|Home | About | Journals | Submit | Contact Us | Français|
Chediak–Higashi syndrome is an autosomal recessive, immune deficiency disorder of human (CHS) and mouse (beige, bg) that is characterized by abnormal intracellular protein transport to, and from, the lysosome. Recent reports have described the identification of homologous genes that are mutated in human CHS and bg mice. Here we report the sequences of two major mRNA isoforms of the CHS gene in human and mouse. These isoforms differ both in size and in sequence at the 3′ end of their coding domains, with the smaller isoform (~5.8 kb) arising from incomplete splicing and reading through an intron. These mRNAs also differ in tissue distribution of transcription and in predicted biological properties. Novel mutations were identified within the region of the coding domain common to both isoforms in three CHS patients: C→T transitions that generated stop codons (R50X and Q1029X) were found in two patients, and a novel frame-shift mutation (deletion of nucleotides 3073 and 3074 of the coding domain) was found in a third. Northern blots of lymphoblastoid mRNA from CHS patients revealed loss of the largest transcript (~13.5 kb) in two of seven CHS patients, while the small mRNA was undiminished in abundance. These results suggest that the small isoform alone cannot complement Chediak–Higashi syndrome.
Chediak–Higashi syndrome (CHS) is an autosomal recessive, immune deficiency disease that maps to human chromosome (Chr) 1q42–q43 (1–3). Affected individuals have giant, perinuclear lysosomes, defective granulocyte, natural killer (NK) and cytolytic T-cell function, and die prematurely of infection or malignancy (4–10). CHS patients also exhibit partial oculocutaneous albinism, platelet storage pool deficiency and late neurologic defects such as peripheral neuropathy and ataxia (11–15). Recently, it has been demonstrated that intracellular protein transport to, and from, the lysosome is disordered in CHS (10,16–18). The defects in trafficking of secretory lysosomes of granular cells (leukocytes, melanocytes, megakaryocytes and cerebellar Purkinje cells) provide a unifying hypothesis that explains the diverse clinical features of CHS (19), and suggest that the CHS gene regulates intracellular protein transport to, and from, the lysosome.
As an antecedent to identification of the human CHS gene, two groups positionally cloned the homologous mouse mutation, beige (bg) (20,21). The clinical and pathologic features of CHS and bg are very similar, and bg maps on proximal mouse Chr 13 within a linkage group conserved with human Chr 1q42–q43 (the position of the CHS locus) (1–3,22–24). Subsequently, the human homolog of the bg gene was mapped to Chr 1q43, and found to be mutated in three CHS patients(20,25). There has been some ambiguity, however, in the literature about the exact structure of the mouse Chediak–Higashi syndrome (bg) gene (designated Lyst) (20,21). We reported a novel mouse cDNA from a bg critical region yeast artificial chromosome (YAC) that was mutated in bg mice (20). Simultaneously, another group published a partial cDNA sequence that had been isolated from the same YAC clone (21). The latter was mutated in two other bg alleles, but differed in sequence from the former. In the present report, this discrepancy was explained by the demonstration that each of the previously reported bg gene sequences are derived from a single gene with alternatively spliced mRNAs. We also report the sequences of two major isoforms of the CHS gene in human. These isoforms arise in mouse and human as a result of incomplete mRNA splicing and differ in pattern of expression and predicted biological properties.
The existence of more than one mRNA isoform of the mouse bg gene (designated Lyst) had been suggested both by Northern blot analysis and by Lyst cDNA sequences that differed at their 3′ end (20,24). Mouse Lyst mRNA isoforms were therefore sought by anchored, nested PCR (3′RACE-PCR). Two fragments (1.25 and 2 kb) were amplified from mouse spleen cDNA using this technique. The 1.25 kb clone contained the 3′ end of the previously described 5893 bp Lyst cDNA that corresponds to a small mRNA isoform (Lyst-II) (20). The 5′ end of the 2 kb clone, however, contained sequence derived from the Lyst-II cDNA, while the 3′ end sequence was from the largest isoform of the bg gene (Lyst-I, previously called BG) (24). Reverse transcription and PCR (RT-PCR) confirmed that nucleotides 1–4706 of Lyst are common to both mRNA isoforms (Fig. 1A). The large isoform cDNA that contains the entire coding domain was assembled from nucleotides 1–4706 of Lyst, the 2 kb 3′RACE-PCR clone, and 6824 nucleotides of BG cDNA. This 11 817 bp cDNA sequence (Lyst-I, GenBank accession no. U70015) corresponds to the largest mRNA observed in Northern blots (20). This 11.8 kb cDNA, however, is truncated at both the 5′ end of the 5′-untranslated region (UTR) and at the 3′ end of the 3′ UTR, and thus is smaller than the largest mRNA observed on mouse Northern blots (~12–13 kb) (20).
