INTRODUCTION

Top Abstract Introduction Results Discussion Methods References

Usher (USH) syndrome refers to a heterogeneous collection of disorders characterized by congenital hearing impairment, retinitis pigmentosa, and variable vestibular dysfunction. USH syndrome has been divided into three clinical types (Kimberling and Moller 1995): Patients with type I disease (USH1) have severe congenital hearing loss and the absence of vestibular function; in type II disease (USH2) the hearing loss is congenital but moderate to severe and there is no disruption of vestibular capacity; patients with type III disease (USH3) are distinguished from USH2 patients by a progressive loss of hearing. Progressive pigmentary retinopathy is a feature of all three types of USH syndrome.

The complex clinical picture is reflected in the genetic heterogeneity of the disease. At least eight different loci have been identified that contribute to these autosomal recessive disorders. The majority of USH2 families exhibit genetic linkage to markers on the long arm of chromosome 1 (USH2A; Kimberling et al. 1990; Kimberling and Moller 1995), whereas the disease locus in one large family segregating USH2 does not (Pieke-Dahl et al. 1993). The USH3 locus has been assigned to the long arm of chromosome 3 at 3q21-25 (Sankila et al. 1995). At least five loci are implicated in the development of type I disease. USH1A has been mapped to 14q32 (Kaplan et al. 1992), whereas the USH1B and USH1C genes have been localized to the long and short arms, respectively, of chromosome 11 (Kimberling et al. 1992; Smith et al. 1992). More recently, two additional USH1 loci have been identified, USH1D at chromosome 10q (Wayne et al. 1996) and USH1E, which maps to chromosome 21q21 (Chaib et al. 1997).

The locus responsible for USH1B (11q13) has been identified as an unconventional myosin VIIA gene (Weil et al. 1995), a finding consistent with the observation that USH syndrome patients exhibit a generalized disorganization of microtubules in the axoneme of sensory hair cells. This gene also appears to be involved in certain cases of hereditary nonsyndromic deafness (DFNB2; Liu et al. 1997; Weil et al. 1997). Whether mutations in genes encoding additional unconventional myosins or proteins that interact with them result in other types of USH syndrome awaits the isolation of the remaining disease loci.

The location of the USH1C locus was initially assigned to 11p14-15.1 (Smith et al. 1992) and was later refined by linkage and haplotype analysis to the 2- to 3-cM interval between D11S889 and D11S861 (Keats et al. 1994). To identify the gene responsible for USH1C we undertook the isolation of the critical region in yeast artificial chromosomes (YACs). Haplotyping of affected patients with additional markers ordered by somatic cell hybrids and the YAC contig narrowed the USH1C locus to between D11S902 and D11S1890. To facilitate subsequent transcript identification, the interval between D11S1310 and D11S1890 has been converted to PACs. Large-scale sequencing of a portion of this contig has resulted in the identification of several transcripts within the USH1C critical region. These transcripts are being tested as candidates for the USH1C locus.

	RESULTS

Top Abstract Introduction Results Discussion Methods References

Generation of a YAC Map through the USH1C Region

The location of Usher syndrome 1C (USH1C) was originally determined to be in the interval flanked by D11S861 and D11S889 on chromosome 11p14-15.1 (Keats et al. 1994). A number of markers believed to map near or within this region (James et al. 1994; Keats et al. 1994; Fantes et al. 1995) were ordered in a panel of somatic cell hybrids (data not shown) as indicated in Figure 1A. Following PCR screening of a chromosome 11 YAC library (Qin et al. 1993), small contigs were assembled around markers for the genes LDHA, SAA1, TPH, KCNC1, and MYOD1, as well as microsatellites D11S419, D11S1310, D11S902, D11S921, D11S1888, and D11S1890 (Fig. 1B). The contigs were verified by Southern analysis using STSs and single copy gene markers as hybridization probes. Overlap detected by hybridization of Alu-PCR products was confirmed by Southern analysis of Alu-PCR fingerprints (Qin et al. 1996). Finally, YACs representing the shortest path through each contig were mapped to 11p14-15.1 by FISH (see Fig. 1B).

