Abstract
The exponential expansion of the publicly available human DNA sequence database has increasingly facilitated cloning by homology of genes for biochemically defined, functionally similar proteins. We hypothesized that an as-yet uncloned human α-glucosidase (human neutral α-glucosidase C or GANC) is a previously uncharacterized member of a paralogous human glycosyl hydrolase gene family 31, sharing sequence homology and related, but not identical, functions with other cloned human α-glucosidases. We now report both the in silico and physical cloning of two alleles of human neutral α-glucosidase (designated GANC on the human gene map). This cloning and correct identification and annotation as GANC was successful only because of the application of the biochemical and genetic information we had previously developed regarding this gene to the results of the in silico method. Of note, this glucosidase, a member of family 31 glycosyl hydrolases, has multiple alleles, including a “null” allele and is potentially significant because it is involved in glycogen metabolism and localizes to a chromosomal region (15q15) reported to confer susceptibility to diabetes.
The rapid rise in the number of cloned genes has resulted in an increasing ability to identify sequence homologies between different genes, either within the same species (paralogs) or between species (orthologs). The publicly available human draft sequence (refs. 1 and 2; http://www.ncbi.nlm.nih.gov/) has provided the potential to identify more easily and clone members of gene families in silico, greatly facilitating their physical cloning. We report one such case, the cloning of a previously described human α-glucosidase (neutral α-glucosidase C or GANC). However, physical cloning and correct annotation of full-length GANC was only made possible by applying the biochemical and genetic information we had previously developed regarding this gene to the in silico information (3–5).
At least six different human α-glucosidases hydrolyzing α-linked glucose in simple and complex carbohydrates have been described biochemically and genetically. These enzymes differ with respect to substrate specificities, pH optima, molecular weights, sites of expression, and chromosomal localizations. Five of these human α-glucosidases have now been cloned: lysosomal acid α-glucosidase (GAA; refs. 6–8), intestinal sucrase-isomaltase (SI; refs. 9 and 10); intestinal maltase-glucoamylase (MGA; ref. 11), the catalytic unit of the endoplasmic reticulum enzyme glucosidase II (GANAB); see Online Mendelian Inheritance in Man (OMIM 601862 and 104160; refs. 12 and 13); and glucosidase I (GCSI; ref. 14). Four of the five cloned human α-glucosidase genes (excluding glucosidase I) share homology, based on sequence similarity and signature sequence for family 31 glycosyl hydrolases, suggesting that they are members of a gene family in a single species or human paralogs for family 31 glycosyl hydrolases (ref. 15; reviewed in ref. 16 and http://afmb.cnrs-mrs.fr/∼cazy/CAZY/index.html).
Neutral α-glucosidase C, the remaining human α-glucosidase, had not as yet been cloned. Neutral α-glucosidase C (GANC) has activity at neutral pH, a characteristic electrophoretic mobility relative to the human lysosomal enzyme acid α-glucosidase and to human glucosidase II as well as to the murine ortholog of GANC, and a molecular weight similar to that of the lysosomal enzyme acid α-glucosidase (4, 5, 17). The gene maps to human chromosome 15 (5), is designated as GANC on the human gene map (ref. 17; OMIM 104180), and exhibits a biochemical genetic polymorphism with four alleles, including a null allele, segregating in the population (3). The known properties of GANC, including 100 kDa molecular mass, its ability to degrade glycogen, and the neutral pH optimum suggested homology to family 31 glycosyl hydrolases and particularly to lysosomal acid α-glucosidase and to the catalytic unit of glucosidase II.
Materials and Methods
Computer-Based Identification of Candidate Clones and Construction and Analysis of a Computer Contig.
The nonredundant sequence database and the human EST database (http://www.ncbi.nlm.nih.gov/) were screened by using as the initial query sequence both human lysosomal acid α-glucosidase, glucosidase II, and the original BLAST program with BLAST X or the gapped BLAST program. This screen identified three ESTs: GB R95789, clone 1; GB AA101464/3, clone 2; and GB AA323316, clone 3. These clones were obtained from commercial sources (American Type Culture Collection, Research Genetics, or Genome Research) and sequenced completely in both directions. A contig was constructed on the computer and compared for homology with other glucosidases (see Fig. 1a and legend).
Figure 1.
