Mutations in a Novel Gene with Transmembrane Domains Underlie Usher Syndrome Type 3

Abstract

Usher syndrome type 3 (USH3) is an autosomal recessive disorder characterized by progressive hearing loss, severe retinal degeneration, and variably present vestibular dysfunction, assigned to 3q21-q25. Here, we report on the positional cloning of the USH3 gene. By haplotype and linkage-disequilibrium analyses in Finnish carriers of a putative founder mutation, the critical region was narrowed to 250 kb, of which we sequenced, assembled, and annotated 207 kb. Two novel genes—NOPAR and UCRP—and one previously identified gene—H963—were excluded as USH3, on the basis of mutational analysis. USH3, the candidate gene that we identified, encodes a 120-amino-acid protein. Fifty-two Finnish patients were homozygous for a termination mutation, Y100X; patients in two Finnish families were compound heterozygous for Y100X and for a missense mutation, M44K, whereas patients in an Italian family were homozygous for a 3-bp deletion leading to an amino acid deletion and substitution. USH3 has two predicted transmembrane domains, and it shows no homology to known genes. As revealed by northern blotting and reverse-transcriptase PCR, it is expressed in many tissues, including the retina.

Introduction

The Usher syndromes (USHs) are characterized by variable degrees of loss of hearing and of loss of eyesight. Three distinct phenotypes have been described. Patients with USH1 have congenital severe-to-profound hearing loss and absence of vestibular function, whereas patients with USH2 have congenital moderate-to-severe hearing loss and normal vestibular function. The onset of retinitis pigmentosa (RP) is prepubertal in patients with USH1 and occurs during the 2d decade of life in patients with USH2 (Smith et al. 1994). Patients with USH3 have postlingual, progressive hearing loss, and the onset of retinitis pigmentosa (RP) symptoms—nyctalopia, progressive constriction of visual fields, and reduction of central visual acuity—usually occurs by the 2d decade of life (Pakarinen et al. 1996). At least 10 loci account for subtypes of USH (see the Hereditary Hearing Loss Homepage). Five genes have so far been identified: myosin VIIA (MYO7A), for USH1B (MIM 276903) (Weil et al. 1995); usherin, for USH2A (MIM 276901) (Eudy et al. 1998); harmonin, for USH1C (MIM 276904) (Bitner-Glindzicz et al. 2000; Verpy et al. 2000); cadherin 23 (CDH23), for USH1D (MIM 601067) (Bolz et al. 2001; Bork et al. 2001); and protocadherin 15 (PCDH15), for USH1F (MIM 602083) (Ahmed et al. 2001).

We previously assigned USH3 to 3q21-q25 (Sankila et al. 1995). Significant linkage disequilibrium between USH3 and the linked markers suggested enrichment of a founder mutation in Finland and allowed us to refine the localization of USH3 to a 250-kb genomic interval between markers 107G19CA7 and D3S3625 (Joensuu et al. 1996, 2000). Haplotype analysis indicated 107G19CA7 to be the proximal historic recombination breakpoint, and a recombination in one family with USH3 placed the USH3 gene proximal to D3S3625. In this study, we (a) sequenced, assembled, and annotated 207 kb of the USH3 critical region, (b) excluded two novel genes—NOPAR and UCRP—and one previously identified gene—H963—as candidates for USH3, and, finally, (c) positionally cloned USH3.

Subjects and Methods

Families with USH3 and Controls

Clinical and genealogical data on the Finnish and Italian families with USH3 that are studied here also have been described elsewhere (Pakarinen et al. 1995; Sankila et al. 1995; Joensuu et al. 1996, 2000; Gasparini et al. 1998). Control DNA samples from 100 anonymous blood donors from eastern Finland (where there is a high prevalence of USH3) and from 100 donors from western and southern Finland (where there is a low prevalence of USH3) were provided by the Finnish Red Cross Blood Transfusion Service. A total of 51 grandparents in families from the Centre d'Étude du Polymorphisme Humain (CEPH) were also studied as controls.