View larger version (43K): [in this window] [in a new window]   Figure 1   Physical maps of the USH1C region. (A) Somatic cell hybrids were used initially to bin markers believed to map between D11S889 and D11S861. The heavy lines represent the approximate extent of chromosome 11 from pter contained in each hybrid cell line. The order within each bin is based on the YAC map below. The placement of markers D11S899, D11S419, and D11S861 on the hybrid panel was determined by Fantes et al. (1995). These markers are for reference only and are not present within the clone contig. (B) YAC contig of USH1C critical and flanking regions. The contig was assembled by SEGMAP using both STS content and Alu-PCR hybridization data. YAC names and sizes (kb) are shown to the left of each clone and in parenthesis, respectively. For some clones (boxed), the cytogenetic position determined by FISH is indicated above the horizontal line representing the YAC. Markers contained within the contig are shown above the contiguous line that represents the genomic map. Markers with the prefix y47, y3, or ySJ are Alu-PCR probes generated from the named YAC clone using one of three Alu primers as described (Qin et al. 1996). Marker names prefixed with yRP and ending with RR, ER, or RL are end probes generated from the named YAC clone. The only link between the LDHA/SAA1-containing clones and the remainder of the contig is through the mega-YAC 804C12 determined by Alu-PCR hybridization. The proximity of these clones with the rest of the contig is consistent with other YAC contigs in this region (Sellar et al. 1994; Ayyagari et al. 1996). The ends of a PAC clone (59M18) were used to show a second link between the TPH YAC clones and KCNC1/S1888/MYOD1 YAC clones.

In an effort to join these contigs, the CEPH mega-YAC library was screened for D11S1310, resulting in a single positive clone (804C12). Alu-PCR products generated from this CEPH mega-YAC were then hybridized to pooled Alu-PCR products of the chromosome 11-specific YAC library (Qin et al. 1996) identifying each of the smaller YAC contigs except those containing D11S419 and D11S1890. End probes were derived from two D11S1890-YACs by a ligation-mediated PCR method (Kere et al. 1992) and found to hybridize to the single D11S921-containing YAC, yRP-2f11. Closure of the gap between the D11S419-contig and the larger contig was not attempted, as D11S419 mapped outside the critical region defined by reconstruction of haplotypes (see below).

The STS/probe content of each YAC, along with clone size, was analyzed by the contig assembly program SEGMAP (Green and Green 1991). This algorithm generated a clone contig consisting of >40 YACs from LDHA proximally to beyond (distal) D11S1890 encompassing ~3.5 mb. The contig contains markers for five genes, five microsatellites, two anonymous STSs (D11S1168, D11S1122), four YAC ends, and 34 Alu-PCR products. The predicted order is CEN-LDHA-SAA1-D11S1168-TPH-D11S1310-D11S1122-(D11S1888/KCNC1)-MYOD1-D11S902-D11S921-D11S1890-TEL.

Reconstruction of USH1C Haplotypes

Using microsatellite markers ordered by the contig, haplotype analysis was carried out on USH1C family members which narrowed the critical disease locus region to between D11S902 and D11S1890 (Table 1). Based on the YAC contig assembled by SEGMAP, the USH1C critical region is ~400 kb (Fig. 1B). Because all affected individuals, and no carrier or unaffected person from the Acadian population tested, were homozygous for allele 3 at D11S921 (Table 1), the disease gene is likely to be located in the middle of the contig close to this marker.

