SUBJECTS AND METHODS

The appropriate institutional review boards approved this study and informed consent was obtained from human subjects. We observed four families with two or more children who had sensorineural hearing loss and normal OHC function (fig 1). The degree of hearing loss was determined by standard pure tone audiometry. The scale used to classify the degree of hearing loss is as follows: 0–20 dB HL is normal, 21–40 dB HL is mild, 41–60 dB HL is moderate, 61–80 dB HL is severe, and above 80 dB HL is profound. The status of the auditory pathway was determined using tympanometry, middle ear muscle reflex (MEMR), and ABR. OHC function was measured using distortion product OAE (DPOAE),³ transient evoked OAE (TEOAE),¹⁸ and/or cochlear microphonics (CM).¹⁹ All subjects diagnosed with hearing loss were examined by an otolaryngologist, a neurologist, and geneticist and underwent testing to rule out syndromic disorders.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Pedigrees of non-syndromic recessive auditory neuropathy (NSRAN) families and their haplotype data for chromosome 2. The OTOF locus is flanked by markers D2S2144 and D2S165. Black symbols are subjects with NSRAN and white symbols are subjects with normal hearing levels. The number directly under the symbol is an identification number. The four digit number below the ID number is that person’s year of birth

DNA extraction and linkage analysis

Blood was drawn from affected children, their sibs, parents and, if possible, their grandparents. DNA was extracted using Puregene DNA isolation kits (Gentra Systems Inc). Genomic DNA was PCR amplified and genotyped using ABI Prism set 1 on an ABI Prism 377 DNA Sequencer according to standard protocol.²⁰ All other primers used in this study, including locus specific primers, were purchased from Integrated DNA Technologies Inc (IDT). Linkage analysis was performed using LINKAGE program version 5.1.²¹ Lod scores were generated using MLINK for two point analyses and LINKMAP for multipoint analyses.

Mutation detection

OTOF (otoferlin) intronic and exonic primers^22,23 (tables 1 and 2) were used to PCR amplify genomic DNA or cDNA (derived from transformed peripheral blood lymphocytes). Sequence data analysis was performed using ABI Sequence Analysis 3.4, Fractura 2.2.0, Navigator 1.1 software, GCG suite (Oxford Molecular Group Inc), and the Lasergene suite (DNASTAR Inc). Mutations were confirmed and controls were screened by restriction enzyme digestion of PCR products.

RESULTS

Clinical features of non-syndromic recessive auditory neuropathy (NSRAN)

Nine children from four families with NSRAN had sensorineural hearing losses ranging from moderate to profound with functioning OHCs. Neurological examinations and nerve conduction studies indicated that none of the children had peripheral neuropathy. There was no evidence that the hearing loss was caused by a recognised syndrome or any known non-genetic factors. MEMRs were absent and tympanometry was normal. All the children reported receiving no benefit from hearing aids and had poorer speech comprehension than expected based on their pure tone audiograms. The clinical details of the affected children are summarised in table 3.

2p23 is a critical region for NSRAN

A genome screen was performed on three NSRAN families (1802, 1721, 2117) and linkage was found for a 21.7 cM region on chromosome 1 (lod=2.905) and a 16.3 cM region on chromosome 7 (lod=2.908).²⁹ No DFNB loci were located in our critical regions; however, while finishing the genome search a new hearing loss gene called OTOF was discovered. The mRNA expression pattern and proposed function made it an ideal candidate for NSRAN. Linkage data were re-evaluated and found to be incomplete for the OTOF locus, so more markers were purchased. During this time, a fourth NSRAN family was obtained, so all four families were genotyped using the five dinucleotide markers from the OTOF locus, D2S2221, D2S2144, D2S165, D2S365, and D2S2347. Haplotype analysis was consistent with linkage for all four families (fig 1). The otoferlin gene (OTOF) was noted to be located between markers D2S2144 and D2S165.²² Two point linkage analysis gave a maximum lod score of 2.908 for the four families at marker D2S2144 and a multipoint analysis produced a maximum lod score of 3.47.