Bacterial Artificial Chromosome (BAC) DNA and Shotgun Sequencing

We shotgun sequenced partly overlapping BACs—specifically, 355L4 and 545J16. Shotgun libraries were prepared and plasmid DNA was extracted by GATC Shotgun Library Construction service. DNA fragments were ligated to the PCR-Blunt II-Topo cloning vector (Invitrogen), and we sequenced the subclones, from both directions, with custom vector primers designed by GATC and with BigDye-terminator chemistry (PE Applied Biosystems), by use of a PE9700 Thermal Cycler (Perkin Elmer). Sequence readings were transferred to a UNIX-based system; edited and filtered by removal of matches to the cloning and sequencing vectors, as well as to Escherichia coli, by the GASP software package (University of Washington Genome Center); and assembled by Phil Green's Phred, Phrap, and Consed sequence-assembly programs. The orientation of contigs, from centromere to telomere, was determined on the basis of the marker order of the physical map described elsewhere (Joensuu et al. 2000).

Computational Analyses and Sequence Annotation by Integrated Gene-Prediction Programs

The complete genomic sequence assembled was masked for repetitive elements, by the RepeatMasker2 program (University of Washington Genome Center). Sequence-similarity searches were performed using BLAST programs (Altschul et al. 1997) against different databases, Human Index for Nonredundant Transcripts (HINT) (Zhuo et al. 2001), UNIGENE (Boguski and Schuler 1995), and the TIGR Human Gene Index (Liang et al. 2000). The sequences were compared with the High-Throughput Genome Sequence and Genomic Survey Sequence databases (National Center for Biotechnology Information), to identify BAC clones or P1-derived artificial clones mapping to the same region. To compare orthologous mouse genomic sequences, homology searches were performed by use of the Celera Assembled and Annotated Mouse Genome. Complete sequence annotation was performed by a modified computational sequence annotation tool, Genotator (Harris 1997), and by the NIX program (UK Human Genome Mapping Project Resource Centre). Transcription-factor and promoter predictions at the 5′-flanking genomic sequences of identified genes were performed by MatInspector/TRANSFAC and Neural Network Promoter Input programs (Baylor College of Medicine Search Launcher). Protein predictions were performed by the PIX program (UK Human Genome Mapping Project Resource Centre), LASERGENE (DNASTAR), TMpred (Swiss Institute of Bioinformatics), SignalP (Center for Biological Sequence Analysis), and PSORT II (Human Genome Center).

DNA and cDNA

We extracted BAC DNA as described elsewhere (Joensuu et al. 2000). Integrated Molecular Analysis of Genomes and Their Expression (IMAGE) cDNA clones for expressed-sequence tags (ESTs) homologous to the USH3 region were obtained from Research Genetics. To obtain the complete sequence of the inserts, cDNA clones were sequenced on both strands by vector-specific primers.

Human-placenta RNA and Marathon Ready Human Retina cDNA were obtained from Clontech. The first-strand cDNA synthesis was performed with 1 μg of total RNA, according to the supplied protocol for the SMART PCR cDNA Synthesis kit (Clontech). Subsequently, 2 μl of the 100-μl reaction mixture was PCR amplified by gene-specific primers. To isolate full-length cDNAs of the predicted genes, we performed rapid amplification of cDNA ends (RACE) PCR with the Marathon Ready Human Retina cDNA libraries, according to the manufacturer's instructions, by a modified PCR with annealing at 65°C for 1 min 30 s, for 20 cycles, followed by nested PCR with annealing at 60°C for 1 min 30 s, for 40 cycles, in a total volume of 50 μl.

We designed combinations of forward and reverse first-amplification primers and nested primers with the program Primer3 (Whitehead Institute for Biomedical Research/Massachusetts Institute of Technology Center for Genome Research). When possible, the primer pairs were designed to span introns in order to minimize false-positive results by contaminating DNA in the reverse-transcriptase PCRs (RT-PCRs). PCR primer sequences and conditions are available on request.

PCR and RT-PCR

We analyzed candidate genes by PCR amplification and direct sequencing of all exon regions with adjacent splice sites, from genomic DNA and, when possible, from total RNA and poly(A) RNA—using the Qiagen Total RNA kit and the poly(A) RNA–extraction kit provided by Amersham, respectively—from Epstein-Barr virus–transformed lymphoblasts extracted from two patients with USH3, two carriers, and two controls. Amplification of genomic DNA was performed as described elsewhere (Joensuu et al. 1996). For RT-PCR, either 200 ng of lymphoblast poly(A) RNA or 1 μg of total RNA was primed by random hexamers and was reverse transcribed for the first-strand synthesis of cDNA. Subsequent nested PCR from cDNA was essentially performed with 2 μl of the 20-μl cDNA reaction mix in a total volume of 25 μl with 0.2 mM of each dNTP, 20 pmol of each primer, and 0.5 U of AmpliTaq Gold DNA polymerase (PE Applied Biosystems). Control reactions containing human genomic DNA and water, without cDNA, were performed. Amplification of GC-rich regions was attempted with 1.2 mM betaine or 7% dimethyl sulfoxide.