PAC Contig of USH1C Critical Region

To facilitate the identification of transcribed sequences in the USH1C critical region, part of the YAC contig was "converted" to large-insert bacterial clones using a PAC library (Ioannou et al. 1994). Initially, Alu-PCR products generated from several YAC clones were pooled and hybridized to high-density PAC filters. After secondary screening with Alu-PCR products from individual YACs, the clones were characterized by pulsed-field gel electrophoresis, STS analysis with microsatellite and gene markers, and Southern hybridization to the original YAC contig. To complete the contig, end probes were rescued from selected PAC clones (Cooper et al. 1997) and used to rescreen the PAC library by hybridization. End probes were also used to verify overlap between adjacent clones by Southern analysis. As they became available, expanded portions of the PAC library were screened by hybridization using microsatellite and PAC end probes to increase contig depth. To help verify the contig, 15 PACs representing the shortest path between D11S1310 and D11S1890 were mapped to 11p14-15.1 by FISH (Fig. 2).

View larger version (37K): [in this window] [in a new window]   Figure 2   PAC clone contig of the USH1C region generated by SEGMAP using STS content mapping and YAC-derived Alu-PCR and PAC end probe hybridization data. Clone names and sizes (kb) are indicated as in Fig. 1B. Marker names ending with SP6R or T7L are PAC end probes generated from the SP6 and T7 ends, respectively, of the indicated clone.

Analysis by SEGMAP generated a 1.3-mb contig consisting of 55 clones and 42 markers including 5 microsatellite markers and 31 PAC end probes. The sulfonylurea receptor 1 (SUR1) gene, a candidate disease gene for familial persistent hyperinsulinemic hypoglycemia (PHHI; Thomas et al. 1995) and the -cell inward rectifying potassium channel (BIR) gene, both of which had been mapped previously to this region (Ayyagari et al. 1996), were positioned in the contig and ordered with respect to other markers. Sequence analysis of a CpG island shared by PACs 169P3 and 306B4 identified a 5 exon from the previously unmapped NEFA protein gene (Barnikol-Watanabe et al. 1994). PCR and Southern analysis located NEFA just centromeric to D11S921 (Fig. 2). The higher resolution of the PAC map (cf. YAC contig) permitted the placement of D11S1888 telomeric to KCNC1 (Fig. 2). Finally, the order of markers defined by this contig is CEN-D11S130-KCNC1-D11S1888-MYOD1-D11S902-SUR1-BIR-NEFA-D11S921-D11S1890-TEL, confirming and refining the order defined by the YAC contig (Fig. 1B).

Large-Scale Sequencing through the USH1C Critical Region

The combination of large-scale genomic sequencing and the availability of the unprecedented resource of dbEST (and the corresponding cDNA clones) has allowed us to greatly increase the density of potential candidate genes in the USH1C region. Three PAC clones (239B22, 306B4, and 169P3) have been sequenced thus far, providing close to 300 kb of sequence in 10 contigs (Table 2; GenBank accession nos. AC000406-AC000407). The sequence data were analyzed with PowerBLAST (Zhang and Madden 1997), a newly developed program that provides graphic representations of BLAST searches, and the exon-prediction algorithm GRAIL.

Known Genes

View larger version (11K): [in this window] [in a new window]   Figure 3   Transcript map of a portion of the USH1C critical region. Known genes and ESTs are shown relative to the sequenced PAC clones. Arrows indicate the direction of transcription; CpG islands are indicated as lollipops.

Expressed Sequence Tags/Novel Genes

Exon and CpG Island Prediction Using GRAIL

Mutation Analysis of NEFA in USH1C Patient cDNA

The predicted amino acid sequence of NEFA indicates the presence of a DNA-binding domain containing a leucine zipper and two helix-loop-helix (HLH) domains that exhibit EF hand motifs, a property of certain Ca²⁺-binding proteins. Barnikol-Watanabe et al. (1994) postulate that these sites may play a role in the modification of protein folding similar to that seen for calmodulin (Strynadka and James 1989). The presence of EF hand domains in NEFA protein suggested the possibility that this protein could interact with unconventional myosins in a manner analogous to calmodulin (Wolenski 1995). If true, NEFA would be an attractive candidate gene considering that the USH1B locus encodes unconventional myosin VIIA (Weil et al. 1995).

