (blm) was differentiated from wild type plants based on pathogenesis-related class 5 and 10 proteins including a novel OsPR10d protein specific to the mutants' response. This study also reported increase in phytoalexins and oxidative stress related marker proteins in blm mutant \[[@b20-ijms-14-03921]\]. In another study, more than 150 protein spots were identified as differentially regulated between normal leaves of wild type and spotted leaves of the spl6 rice mutant, indicating the potential of proteomics to elucidate molecular response of rice \[[@b21-ijms-14-03921]\]. Proteomics of rice mutants, will certainly help to elucidate different proteins potentially involved in rice interaction with BPH and explain rice defense strategies against biotic stress \[[@b22-ijms-14-03921]\] This approach could be useful to explore QTL dependent resistance in rice cultivars such as IR64 and its mutants. IR64 is a modern rice variety developed at International Rice Research Institute (IRRI) that carries the major gene Bph1 and other minor genes located in a QTL responsible for resistance to BPH. The durable nature of BPH resistance in IR64 is thought to be due to synergy with minor genes, which contribute to a combined resistance through the mechanisms of antixenosis, antibiosis and tolerance \[[@b13-ijms-14-03921]\]. The mutants of this cultivar have been developed at IRRI \[[@b23-ijms-14-03921]\] and used for elucidating various physiological responses of rice. The objective of the present study is to describe the proteomic responses of indica rice IR64 and two of its chemically generated mutants, one resistant and one susceptible to BPH infestation. Previous study with these IR64 mutants found no growth or yield penalty under normal field conditions \[[@b23-ijms-14-03921]\]. The contrasting phenotypes expressed by mutants that are essentially near-isogenic offer an opportunity to perform genetic analysis in response to BPH infestation and identify specific genes or proteins related to rice resistance. We performed a time-series analysis of gradual BPH stress on IR64 to identify BPH induced proteins. These proteins were further compared between wild type IR64 and the mutants to explain potential role of differentially altered proteins with BPH infestation. 2. Results ========== 2.1. Rice Phenotype during BPH Stress ------------------------------------- Using a modified seedbox screening technique \[[@b13-ijms-14-03921]\] ten-day-old seedlings were uniformly infested with 3--4 second-instar BPH nymphs with free choice to settle on their preferred host. Hopperburn symptoms were observed at different intervals ([Table 1](#t1-ijms-14-03921){ref-type="table"}). Following infestation, continuous feeding by growing second generation BPH nymphs caused wilting of the seedlings, leading to hopperburn (browning of stem and leaves) symptoms first on D1131, followed by IR64 and finally on D518 ([Figure 1](#f1-ijms-14-03921){ref-type="fig"}). Early on infestation (T1 and T2), damage symptoms were not detected on infested plants. This is likely due to a low number of nymphs that were initially released on plants, which did not cause enough damage and plants were able to overcome low level of insect stress. The difference in phenotype among the mutants and IR64 was more obvious at T3 and T4 (28 DAI and 34 DAI, respectively). The average leaf damage rate was recorded on a modified 1--9 scale (1 = resistant, 9 = highly susceptible) \[[@b23-ijms-14-03921]\]. Leaf damage at T3 was lowest for D518 (3.5), intermediate for IR64 (5.2), and highest for D1131 (6.8). 2.2. Proteome Analysis of BPH Induced Proteins in IR64 ------------------------------------------------------ The proteome response of wild type IR64 during BPH infestation over 5-week period after infestation was first studied. This is a condition that simulates natural infestation on rice under field conditions. Among 1500 protein spots visualized on silver stained 2-D polyacrylamide gel (3--10 pH), 65 protein spots were found altered (p \< 0.001) with BPH infestation ([Figure 2](#f2-ijms-14-03921){ref-type="fig"}) at pI 4--7, whereas the remaining spots were detected with pI \> 7.0 (figure not shown). Mixed models ANOVA using BPH induced proteins in the control and BPH infested IR64 treatments shows that a larger cohort of these proteins was changed only during T3 and T4 stage, indicating higher stress response at the later stage ([Figure 3](#f3-ijms-14-03921){ref-type="fig"}). Since the effect of BPH stress was more evident at T3 (28 DAI), we compared the protein abundance at T3 in isolation using control and BPH infested plants. Comparison of protein abundance (spot volume of infested/control at T3 showed that a total of 36 proteins increased \>1.5 fold while 29 proteins showed \<0.5 fold decrease with BPH infestation. The protein abundance showed a reduction through time as the plants entered senescence at T4 (34 DAI). Based on matrix assisted laser desorption ionization time-of-flight (MALDI-TOF) and quadrupole time-of-flight (Q-TOF) mass spectrometry, the identity of 52 proteins was generated; 27 proteins with increased abundance and 25 proteins with decreased abundance ([Table 2](#t2-ijms-14-03921){ref-type="table"}). Peptide mass of the remaining 13 of total 65 protein spots did not match with any known proteins in the NCBI protein database. These BPH responsive proteins were classified into 11 functional categories \[[@b24-ijms-14-03921]\] of which 39% belonged to energy category, whereas 16% were stress and plant defense related. The identity and function of 20% of BPH responsive protein spots in IR64 are not known. In general, the dominating category of BPH affected the functional group involved photosynthesis and metabolism related proteins. BPH induced proteins related to photosynthetic processes were identified as Rubisco activase (Ract), various rubisco large subunits, ferredoxin \[(flavodoxin-NADP(H)\] reductase (FNR) and oxygen evolving enhancer protein 3 (OEE3) in IR64. This indicates that photosynthesis was one of the common responses to BPH infestation. Likewise, oxidative stress response proteins such as ascorbate peroxidase (APX), GSH dependent dehydro-ascorbate reductase, and