We electrophoresed PCR products on either 1%–2% SeaKem agarose gel or 0.7 × MDE SSCP (FMC) polyacrylamide gels and visualized them by ethidium-bromide or silver staining (Bassam et al. 1991). Direct sequencing of the purified amplification products (PCR purification kit; Qiagen) was performed by automated sequencing (ABI 377-sequencer; PE Applied Biosystems) by use of the PCR primers.

Analysis of Novel Single-Nucleotide Polymorphisms (SNPs)

By genomic sequencing of NOPAR, in families with USH3, we identified novel SNPs—RSNP2 and TSNP1, in NOPAR introns 1 and 9, respectively. Intronic forward and reverse primers used in the genotyping of the families with USH3 were as follows: TSNP1 (forward, 5′-TGGTAGGTAATCTTAGTCCAACAA-3′; reverse, 5′-TTTTGCCTCACATAGAATGGTG-3′), 390 bp; and RSNP2 (forward, 5′-CATGCACCAAGCTCTCACAT-3′; reverse, 5′-AAGCTGAGGGAAGAAACTGC-3′), 264 bp. The alleles were detected either by direct sequencing or by SSCP analysis, in 0.7 × MDE gel, for 18 h at 5 W, at room temperature. The SSCP alleles were visualized by silver staining. Haplotypes were manually constructed under the assumption that there were a minimum number of recombinations.

Isolation of the Candidate Gene USH3

The interexon 5′- and 3′-UTR primers used in combining the exons from cDNA synthesized, by PCR (conditions were as described above), from human-retina total RNA or the Marathon Ready Human Retina cDNA library (Clontech) were as follows: E3F (5′-CTCCTGTGGCTGTCTTGTCA-3′) and W27577R (5′-TGGTGGGTTTGCCTCTTAGT-3′); E2F (5′-AAGCAATCCCAGTGAGCATC-3′) and E3R (5′-TGACAAGACAGCCACAGGAG-3′); E1F (5′-TCACTATCTGAAACTATCTTGTTGT-3′) and E2R (5′-TAGGGGACCATGCAGAGTTT-3′); and E1F2 (5′-TGCCTCCCCACCATTCACCA–3′) and E2R2 (5′-CATGAAGAAGGCTGTCCCCACCA-3′).

Mutation Detection

Using genomic DNA, we amplified genomic fragments encompassing five USH3 exons, by use of the following intronic primers: E1F (see above) and P1R (5′-AAGCCCCTGAACTTTATAGG-3′), 910 bp; P1BF (5′-TTGTGGCCATTTTTGGAGAT-3′) and P1BR (5′-CCCCAAACATGTATCAAGTGC-3′), 207 bp; P2F (5′-TCAGAAGGATTTTAGTGATGTTTGA-3′) and P2R (5′-TCTTTTTGACATATTGAAAAGCACA-3′), 352 bp; P3F (5′-ATGTCAATGGGGATGATGGT-3′) and P3R (5′-GGAGCCCATTCAGAAAATGA-3′), 291 bp; and P4F (5′-TTCCCCTGAATTACCCATCA-3′) and P4R (5′-AGCATCTGGAAACTCGGTGT-3′), 339 bp. PCR products were purified and sequenced as described above.

The Fin_major mutation and the Italian 3-bp deletion were screened by direct sequencing of the 291-bp product amplified by primers P3F and P3R. For the Fin_minor mutation, we used SSCP; a 198-bp fragment was PCR amplified by primers P2F2 (5′-TCCCAGTGAGCATCCACGTC-3′) and P2R2 (5′-TGAAAAGCACATTTGTCTTCAGAGG-3′), and the alleles were separated on 0.7 × MDE gels, for 18 h at 5 W, at room temperature and were visualized by silver staining.