Analysis of a P1 genomic clone (number 8592) (26) containing the entire bg gene revealed that the 11 817 bp Lyst-I cDNA results when splicing occurs from Lyst exon σ (containing nucleotide 4706) to downstream exon τ (Fig. 1A and B). In contrast, incomplete splicing and reading through intron σ′ interposed between exons σ and τ yields the truncated Lyst-II isoform, of length 5893 bp cDNA (Lyst-II, Fig. 1A and B, GenBank accession No. L77884). Lyst intron σ′ encodes 37 in-frame amino acids followed by a stop codon and a polyadenylation signal. Lyst-II corresponds to one of several smaller mRNAs observed on Northern blots (20). Full-length cDNAs corresponding to the Lyst-I and Lyst-II isoforms were both amplified readily from mouse bone marrow RNA by RT-PCR (data not shown). The putative Lyst-I and Lyst-II proteins are of relative molecular mass 425 287 (Mr 425 kDa) and Mr 172.5 kDa, respectively.
cDNAs corresponding to the human homolog of the largest mRNA isoform of the bg gene were obtained by identification of human expressed sequence tags (ESTs) similar in sequence to mouse Lyst-I by database searches (GenBank accession Nos L77889, W26957 and H51623) (Fig. 1C). Intervening cDNA sequences were isolated using RT-PCR, and a partial human cDNA sequence (GenBank accession No. U70064; 7.1 kb) was assembled by alignment of these clones with the largest mouse bg gene cDNA (Fig. 1C). The predicted human and mouse peptides shared 82% amino acid identity over 1990 amino acids. The predicted human amino acid sequence contains a six amino acid insertion relative to that of mouse at residue 1039. While this paper was being reviewed, another group published the complete sequence of the human CHS cDNA (25). Our partial cDNA sequence differs in four nucleotides and three predicted amino acids from the latter (25). This 13.5 kb cDNA sequence corresponds to the largest mRNA observed on Northern blots of human tissues (CHS-I, legend of Fig. 2) (20). These Northern blots also demonstrated the existence of a smaller human transcript (~4.5 kb) that was similar in size to the small mouse Lyst mRNA, and that appeared to differ in distribution of expression in human tissues from the large isoform (20). Assuming that the genomic derivation of the human small isoform might be the same as the mouse small isoform, the 3′ end of the human small isoform was identified by cloning human intron σ′ using PCR of human genomic DNA with primers derived from exon σ and intron σ′. As in the mouse, the sequence of the 5′ end of human intron σ′ contained 17 codons in-frame with exon σ, followed by a stop codon (CHS-II, legend of Fig. 2). cDNA corresponding to this human short isoform was amplified from human peripheral blood RNA by RT-PCR with primers from a 5′ exon and intron σ′, indicating that this intron was indeed present in CHS-II mRNA (data not shown). Nucleotides 1–5905 of human small and long isoform cDNAs are identical, and are followed either by intron σ′ sequence in the short isoform (GenBank accession No. U84744), or by exon τ, etc. in the long isoform (Fig. 2 legend). The predicted intron-encoded aminotermini of the mouse and human short isoform peptides shared 65% identity.
Additional splicing complexity of small CHS isoforms exists: exon α, and/or exon β, and/or exon γ also can be alternatively spliced, and are absent in small CHS isoform mRNAs of ~3.5, 4.5 and 5.5 kb, that correspond to additional bands observed on human Northern blots (20).
The only significant sequence similarity of the human short isoform to known proteins was with the stathmin family (20). Identity with mouse Lyst-II in this region (amino acids 376–540) was 92% (and similarity was 99%) (Fig. 4).
Analysis of Northern blots of mouse mRNA had suggested that the relative abundance of the mouse large and small transcripts differed from tissue to tissue (20). The relative abundance of mRNA isoforms of the homologous human gene in human tissues at different developmental stages was examined by sequential hybridization of a poly(A)+ RNA dot-blot with several cDNA probes. The quantity of poly(A)+ RNA loaded on the blot was normalized to eight housekeeping genes (phospholipase, ribosomal protein S9, tubulin, a highly basic 23 kDa protein, glyceraldehyde-3-phosphate dehydrogenase, hypoxanthine guanine phosphoribosil transferase, β-actin and ubiquitin).