CuZn superoxide dismutase (SOD) were identified as BPH stress response proteins in IR64. Abundance of multiple spots of ribulose bisphosphate carboxylase large (rubisco, rbcl) subunits (4 spots), ascorbate peroxidase (APX) (5 spots), unnamed protein (2 spots), oxygen evolving enhancer protein 3 (2 spots), and enolase (2 spots) may represent post translational modifications during BPH stress or presence of multiple gene copies of these proteins in rice. Abundance of several oxidative stress-response proteins, drought (\#LD7) and two salt stress (\#23 and \#27) response proteins was altered with BPH stress as observed at T3 ([Figure 4](#f4-ijms-14-03921){ref-type="fig"}). Repeated measures analysis with individual spot abundance in control and BPH infested plants indicated that the spots \#13, \#14 and \#28 were consistently increased (p \< 0.05) with BPH stress over time whereas spots \#12, \#21, \#23, \#49 and \#LD7 showed significant decrease as compared to the control (p \< 0.05) over time ([Figure 4](#f4-ijms-14-03921){ref-type="fig"}). Although the protein "\#LD7" (S-like RNase) was less changed with BPH infestation as compared to the control plants, protein levels increased through time during infestation (p \< 0.05). The abundance of protein spots \#23 and \#27, which showed similarity to salt stress root protein "RS1" (Gi34904362) \[[@b25-ijms-14-03921]\], also changed differentially with BPH infestation at different times, particularly at T3 and T4 (p \< 0.05). At T3, the abundance of protein \#23 decreased \> 2 times (p \< 0.05) than in control plants, while the protein spot \#27 which remained suppressed in control plants, was however more abundant with infestation through all four time points ([Figure 4](#f4-ijms-14-03921){ref-type="fig"}). 2.3. Rice Proteins Induced in BPH Infested Plants ------------------------------------------------- The abundance of 16 protein spots (spot \#20, \#32, \#38, \#39, \#−39a, \#40, \#43, \#45, \#47, \#50, \#53, \#57, \#59--61 and \#64) was observed ([Figure 5](#f5-ijms-14-03921){ref-type="fig"}) at different time points only in BPH infested plants. Interestingly, a change in the protein levels of the spot \#20 (proteophosphoglycan, PPG), spot \#50 and spot \#64 (EFTu1) was also observed at T1 and or T2 indicating that these proteins accumulate in IR64 during early BPH-induced stress ([Figure 5](#f5-ijms-14-03921){ref-type="fig"}). Induction of proteophosphoglycan (\#20), putative 1,4-benzoquinone reductase (\#47), Putative defective chloroplasts and leaves (DCL) protein (\#53), Putative FH protein NFH2 (\#59), hypothetical protein P0677B10.12 (\#60), putative oxygen evolving enhancer protein 3-1, chloroplast precursor (\#61) and chloroplast translation elongation factor Tu1 (\#64) have not been reported earlier in BPH-rice interactions and may have role in rice resistance to BPH infestation. The highest levels of these proteins was observed with spot \#64 (spot density = 12.58 ± 1.52) at T3 as compared to the abundance of other proteins whereas the spot \#39 (0.20 ± 0.06) was least induced with BPH infestation. The abundance of all these BPH induced proteins, except spots \#32 (enolase), \#43 (unknown), and \#47 (putative 1,4-benzoquinone reductase) showed declining trend at T4 as the plants started to senesce. A few proteins identified in this study were also non-rice proteins (\#38, \#39, \#39a and \#68). Spot \#38 was identified as "ATP-dependent DNA helicase UvrD (Shewanella denitrificans OS217)". Also \#39a with molecular weight of 97.5 kDa showed similarities to leech derived protease inhibitor protein (LDPI) and \#39 showed a similarity with "Vitellogenin" from BPH. Spot \#68 matched to "Succinyl-CoA ligase \[ADP-forming\] subunit beta OS = Mesorhizobium sp. (strain BNC1)". These proteins could be either BPH associated proteins injected into rice sheath during feeding or environmental contaminants that colonized BPH wounded rice plants. 2.4. Comparative Proteomics of IR64 and Mutants ----------------------------------------------- To understand the defense response of rice against BPH infestation, the protein levels in control and BPH infested IR64 were compared with gain (D518) and loss of resistance (D1131) mutants of IR64 at T3. These mutants were previously identified during a screening of chemically generated IR64 mutants against BPH using a modified seedbox screening technique \[[@b23-ijms-14-03921]\]. Field performance of these mutants did not show compromise in agronomical traits due to mutations. By comparing the protein abundance (protein volume in BPH infested/control) between IR64 and the mutants, 22 proteins were identified that showed differential abundance ([Table 3](#t3-ijms-14-03921){ref-type="table"}). Ten proteins were altered in a unique manner in the susceptible mutant (D1131) when compared to IR64 and resistant mutant (D518). Among these proteins, eight proteins (spot \#7, \#43, \#45, \#47, \#53, \#57, \#59, \#B) were significantly increased (p \< 0.05) whereas two proteins (\#21 and \#32) were highly decreased in D1131 (p \< 0.05) than in D518 and IR64. The protein \#27 generally increased with BPH stress, however showed little change in D1131 whereas comparatively, the abundance of this protein was in greater quantities in IR64 and D518 following BPH infestation (p = 0.018). In contrast, twelve proteins were linked to a D518 related response to BPH ([Table 3](#t3-ijms-14-03921){ref-type="table"}). Three proteins (\#35; \#38 and \#40) were significantly (p \< 0.05) reduced in D518 during BPH stress; another five proteins (\#9, \#21; \#29, \#30 and \#31) were least affected in D518 whereas the same proteins were decreased in IR64 and D1131 (p \< 0.05). Similarly, two proteins (\#27 and \#LD7) showed higher levels (p \< 0.05) in D518 as compared to IR64 and D1131 The abundance of protein \#64 was higher than D1131 but this difference was not significant than IR64. Two proteins (\#8 and \#41) though increased in abundance, but to a lesser extent (p \< 0.05) in D518 compared to IR64 and D1131. The abundance of spot "LD7" exceptionally increased in D518 but reduced in IR64 and D1131 with BPH stress. When compared over time after BPH infestation, the protein