Because a founder effect is likely for USHIC in the Acadian families segregating the disease (Smith et al. 1992), only one mutant allele is expected to be present in this population. For this reason, we opted to do mutation analysis by sequencing to approach 100% detection efficiency. Primers were designed to amplify the 1.6-kb NEFA cDNA in four overlapping amplimers of ~450 bp each. Following PCR, the products were sequenced in both directions. No changes in the nucleotide sequence of the amplified NEFA cDNA were observed in USH1C patients compared to normal.

	DISCUSSION

Top Abstract Introduction Results Discussion Methods References

The USH1C locus was originally mapped in French-Acadian families by linkage and haplotype analysis between D11S861 and D11S899 (Keats et al. 1994). Toward the positional cloning of the USH1C disease gene, we have constructed a 3.5-Mb YAC contig encompassing loci mapping between these flanking markers (Weissenbach et al. 1992; James et al. 1994; Keats et al. 1994; Fantes et al. 1995). Although this region has been isolated previously using CEPH mega-YACs, the order of many critical markers could not be established definitely because of apparent deletions or other rearrangements in the YAC clones (Ayyagari et al. 1996). Furthermore, many of the YACs were chimeric, limiting their downstream utility (Ayyagari et al. 1996). Except for a single mega-YAC, the contig presented here consists of smaller chromosome 11-specific YACs, which have a lower incidence of rearrangement and chimerism (Qin et al. 1993, 1996). Many of the previously unordered markers in earlier mega-YAC contigs (Chumakov et al. 1995; Ayyagari et al. 1996) were ordered in the present clone contig (Fig. 1B). This order was consistent with that determined by the single-linked mega-YAC contig but not the double-link contig constructed by The Whitehead Institute/MIT Center for Genome Research (1997). Our marker order is also at odds with the placement of D11S921 and D11S1890 centromeric to D11S902 and D11S1888 by radiation hybrid mapping; however, the latter order was not established at 1000:1 odds (James et al. 1994). The order determined in the YAC contig shown in Figure 1B is consistent with that determined by somatic cell hybrid mapping, although it could be argued that these two methods are not independent, as both the hybrid mapping panel and the chromosome 11 YAC library were derived from the monochromosomal hybrid J1 (Koa et al. 1976; Qin et al. 1993). However, the order was confirmed by STS-content mapping using PAC clones (Fig. 2) that were made from an independent DNA source (Ioannou et al. 1994).

The confirmed order of microsatellite markers (CEN-S1310-S1888-S902-S921-1890-TEL) allowed reconstruction of haplotypes of affected individuals that pointed to the 400-kb region between D11S902 and D11S1890 (Fig. 1B; Table 1) as harboring a gene involved in USH1C. This localization represents a significant (five- to eightfold) reduction in the size of the critical region for this disease. Based on earlier mapping studies (Ayyagari et al. 1996), the genes coding for the potassium channel protein KCNC1 and myogenic factor 3 (MYOD1) were considered as candidate loci for USH1C. Our refinement of the critical region to between D11S902 and D11S1890 now excludes these two loci as the disease-associated gene (Figs. 1B and 2), consistent with failure to detect mutations in the KCNC1 gene in USH1C patients (Marietta et al. 1997).