Steps in the in silico and physical cloning and expression of GANC. (a) Computer-based identification of candidate clones and construction and analysis of a computer contig. Sequence for human acid α-glucosidase and the catalytic subunit of glucosidase II was used as “bait” in a TBLASTX and BLASTN search of the human EST database for “orphan” ESTs with homology to the two glucosidases. This procedure identified GB R95789 (clone 1) that contained a sequence tagged site (STS) WI-17962 mapping to chromosome 15q15 in radiation hybrid panels 3 and 4, which is consistent with the earlier mapping of GANC to chromosome 15 in somatic cell hybrids (5). The search simultaneously identified GB AA101464/3 (clone 2), a putatively “reversed” EST that contained sequence for the consensus catalytic site for family 31 glycosyl hydrolases (15) and an STS G31167 that also mapped to chromosome 15q15 in the radiation hybrid maps 3 and 4. Clone 2 additionally identified GB AA323316 (clone 3) that overlapped a conserved sequence in clone 2. Complete sequencing of all clones in both directions and comparison for homology against the sequence database revealed two areas that did not show homology. (b) Deletion of these two regions and “pasting together” the computer construct resulted in an ORF of an appropriate size. After deletion of the two areas of apparent nonhomology (confirmed by absence of these areas from sequence of cDNA obtained by RT-PCR of mRNA), the computer-edited cDNA gave an appropriate size ORF from the initiating methionine to a stop codon (2.745 kb and 914 codons) and homology throughout the ORF (see Table 2). (c) To facilitate deletion of the two areas of nonhomology from a cDNA containing the coding region, the initial construct tested for expression of neutral glucosidase was a hybrid. This hybrid contained a 5′ fragment from clone 2, (beginning 8 bp 5′ of the ATG with an added EcoRI site), ligated at an endogenous SacI site (2,371 bp) to a 3′ fragment derived by RT-PCR of mRNA from cell line GM02756 and extending 15 bp 3" of the stop codon with an added Xba site. (The first methionine in the construct was at bp 753–755 of the original full-length cDNA). We previously had determined that GM02756 expressed two different electrophoretically distinguishable alleles for GANC and that GM02756, therefore, did not contain a null allele lacking expression of enzyme activity. (d) Two allelic cDNA clones, differing in sequence at several sites, were obtained by RT-PCR of mRNA from cell line GM02756. The differing cDNA clones were found to express electrophoretically distinguishable allozymes (see Fig. 2 and Table 1).
Construction of cDNAs Containing the Coding Sequence in an Expression Vector.
The initial cDNA construct that was expressed contained a 5′ fragment from clone 2 ligated to a 3′ fragment obtained by RT-PCR to eliminate two areas of nonhomology found by analysis of the computer contig (Fig. 1c). In detail, a 2.1 kb EcoRI/SacI fragment from clone 2, (deleting the first 263 bp of clone 2 and starting 490 bp 5′ of the putative ATG and extending 1,619 bp 3′ of the ATG) was cloned into pUC19 (clone B). To eliminate the major portion of the 490 bp of 5′ untranslated sequence, an EcoRI site was introduced into clone B by PCR just 5′ of the putative ATG (by using a sense amplification primer containing an added EcoRI site and extending from −8 to + 10, relative to the ATG, and an antisense primer 3′ of a unique EcoRV site). The PCR fragment was digested with EcoRI and EcoRV and exchange-ligated into the EcoRI/EcoRV digested clone B to give clone B-7. The remaining 3′ end of the cDNA was obtained by RT-PCR by standard methods with mRNA from a cell line (GM02756) known to contain two different alleles expressing enzyme activity for GANC (3) and a sequence-based sense primer 34 bp 5′ of the unique SacI site (bp 2,337–2,358) and an antisense primer in the 3′ UT (ending 15 bp 3′ of the TGA stop codon), with an added XbaI site. The clone (B-7) containing the 5′ end was digested at the endogenous SacI site and at a 3′ XbaI site in the vector cloning site. The 3′ RT-PCR product, digested at the endogenous SacI site and the XbaI site in the primer, was ligated into the B7 clone to yield B7-7. The full-length insert was released with EcoRI/XbaI digestion and cloned into the expression vector pCDNA 3 (Invitrogen) to yield clone B-7-7-9. The PCR-amplified section of the 5′ clone, all of the ligation joints, and the complete 3′ end of several clones were sequenced to confirm the absence of cloning artifacts. Differences between the consensus sequence of the original EST clones and the absence of the areas of nonhomology were confirmed by sequence analysis of RT-PCR products of mRNA from the human cell line that expressed two different alleles for GANC (GM02756; National Institutes of Health Mutant Repository, Camden, NJ). For cloning of two alleles of GANC, mRNA was isolated (by using an Invitrogen micro fast-track kit according to the manufacturer's directions) from the cell line GM02756 previously determined to express two different alleles for GANC (3). RT-PCR was performed by using an antisense primer ending 15 bp 3′ of the stop codon with an added XbaI site and a sense primer 5′ of the ATG with an added EcoRI site and cloned into the PCR zero blunt vector (Invitrogen). Clones were sequenced with M13 and gene-specific primers, and the insert was then cloned into PCDNA3 for expression (Fig. 1d).