spot \#LD7 remained unchanged at T1 and T2, increased to greater quantities at T3 and decreased thereafter at T4 in D518. From the biplot analysis, it is clear that the variation in the levels of specific proteins was associated with specific factors. For example, the variation in the abundance of proteins \#12, \#29, \#23, \#35, \#48 and \#49 were associated with the control. Furthermore, all proteins whose eigenvectors are travelling in the same direction as the thick eigenvectors, are associated with that factor. Likewise, \#LD7 was associated with D518 and to a lesser extent IR64. Several proteins including \#64, \#28, \#13, \#32 were associated with the "BPH infested" treatment. Variability in the protein 11 (unknown protein) was the only one clearly associated with D1130 ([Figure 6a](#f6-ijms-14-03921){ref-type="fig"}). Similarly, broader random experimental factors can be included to evaluate responses to covariates ([Figure 6b](#f6-ijms-14-03921){ref-type="fig"}). Variation in the abundance of proteins located on the right side of the biplot ([Figure 6b](#f6-ijms-14-03921){ref-type="fig"}) indicates that change in protein levels was associated with the progression of "time" and the presence of "BPH infestation", whereas proteins on the left side of the biplot were associated with the lack of treatment or control as well as the earlier time points. Interestingly, levels of protein \#a (unknown protein) was strongly associated with D518 and conversely, variation in the induction of protein \#4 (Putative ribosomal protein s12) was associated to a lesser degree with D1131 and to a greater degree BPH infestation. IR64 did not explain a significant proportion of variation in the protein abundance data. 3. Discussion ============= Rice resistance to brown planthopper (BPH) is intricate involving genetically controlled defense mechanisms. Despite the existing knowledge of a large collection of rice genes, the molecular response involved in rice stress physiology particularly during interactions with BPH remained elusive. Mutants are valuable source of genetic diversity for gene discovery that could provide valuable information to explain plant defense mechanisms \[[@b18-ijms-14-03921]--[@b20-ijms-14-03921],[@b26-ijms-14-03921]\]. We used mutants of the indica rice IR64 that differ in their response to BPH infestation to facilitate the understanding of rice resistance mechanisms to this economically important pest of rice. The time dependent differential change in the levels of BPH response proteins in rice helped to discriminate wild type with the mutants and revealed candidate proteins involved in plant resistance against BPH infestation. Initially, the response of wild type IR64 was determined during BPH infestation, and proteins related to various functional categories were identified in BPH infested IR64; nevertheless photosynthesis, metabolism, and oxidative stress related proteins were predominantly altered ([Table 2](#t2-ijms-14-03921){ref-type="table"}). It has been reported that BPH infestation reduces photosynthetic activity in rice due to excessive loss of plant assimilates, decreased leaf area and wilting \[[@b11-ijms-14-03921],[@b27-ijms-14-03921]\]. Phloem feeding insects are generally known to alter the expression of genes required for photosynthesis \[[@b14-ijms-14-03921],[@b28-ijms-14-03921]\]. However, the role of housekeeping proteins such as those related to photosynthesis cannot be ruled out in defense against insects as housekeeping genes could shift their role towards defense metabolism to manage the increased energy demands during stress \[[@b29-ijms-14-03921],[@b30-ijms-14-03921]\]. For instance, photosynthesis-related genes altered during plant-insect interaction contributed towards defense needs while protecting the basic photosynthetic capacity \[[@b29-ijms-14-03921],[@b30-ijms-14-03921]\]. We also found a number of Rubisco large subunit fragments (RLSU) with BPH infestation. Similar observations have been reported with abiotic and biotic stresses in rice \[[@b31-ijms-14-03921],[@b32-ijms-14-03921]\]. Presence of several Rubisco large fragments (rbcl) with various experimental molecular weights and pIs could also be due to oxidative stress induced fragmentation of the major Rubisco protein which is an abundant source of macronutrients such as nitrogen in senescing leaves \[[@b31-ijms-14-03921],[@b33-ijms-14-03921],[@b34-ijms-14-03921]\]. This supply of nitrogen during stress might serve as fuel for metabolic processes increased during BPH feeding stress. We also observed changes in the levels of several antioxidant proteins that are known to scavenge excessive reactive oxygen species generated under stress \[[@b9-ijms-14-03921],[@b31-ijms-14-03921],[@b35-ijms-14-03921],[@b36-ijms-14-03921]\]. Some of these oxidative enzymes can be antinutritive to insects \[[@b37-ijms-14-03921],[@b38-ijms-14-03921]\]. Increased levels of oxidative enzyme activity might have adverse effect on the BPH performance thus helping to reduce damage. Similarly, generation of ROS can also act as stress signals to induce defense related genes during insect infestation \[[@b39-ijms-14-03921]\]. Few ascorbate peroxidase (APX) isoforms were found to be induced as early as 13 DAI ([Figure 4](#f4-ijms-14-03921){ref-type="fig"}), indicating their primary importance during BPH infestation and implication in defense signaling. Moreover, we observed differential levels of APX related proteins in BPH infested IR64 as three of the APXs were increased whereas two were decreased during the infestation which is in agreement with previous studies on differentially induced ascorbate peroxidase isozymes during oxidative stress \[[@b40-ijms-14-03921]\]. Induction of proteins during stress is important in dealing with the stress-induced metabolic homeostasis through readjusting metabolic pathways and reallocation of plants' resources for defense \[[@b41-ijms-14-03921]--[@b43-ijms-14-03921]\]. During such response, proteins may be reduced or increased in activity as evidenced in this study. We observed 64 proteins induced with BPH infestation and 52 of these were identified ([Table 2](#t2-ijms-14-03921){ref-type="table"}), some of these might have