Using a probe that hybridized only to the largest mRNA isoform on Northern blots (20), transcripts were most abundant in thymus (adult and fetal), peripheral blood leukocytes, bone marrow and several regions of the adult brain (Fig. 2). However, the largest mRNA isoform was not detected in fetal brain. There was also negligible transcription of this isoform in heart, lung, kidney and liver in all developmental stages.
A somewhat different pattern of relative expression was evident upon rehybridization of the blot with a probe derived from the 5′ end of the coding domain of CHS (Fig. 2), a region that hybridized to all mRNA isoforms on Northern blots (20). Consistent with the transcription pattern of the largest isoform, this probe detected abundant expression in peripheral blood leukocytes, thymus (adult and fetal) and bone marrow, but negligible expression in skeletal muscle. However, several tissues with abundant large isoform transcripts exhibited considerably less relative expression with the 5′ probe, including most regions of the adult brain, fetal and adult thymus, and spleen. Furthermore, several tissues with negligible large isoform relative transcription exhibited increased relative expression with the 5′ probe, including adult and fetal heart, kidney, liver, and lung, and adult aorta, thyroid gland, salivary gland and appendix, and fetal brain.
As an initial screen for mutations in CHS patients, we analyzed Northern blots of poly(A)+ RNA from CHS patients. The largest mRNA species (~13 kb) was greatly reduced in abundance or absent in lymphoblastoid mRNA of patients P1 and P3, respectively (Fig. 3A), while the smaller transcript (~4.4 kb) was both present and undiminished in abundance in all three patients. While the selective loss of the larger transcript in these two patients cannot be excluded definitively, rehybridization of this blot with an actin probe suggested that absence of the larger transcript was not due to uneven gel loading or RNA degradation (data not shown). Fibroblast poly(A)+ RNA from three other CHS patients (369, 371 and 373) showed a moderate reduction in the ~13 kb mRNA (51–60% of control by densitometry), while the ~4.4 kb mRNA was essentially unaltered in abundance (103–147% of control) (data not shown).
Single-strand conformation polymorphism (SSCP) analysis was undertaken using cDNA samples derived from lymphoblastoid or fibroblast cells lines from CHS patients. Anomalous bands were detected in PCR products from the 5′ end of the open reading frame (ORF) in two unrelated CHS patients (371 and 373) different from those with probable selective loss of the larger transcript on Northern blots (Fig. 3B). Sequence analysis identified a C→T substitution at nucleotide 148 of the coding domain in patient 373 (Fig. 3C). Four of nine cDNA clones derived from patient 373 contained this mutation. Restriction enzyme digestion confirmed this mutation: TaqI digestion of cDNA (nucleotides 520–808) showed loss of this restriction site in patient 373 to be heterozygous (data not shown). The C→T substitution creates a stop codon at amino acid 50 (R50X). The mutation in patient 373 occurs within a CpG dinucleotide, which represents a common hotspot for mutations when methylated.
Patient 371 had been shown previously to have a frameshift mutation with a G insertion at nucleotide 118 of the coding domain (Fig. 3C) (20). While analysis of genomic DNA had shown patient 371 to be heterozygous for this mutation(24), each of five cDNA clones isolated from lymphoblasts of this patient contained the insertional mutation. Allele-specific oligonucleotide (ASO) hybridization of cDNA from this patient failed to detect a signal with an oligonucleotide corresponding to the normal allele, suggesting that the patient is transcriptionally hemizygous for the G insertion mutation. Alternatively, the lack of ASO hybridization may reflect a base mismatch (such as a polymorphism) within the patient’s cDNA.
Novel mutations were identified in two other CHS patients: a homozygous C→T substitution at nucleotide 3085 of the coding domain that created a stop codon at amino acid 1029 (Q1029X) was found in patient 370, and a heterozygous frameshift mutation was found in patient 369. Nucleotides 3073 and 3074 of the coding domain were deleted in two of five cDNA clones isolated from the latter patient. This deletion results in a frameshift at codon 1026 and termination at codon 1030.