Northern Blot and In Situ Hybridization

A human multiple-tissue northern blot from 12 adult tissues was hybridized by use of an α[³²P]-dCTP random-labeled cDNA probe (USH3 nucleotides −33–321; 354 bp) comprising exons 1–3 and part of the 5′ UTR. The membrane was washed at 65°C with 2 × SSC (saline sodium citrate) and 0.05% SDS, then exposed to x-ray film at −80°C for 2 d. In situ hybridization and the production of cRNA probes were performed as described by Wilkinson and Green (1990). In short, 4% paraformaldehyde–fixed paraffin sections from embryonic 16-d-old mouse embryos and from dissected adult eyes and inner ears were used for analysis. A 262-bp fragment from the Marathon Ready Mouse Brain cDNA library (Clontech) was produced by primers mouse/p1 (5′-ATCCCCGTAAGCATCCACAT-3′) and mouse/p2 (5′-TTCTGTAGGCATAGGTCCCTTC-3′). The amplified fragment was cloned into the pGEM-T Easy vector (Promega) according to the manufacturer's instructions. Subsequently, [³⁵S]-labeled antisense and control-sense cRNA probes were synthesized by in vitro transcription. The control-sense probe did not produce any hybridization signal above background level.

Results

Construction of the Genomic Contigs and Transcript Map

We chose two partly overlapping BAC clones (355L4 and 545J16) flanking both 107G19CA7 and most of the previously defined USH3 critical region toward D3S3625, for shotgun sequencing. Database-derived BAC clone RP11-385G14 sequence was aligned with genomic contigs of BAC clone 545J16, to extend the distal sequences. Gap filling was performed by primer walking from the contig end sequences and by PCR across the gaps. A final sequence contig of 207 kb, which is depicted in figure 1, was assembled.

Figure 1.

Schematic representation of physical and transcript maps covering the USH3 region, depicting the initial 250-kb USH3 region and the extended 133-kb USH3 region, with partly overlapping BAC clones. Polymorphic markers and sequence-tagged sites are indicated by vertical lines. Previously mapped BAC clones are shown in dark blue, and the 207-kb sequence contig assembled is indicated by a purple line. Database-derived and partly assembled BAC sequences from clones RP11-385G14 and RP11-251C9 are indicated by green dashed arrows. The centromere is on the left. A pair of previously known genes—KIAA0001 and H963—and a group of three novel genes—NOPAR, UCRP, and USH3—are indicated by yellow and light-blue arrows, respectively, showing the orientation of transcription. ESTs representing parts of the 3′ UTRs of the transcripts are indicated by vertical arrows. H963, NOPAR, UCRP, and USH3 were subjected to mutation analyses in this study.

Use of nucleotide-nucleotide BLAST for searches of the contig identified 30 distinct or partly overlapping human ESTs and 11 orthologous mouse or rat ESTs (table 1). Starting from the identified ESTs and exons predicted by GRAIL and GENESCAN, we used a combination of RT-PCR, RACE experiments, and direct PCR of cDNA libraries, to assemble two novel genes: NOPAR (for no opposite-paired repeat) and UCRP (for USH critical region pseudogene) (fig. 1). We identified two alternatively spliced transcripts—NOPAR and NOPAR2—which encode putative polypeptide isoforms, of 721 and 756 amino acids, respectively, showing 61% identity to the amino terminus of TRAP230 (Ito et al. 1999), also called the “human OPA-containing gene,” or “HOPA” (Philibert et al. 1999). The NOPAR coding region comprises approximately one-third of the length of TRAP230/HOPA and lacks the 3′ OPA repeat contained in TRAP230/HOPA. Northern blot analysis of NOPAR showed widespread expression in adult tissues (data not shown). Transcription in human retina was shown by RT-PCR.

Table 1.

