a, The 2-bp deletion identified in exon 9b of patient 1. b, The T-to-A nonsense mutation in exon 4 of patient 2 (stop codon is underlined). c, Upper panel, intronic mutation in patient 3; lower panel, sequencing of the RT-PCR product in patient 3, showing exon-6 skipping. d, Upper panel, The G-to-A point mutation at the splice-donor site in intron 5 of patient 6; lower panel, sequencing of the RT-PCR product in patient 6, showing a 6-bp insertion at the junction between exons 5 and 6. The inserted sequence (underlined) is derived from the first six nucleotides of intron 5 (underlined), which creates a stop codon (dotted). Most probably, nucleotides 7 and 8, GT (double underlined), in intron 5 are alternatively recognized as a splice-donor site. e, Upper panel, the 10-bp deletion in patient 9. This mutation deletes four nucleotides from intron 5 and six from exon 6. Lower panel, sequence of the RT-PCR product in patient 9 reveals the deletion of the 10 nucleotides at the 5′ end of exon 6. The 'AG' (double underlined) in exon 6 immediately after the genomic DNA deletion appears to be recognized as an alternative splice-acceptor site.

Figure 2 Representation of the mutant LAMP-2b. 'Normal' represents the LAMP-2b amino-acid sequence. Because patient 1, who developed full DD symptoms, has a mutation in exon 9b, the LAMP-2b isoform is most probably responsible for DD. Residue 265, one of the putative disulphide-bond-creating cysteines, and two glycosylation sites are lost by exon-6 skipping in patient 3, which probably results in a significant change in the secondary structure. Black boxes represent the loops created between two cystein residues flanking each of the four loop regions; Y depicts a glycosylation site; hatched boxes show altered amino acids caused by frameshift mutations; dashes represent missing amino acids due to exon-6 skipping in patient 3. All of the other mutant proteins lack the transmembrane domain (grey bar) because of nonsense or frameshift mutations. Thus, those molecules cannot function as lysosomal 'membrane' proteins.

a, Skeletal muscle. Muscle extracts from patients 1-6 and 10 and a control were blotted and labelled with antibodies against LAMP-2 (top panel) and β-actin (bottom panel). LAMP-2 is deficient in all muscles examined except the sample from patient 1 that showed a trace amount of protein, which is probably the LAMP-2a isoform. In support of this idea, LAMP-2b is the principal LAMP-2 species in skeletal muscle6. Numbers denote patients as in Table 1; C, control. b, Heart. Protein was extracted from cardiac muscle of patient 4 and a control, blotted onto a nitrocellulose membrane, and labelled with the antibodies against LAMP-2 (too panel) and β-actin (bottom panel). LAMP-2 is deficient also in cardiac muscle. C, control; 4, patient 4.

Serial transverse sections of the biopsiedmuscle from six DD patients, and samples from a XMEA patient and a normal control, were immunostained with antibodies against LAMP-2 and limp-I. LAMP-2 is absent in all DD muscles (patient 6 shown here), whereas it is present in the XMEA muscle. The antibody against limp-I highlighted vacuoles in both DD and XMEA muscles. LAMP-1 antibody faintly stained both DD and XMEA muscles, suggesting that the expression level of LAMP-1 might be lower in skeletal muscle than LAMP-2 or limp-I (data not shown). The muscle from the patient without a lamp-2 mutation also showed LAMP-2 deficiency on immunohistochemistry similar to genetically proven DD patients with primary LAMP-2 deficiency (data not shown). The immunohistochemical stains of LAMP-2