The tubby and rd5 mouse phenotypes display both retinal and cochlear degeneration and have been suggested as murine models for Usher syndrome, as these mutations map to distal mouse chromosome 7, which is partially syntenic to 11p15 (Heckenlively et al. 1995; Ohlemiller et al. 1995; Chung et al. 1996). However, based on both genetic and physical mapping studies, a relationship with USH1C seems unlikely. Chung et al. (1996) mapped the mouse tub locus near Hbb >14 cM distal to Sur, the human homolog of which maps within the USH1C critical (this study; Ayyagari et al. 1996). Kleyn et al. (1996) identified tub exons in a mouse P1 clone that also contained the Rbtn1 gene. In humans, pulsed-field gel electrophoresis locates RBTN1 >6 mb telomeric to MYOD1 (Higgins et al. 1994), which maps very close to the USH1C locus (Figs. 1 and 2). Thus, USH1C and tub appear to be distinct genetic loci. Similarly, the Rd5 mouse is unlikely to be the murine homolog of USH1C, as the Rd5 locus maps only 2 cM from Hbb (Hechenlively et al. 1995). The human STEP (striatum-enriched phosphatase) gene was mapped to 11p15.1-15.2 and was suggested as a potential candidate for USH1C (Li et al. 1995). PCR analysis of the PAC contig shown in Figure 2 with primers for STEP was negative indicating that this gene does not map within the USH1C critical region.

Mutations in genes coding for unconventional myosins result in USH1B (Weil et al. 1995), DFNB2 (Liu et al. 1997; Weil et al. 1997), and the mouse Snell's waltzer mutant (Avraham et al. 1995). To determine whether a similar situation exists in USH1C, the shortest path of PACs through the critical region was hybridized at normal and reduced stringency (2× SSC, 0.5% SDS at 55°C) with a cDNA clone corresponding to the unconventional myosin VIIA gene (USH1B locus). Consistent with a similar experiment (Ayyagari et al. 1996), no specific hybridization was observed (not shown).

The PAC contig has provided reliable DNA templates for large-scale genomic sequencing. PowerBLAST and GRAIL analyses of the available sequence have allowed the construction of a transcript map encompassing roughly two-thirds of the USH1C critical region. This map permitted the precise localization and transcriptional orientation of two genes (BIR, SUR1) that had been shown previously to map to this region (Ayyagari et al. 1996) as well as the unmapped, but previously described gene NEFA (Barnikol-Watanabe et al. 1994). Large-scale sequencing and computational analyses also identified a collection of potential novel genes in the USH1C critical region as evidenced by clusters of ESTs, CpG islands, and GRAIL-predicted exons (Table 2; Fig. 3). Because of their localization, these genes and novel transcripts represent candidate disease genes for USH1C.

By Northern blot, SUR1 and BIR are primarily expressed in the islet cells of the pancreas (Inagaki et al. 1995), although we have detected low-level expression of these two genes in retina by RT-PCR (not shown), making them candidate genes for USH1C. However, loss-of-function mutations in the SUR1 gene have been identified in some patients with PHHI (Thomas et al. 1995), a condition typified by unregulated secretion of insulin. Given the precedents of other syndromic and nonsyndromic forms of hearing loss [e.g., MYOVIIA in USH1B (Weil et al. 1995), PAX3 in Waardenburg syndrome (Baldwin et al. 1992; Tassabehji et al. 1992), and Cx26 in DFNB1 (Kelsell et al. 1997)], it is likely that alterations in the USH1C gene will also be loss-of-function mutations. There is no apparent phenotypic overlap between PHHI and USH1C, making it unlikely that the SUR1 gene is involved in USH. It has been suggested that the BIR protein may interact with the SUR1 gene product in the formation of one or more pancreatic -cell potassium channels (Inagaki et al. 1995). If true, then the lack of characteristics in common between PHHI and USH1C patients also argues against BIR being the locus responsible for USH1C.

The NEFA gene maps to a PAC clone that also contains D11S921, a marker found to be homozygous for allele 3 in all patients affected with USH1C (Fig. 3; Table 2). Its map location as well as some speculative properties of the NEFA protein, prompted us to test this gene for mutations in patients. However, sequence analysis of the NEFA cDNA of a USH1C patient failed to reveal any mutation in the protein coding region. This result, however, does not exclude the involvement of NEFA and USH1C, as mutations affecting regulatory regions have not been assessed. Further assessment of this gene as a candidate for USH1C is warranted.