Location of ESTs in the 207-kb Contig

EST	Location(s)	GenBankAccession Number	dbEST Details
1	24264–24640	R75628	Homo sapiens cDNA clone IMAGE: 143509 3′
2	65074–65380	R79610	H. sapiens cDNA clone IMAGE: 146349 5′
3	65715–65825	T39157	H. sapiens cDNA clone IMAGE: 60197 5′
4	66204–66483	T40442	H. sapiens cDNA clone IMAGE: 60197 3′
5	66139–66502	R79611	H. sapiens cDNA clone IMAGE: 146349 3′
6	78227–78422, 107727–107829, 114184–114230	AI501968	Rattus norvegicus cDNA clone UI-R-C0-jb-b-08-0-UI 3′
7	78227–78422, 107727–107829, 114184–114251	AW920576	EST351880 Bento Soares R. norvegicus cDNA clone RGIGZ01 5′
8	79115–79401	AW897677	CM1-NN0062-280400-202-d05 NN0062 H. sapiens cDNA
9	79118–79401	AW897669	CM1-NN0062-280400-202-b05 NN0062 H. sapiens cDNA
10	99994–100283	AA776659	H. sapiens cDNA clone IMAGE: 970436 3′
11	78227–78422, 107727–107829, 114184–114241	BE948099	UI-M-BH3-aws-h-03-0-UI.s1 Mus musculus cDNA clone UI-M-BH3-aws-h-03-0-UI 3′
12	136681–136758	AF062710	AF062710 H. sapiens cDNA clone HA0028
13	136681–136758	AI110602	H. sapiens cDNA HA0028
14	181289–181421	AA931311	H. sapiens cDNA clone IMAGE: 1565326 3′
15	186110–186218	BE843843	RC0-TN0079-310700-021-b10 TN0079 H. sapiens cDNA
16	186131–186641	AA170490	M. musculus cDNA clone IMAGE: 618166 5′
17	186185–186518	AW591154	H. sapiens cDNA clone IMAGE: 2703652 3′
18	186193–186446	BB209769	BB209769 M. musculus cDNA clone A430094C06 3′
19	186193–186773	BE283279	601101050F1 M. musculus cDNA clone IMAGE: 3493212 5′
20	186235–186446	AI172577	R. norvegicus cDNA clone UI-R-C2p-nz-h-03-0-UI 3′
21	186313–186789	AA162789	M. musculus cDNA clone IMAGE: 598871 5′
22	186665–187193, 190857–190910	AV653286	AV653286 GLC H. sapiens cDNA clone GLCDJC11 3′
23	186660–187193, 190857–190910	AV653266	AV653266 GLC H. sapiens cDNA clone GLCDJB03 3′
24	186676–187016	AV757663	H. sapiens cDNA clone: BMFBHF02, 5′ end
25	186903–187075	AA184698	M. musculus cDNA clone IMAGE: 634121 3′
26	186952–187148	AI592667	M. musculus cDNA clone IMAGE: 634121 5′
27	187086–187195, 190857– 190954	AA353758	EST61951 H. sapiens cDNA 5′
28	200484– 201179	AU138895	H. sapiens cDNA clone: PLACE1009521, 5′
29	204146– 204843	AL042755	H. sapiens mRNA; EST DKFZp434C0522_r1
30	199885– 200422	AU157802	H. sapiens cDNA clone: PLACE1009521, 3′
31	205968– 206348	AL042756	H. sapiens mRNA; EST DKFZp434C0522_s1
32	199885–200231	AI167543	Soares_NhHMPu_S1 H. sapiens cDNA clone IMAGE: 1661439 3′
33	199883–200242	AA027011	Soares_pregnant_uterus_NbHPU H. sapiens cDNA clone IMAGE: 469358 3′
34	199885–200183	AI277229	Soares_placenta_8to9weeks_2NbHP8to9W H. sapiens cDNA clone IMAGE: 1893709 3′
35	206135–206348	AI889440	H. sapiens cDNA clone IMAGE: 2444461 3′
36	200575–200753	AA026830	H. sapiens cDNA clone IMAGE: 469358 5′
37	200005–200189	AI686699	H. sapiens cDNA clone IMAGE: 2268648 3′
38	200580–200783	R35752	Soares placenta Nb2HP H. sapiens cDNA clone IMAGE: 136983 5′
39	202796–202998	AA056078	Soares retina N2b4HR H. sapiens cDNA clone IMAGE: 381530 3′
40	201215–201642	AI050350	M. musculus cDNA clone IMAGE: 1379171 5′
41	199880–200139	R35645	H. sapiens cDNA clone IMAGE: 136983 3′

The 2,045 bp of UCRP cDNA obtained consists of five exons spanning >70 kb genomic DNA. However, the open reading frame (ORF) is disrupted, and the gene thus is probably an unprocessed pseudogene. The region also contained two previously identified genes: KIAA0001 (Nomura et al. 1994), which was previously excluded as USH3 (Joensuu et al. 2000), and platelet-activating–receptor homolog mRNA H963 (Jacobs et al. 1997).