Despite mutation analysis of >6 kb of ORF, the mutations in CHS patients P1 and P3 (that appear to have selective loss of the larger transcript on Northern blots) have not been identified.
Lymphoblasts from all of these patients (369, 370, 371, 373, P1 and P3) contain the giant perinuclear lysosomal vesicles that are the hallmark of CHS. Patients 369, 370 and 371 had typical clinical presentations of CHS, with recurrent childhood infections and oculocutaneous albinism. The parents of patients 369 and 370 are known not to have been consanguinous. In contrast, the clinical course of patient 373 was milder: This patient has not had systemic infections and remains alive at age 37. Patient 373 does, however, have hypopigmented hair and irides as well as peripheral neuropathy.
We have demonstrated here that each of the previously reported bg gene sequences (20,21) are derived from a single gene with alternatively spliced mRNAs. The previously reported sequences are derived from non-overlapping parts of two Lyst mRNA isoforms with different predicted C-terminal regions. By sequencing RT-PCR products, we have shown that nucleotides 1–4706 of Lyst also represent the previously undetermined 5′ region of the largest Lyst isoform. Alternative splicing at nucleotide 4706 results in bg gene isoforms that contain different 3′ regions. Splicing from exon σ (containing nucleotide 4706) to exon τ results in an ~12 kb mRNA (Lyst-I) that corresponds to the largest band observed on Northern blots. Incomplete splicing at nucleotide 4706 results in a 5893 bp cDNA (Lyst-II) that contains intron-derived sequence at the 3′ end. Lyst-II corresponds to a smaller mRNA observed on Northern blots (20). While several other genes generate an alternative C-terminus by incomplete splicing (27–31), the bg gene is unique in that the predicted structures of the two C-termini are quite different. The C-terminus of Lyst-I contains a ‘WD’-repeat domain that is similar to the β-subunit of heterotrimeric G proteins and which may assume a propeller-like secondary structure (32). This domain is absent in Lyst-II.
Northern blots of human tissues had suggested that transcription of the homologous human gene, CHS, has a similar complexity to the mouse (20). We recently identified two human ESTs homologous to mouse Lyst and described a mutation in one of these in a CHS patient (20). Subsequently, another group published the cDNA sequence of the largest CHS isoform, and identified mutations in this gene in two additional patients with CHS (25). Here we have described the identification of a second isoform of the human CHS gene. This mRNA encodes a protein of 1531 amino acids that is homologous to mouse Lyst-II. Like the latter, this human mRNA arises from incomplete splicing and retention of a transcribed intron that encodes the C-terminus of the predicted CHS protein. The mouse and human codons unique to this short isoform share 65% amino acid identity. The stop codon, however, is not conserved precisely between the human and mouse short isoforms. While mouse Lyst-II is predicted to contain a C-terminal prenylation motif (CYSP), translation of human short isoform is predicted to terminate 22 codons earlier and to lack this motif.
However, the other predicted structural features of the human short isoform were conserved with mouse. The most notable of these was a region similar in sequence to stathmin (amino acids 376–540) (20). While the mouse and human CHS genes had an overall amino acid identity of 81%, identity in the stathmin-like domain was 92% (and similarity was 99%) (Fig. 4). Stathmin is a coiled-coil phosphoprotein that regulates microtubule polymerization in a phosphorylation-dependent manner, and acts as a relay for intracellular signal transduction (33–35). This region of the CHS gene may encode a coiled-coil protein interaction domain and may regulate microtubule-mediated lysosome trafficking. Intriguingly, a defect in microtubule dynamics has been documented previously in CHS (36) and intact microtubules are required for maintenance of lysosomal morphology and trafficking (37–40).
Comparison of the relative abundance of CHS gene transcripts in human tissues at different developmental stages revealed an overlapping but distinct pattern of expression (Fig. 2). A quantitative estimate of the relative expression of the smaller mRNA isoforms was obtained by comparing the relative hybridization intensity obtained with a probe specific for the large isoform with that obtained with a probe that hybridizes to all CHS transcripts. Large isoform transcripts predominated in thymus, fetal thymus, spleen and brain (with the exception of amygdala, occipital lobe, putamen and pituitary gland). Large and small mRNA transcripts were abundant in the latter brain tissues, peripheral blood leukocytes and bone marrow. However, in several tissues, only the small isoforms were expressed, including heart, fetal heart, aorta, thyroid gland, salivary gland, kidney, liver, fetal liver, appendix, lung, fetal lung and fetal brain. The developmental pattern of mRNA isoform expression in brain was particularly interesting, since only the small isoforms were expressed in fetal brain, whereas the largest isoform predominated in many regions of the adult brain.