NSRAN mutations in OTOF

All of the known OTOF exons and approximately 100 bases of the introns flanking these exons from each of the proband’s DNA from the four families was sequenced and examined for mutations. All mutations are identified by their position in GenBank file AF183185. Once a mutation was identified, DNA from the rest of the family was sequenced to determine mutation segregation. A control sample of 94 normal hearing, unrelated persons was tested using restriction enzyme digestion. The results are summarised in table 4. In family 1802, the affected subjects were observed to be compound heterozygous for a frameshift mutation and a missense mutation. The frameshift mutation is the result of deletion of a guanine at position 1778 (exon 16), which causes a stop codon after four additional amino acids. The missense mutation, 6141G>A, substitutes a glutamine for an arginine at the second codon in exon 48 (fig 2). In family 1721, both children and their father were observed to be heterozygous for a conserved G to C transversion in the donor splice-site (GT) of intron 39 (IVS39+1G>C). Using RT-PCR, it was found that this leads to an mRNA with 30 new codons, the twentieth of which codes for a stop codon and is predicted to result in a truncated protein (fig 3). Family 2117 was heterozygous for the missense mutation in exon 48, 6285C>G, which changes the amino acid proline to an arginine in codon 50 of exon 48. These missense mutations are believed to be pathological because, as seen in fig 2, these amino acids are highly conserved in mouse and man for otoferlin, as well as the myoferlin and dysferlin genes. None of these mutations was observed in a set of 94 unrelated controls (188 chromosomes). Mutation segregation was consistent with autosomal recessive inheritance in all cases.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Comparison of the carboxy-terminus amino acid sequence between ferlin-like proteins. Comparison between the deduced amino acid sequences encoded by exon 47 of the brain isoform, exon 48 of the cochlear isoform in human and mouse, and the corresponding predicted amino acid sequence from human dysferlin and human myoferlin cDNA. The transmembrane region is underlined. The amino acids in bold are those changed in human exon 48 by the missense mutations described in table 4.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

cDNA sequence from family 1721 of the allele with the splice site mutation IVS39+1G>C (bold). The coding region in underlined and translated.

DISCUSSION

Mutations in the gene encoding otoferlin cause a non-syndromic recessive auditory neuropathy. The evidence for this assertion is threefold. First, NSRAN families show linkage to a small region on chromosome 2 that contains the otoferlin gene. Second, the gene encoding otoferlin is expressed in the cochlea in the appropriate cell type, the cochlear inner hair cells (IHCs). Third, mutations have been found in the otoferlin gene in three of four families with NSRAN. Our conclusion not only raises several interesting questions but also has important implications with regard to the diagnosis and treatment of children with hearing loss.

Knowledge of the tissue specific expression of otoferlin is currently limited to results, primarily from mouse, based on in situ hybridisation, northern analysis, and RT-PCR studies.²³ In situ data indicate that otoferlin mRNA from mouse embryos is expressed in IHCs, OHCs, and spiral ganglion cells. In the cochlea, it persisted in only IHCs for the adult mouse. Otof was not expressed in the supporting cells of the organ of Corti. In the mouse vestibule, Otof is expressed in the neuroepithelia of the saccule, utricle, and semicircular canals; however, no vestibular abnormality was observed in our families. The mRNA was detected in other tissues by RT-PCR, primarily in the brain but also in heart, kidney, and liver.