Refinement of the USH3 Critical Region

The exons and flanking intronic sequences of NOPAR, UCRP, and H963 were examined for mutations in patients with USH3; however, no potential disease-causing mutations were detected. Genomic sequencing of the NOPAR introns in patients with USH3 indicated the presence of two novel SNPs, RSNP2 (nucleotide C/G) and TSNP1 (nucleotide T/C), in introns 1 and 9, respectively (fig. 1). Haplotype analysis revealed conservation of the segment RSNP2–TSNP1 in all 48 USH3 chromosomes studied (data not shown). This supported our previous haplotype analysis placing USH3 distal to 107G19CA7 (Joensuu et al. 2000); however, a recombination event in one family excluded the RSNP2-TSNP1-tel segment as the site of the causative mutation (fig. 2). No additional transcripts in the remaining 60-kb critical region between markers 107G19CA7 and RSNP2 were found. This implicated the region immediately centromeric to 107G19CA7 and extending to 25B8CA2, as part of the USH3 critical region.

Figure 2.

Pedigree of a Finnish family with USH3, segregating a paternal Fin_major mutation (maj)—c.300T→G—and a maternal Fin_minor mutation (min)—c.131T→A. Alleles at seven microsatellite markers, two SNPs, and USH3 are shown; wt = wild type. The chromosomes assumed to carry the disease allele are in boldface type. The father carries a conserved ancestral haplotype, whereas the mother’s disease-associated haplotype shares alleles at the RSNP2–D3S1594 segment only. A recombination in the maternal chromosome of individual II-1 excludes the segment RSNP2–D3S1279 as the site of the USH3 mutation. At the bottom is the result of an SSCP analysis used in the detection of the Fin_minor mutation. A mobility shift caused by the mutation was detected in individuals I-2, II-2, and II-4.

Identification of the USH3 Gene

We assembled database-derived RP11-251C9 sequences between 107G19CA7 and 25B8CA2. BLAST analyses of the newly constructed 133-kb contig revealed 11 ESTs (data not shown). Starting from a retina-derived EST 35f2 and exons predicted in its vicinity, a contiguous cDNA sequence of 1,444 bp was obtained by RT-PCR of human retina-derived total RNA (a gift from F. P. M. Cremers) and by PCR of Marathon Ready Human Retina cDNA. Sequence analysis suggested a 360-bp ORF comprising four exons that span ∼18 kb at the genomic level. By comparison of the genomic sequence with the USH3 cDNA sequence, the exon-intron boundaries and their flanking sequences were confirmed and were in agreement with the consensus GT-AG splicing rule. A splice variant, USH3 isoform b, comprising an additional 87-bp exon (1b) with an in-frame stop codon between exons 1 and 2 was also detected. The first in-frame ATG codon in exon 1 conforms with the Kozak motif (Kozak 1987). No splice acceptor was found at the 5′ end of exon 1, suggesting that this exon is the first. There is a 5′ UTR of ⩾392 bp and a 3′ UTR of ⩾692 bp. Computational promoter analysis revealed five possible promoters for USH3. The putative promoter regions lack TATA boxes but include CAAT and GC boxes. Three alternative polyadenylation signals were found, at 156 bp, 764 bp, and 3,568 bp downstream of the termination codon, consistent with both the northern blot analysis showing three transcripts, of 4.5 kb, 1.5 kb, and 1.0 kb, respectively, and expression in many adult-human tissues (fig. 3). Expression was also detected in adult-human retina, by RT-PCR. Only human lymphoblasts did not show a signal, as confirmed by both RT-PCR and northern blotting.

Figure 3.

USH3 expression. Northern analysis of an adult-human multiple-tissue blot was performed by use of a probe comprising the coding region of USH3. Two signals, of ∼4.5 kb and ∼1.0 kb, are detected in all tissues. In addition, a weak signal, of ∼1.5 kb, is detected in spleen mRNA. The blot was hybridized with a β-actin probe as a control (bottom).

The analysis with the Celera Assembled and Annotated Mouse Genome indicated several genomic fragments highly homologous to exons 2 and 3 of USH3. In situ hybridization with [³⁵S]-labeled riboprobes corresponding to the orthologous mouse sequence showed a low-level, ubiquitous expression in several mouse tissues (data not shown).

The candidate gene USH3 encodes a novel, putative transmembrane protein of 120 amino acids with a calculated molecular mass of 13.4 kD (fig. 4). It shows no homology to previously known genes or proteins. Bioinformatics analyses suggest a cytosolic amino terminus, two helical transmembrane domains (residues 25–41 and 63–79), and a putative endoplasmic-reticulum membrane–retention signal, TKGH, in the carboxy terminus.