role in higher energy demands during stress. This seems plausible as many of these proteins ([Table 2](#t2-ijms-14-03921){ref-type="table"}, [Figures 4](#f4-ijms-14-03921){ref-type="fig"} and [5](#f5-ijms-14-03921){ref-type="fig"}), except for few non-rice proteins (\#20, \#38, \#39, \#39a, \#40, \#41, \#68), are plant stress response proteins. These induced proteins could be by-products of stress metabolism or post translation modification but may also represent molecules needed in signal transduction or acclimation response of plants during stress \[[@b42-ijms-14-03921]\]. Fifteen proteins ([Figure 5](#f5-ijms-14-03921){ref-type="fig"}) were observed only in BPH infested plants whereas these proteins were absent in controls. BPH induced proteins, some of which are still unknown, are potentially involved in rice defense during BPH stress. Induction of several other proteins (\#23, \#27 and \#LD7) during BPH stress showed rice response similar to that observed in abiotic stress such as drought and salinity \[[@b25-ijms-14-03921]\]. Excessive loss of phloem sap and impaired water movement during BPH infestation leads to wilting like condition "hopperburn" which is the susceptible response of rice to BPH \[[@b9-ijms-14-03921],[@b11-ijms-14-03921]\]. Phloem feeding insects generally reduce foliar water potential in plants as a result of extensive feeding and results in the induction of transcripts associated with water stress \[[@b28-ijms-14-03921],[@b44-ijms-14-03921],[@b45-ijms-14-03921]\]. Any counter activity such as altered levels of abiotic stress related proteins that could to delay wilting may help to overcome BPH stress. Up-regulation of drought induced S-like RNase and salt stress induced proteins in BPH infested rice points the need for exploring these proteins in rice defense response to BPH stress. Comparative analysis was performed to differentiate the proteome response of mutants from the IR64. Defensive response of mutants was demonstrated by differential pattern of proteins induced with BPH infestation. For example, abundance of stress induced glyoxalase I, known with plant defense activity \[[@b46-ijms-14-03921]\], was reduced in D1131 and IR64 but not to the same extent in D518 ([Table 3](#t3-ijms-14-03921){ref-type="table"}). A similar response was evident with GSH-dependent dehydro ascorbate reductase in D518. The protein EFTu1, similar to 45- kDa heat shock proteins with chaperone like activity \[[@b47-ijms-14-03921],[@b48-ijms-14-03921]\], was induced earlier (T2) and more intensely in D518 and IR64 (S [Figure 1](#f1-ijms-14-03921){ref-type="fig"}) and its abundance was greater in D518 followed by IR64 and then D1131. EFTu1 has been reported as an important component of thermo-tolerance in maize and other environmental stresses \[[@b48-ijms-14-03921]\]. Another two proteins, S-like RNase and spot \#27 were also more abundant in D518 in contrast to moderate levels of these proteins in IR64 and susceptible mutant D1131 ([Table 3](#t3-ijms-14-03921){ref-type="table"}). Higher levels of these proteins in D518 could be important in providing defense to D518 against increasing BPH stress. Similarly, abundance of certain proteins was highly reduced in D518 during BPH infestation whereas the decrease in protein levels was slow in IR64 and D1131 suggesting for higher metabolic shift or adjustment of metabolic pathways in the resistant mutant. On the contrary, some proteins were in greater quantities in D1131 than IR64 and D518 and may represent a susceptible response during BPH infestation ([Figure S1](#s1-ijms-14-03921){ref-type="supplementary-material"}). Several antioxidant enzymes and their isoforms were affected with BPH stress. Differential modulation of antioxidant proteins in a resistant and susceptible rice line infested with BPH was previously reported \[[@b9-ijms-14-03921]\]. However, we could not differentiate IR64 resistance solely from its mutants based on antioxidant proteins such as APX as levels of these proteins were not different. Differential induction of drought induced S-like RNase and salt stress induced proteins (spot \#23 and \#27) suggests for the relationship between rice resistance to BPH and abiotic stress that urges for exploring abiotic stress tolerant varieties against BPH and vice versa. S-like RNase genes constitute an important family of RNA-degrading enzymes that have been associated with phosphate starvation, ethylene responses, senescence and programmed cell death and defense against multiple stresses \[[@b25-ijms-14-03921],[@b49-ijms-14-03921]--[@b51-ijms-14-03921]\]. Sticky digestive liquid from a carnivorous plant, Drosera adelae, contained an abundant amount of S-like RNase which assists plants to obtain phosphates from trapped insects which help to defend them against microbes \[[@b52-ijms-14-03921]\]. Induced S-like RNase has shown to prevent the growth of fungal hypha in tobacco \[[@b53-ijms-14-03921]\]. It is likely that increased abundance of S-like RNase may play a role to protect the resistant cultivar D518 from BPH perhaps by inhibiting stylet or ovipositor movement in phloem sheath and reduced settling, feeding and egg laying has previously been observed \[[@b23-ijms-14-03921]\]. Further studies in this area will elucidate mechanisms that S-like RNase and other proteins might play in rice resistance to BPH. One option is to investigate the interaction of BPH induced rice proteins with in silico structure analysis and molecular docking (to reveal complexity of rice response to BPH stress particularly for possible links to phosphate (Pi) starvation, plant-microbe interaction and drought. Further experiments with in silico and transgenic approach will help to elucidate the precise role of BPH induced proteins in rice defense to BPH. 4. Experimental Section ======================= 4.1. Insect Culture and Plant Material -------------------------------------- Brown planthopper (BPH), Nilaparvata lugens (Stal) populations were continuously maintained on the susceptible variety "Taichung Native 1" (TN1) at the International Rice Research Institute (IRRI), Los Baños, Philippines. The parent BPH population was collected from rice fields around IRRI, Laguna. Gravid females were used to get a synchronized hopper stage for infestation. The Indica rice cultivar IR64 along with its two mutants, i.e., D518 (gain-of resistance) and D1131 (loss-of-resistance) generated through the chemical and radioactive mutagenesis of IR64 \[[@b23-ijms-14-03921]\] were used for this study. The mutant D518 shows enhanced resistance during BPH infestation whereas D1131 is susceptible. The mutants were used following six generations of selfing and after confirmed field evaluation showing absence of any deleterious effect of mutations. The field trials of these mutants revealed no agronomical differences from IR64 \[[@b23-ijms-14-03921]\] whereas analysis using IR64 specific molecular markers suggested that the mutants are essentially near-isogenic (unpublished data). The experimental plants were maintained under greenhouse conditions at 28 ± 2 °C with a photoperiod of 16 h day/8 h night cycle. 