Transient Expression of the Cloned cDNAs and In Situ Visualization of Neutral α-Glucosidase C Activity.
The cDNA constructs in the pCDNA3 expression vector were transiently expressed in three different cell lines; murine 3T3 cells, a simian virus 40-transformed human cell line deficient for acid glucosidase (TR4912; derived from GM 04912, a cell line obtained from the National Institutes of Health Mutant Cell Repository, Camden, NJ; ref. 18) and in monkey kidney COS cells. Previously described standard CaPO4-based methods were used for transfection (19). Cell lysates from transfected and nontransfected cells and lymphoid line cells were prepared and analyzed as described (3, 5, 17). In brief, cell lysates from transfected and nontransfected cells were electrophoresed in starch gels at 4°C to separate isozymes and/or allozymes primarily by their charge, overlaid with 3 MM filter paper saturated with substrate (4 methyl umbelliferyl⋅d-glucopyranoside) at either acid pH (not shown) or neutral pH 7.5 and incubated at 37°C. Areas of enzyme activity were detected as fluorescent bands, as described (17).
Results
Based on the hypothesis that GANC shares homology with other human family 31 paralogs, we searched the EST database for candidate clones that had homology to α-glucosidases but did not code for any of the four previously cloned family 31 mammalian or human glycosyl hydrolases. Three candidate human EST clones were identified (Materials and Methods, and Fig. 1), two of which contained sequence tagged sites (STSs) that had been mapped to human chromosome 15q by analysis of radiation hybrid panels 3 and 4, the chromosome to which we had previously mapped GANC, based on segregation of the human neutral glucosidase C isozyme in human-mouse hybrids (5).
Complete sequencing and analysis of the three EST clones revealed two small areas lacking homology to other glucosidases, one in clone 2 and one in clone 3 (Fig. 1a). RT-PCR of mRNA from a diploid normal human cell line, GM02756, both of whose GANC alleles were known to encode functional, electrophoretically distinguishable allozymes, revealed that neither of the areas of nonhomology were present in the mRNA. Omitting the two areas of nonhomology and “ligating” the remaining sequence in silico resulted in a 5,022-bp transcript containing a 2,745 bp ORF (including the ATG and stop codon) that exhibited high homology to α-glucosidases throughout (not shown). The ATG of the ORF was preceded by a large, 752 bp 5′ untranslated region and was followed by a large, 1,525 bp 3′ untranslated sequence (20) without homology to other human glucosidases (Fig. 1). The derived cDNA predicted a protein of 914 amino acids and a molecular mass of 104 kDa, consistent with that of 92.2 ± 9.8 kDa previously determined for the partially purified GANC (4). This molecular mass is also similar in size to that of the paralogs lysosomal acid α-glucosidase and the catalytic unit of glucosidase II (952 and 944 aa) and half the size of the duplicated paralogs sucrase-isomaltase and maltase-glucoamylase.
To demonstrate that the cDNA sequence we had derived in silico represented the in vivo GANC mRNA, we physically cloned and expressed three different cDNAs. Two of the expressed clones were totally derived by RT-PCR of mRNA from the cell line GM02756, known to express two functional, electrophoretically distinct allelic forms of GANC (see Fig. 1d). Two different classes of clones could be distinguished, initially by appearance of a new site for EcoRI, and then by differences in sequence (see Table 1). Transient expression of completely sequenced clones representing each class resulted in the appearance of bands of neutral α-glucosidase enzyme activity that comigrated with either the “type 2” or the “type 3” alleles expressed by the GM02756 cell line (refs. 3–5; Fig. 2). Furthermore, transfection of 3T3 cells with a hybrid cDNA derived in part from the EST clone and in part from RT-PCR yielded α-glucosidase activity with biochemical properties consistent with human as opposed to murine GANC (5).