Analysis of its primary structure identifies two domain types that offer an indication of what otoferlin might be doing and what other molecules might interact with it. Otoferlin has six C2 motifs (Ca⁺⁺ binding domains) and a transmembrane domain at its C terminus (fig 4). Because of the presumed role of otoferlin’s orthologue, the Caenorhabditis elegans fer-1 gene in membrane fusion,²⁴ and the sequence similarity of its C2 domains to synaptotagmin I, it is likely that otoferlin plays a role in membrane trafficking activated by increased local Ca⁺⁺ concentration. It has been suggested that otoferlin may play a synaptotagmin-like role in IHC synaptic vesicle fusion.²² However, the otoferlin missense mutations found in the NSRAN families do not involve the C2 domains but are adjacent to the transmembrane domain (fig 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Otoferlin mutations and functional domains. A diagram of otoferlin’s structure showing the location of all mutations found to date. (1) Yasunaga et al,²² (2) Adato et al,²⁸ (3) Yasunaga et al,²³ and (4) Houseman et al.²⁷

Alternatively spliced forms of otoferlin have been found in human and mouse.²³ In humans, spliced forms were seen in the cochlea but not the brain and other isoforms were found both in the brain and cochlea. Exon 47 is skipped in cochlea, heart, kidney, and total fetus leading to a different carboxy-terminal peptide sequence from that of the brain isoform, which expresses exon 47. In mRNA with exon 47, translation terminates in that exon, leaving exon 48 untranslated. In mRNA without exon 47, translation continues into exon 48 where it is terminated.

Exon 47 of the brain isoform and exon 48 of the cochlear isoform specify the C-terminal 60 amino acids of otoferlin. The amino acid sequences are 63% identical, differing in 22 amino acids (fig 2). This transmembrane domain and the flanking sequence may play an important role in targeting this protein to a specific location in the cell or specify the interaction of otoferlin with specific ligands.

The hearing loss described in previous DFNB9 reports was characterised by pure tone audiometry or ABR,^{22,23,25–28} but there was no report of testing for OHC function. The hearing loss families with OTOF mutations that we report here are distinguishable by the relative preservation of OHC function. It is not clear whether the preservation of OHC function is a consistent and stable phenotype for hearing loss caused by OTOF mutations. In family 1802, there is clear evidence that OAEs diminish in older children.²⁹ While this may be the result of damage from hearing aid use, it is also possible that it represents an effect of the OTOF mutation on the OHCs. In either case, such a trend would make the phenotype of older subjects with NSRAN indistinguishable from other forms of NSRHL. If OAEs were extinguished progressively in NSRAN patients, it would suggest that this form of hearing loss is more common than previously estimated.

It is possible that the preservation of OHC function in the hearing loss families reported here is the result of a specific class of OTOF mutations. All previously reported OTOF mutations have been found in consanguineous families and are present in the homozygous state. Four OTOF mutations were found, two nonsense^22,27 and two splice site mutations.^23,28 These mutations would be expected to completely “knock out” the otoferlin function in all tissues, and we would predict a profound pan-cochlear hearing loss.

Affected subjects in NSRAN family 1802 are heterozygous for a frameshift (1778delG) and a missense mutation (6141G>A). The deletion would be expected to knock out this otoferlin allele in all tissues since a stop codon occurs after four additional amino acids in exon 16. However, its companion is a missense mutation in the cochlear specific, alternatively spliced coding region of exon 48. This otoferlin allele would be predicted to have no effect on the primary structure of the brain isoform. In tissues expressing the cochlear isoform, such as the IHCs, a dysfunctional protein would be created which could disrupt the protein’s subcellular localisation or otoferlin’s interaction with other proteins critical to IHC function and, thus, an AN phenotype might be observed.

The mutations found in families 1721 and 2117 also support this hypothesis, but their companion mutations have yet to be found. While every known exon has been sequenced in these two families, it is possible that the other mutations are in the promoter, introns, or enhancers or hidden because of the lack of amplification of the mutated allele. Despite the suggestive linkage of family 2821, hearing loss in this family could be caused by mutations in a different NSRAN gene.

The discovery of a cause for NSRAN will allow a rapid genetic diagnosis of this form of hearing loss and will probably improve our ability to provide appropriate communication skills and tools in a timely manner to those affected. Understanding the mechanism by which mutations in otoferlin can cause NSRAN will further define the mechanism of hearing and the specific roles IHCs and OHCs play in this phenomenon.