Figure 4.

Predicted schematic structure and regions of secondary structure in the human USH3 gene and in USH3 protein. A, The five exons of USH3, depicted by numbered boxes (E1–E4). The lengths of the exons are shown below the boxes, and those of the three separating introns are shown above the lines. The translation-initiation site and the first stop codon are indicated. Three polyadenylation signals (Poly(A)) and their predicted locations downstream of the termination codon are indicated by arrows. B, One-hundred-twenty-amino-acid protein, encoded by USH3, with two predicted transmembrane domains, at residues 25–41 and 63–79. The combination of putative secondary structures, such as α-helices, β-pleated sheets, and β-turns, features possible functional protein domains. An alternatively spliced transcript (bottom) of USH3 predicts a 30-amino-acid protein.

USH3 Mutations

We identified three causative mutations in Finnish and Italian patients with USH3 (fig. 5). The Finnish founder mutation is a nonsense mutation, c.300T→G in exon 3 (Y100X, or Fin_major), predicting the truncation of 21 amino acids. Fin_major was found in 52 patients with USH3 who were homozygous for a conserved ancestral haplotype, whereas all 36 parents analyzed and 26 of 45 unaffected siblings were heterozygous for the mutation. The second Finnish mutation is a c.131T→A transversion at codon 44 in exon 2, resulting in an amino acid substitution of lysine for methionine (M44K, or Fin_minor). In two Finnish families, four patients are compound heterozygous for Fin_major and Fin_minor. These patients have inherited the common ancestral haplotype, from one parent, and a partly distinctive haplotype, from the other parent (fig. 2). The Fin_minor mutation originates from the same region in eastern Finland where USH3 is known to be prevalent. In a consanguineous Italian family with USH3 showing linkage to the 3q21-q25 markers (Gasparini et al. 1998), we detected a homozygous 3-bp deletion, c.231–233delATT in exon 3, resulting in the substitution of one methionine for isoleucine and leucine at codons 77 and 78.

Figure 5.

Detection of three different mutations segregating in families with USH3. A, Sequence analysis of a genomic PCR fragment comprising exon 3 in a normal control, in a patient with USH3 who is homozygous for the ancestral Fin_major mutation—c.300T→G (resulting in Y100X)—in a heterozygous carrier, and in a compound heterozygous patient (Fin_major/Fin_minor). Positions of mutated nucleotides are indicated by arrows and asterisks. B, Finnish mutation Fin_minor—c.131T→A (resulting in M44K) in exon 2: sequence chromatograms of a control and of a compound-heterozygous patient with USH3 (Fin_major/Fin_minor). C, Sequence chromatograms of a control, of a homozygous patient, and of a heterozygous carrier, representing the Italian mutation—a 3-bp deletion, c.231–233delATT—in exon 3. The deletion results in the substitution of one methionine for isoleucine and leucine. The deleted nucleotides are indicated below the normal control sequence.

By direct sequencing (in the case of Fin_major and the Italian mutation, c.231–233delATT) and by SSCP analysis (in the case of Fin_minor), none of either 100 anonymous blood donors from southern and western Finland or 51 grandparents in families from the CEPH were found to carry any of the three mutant alleles. Of 100 anonymous blood donors from eastern Finland, 1 was heterozygous for Fin_major. This was expected, because USH3 is clustered in eastern Finland (Pakarinen et al. 1995).

Discussion

The association of RP with sensorineural deafness is named “USH.” More than 60 genes responsible for RP and for related disorders have been identified (Retnet), and, to date, ∼20 genes for nonsyndromic deafness are known. Products of RP genes include phototransduction proteins, photoreceptor structural proteins, transcription factors, and proteins involved in metabolism of the photoreceptor and of the retinal pigment epithelium. The recent identification of several deafness genes by molecular genetic studies has enabled the investigation of the molecular basis of normal and pathological auditory function. The genes encode proteins that have a role in hair-cell transduction, in ionic homeostasis in the cochlear duct, and in integrity of the tectorial membrane (Steel and Kros 2001).