Table 1.
Polymorphic differences (SNPs) in sequence of GANC cDNAs
| Polymorphic differences in cDNA and location | Computer construct | Hybrid cDNA | Type 3* cDNA | Type 2** cDNA |
|---|---|---|---|---|
| 883 G/A (exon 4) CGG>CAG; Arg44Gln | G (Arg) | G (Arg) | A (Gln) | A (Gln) |
| 1013 C/T (exon 5) AAC>AAT; Asn87Asn | C (Asn) | C (Asn) | C (Asn) | T (Asn) |
| 1049 G/A (exon 5) CCG>CCA; Pro99Pro | G (Pro) | G (Pro) | A (Pro) | G (Pro) |
| 1211 A/G (exon 6) ATA>ATG; Ile153Met | A (Ile) | A (Ile) | A (Ile) | G (Met) |
| 2081 A/T (exon 13) GAT>GAA; Asp443Glu | A (Glu) | A (Glu) | T (Asp) | T (Asp) |
| 3286 T/C (exon 24) TTT>TCT; Phe845Ser | T (Phe) | C (Ser) | C (Ser) | C (Ser) |
| 3295 A/G (exon 24) CAG>CGG; Gln848Arg | A (Gln) | G (Arg) | G (Arg) | A (Gln) |
Figure 2.
Transient expression in Cos cells of cloned α-glucosidase C alleles. Cos cells were transiently transfected with the various cDNA clones in an expression vector. Cell lysates were prepared from nontransfected cells and from cells transfected with the two clones containing differing alleles derived from mRNA of the GM02756 cell line that is heteroallelic for type 2 and type 3 GANC alleles (see Fig. 1d and Table 1). Areas of α-glucosidase activity were visualized under long-wave UV light as fluorescent bands. GANAB is the gene name for human α-chain (catalytic unit) of glucosidase II on chromosome 11; GANC is the official gene name for human neutral α-glucosidase C on chromosome 15. Lane 1: cloned human GANC cDNA expressed in COS cells, showing activity comigrating with the product of human GANC type 3 allele. Lane 2: cloned human GANC cDNA expressed in COS cells, showing activity comigrating with the human type 2 allele. Lane 3: human cell line heterozygous for type 2 and type 3 GANC alleles and the source of mRNA for cloning the allelic cDNAs. Lane 4: human cell line null for GANC (fourfold greater amount of cell lysate applied).
Comparison of the sequence of the EST-derived cDNA with that of the two clones derived totally by RT-PCR (and the additional hybrid cDNA) revealed seven polymorphic sites, two of which were silent (Table 1). The two GM02756 clones differed from each other at four sites, two of which predicted amino acid differences (Ile at codon 153 and Arg at codon 848 for type 3 vs. Met at codon 153 and Gln at codon 848 for type 2). Both clones also differed from the computer-derived consensus sequence at three additional sites predicting amino acid differences (Arg-44-Gln; Glu-443-Asp, Phe-845-Ser). All of these sequence variations were true polymorphisms and were not due to cloning, sequencing, or PCR errors, because we could define both alleles for all of these single nucleotide polymorphisms (SNPs) in genomic DNA of additional individuals.
GANC shows homology at the catalytic site (WXDMNE) to the four previously cloned human α-glucosidases: lysosomal acid α-glucosidase (the original identified human family 31 glycosyl hydrolase), the two duplicated α-glucosidases, sucrase-isomaltase and maltase-glucoamylase, and the catalytic unit of glucosidase II (synonymous with GANAB) (Table 2). The greatest homology was to the catalytic unit of glucosidase II, an enzyme that is also active at neutral pH. Similarly, GANC showed overall homology to the previously cloned α-glucosidase paralogs, with similarity ranging from 67 to 41%, but again, with the highest overall homology to the catalytic unit of glucosidase II (Table 2).
Table 2.