Novel mutations were identified within the coding domain of the CHS gene in three CHS patients (Table 1). The genetic lesions in two CHS patients (370 and 373) were C→T substitutions that resulted in premature termination (Q1029X and R50X, respectively). Another patient had a frameshift mutation. This patient was heterozygous for a dinucleotide deletion that results in premature termination at codon 1030. Interestingly, all bg and CHS mutations identified to date are predicted to result in the production of either truncated or absent proteins (20,21,25). Unlike Fanconi anemia, type C, there does not appear to be a correlation between the length of the truncated proteins (which may or may not be stable) with clinical features or disease severity in CHS patients (41). However, until the other mutant allele in the compound heterozygote patients is identified, and the exact effects of each mutation at the protein level are characterized, such correlation is imprecise.
An important question that remains to be answered definitively is which transcripts are biologically relevant in regulating lysosomal trafficking and in preventing CHS. The distribution of expression of mRNA isoforms in human tissues was consistent with the pattern of clinical features observed in CHS patients: transcripts were most abundant in peripheral blood leukocytes, bone marrow, thymus, lymph node, spleen and brain, and common clinical features among CHS patients are immune deficiency, platelet storage pool deficiency, neurologic manifestations and albinism. Northern blots demonstrated an ~13 kb mRNA (corresponding to the largest isoform) to be severely reduced in abundance in two CHS patients. A 4.4 kb band (corresponding to a smaller isoform), however, was present in mRNA from these patients in normal abundance. These results suggest that in some patients CHS results from loss of the protein encoded by the largest isoform. All of the mutations identified in CHS patients to date are within the region of the coding domain that is common to all isoforms (20,25). However, the mouse bg8J mutation results in the generation of a premature stop codon in Lyst-I that is unlikely to affect Lyst-II mRNA processing (21). Together, these results suggest that defects in the full-length mRNA alone can elicit CHS and that expression of the smaller isoform alone cannot compensate for loss of the largest isoform.
Segments of the human CHS sequence were obtained by an anchored, nested PCR (5′RACE-PCR) using liver cDNA as a template (Clontech Laboratories, Palo Alto, CA), by RT-PCR using total RNA and by sequencing of human ESTs similar in sequence to mouse Lyst. For the 5′RACE-PCR, two nested primers were used that were derived from a human EST (GenBank accession No. W26957) and had the following nucleotide sequences: 5′-CCAAGATGAAAGCAGCCGATGGGGAAAACT-3′ and 5′-TCAGCCTCTTTCTTGCTCCGTGAAACTGCT-3′. For RT-PCR experiments, total RNA was prepared from the promyelocytic HL-60 cell line. RT-PCR was performed with Expand polymerase (Boehringer Mannheim, Meylan, France) with the following primer pairs: 5′-AGTTTATGAGTCCAAATGAT-3′ and 5′-GAATGATGAAGTTGCTCTGA-3′ (bp 490–2034); 5′-CAGCAGTTCTTCAGATGGA-3′ and 5′-ATCTTTCTGTTGTTCCCCTA-3′ (bp 1891–3050); and 5′-TAGGGGAGCAACAGAAAGAT-3′ and 5′-GCTCATAGTAGTATCACTTT-3′ (bp 3320–4722). The primers used to amplify the cDNA between base pairs 1891 and 3050 were derived from the mouse Lyst sequence (20). Human primers were designed from the sequence of the PCR product (1159 bp) and used to amplify the flanking sequences. Human CHS intron σ′ was PCR amplified from human genomic DNA (100 ng) with the primers 5′-CCGCTCGAGTAGGATCTTTAAGGTGAATAAC-3′ and 5′-GTGATACTACTATGAGCCCTTCACAGTATC-3′. PCR conditions were: 94°C for 1 min, then 32 cycles of 94°C for 15 s, 63°C for 20 s and 72°C for 30 s, followed by 10 min at 72°C.