As indicated in Table 2, PowerBLAST analysis identified the microsatellite markers AFMb340wf5 within the NEFA gene. A computer search of the available genomic sequence identified four additional loci with sufficient dinucleotide repeats to be polymorphic. Additional haplotyping of USH1C patients with these markers should prove useful in further narrowing the critical region.

The PAC clone contig presented here should greatly facilitate the identification of the gene associated with USH1C. The ease (cf. YACs) in obtaining high-quality DNA from clones representing the USH1C critical region has resulted in the identification of several ESTs or potential candidate genes by large-scale sequencing and computer analysis. These same reagents can be used for complementary gene finding approaches, such as exon trapping and cDNA selection. The identification and characterization of the USH1C disease gene and its product will aid in our understanding of both syndromic and nonsyndromic hereditary hearing loss in general and in the pathology of USH syndrome in particular.

	METHODS

Top Abstract Introduction Results Discussion Methods References

Patient Identification

Individuals with the presumed diagnosis of USH1 were examined at the Catholic Deaf Center (LaFayette, LA). The diagnosis of USH1 was established by clinical criteria (Smith et al. 1994). French-Acadian USH1C families investigated in this study are described in Smith et al. (1992).

Genotyping

To determine individual genotypes, 30 ml of blood was collected from each person in EDTA-containing tubes. Genomic DNA was prepared from peripheral lymphocytes (Grimberg et al. 1989) and amplified by PCR as described previously (Smith et al. 1992). Markers selected for analysis were known to be tightly linked to the USH1C locus (this study; Smith et al. 1992; Keats et al. 1994).

Hybrid Panel Analysis

Genomic DNAs (100 ng) from J1-4b, J1-8, J1-37 (Koa et al. 1976), NW (Gessler et al. 1989), human (GM00131), and hamster (CHW1102) were tested by PCR for the presence of D11S416, D11S902, D11S921, D11S1310, D11S1888, and D11S1890. Primer pairs were obtained from GDB. MYOD1, and KCNC1, were mapped by Southern hybridization using the same DNA and plasmid probes Myf3 and HNGK4, respectively.

YAC Library Screening

A 4X chromosome 11 YAC library (Qin et al. 1993) was screened by PCR using microsatellite markers and the pooling scheme described in Qin et al. (1996). DNA pools for the CEPH-A library (Albertson et al. 1990) were obtained from Research Genetics and screened with primers for D11S1310. In some cases, small YAC contigs nucleated with an STS marker were expanded using an Alu-PCR hybridization strategy (Qin et al. 1996).

PAC Library Screening

PAC clones were identified in libraries RPCI-1, 3, 5, and 6, constructed as described (Ioannou et al. 1994), but in the pCYPAC2 or pPAC4 (RPCI-6) vectors. Pools of radiolabeled probes were hybridized to high-density filters each containing ~18,000 unique clones. Positive clones were grown in 96-well plates, transferred to filter replicates, and hybridized to individual probes as described (Ioannou et al. 1994). A variety of probes were used to detect clones, including PCR products generated from STSs, Alu-PCR products derived from YACs (Qin et al. 1996), and end fragments isolated from PACs (see below). All probes were labeled by random-priming (Feinberg and Vogelstein 1983) and hybridized as described (Church and Gilbert 1984).

YAC and PAC End Isolation

The ends of YAC inserts were isolated by the ligation-mediated PCR method of Kere et al. (1992). PAC insert ends were rescued by a modification of this method using vector primers immediately adjacent to the BamHI cloning site in pCYPAC2 or pPAC4 (Cooper et al. 1997).