4.2. Plant Phenotype to BPH Infestation --------------------------------------- Phenotypic response of IR64 to BPH infestation was determined using a modified seedbox screening technique under greenhouse conditions \[[@b13-ijms-14-03921]\]. This technique provides free choice to BPH nymphs to colonize the plants in the seedbox. Briefly, pre-germinated seeds were sown in seedboxes (45 cm × 35 cm ×15 cm) containing heat-sterilized soil in six equally spaced rows (two rows for each entry) and 15 seedlings per row. Each row (mutant or wild type) was randomized within a seedbox and replicated in three independent seedboxes. Ten-day-old seedlings were uniformly infested with 3--4 second-instar BPH nymphs per plant and allowed to settle on plants of their choice. Hopperburn symptoms were observed 34 days after infestation (DAI). 4.3. Proteomics Response after BPH Infestation ---------------------------------------------- Since phenotypic response of IR64 differed with two mutants, a no-choice setup was planned to allow equal number of BPH stress to feed on these genotypes. Fifteen seeds of mutants or wild type plants were sown in individual nine inch circular pots using three technical and three biological replicates. The seedlings were maintained in the greenhouse and before infestation with three nymphs per plant 10 days after sowing, pots were randomized between entries and covered with mylar cage and infested. Control plants were not infested but were covered with mylar cage and arranged randomly. For protein extraction, the plants from three experimental and biological replicates were sampled at four time points after infestation. For the first sample (T1), plant tissue was harvested 2 DAI when the infested nymphs were still in 3^rd^--4^th^ instar stage; the second sampling (T2) was done at 13 DAI when the majority of nymphs were at the adult stage; the third sampling (T3) was performed 28 DAI following the emergence of second generation nymphs; the fourth sampling (T4) was done when the susceptible mutant (D1131) started wilting (34 DAI). For protein analysis, a 10 cm sample above ground portion of leaf sheath was harvested and stored immediately in liquid nitrogen. For control, plants were harvested at same time points using non-infested plants. 4.4. Protein Analysis --------------------- Protein extraction. Total leaf sheath proteins were extracted in a precipitation solution (10% Trichloroacetic acid, 89.93% Acetone, 0.07% Dithiothreitol) using a modified method of Damerval et al. \[[@b54-ijms-14-03921]\]. The protein concentration was determined using a Protein-Assay-Kit (Bio-Rad) following the manufacturer's instructions. Protein separation and image analysis of 2D Gels. Gel electrophoresis was performed using non-linear (NL) 18-cm IPG strips with pH 4--7 and 3--10 (Amersham Pharmacia Biotech, Uppsala, Sweden). The IPG strips were rehydrated overnight in 350 μL of rehydration buffer and 100 μg of sample protein. The isoelectric focusing (IEF) of proteins was performed on a Multiphor II Electrophoresis unit (Amersham Biosciences) at 20 °C with constant 200 V for the first hour, 500 V for next 2 h and finally 16 h at 2950 V. Proteins from DTT/IAA equilibrated IEF strips were separated on 15% sodium dodecyl sulfate (SDS) polyacrylamide gels using a Protean-II Multi cell (Bio Rad: Hercules, CA, USA) at 4 °C. The gels were stained with silver nitrate (Sigma Aldrich) for scanning or spot quantification analysis whereas coomassie blue stain (Sigma Aldrich) was used for protein identification with mass spectrometry using standard staining protocols. The gels were scanned with a GS-800 Calibrated Densitometer (Bio-Rad) at a resolution of 600. For spot detection, protein quantification and spot analysis, Melanie-3 image analysis software (GeneBio, Geneva, Switzerland) was used. Spot detection parameters were as follows: number of smooths, 5; Laplacian threshold, 5; partial threshold, 1; saturation, 90; peakness increase, 100; minimum perimeter, 10. The Melanie software automatically normalized the spot intensity (the relative volume) i.e., the volume divided by the total volume over the whole image (Melanie 3 user manual). The percent spot volume detected by software was used to match spots for intensity differences and predict BPH induced proteins.. The protein spots were categorized as BPH altered (increased or decreased in abundance) if protein abundance in a rice line increased or decreased with BPH infestation compared to mean control value. Abundance ratio (protein volume in infested plants/control plants) was compared with control at a time point to determine fold change in proteins. An arbitrary cutoff was used to express highly altered proteins \[\>1.5 (increased), \<0.5 (decreased) or \>0.5 and \<1.5 (least altered)\]. 