Total cellular RNA was isolated from mouse bone marrow cells with Trizol (Life Technologies Inc., Bethesda, MD). Five μg of RNA was reverse transcribed using Superscript reverse transcriptase (Life Technologies Inc.), with first strand cDNA synthesis priming by an oligo(dT) primer and a gene-specific primer (5′-CACAGTCATGGGACTGCTAA-3′) for Lyst-I and Lyst-II, respectively. One μl of cDNA was used as template for PCR amplification using KlenTaq DNA polymerase (Clontech, Palo Alto, CA). PCR conditions were: 94°C for 1 min, then 30 cycles of 94°C for 15 s, 68°C for 10 or 5 min (for Lyst-I and Lyst-II, respectively), followed by 20 or 10 min (for Lyst-I and Lyst-II, respectively) at 68°C. The same upstream primer (5′-AGCGGAGGTGAAGCCTTATGCTGAGACAGT-3′) was used for PCR amplification of both isoforms. Isoform-specific downstream primers used were 5′-TTATGTCCTGTGGGGACACTCCTTC-3′ (for Lyst-I) and 5′-ACAGAGCATCCCCACTTCCCTATCTAAAGT-3′ (for Lyst-II).
PCR products were cloned using a TA cloning kit (Invitrogen Corporation, San Diego, CA) and both strands were cycle sequenced. The sequences were analyzed with the GCG Package (42) and searches of the National Center for Biotechnology Information database were performed using the BLAST network server (43) (National Library of Medicine, via INTERNET) and the Whitehead Institute Sequence Analysis Programs (MIT, Cambridge, MA).
Preparation of mouse, human, bacteriophage P1 and yeast DNA samples, digestion with restriction endonucleases, agarose gel electrophoresis and Southern transfers were performed using standard techniques (44,45). An EcoRI monochromosomal somatic cell hybrid blot was obtained from BIOS Laboratories (New Haven, CT). Isolation of poly(A)+ RNA from fibroblast and Epstein–Barr virus-transformed B lymphoblast cell lines, formaldehyde agarose gel electrophoresis and Northern blotting were performed according to standard procedures (44). A human RNA dot-blot containing poly(A)+ RNA from 50 human adult and fetal tissues was obtained from Clontech Labs. The quantity of poly(A)+ RNA on each dot was normalized for eight housekeeping genes, and varied from 100 to 500 ng. Membranes were hybridized with various CHS or actin probes labeled with [α-32P] dCTP (45). Autoradiography or phosphorimaging was performed using X-ray film or a Molecular Dynamics phosphorimager (Sunnyvale, CA), respectively.
Detection of nucleotide changes by SSCP was performed as described by Orita et al. (46). Briefly, each PCR product was mixed with an equal volume of denaturing buffer and heated to 95°C for 3 min, after which the samples were loaded onto 0.8 mm thick, 10% native polyacrylamide gels. Gels were run at ambient temperature at 9 W for 6–10 h, depending on the size of the PCR product. Bands were visualized by silver staining (47). Primer sequences are available upon request.
PCR products spanning the mutation site in patient 371 were transferred to nylon membranes using a slot-blot apparatus. Approximately 5 ng of each PCR product was treated with a denaturing solution (0.5 M NaOH, 1.5 M NaCl), split in half and loaded in duplicate. Two 17mer oligonucleotides were synthesized that span the region containing the mutation. One contained the sequence of the normal allele (5′-CGCACATGGCAACCCTT-3′), while the other contained the sequence of the mutant allele (5′-GCACATGGGCAACCCTT-3′). These were end-labeled with [γ-32P] dATP using T4 polynucleotide kinase and hybridized to the membranes at 50°C. Hybridization and wash buffers were as described (48). Membranes were washed sequentially at 45, 55 and 65°C for 10 min each and exposed to X-ray film.
We thank K. Achey and S. Certain for technical assistance and Dr Seth Orlow for provision of mouse melanocyte RNA. We thank Dr Roberto Solari and Dr Charlie Hodgman for sequence alignments. S.F.K. is supported by the American Cancer Society, the Arthritis Foundation, Glaxo-Wellcome Research and Development, the Howard Hughes Medical Institute and the National Institutes of Health (AI39651 and AI39824). S.J.B. is supported by the Vanderbilt Cancer Center and the National Institutes of Health (5P30-AR 41943). G.S.B. is supported by grants from INSERM, l’Association Vaincre les Maladies Lysosomiales, l’Association Française contre les Myopathies, le Ministère de l’Education Nationale de l’Enseignement Supérieur et de la Recherche and l’Assistance Publique Hôpitaux de Paris.