Homology of GANC to other human α-glucosidase paralogs
| Homology at the catalytic site | Catalytic site | ||
| Lysosomal acid α-glucosidase | FHDQVPFD | GMWIDMNE | PSNF |
| Maltase-glucoamylase: maltase | FHNQVIFD | GIWIDMNE | VSNF |
| glucoamylase | ----VPFD | GMWIDMNE | PSSF |
| Sucrase-isomaltase: isomaltase | FHQEVQ-D | GLWIDMNE | VSSF |
| sucrase | KFD | GLWIDMNE | PSSF |
| Glucosidase II (GANAB) | YEGSAPNL | FVWNDMNE | PSVF |
| Neutral α-glucosidase C (GANC) | YQGSTDIL | FLWNDMNE | PSVF |
| Consensus: | GMWXDMNE* | ||
| FI S | |||
| V A | |||
| L | |||
| F | |||
| Overall homology | % Identity | % Similarity** |
| Glucosidase II (GANAB) | 49 | 67 |
| Lysosomal acid α-glucosidase | 31 | 47 |
| Sucrase-Isomaltase: isomaltase | 30 | 47 |
| sucrase | 25 | 41 |
| Maltase-glucoamylase: maltase | 28 | 45 |
| glucoamylase | 27 | 43 |
Prosite: PD0C00120. Consensus based on additional orthologs and paralogs; X indicates numerous different amino acids at the indicated position.
Identity, same amino acid; similarity, conservative substitution.
Alignment of the full-length cDNA sequence with human genomic DNA sequences from bacteria artificial chromosome (BAC) clones AC022468 and AC012651 and with the draft sequence (http://genome.ucsc.edu/) suggested that the GANC gene consists of 25 exons distributed over 80 kb. We confirmed both this proposed exon–intron structure and the accuracy of our cDNA sequences by amplifying and sequencing 23 of the 25 exons and their flanking intronic regions from genomic DNAs. The first exon was noncoding, whereas the second 343-bp exon was predominantly noncoding, containing only the first 10 codons, including the initiation of translation within a Kozac consensus sequence. The terminal exon 25 was only sequenced from genomic DNA to just 3′ of the stop codon, whereas only a small part of the flanking intron for exon 19 could be sequenced from PCR products of genomic DNA. The additional SNPs noted in the sequence of the ESTs and of the newly cloned cDNAs (Table 1) also were found in genomic DNA from additional individuals (not shown).
Correct annotation of GANC present in the draft sequence and even earlier in the two BACs used for the draft sequence was probably inhibited by the artifactual portions of the initial EST sequences as well as by several features of the exonic structure. The area of “nonhomology” in the cDNA of EST clone 2 (Fig. 1, AA101464) was apparently caused by priming from intron 15 on the antisense strand with inclusion of +59 to +1 bp from intron 15. The area of “nonhomology” in cDNA of the EST clone 3 (Fig. 1, AA332316) represented inclusion of 98 bp of intron 20 (1.264 kb downstream of exon 20), predicting an exon with two in-frame stop codons (TAA) and, therefore, a nonfunctional transcript. Although such events can be interpreted as alternate splicing, they can also be found as artifacts during construction of cDNA libraries for extensively studied, known genes. Our RT-PCR results would indicate that the inclusions of intronic sequence seen were artifacts. The features of the intron/exon structure that may have interfered with correct prediction of the mRNA based on algorithms were the relatively small size of several of the exons, (ranging from 46 to 80 bp), some of which were buried within large surrounding introns, leading to exclusion from the predicted mRNA sequences. Similarly, the ATG, sitting at the terminal portion of a large, primarily noncoding exon appears to cause difficulty for identification by current algorithms (ref. 21; GenBank database accession nos. AF545044–AF545047).
The GANC cDNA maps to human chromosome 15q15.1-.2 on the draft sequence (http://genome.ucsc.edu/), flanked on the 5′ end by a predicted gene (DKFZP564G2022). We confirmed, by sequence analysis of PCR products from genomic DNA, that this putative gene begins only 168 bp upstream of the start of the GANC cDNA and is encoded on the reverse strand with transcription toward the centromere. By contrast, GANC is encoded on the sense strand with transcription toward the telomere, suggesting overlap or close proximity of the two promoter regions On the 3′ end. GANC is flanked by calpain 3 (CAPN3), which begins 6.151 kb 3′ of GANC, with transcription of both genes directed toward the telomere.