4.5. Protein Identification --------------------------- The proteins spots from Coomassie Brilliant Blue (G-250) stained gels were manually excised using a sterilized scalpel and submitted to the Australian Proteome Analysis Facility (APAF) Macquarie University, Sydney, Australia \[[@b55-ijms-14-03921]\] for characterization. Protein samples were analyzed with matrix assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry using a Micromass Tofspec time-of-flight mass spectrometer (Micromass, Manchester, UK) at APAF following standard procedures. If proteins could not be identified with MALDI-TOF, a further analysis was performed on Q-TOF LCMS. For protein identification, peak lists were used and peptide masses were searched against SWISS-PROT and NCBInr databases using the Mascot search engine \[[@b56-ijms-14-03921]\] supported by Matrix Science Ltd., London. In MS/MS Ion Search, following parameters were used for database queries on monoisotopic peptide masses using the Viridiplantae and Oryza sativa as taxonomic categories; peptide mass tolerance of 150 ppm; fragment mass tolerance: ±0.6 Da; variable modifications: Oxidation (M), Propionamide (C); and the maximum number of missed tryptic cleavages, 1. Peptide masses that yielded a significant ion score (p \< 0.05) were considered positively identified. 4.6. Statistical Analysis ------------------------- Data analysis was performed with Statistical Analysis Software (SAS) (Version 9.1) and JMP-IN (Version 5.1) (SAS Institute, Cary, NC, USA) using protein abundance values in control and BPH infested plants of three genotypes (wild type IR64 and two mutants) and compared at each respective time point. Protein abundance ratio in relation to each control group (IR64 or mutants) was calculated by dividing the spot abundance in the BPH infested plants by the mean spot abundance of the control plants and expressed as fold change with statistical significance at p-value lower than 0.05. A 2-way ANOVA was used to compare the protein abundance between IR64 and the mutants and the means were separated with the Tukey's HSD multiple means comparison test (p \< 0.05). Ordination statistics were performed on protein abundance and genotypes to measure interactions between the BPH and rice proteins (Canoco V.4.5) \[[@b57-ijms-14-03921]\]. Initially, detrended correspondence analysis (DCA) was performed to measure eigenvector length of expressed proteins variables (control, infested) \[[@b26-ijms-14-03921]\]. Redundancy analyses (RDA) were performed and the significance of the first two axes, as well all four axes, were tested using a Monte Carlo test with 1000 permutations in reduced space. The reason RDA was chosen in this particular instance rather than another multivariate method, is that the variable data showed linear responses as opposed to unimodal responses. Multivariate biplots allow one to explore trends through numerical data analysis above and beyond simple hypothesis testing. Where relationships and covariation between variables is not evident with simple univariate statistics, multivariate methods clearly show the abundance of specific proteins as variables in relation to experimental factors. In this case it is clear that specific proteins covary with specific treatments, and the treatments themselves also show covariation. 5. Conclusions ============== BPH infestation on rice cv. IR64 altered the induction of several proteins involved in various functional categories. A differential induction in proteins was evident both in resistant and susceptible mutant of IR64. Overall, D518 essentially resists against BPH attack via increased activity of proteins related to metabolism (Glyoxalase I, Probable ATP synthase 24 kDa subunit, Enolase), stress response (S-like RNAse, GSH-dependent dehydro ascorbate reductase, Salt stress root protein "RS1") and protein synthesis (Chloroplast translation elongation factor Tu1) ([Table 3](#t3-ijms-14-03921){ref-type="table"}). Altered abundance of proteins, in particular lower levels of stress related proteins might have role in susceptibility of D1131 ([Table 3](#t3-ijms-14-03921){ref-type="table"}). Moreover, the resistant plant also appears to compensate through a timely induction of some of these proteins thus providing a leading edge over the susceptible plants. Differential response of the mutants to BPH feeding thus leads to altered hopperburn symptoms on the rice plants ([Figure 7](#f7-ijms-14-03921){ref-type="fig"}). The complex plant response to BPH also insists on refocusing the research for rice defense towards other metabolic pathways like photosynthesis and their possible interaction to understand rice resistance mechanisms to BPH infestation and to develop resistance breeding program. Further experiments to explore a defined biological interaction between differentially induced proteins with other housekeeping proteins may explain how resistant mutant would overcome BPH stress than susceptible mutant D1131 or moderately susceptible IR64. The authors would like to thank Swiss Agency for Development and Cooperation (SDC) for Ph.D. support to JSS. We are also grateful to Reyeul Quintana, Rodante Abas, and Alberto Naredo, Carmencita Bernal, and Angelita Romena for their help with the insect colony maintenance and with the greenhouse screening of rice mutants. ###### Supplementary Information (PDF, 243 KB) ###### Click here for additional data file. ![Phenotype of wild type IR64 and mutant plants exposed to brown planthopper (N. lugens) infestation under greenhouse conditions during seedbox screening (free choice). Pre-germinated seeds were sown in the heat sterilized soil in seed boxes a density of 15 seedlings per row. Hopperburn symptoms appeared first on D1131, followed by IR64 and lastly on D518. The experiment was repeated 3 times.](ijms-14-03921f1){#f1-ijms-14-03921} ![2-D gel electrophoresis of IR64 leaf sheath proteins following brown planthopper (N. lugens) infestation (left panel) and control (right panel) condition. Total plant proteins extracted using TCA-Acetone method were separated on 15% SDS PAGE using non linear (NL) 18-cm IPG strips. The gels were stained with silver nitrate for protein detection. The red boxes represent down regulated proteins whereas green boxes represent up regulated proteins after BPH infestation.](ijms-14-03921f2){#f2-ijms-14-03921} ###### Abundance of brown planthopper (N. lugens) responsive proteins in IR64 at different days after BPH infestation (DAI) (T1 = 2 DAI; T2 = 13 DAI; T3 = 28 DAI; T4 = 34 DAI). The figure shows log~2~ values of proteins \[BPH infested (T)/control (C)\] at different time points. (n = 3; p \< 0.05). The protein legends in the figure represent induction response of IR64 proteins (log 