Clone Verification and Contig Assembly

After preparing individual YAC DNA in agarose by the lyticase/lithium dodecyl sulfate method (Anand et al. 1990), each clone was tested by PCR with the STS used to identify it. Overlap detected by hybridization of Alu-PCR products was confirmed by Southern analysis of Alu-PCR fingerprints. YACs were sized by pulsed-field gel electrophoresis using a CHEF-DRII apparatus (Bio-Rad) in 1% agarose gel (0.5× TBE) with a 10- to 50-sec switch time ramp (200 V, 29 hr), blotting to Gene Screen Plus (NEN), and hybridization with human Cot1 DNA (GIBCO BRL).

PAC DNA was prepared by an alkaline/SDS lysis procedure. PAC clones identified by Alu-PCR products from YACs were labeled by random priming and mapped back to YAC clones by Southern blotting. All other PAC clones were verified by Southern hybridization with single-copy plasmid probes and gel-purified STS-generated PCR products. Isolated end probes were used to confirm overlap between clones by Southern hybridization and to extend initial contigs by rescreening high-density PAC filters. PAC inserts were sized by NotI digestion of clone DNA and separation in CHEF gels as described above but with a switch time of 8 sec and running time of 22 hr.

YAC clones were mapped by FISH as part of a larger project (Qin et al. 1993, 1996). PAC clones were localized by FISH essentially as described (Sait et al. 1994) using between 100 and 200 ng of clone DNA. Contigs were assembled using the program SEGMAP (C. Magness, Y. Xu, and P. Green, unpubl.).

Large-Scale Sequencing

Large-insert bacterial clones were shotgun cloned into M13 phage and sequenced. The single-pass sequence traces from 1000 to 1500 clones per PAC were analyzed by Phred (P. Green and B. Ewing, University of Washington, Seattle) and assembled into contigs using Phrap (P. Green). Genomic DNA sequence was assessed for gene coding potential using the client program PowerBLAST (ftp://ncbi.nlm.nih.gov/pub/sim2/PowerBLAST/) and a network version of GRAIL (http://avalon.emp.ornl.gov/GRAIL-bin/EmptyGrailForm). The latter web page provided for the identification of CpG islands as defined by Gardiner-Garden and Frommer (1987).

Mutation Analysis of the NEFA Gene

Total RNA was isolated from lymphoblast cell cultures of affected and nonaffected individuals using the RNeasy minikit (Qiagen). Following treatment with RNase-free DNase (GIBCO BRL), first-strand cDNA synthesis was carried out using an oligo(dT) primer and Superscript reverse transcriptase (GIBCO BRL) as described by the manufacturer. PCR was then performed using eight oligonucleotides designed to amplify all of the NEFA cDNA (GenBank accession no. X76732) in four amplimers, each 400-500 bp in length. The primer sequences were F1, CGCCGACACCCGGCCAAGAAC; R1, CCAGTTCTTTGCTTAGCCTCCC; F2, CTCAAGCAAGTGATTGATGTGCTGG; R2, GCTTCCTGGGTGATTAACTTTAGG; F3, GGAACACTATGACAAGACTCGTCA; R3, ACTCCTCCAGAGTCACCAATCTG; F4, GAAAGGCTTAGAATGAGGGAACATG; and R4, GAAATAGATGTTGAGTTAACAGC. Amplifications were done in GeneAmp PCR buffer with 1.5 mM MgCl₂ (F1-R1 and F4-R4) or in TNK50 (Kere et al. 1992) (F2-R2 and F3-R3) using a touchdown program with two cycles each at annealing temperatures ranging from 63°C to 57°C (1 min each), followed by 25 cycles at an annealing temperature of 56°C. Together, these four amplimers cover the entire NEFA cDNA except for 50 bp of the 5 UTR and 106 bp of the 3 UTR. The four PCR products were cloned into pCR-Script Cam (Strategene). The clones were sequenced using fluorescently tagged dideoxy chain terminators and an ABI 373A sequencer. The sequence of the NEFA cDNA from an individual with USH1C was compared with the wild-type sequence using the computer program GeneWorks (Oxford Molecular Group).