Discussion
We have cloned a neutral α-glucosidase that is a new member of a human paralogous gene family, belonging to the glycosyl hydrolase gene family 31, initially by in silico manipulation of sequences of EST clones and then physically by the cloning and expression of two cDNAs coding for alleles of GANC. Both cDNAs express α-glucosidase activity at neutral pH, exhibit the expected differences in electrophoretic mobility for two of the known genetic variants, are of appropriate size, and map to chromosome 15q. These biochemical and genetic properties allow us to annotate correctly the cDNAs and the corresponding genomic DNA as GANC (OMIM 104180). The molecular basis for the allelic differences was identified as single-base changes (SNPs), some of which resulted in amino acid differences. Also, we have experimentally determined the exon and flanking intron sequence and boundaries and placed the cDNA on the genome sequence of chromosome 15 at 15q15. GANC is flanked on the 5′ end by DKFZP564G2022, a gene transcribed toward the centromere on the antisense strand and on the 3′ end by CAPN3, transcribed toward the telomere in tandem with GANC on the sense strand.
Correct annotation is now a critical next step in analysis of the human genome and, as is evident from the example reported here, is likely to require a more individualized process that will benefit from detailed molecular and biochemical approaches. The observations reported here demonstrate the utility of biochemical, genetic, and molecular data in the final stages of genome annotation, as well as perusal and knowledge of prior published data, annotated in several linked databases such as OMIM that, for GANC, contained all of the necessary information. Full-length cDNA transcripts facilitated the verification of the correct annotation for GANC and had led to our initial report in an abstract.‡ During final preparation of this manuscript, a putative “full length” cDNA (AK074037) encoding a proposed protein named FLJ00088 was reported and assigned GenBank accession number BAB84863. However, this cDNA encompasses 4,476 of the 5,022 bp reported here and lacks exon 1 and part of exon 2. The initiating methionine was contained in FLJ00088 but was only identified as the initiating first codon by the expression studies reported here. The initiating ATG is at codon 12 of FLJ00088 (bp 208–210), corresponding to bp753–755 of the GANC sequence. In the light of our studies, we propose that FLJ00088 be renamed GANC.
The localization of GANC next to calpain 3 is of note because linkage studies of type 2 diabetes recently provided evidence for an interacting susceptibility locus in the Mexican American population on chromosome 15 that includes the calpain 3 (CAPN3) locus (22, 23). CAPN3 (mutations in which result in LGMD2) is a paralog of calpain 10, a putative diabetes susceptibility gene on chromosome 2 (24, 25). We note that GANC sitting next to CAPN3 on chromosome 15 is also a candidate for a role in diabetes. GANC is an enzyme involved in degradation of glycogen; there are known multiple alleles for this gene, at least one of which codes for absence of GANC activity. Alternatively, the known genetic variation in GANC may be associated with differences in severity of glycogen storage disorders and/or normal muscle function.
Acknowledgments
We are grateful for helpful discussions with Drs. J. and K. Hirschhorn, G. Weissmann, S. Brown, and P. D'Eustachio. We thank C.-K. Jiang for her help in the final expression studies. This work was supported by a grant from the March of Dimes National Foundation (to R.H.).
Abbreviation
- GANC
human neutral α-glucosidase C
Footnotes
References
- 1.International Human Genome Sequencing Consortium. Nature. 2002;409:860–921. [Google Scholar]