2 value) after BPH infestation ![](ijms-14-03921f3a) ![](ijms-14-03921f3b) ![Relative protein abundance of brown planthopper (N. lugens) altered stress- and defense-related proteins in BPH infested and control IR64 at different days after infestation (DAI) (T1 = 2 DAI; T2 = 13 DAI; T3 = 28 DAI; T4 = 34 DAI). The protein abundance was quantified with Melanie3 software. Mixed models ANOVA was used for repeated measures analysis of proteins. Mean ± SE (n = 3).](ijms-14-03921f4){#f4-ijms-14-03921} ![Relative abundance of brown planthopper (N. lugens) responsive proteins in IR64 at different days after BPH infestation (DAI) (T1 = 2 DAI; T2 = 13 DAI; T3 = 28 DAI; T4 = 34 DAI)). Mean ± SE (n = 3).](ijms-14-03921f5){#f5-ijms-14-03921} ![Redundancy analysis (RDA) biplot of protein abundance over the duration of the experiment. All factors are illustrated as thick vectors and include Control, Treatment (BPH infested), Loss of resistance (D1131), Gain of Resistance (D518), Wild type (IR64), and Time. Proteins are illustrated as thin vectors and consist of the proteins levels which are listed as a number as described in [Table 2](#t2-ijms-14-03921){ref-type="table"}. Eigen values (lambda) are 0.324, 0.050, 0.010, and 0.004 using data at T3 (a) and all 4-time points (b) Monte Carlo test (1000 permutations) for all canonical axes: F-ratio = 8.490, P = 0.001.](ijms-14-03921f6){#f6-ijms-14-03921} ![A summarized figure of brown planthopper (N. lugens) induced IR64 proteins. Abundance of various proteins associated with rice resistance is altered following BPH infestation. The resistant lines such as D518 may induce specific genes earlier and more intensely than susceptible lines that interact with other proteins thus leading to their enhanced level of resistance against BPH.](ijms-14-03921f7){#f7-ijms-14-03921} ###### Comparative reaction of IR64 and mutants to brown planthopper (BPH) infestation at different times (T1 = 2 days; T2 = 14 days; T3 = 28 days; T4 = 34 days). The infested plants were observed for BPH feeding damage and rated using a 1--9 scale (1 = Resistant, no damage symptoms; 3 = Slight damage, pale outer leaves; 5 = wilting on 50% leaves, slight stunting; recovery possible if insects removed; 7 = Severe hopperburn, only one or two leaves green, no recovery possible; 9 = Highly susceptible, complete wilting). (n = 15, Mean ± SE). Rice line BPH damage (1--9 scale) ----------- ------------------------- ------------ ------------ ------------ IR64 1.0 ± 0.0 1.6 ± 0.55 3.6 ± 0.55 5.2 ± 0.85 D518 1.0 ± 0.0 1.4 ± 0.48 3.0 ± 0.76 3.6 ± 0.56 D1131 1.0 ± 0.0 1.8 ± 0.59 4.8 ± 0.65 6.8 ± 0.66 ###### List of 52 leaf sheath proteins induced during BPH stress on rice variety IR64. Spot PM (%C) Identity/source Accession Exp. (Theo.) M~r~ Exp. (Theo.) pI Mascot score Fold change P-value ---------------------------------------------------- --------- --------------------------------------------------------------------------------------- -------------- --------------------- ------------------- -------------- ------------- ----------- Energy/pentose phosphate 1 2(4) Rubisco large subunit gi11955 17.2(52.8) 4.5(6.13) 64 \>10 ↑ 0.047 2 1(4) Rubisco large subunit gi476752 17.3(45.1) 4.6(8.4) 104 \>10 ↑ 0.006 3 2(5) Ribulose bisphosphate carboxylase/oxygenase large chain gi11466795 17.2(52.8) 5.1(6.2) 98 4.56 ↑ 0.008 5 3(9) Rubisco large subunit from chromosome 10 chloroplast insertion gi37533338 23.7(52.8) 5.4(6.4) 128 1.53 ↑ 0.130 10 5(13) Rubisco large subunit gi476752 24.9(45.1) 6.1(8.4) 174 \>10 ↑ 0.005 RA[\](#tfn1-ijms-14-03921){ref-type="table-fn"} -- Ribulose bis phosphate carboxylase/oxygenase activase P93431 47(42.07) 5.0(5.0) -- 11.45 ↓ 0.0019 Rb 3(9) Rubisco large subunit gi2734976 34.1(43.7) 6.3 332 3.35 ↑ 0.006 Energy/photosynthesis* 61 1(33) Putative oxygen evolving enhancer protein 3-1 chloroplast precursor gi50938199 18.7(22.9) 9.8(9.8) 114 Ind ↑ 0.0053 34 13 (38) Ferredoxin-NADP (H) oxidoreductase Q6ZFJ3_ORYSA 36.0(40.8) 5.9(7.9) 90 4.21 ↓ 0.0114 63 3(44) Putative oxygen evolving enhancer protein 3-1 chloroplast precursor gi50938199 14.5(22.9) 9.9(9.8) 400 2.57 ↓ 0.0053 Energy/glycolysis 32 2(8) Enolase gi33113259 37.8(47.9) 5.5(5.4) 77 Ind ↑ 0.0009 37 14 (49) Enolase gi780372 39.9(47.9) 6.7(5.4) 104 7.95 ↑ 0.0522 TP 8(25) Triose phosphate isomerase, cytosolic P48494 27.5(27.1) 5.6(5.4) 70 \<10 ↓ \<0.0001 9 6(33) Glyceraldehyde-3- phosphate dehydrogenase, cytosolic G3PC_HORV 37.4(33.2) 6.7(6.2) 258 9.6 ↓ 0.0090 44 6(27) Putative dihydrolipoamide dehydrogenase precursor gi34894800 57.0(52.6) 6.6(7.2) 111 \<10 ↓ 0.0011 35 3(17) Formate dehydrogenase gi51536124 41.2(41.3) 6.6(6.7) 100 \<10 ↓ 0.0005 FB[\](#tfn1-ijms-14-03921){ref-type="table-fn"} -- Fructose bisphosphate aldolase Q40677 37.7(36.4) 5.7(5.8) -- 1.31 ↓ 0.1620 Energy/electron transport* 58 3(34) Putative H(+)− transporting ATP synthase gi50912809 25.5(26.2) 4.3(4.9) 305 3.57 ↓ 0.202 30 8(50) Probable ATP synthase 24 kDa subunit gi50905037 28.1(27.2) 5.5(6.5) 300 4.89 ↓ 0.0007 Plant defense 13 6(32) L-Ascorbate peroxidase 2, cytosolic Oryza sativa subsp. japonica (Rice) APX2_ORYSJ 26.3(27.1) 5.3(5.2) 65 \>10 ↑ 0.0005 14 5(36) Putative ascorbate peroxidase gi50920595 26.2(27.1) 5.2(5.4) 94 3.16 ↑ 0.0604 28 5(39) Ascorbate peroxidase gi50940199 28.0(27.1) 5.5(5.2) 239 5.55 ↑ 0.0007 49 5(36) Putative ascorbate peroxidase gi50920595 22.6(27.1) 6.5(5.4) 71 3.49 ↓ 0.0041 12 5(50) Ascorbate peroxidase gi50940199 29.1(27.1) 5.2(5.2) 419 \<10 ↓ \<0.0001 SOD [\](#tfn1-ijms-14-03921){ref-type="table-fn"} -- Superoxide dismutase P93407 17.7(15.7) 5.8(5.3) -- 1.47 ↓ 0.0652 Stress induced* LD7 3(23) Drought induced S-like RNase protein gi17105171 28.2(28.4) 5.1(5.2) 187 1.6 ↓ 0.1033 27 4(38) Unnamed protein product (Salt stress induced protein) gi34904362 29.1(21.8) 4.9(4.9) 153 Ind ↑ 0.0003 21 [\](#tfn1-ijms-14-03921){ref-type="table-fn"} -- GSH-dependent dehydro ascorbate reductase BAA90672 27.0(27.1) 6.1(5.4) -- 4.03 ↓ 0.0009 23 3(40) Unnamed protein product (Salt stress induced protein) gi34904362 30.3(21.8) 4.9(4.9) 179 4.6 ↓ 0.0029 Protein synthesis* 64 3(23) Chloroplast translation elongation factor Tu1 gi50910077 43.7(50.4) 4.3(6.19) 306 Ind ↑ 0.0012 4 3(45) Putative ribosomal protein