Five genes underlying USH have been identified in previous studies. Mutations of myosin VIIA cause nonsyndromic deafness as well as USH1B (Weil et al. 1995; Liu et al. 1997a, 1997b). The shaker-1 mice with myo7a mutations show progressive disorganization of the stereocilia bundle (Self et al. 1998). Although there is no retinal degeneration in shaker-1 mice, abnormal accumulation of opsin in the connecting cilium of photoreceptors and a failure of melanosome transport in the retinal pigment epithelium have been detected (Liu et al. 1998, 1999; Liu and Williams 2001). It is thought that harmonin interacts with myosin VIIA in the sensory cells of the inner ear, and harmonin mutations underlie USH1C as well as autosomal recessive nonsyndromic deafness DFNB18 (Verpy et al. 2000). Surprisingly, no harmonin expression was detected in neonatal-mouse eyes (Verpy et al. 2000), whereas anti-harmonin staining was found in portions of the developing-human eyes (Bitner-Glindzicz et al. 2000). Mutations in two novel cadherin-related genes—CDH23 and PCDH15—are responsible for USH1D and USH1F, respectively (Ahmed et al. 2001; Bolz et al. 2001; Bork et al. 2001). Disorganized stereocilia in the corresponding deaf-mouse mutants waltzer (Di Palma et al. 2001) and Ames waltzer (Alagramam et al. 2001) indicate that cadherins are involved in either the lateral links or the tip links that join adjacent stereocilia, although no gross retinal pathology has been detected in these mouse models. Humans with CDH23 mutations show variable retinal symptoms or, as in autosomal recessive nonsyndromic deafness DFNB12, no retinal phenotype at all (Bolz et al. 2001; Bork et al. 2001). USH2A, as well as nonsyndromic RP, is caused by mutations in usherin, which encodes a putative extracellular-matrix protein (Eudy et al. 1998; Rivolta et al. 2000).

Our results demonstrate that a gene encoding a 120-amino-acid protein with two predicted transmembrane domains and no homology to previously known proteins underlies USH3. In contrast to previously characterized USH genes that show distinct expression patterns in the sensory structures of the inner ear and of the retina, USH3 shows ubiquitous expression in several tissues. However, two genes involved in nonsyndromic X-linked RP—RP2 and RPGR—are also ubiquitously expressed but, nonetheless, have an important and specific role in the retina (Roepman et al. 1996; Schwahn et al. 1998). Patients with USH3 are born with normal hearing and vision, and the progressive sensory defects start later in life, leading to profound hearing loss, variable vestibular dysfunction, and severe RP with nyctalopia, constriction of the visual fields, and loss of central visual acuity.

Previously, we assigned, by linkage, a USH3 locus to 3q21-q25 and established the existence of a USH subtype characterized by postlingual, progressive hearing loss (Sankila et al. 1995). That cases of USH3 comprise ∼40% of all cases of USH in Finland has been explained by a founder effect. Families with USH showing linkage to the USH3 region have also been reported in the United States and Sweden (Kimberling et al. 1995), Spain (Espinos et al. 1998), Israel (Adato et al. 1999), and Italy (Gasparini et al. 1998). Using the significant linkage disequilibrium observed in the Finnish families with USH3, we were able to refine the USH3 critical region to 250 kb (Joensuu et al. 2000). The entire segment was subcloned, sequenced, and assembled, resulting in a 207-kb–sequence contig with 30 human ESTs. Three genes (NOPAR, UCRP, and H963)—two of which were novel—were identified and excluded as candidates, by sequencing. Of these, UCRP is probably an unprocessed pseudogene; however, interestingly, NOPAR shows significant homology to a thyroid-hormone receptor–associated protein, TRAP230 (Ito et al. 1999). The candidate gene that we identified, USH3, lies outside the initial critical region defined by haplotype analyses. This is explained by the fact that two mutations instead of the expected one founder mutation were found to segregate in the Finnish families with USH3. The Finnish founder mutation, c.300T→G in exon 3, was found in all USH3 chromosomes with the Finnish ancestral haplotype. It leads to a premature stop codon, which most probably affects mRNA stability or produces a truncated protein. In two Finnish families, patients were heterozygous for Fin_major and a c.131T→A transversion resulting in an amino acid substitution of lysine for methionine. In an Italian family with USH3, a homozygous 3-bp deletion, c.231–233delATT in exon 3, was detected. Among 200 Finnish controls and 51 individuals from the CEPH families, 1 Finnish individual was, as expected, heterozygous for the Fin_major mutation, whereas neither of the other two mutations was detected. Although the biochemical mechanisms by which USH3 causes retinal and cochlear disease remain to be elucidated, the identification of USH3 will already enable exact and even presymptomatic diagnosis of the disease.