- 2.Venter J C, Adams M D, Myers E W, Li P W, Mural R J, Sutton G G, Smith H O, Yandell M, Evans C A, Holt R A, et al. Science. 2001;291:1304–1351. doi: 10.1126/science.1058040. [DOI] [PubMed] [Google Scholar]
- 3.Martiniuk F, Hirschhorn R. Am J Hum Genet. 1980;32:497–507. [PMC free article] [PubMed] [Google Scholar]
- 4.Martiniuk F, Hirschhorn R. Biochem Biophys Acta. 1981;658:248–261. doi: 10.1016/0005-2744(81)90295-3. [DOI] [PubMed] [Google Scholar]
- 5.Martiniuk F, Hirschhorn R, Smith M. Cytogenet Cell Genet. 1980;27:168–175. doi: 10.1159/000131478. [DOI] [PubMed] [Google Scholar]
- 6.Martiniuk F, Mehler M, Pellicer A, Tzall S, LaBadie G, Hobart C, Ellenbogen A, Hirschhorn R. Proc Natl Acad Sci USA. 1986;83:9641–9644. doi: 10.1073/pnas.83.24.9641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Martiniuk F, Mehler M, Tzall S, Meredith G, Hirschhorn R. DNA Cell Biol. 1990;9:85–94. doi: 10.1089/dna.1990.9.85. [DOI] [PubMed] [Google Scholar]
- 8.Hoefsloot L H, Hoogeveen-Westerveld M, Kroos M A, van Beeumen J, Reuser A J J, Oostra B A. EMBO J. 1988;7:1697–1704. doi: 10.1002/j.1460-2075.1988.tb02998.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Chantret I, Lacasa M, Chevalier G, Ruf J, Islam I, Mantei N, Edwards Y, Swallow D, Rousset M. Biochem J. 1992;285:915–923. doi: 10.1042/bj2850915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wu G D, Wang W, Traber P G. J Biol Chem. 1992;267:7863–7870. [PubMed] [Google Scholar]
- 11.Nichols B L, Eldering J, Avery S, Hahn D, Quaroni A, Sterchi E. J Biol Chem. 1998;273:3076–3081. doi: 10.1074/jbc.273.5.3076. [DOI] [PubMed] [Google Scholar]
- 12.Trombetta E S, Simons J F, Helenius A. J Biol Chem. 1996;271:27509–27516. doi: 10.1074/jbc.271.44.27509. [DOI] [PubMed] [Google Scholar]
- 13.Trombetta E S, Fleming K G, Helenius A. Biochemistry. 2001;35:10717–10722. doi: 10.1021/bi010629u. [DOI] [PubMed] [Google Scholar]
- 14.Kalz-Fuller B, Bieberich E, Bause E. Eur J Biochem. 1995;231:344–351. doi: 10.1111/j.1432-1033.1995.tb20706.x. [DOI] [PubMed] [Google Scholar]
- 15.Henrissat B. Biochem Soc Trans. 1998;26:153. doi: 10.1042/bst0260153. [DOI] [PubMed] [Google Scholar]
- 16.Hirschhorn R, Reuser A J J. In: The Metabolic and Molecular Basis of Inherited Disease, Eighth Edition. Scriver C R, Beaudet A L, Sly W S, Valle D, editors. New York: McGraw–Hill; 2001. pp. 3389–3420. [Google Scholar]
- 17.Swallow D M, Corney G, Harris H, Hirschhorn R. Ann Hum Genet. 1975;38:391–406. doi: 10.1111/j.1469-1809.1975.tb00629.x. [DOI] [PubMed] [Google Scholar]
- 18.Martiniuk F, Bodkin M, Tzall S, Hirschhorn R. Am J Hum Genet. 1990;47:440–445. [PMC free article] [PubMed] [Google Scholar]
- 19.Huie M L, Hirschhorn R, Chen A, Martiniuk F, Zhong N. Hum Mutat. 1994;4:291–293. doi: 10.1002/humu.1380040410. [DOI] [PubMed] [Google Scholar]
- 20. Mignone, F., Gissi, C., Liuni, S. & Pesole, G. (2002) Genome Biol.3 (Feb. 28), http://genomebiology.com/search/results.asp. [DOI] [PMC free article] [PubMed]
- 21.Mount S M. Am J Hum Genet. 2000;67:788–792. doi: 10.1086/303098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hanis C L, Boerwinkle E, Chakraborty R, Ellsworth D L, Concannon P, Stirling B, Morrison V A, Wapelhorst B, Spielman R S, Gogolin-Ewens K J, et al. Nat Genet. 1996;13:161–166. doi: 10.1038/ng0696-161. [DOI] [PubMed] [Google Scholar]
- 23.Cox N J, Frigge M, Nicolae D L, Concannon P, Hanis C L, Bell G I, Kong A. Nat Genet. 1999;21:213–215. doi: 10.1038/6002. [DOI] [PubMed] [Google Scholar]
- 24.Horikawa Y, Oda N, Cox N J, Li X, Ohro-Melander M, Hara M, Hinokio Y, Lindner T H, Mashima H, Schwarz P E, et al. Nat Genet. 2000;26:163–175. doi: 10.1038/79876. [DOI] [PubMed] [Google Scholar]
- 25.Fullerton S M, Bartoszewicz A, Ybazeta G, Horikawa Y, Bell G I, Kidd K K, Cox N J, Hudson R R, Di Rienzo A. Am J Hum Genet. 2002;70:1096–1106. doi: 10.1086/339930. [DOI] [PMC free article] [PubMed] [Google Scholar]