s12 gi50934241 17.2(14.8) 5.3(5.3) 290 2.64 ↑ 0.0390 22 8(34) Putative ribosome recycling factor, chloroplast precursor XP_478772.1 26.4(29.7) 6.0(9.3) 66 2.77 ↓ 0.0007 Protein destination and storage CP 6(23) Putative clp protease OS02g42290 30.2(31.9) 5.7(6.7) 70 1.77\--↑ 0.0051 24 3(20) Putative chaperonin 21 precursor gi51091339 27.2(25.4) 4.9(5.9) 60 4.58 ↓ 0.0148 Growth and division 41 4(20) (O65316) Actin (Mesostigma viride) ACT_MESVI 68.3(41.5) 5.8(5.3) 181 7.27 ↑ 0.0097 Secondary metabolism 47 3(21) Putative 1,4-benzoquinone reductase gi34910128 24.7(21.7) 6.3(6.0) 79 Ind ↑ 0.0004 26 4(20) Putative NADPH-dependent mannose 6-phosphate reductase gi50904895 36.3(35.4) 6.2(5.9) 142 \>10 ↓ 0.0106 31 4(23) Glyoxalase I gi16580747 34.0(32.5) 5.5(5.5) 173 9.12 ↓ 0.0004 Miscellaneous 20 9(15) Putative proteophosphoglycan gi50918953 74.0(96.8) 4.4 (10.5) 104 Ind ↑ 0.0001 53 8(41) Putative defective chloroplasts and leaves (DCL) protein Oryza sativa Q6UUF7_ORYSA 30.8(21.3) 6.9(9.0) 69 Ind ↑ 0.04 59 18 (27) Putative FH protein NFH2.-Oryza sativa (japonica cultivar-group) Q8S0F0_ORYSA 13.3(10.2) 4.7(8.9) 66 Ind ↑ 0.0099 60 4(53) Hypothetical protein P0677B10.12 Q67VJ8_ORYSA 52.5(12.5) 5.1(9.3) 68 Ind ↑ 0.0185 69 12 (44) Putative glyceraldehyde-3- phosphate dehydrogenase (Phosphorylating) Oryza sativa gi115459078 37.2(36.5) 7.8(7.68) 94 1.75 ↑ 0.0341 40 9(40) hypothetical protein OsJ_015102 \[Oryza sativa\] gi125591269 70.0(25.3) 6.4(11.0) 66 Ind ↑ 0.0185 B 6(24) hypothetical protein OsJ_012934 gi125589101 32.0(35.2) 6.1(5.3) 76 1.37 ↓ 0.0277 39 5(3) Vitellogenin \[Nilaparvata lugens\] gi342318865 72.5(22.7) 7.9(8.5) 64 Ind ↑ 0.0372 39a 9(12) Chain E, Leech-Derived Tryptase Inhibitor TRYPSIN COMPLEX gi3318722 97.5(23.4) 6.7 240 Ind ↑ 0.0139 17 2(18) Putative DREPP2 protein gi50906969 32.0(24.0) 4.8(4.7) 105 8.47 ↓ 0.0003 62 11 (36) hypothetical protein OsI_021661 Q5Z6P9_ORYSA 47.0(43.0) 4.7(4.7) 109 3.55 ↓ 0.0095 38 19 (26) ATP-dependent DNA helicase UvrD Shewanella denitrificans OS217 Q3P3H8_9GAMM 70.0(81.6) 6.6(5.9) 78 Ind ↑ 0.0050 42 12 (38) Os12g0420200 \[Oryza sativa (japonica cultivar-group)\] gi115488340 68.7(41.5) 6.3(8.5) 135 1.87 ↓ 0.0476 68 8(22) Succinyl-CoA ligase \[ADP-forming\] subunit beta OS = Mesorhizobium sp. (strain BNC1) SUCC_MESSB 35.5(42.2) 7.5(5.0) 74 1.59 ↓ 0.0327 Notes: = Proteins identified by Salekdeh et al. 2002 \[[@b25-ijms-14-03921]\]; PM = Peptides matched; %C = Percent coverage; Exp. = Experimental; Theo. = Theoretical; Mr = molecular weight; pI = isoelectric point; Ind = Proteins induced only in BPH infested plants. ###### Comparative abundance of BPH induced proteins between IR64 and the mutants (D518 and D1131) at 28 DAI (Time 3). Superscript letters indicate significant difference in abundance between IR64 and the mutants. Values with same superscript letters are not different (p \> 0.05), (n = 3, Mean ± SE). Protein Spot D518 D1131 IR64 Prob. \> F ------------------------------------------------------------------------- ------ ---------------------------------------------------------------- ---------------------------------------------------------------- ---------------------------------------------------------------- ---------------- Unknown 7 1.06 ± 0.07 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 2.18 ± 0.30 ^a^,^▲^ 1.17 ± 0.14 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.014 GSH-dependent dehydro ascorbate reductase 21 0.81 ± 0.11 ^a^,^▼^ 0.47 ± 0.12 ^b^,^▼▼^ 0.73 ± 0.01 ^a^,^▼^ 0.059 Enolase 32 0.77 ± 0.19 ^a^,^▼^ 0.31 ± 0.12 ^b^,^▼▼^ 1.24 ± 0.15 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.011 Unknown 43 1.04 ± 0.20 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 1.81 ± 0.11 ^a^,^▲^ 1.10 ± 0.21 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.042 Unknown 45 1.68 ± 0.34 ^a^,^b^,^▲^ 3.46 ± 0.67 ^a^,^▲▲^ 1.58 ± 0.54 ^b^,^▲^ 0.085 Putative 1,4-benzoquinone reductase 47 0.89 ± 0.19 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 1.85 ± 0.20 ^a^,^▲^ 0.62 ± 0.05 ^b^,^▼^ 0.004 Putative defective chloroplasts and leaves (DCL) protein Oryza sativa 53 0.67 ± 0.19 ^b^,^▼^ 1.76 ± 0.17 ^a^,^▲^ 0.72 ± 0.25 ^b^,^▼^ 0.018 Unknown 57 0.51 ± 0.06 ^b^,^▼^ 3.10 ± 1.42 ^a^,^▲^ 0.58 ± 0.13 ^b▼^ 0.008 Putative FH protein NFH2 Oryza sativa (japonica cultivar-group) 59 0.61 ± 0.06 ^a^,^▼^ 1.14 ± 0.10 ^b^ 0.55 ± 0.12 ^b^,^▼^ 0.011 Hypothetical protein OsJ_012934 B 1.18 ± 0.09 ^b^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 1.67 ± 0.10 ^a^,^▲^ 0.74 ± 0.07 ^c^,^▼^ 0.002 S-like Rnase LD7 2.07 ± 0.37 ^a^,^▲^ 0.37 ± 0.07 ^b^,^▼^ 0.62 ± 0.15 ^b^,^▼^ 0.005 Unknown 8 2.58 ± 0.42 ^b^,^▲^ 9.15 ± 1.08 ^a^,^▲▲▲^ 9.71 ± 2.44 ^a^,^▲▲▲^ 0.030 Glyceraldehyde-3-phosphate dehydrogenase, cytosolic 9 0.27--0.05 ^a^,^▼▼^ 0.11 ± 0.01^b▼▼▼^ 0.10 ± 0.02 ^b^,^▼▼▼^ 0.024 Salt stress root protein "RS1" 27 4.09 ± 0.24 ^a^,^▲▲^ 1.95 ± 1.36 ^b^,^▲^ 2.75 ± 0.17 ^a^,^b^,^▲^ 0.018 Unknown 29 0.37 ± 0.05 ^a^,^▼^ 0.19 ± 0.03 ^b^,^▼▼▼^ 0.08 ± 0.02 ^b^,^▼▼▼^ 0.008 Probable ATP synthase 24kDa subunit 30 0.43 ± 0.01 ^a^,^▼^ 0.18 ± 0.02 ^b^,^▼▼▼^ 0.20 ± 0.02 ^b^,^▼▼▼^ 0.061 Glyoxalase I 31 0.94 ± 0.09 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.47 ± 0.07 ^a^,^b^,^▼▼^ 0.33 ± 0.01 ^b^,^▼▼^ 0.087 Formate dehydrogenase 35 0.46 ± 0.09 ^b^,^▼^ 1.28 ± 0.17 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 1.16 ± 0.21 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.015 ATP-dependent DNA helicase UvrD Shewanella denitrificans OS217 38 0.49 ± 0.15^b^,^▼^ 1.39 ± 0.21^a^,^▲^ 1.04 ± 0.32 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.011 Hypothetical protein OsJ_015102 40 0.33 ± 0.07 ^b^,^▼^ 1.20 ± 0.12 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.99 ± 0.29 ^a^,[\](#tfn2-ijms-14-03921){ref-type="table-fn"} 0.041 (O65316) Actin (Mesostigma viride*) 41 1.81 ± 0.36 ^b^,^▲^ 5.94 ± 1.08 ^a^,^▲▲^ 7.27 ± 1.33 ^a^,^▲▲^ 0.021 EFTu1 64 14.05 ± 1.49 ^a^,^▲▲▲^ 7.78 ± 0.83 ^b^,^▲▲^ 12.58 ± 1.52 ^a^,^▲▲▲^ 0.065 No significant change in protein abundance compared to 1 (protein volume in BPH infested/control); (^▲^) Increase in protein abundance; (^▲▲▲^) Highly increased in abundance; (^▼^) Decrease in protein abundance; (^▼▼▼^) Highly decreased in abundance.	{ "pile_set_name": "PubMed Central" }