Human Genetics

, Volume 118, Issue 1, pp 29–34

Targeted disruption of mouse Coch provides functional evidence that DFNA9 hearing loss is not a COCH haploinsufficiency disorder


  • Tomoko Makishima
    • Section on Gene Structure and FunctionNational Institute on Deafness and Other Communication Disorders, National Institutes of Health
  • Clara I. Rodriguez
    • Cancer and Developmental Biology LaboratoryNational Cancer Institute, National Institutes of Health
  • Nahid G. Robertson
    • Department of PathologyBrigham and Women’s Hospital and Harvard Medical School
  • Cynthia C. Morton
    • Department of PathologyBrigham and Women’s Hospital and Harvard Medical School
    • Cancer and Developmental Biology LaboratoryNational Cancer Institute, National Institutes of Health
    • Section on Gene Structure and FunctionNational Institute on Deafness and Other Communication Disorders, National Institutes of Health
    • Hearing Section, National Institute on Deafness and Other Communication Disorders, National Institutes of Health
Original Investigation

DOI: 10.1007/s00439-005-0001-4

Cite this article as:
Makishima, T., Rodriguez, C.I., Robertson, N.G. et al. Hum Genet (2005) 118: 29. doi:10.1007/s00439-005-0001-4


Dominant progressive hearing loss and vestibular dysfunction DFNA9 is caused by mutations of the human COCH gene. COCH encodes cochlin, a highly abundant secreted protein of unknown function in the inner ear. Cochlin has an N-terminal LCCL domain followed by two vWA domains, and all known DFNA9 mutations are either missense substitutions or an amino acid deletion in the LCCL domain. Here, we have characterized the auditory phenotype associated with a genomic deletion of mouse Coch downstream of the LCCL domain. Homozygous Coch−/− mice express no detectable cochlin in the inner ear. Auditory brainstem responses to click and pure-tone stimuli (8, 16, 32 kHz) were indistinguishable among wild type and homozygous Coch−/− mice. A Coch-LacZΔneo reporter allele detected Coch mRNA expression in nonsensory epithelial and stromal regions of the cochlea and vestibular labyrinth. These data provide functional evidence that DFNA9 is probably not caused by COCH haploinsufficiency, but via a dominant negative or gain-of-function effect, in nonsensory regions of the inner ear.


COCHCochlinDFNA9DeafnessHearing loss


Mutations in COCH cause dominant progressive hearing loss and vestibular dysfunction DFNA9 (Bom et al. 1999; de Kok et al. 1999; Fransen et al. 1999, 2001; Kamarinos et al. 2001; Nagy et al. 2004; Robertson et al. 1998; Usami et al. 2003; Verhagen et al. 2000; Verstreken et al. 2001). COCH encodes cochlin, a secreted protein with a signal peptide sequence and LCCL domain at its amino terminus, followed by two von Willebrand factor type A (vWA) domains (Robertson et al. 1998). The LCCL (Limulus factor C, cochlin, and late gestation lung protein, Lgl1) domain has been suggested to serve either a structural or host defense function (Trexler et al. 2000). vWA domains are known to bind to fibrillar collagens, glycoproteins and proteoglycans, and are found in proteins functioning in hemostasis, the complement system, cellular adhesion and extracellular matrix (Colombatti and Bonaldo 1991; Colombatti et al. 1993). Consistent with this latter observation, cochlin is the major noncollagen component of extracellular matrices of the inner ear (Ikezono et al. 2001).

COCH mRNA and cochlin protein are expressed by fibrocytes in nonsensory regions of the inner ear: the spiral limbus and spiral ligament of the cochlea and throughout the vestibular end organs (Robertson et al. 2001). In DFNA9 temporal bones there is a unique pattern of degeneration and acellularity of these same regions, which accumulate eosinophilic acellular material in their extracellular matrices (Khetarpal 2000; Merchant et al. 2000; Robertson et al. 1998). These findings raise the possibility that mutant cochlin itself aggregates, or causes other molecules to aggregate, to form the unique eosinophilic deposits (Robertson et al. 1998,2001,2003). It has been speculated that these deposits lead to hearing loss and vestibular dysfunction via strangulation and degeneration of auditory and vestibular neuritic processes (Khetarpal 2000; Robertson et al. 2001).

All reported DFNA9 mutations result in either missense substitutions or an amino acid deletion in the LCCL domain (Bom et al. 1999; de Kok et al. 1999; Eavey et al. 2000; Fransen et al. 2001; Kamarinos et al. 2001; Nagy et al. 2004; Robertson et al. 1998; Verhagen et al. 2000; Verstreken et al. 2001). DFNA9 mutations cause misfolding of the LCCL domain of cochlin expressed in bacteria (Liepinsh et al. 2001), and the extracellular deposition of DFNA9 mutant cochlin expressed in mammalian cell lines is altered (Grabski et al. 2003) even though it is synthesized, glycosylated and secreted in the same manner as wild-type cochlin (Robertson et al. 2003). Taken together, these studies suggest that DFNA9 mutations act via a dominant negative or gain-of-function effect in the inner ear.

Recently, a partial genomic deletion allele (Coch−/−) of mouse Coch was created in order to investigate its function in the uterus (Rodriguez et al. 2004). The two vWA domains, which were hypothesized to mediate a critical role in implantation, were deleted by targeted recombination. Although Coch is highly upregulated in the uterus by LIF (leukemia inhibitory factor) at the time of implantation, homozygous Coch−/− mutant mice showed normal implantation and reproduction. Here, we have used this mouse model and a related Coch-LacZΔneo reporter allele to explore the functional requirement for Coch in the ear and the mechanism of DFNA9 hearing loss.

Materials and methods

Mouse strains

Mice were maintained in accordance with the Guide for the Care and Use of Laboratory Animals (National Research Council, Washington, DC). Coch-LacZ and Coch−/− mice have previously been described (Rodriguez et al. 2004). Coch-LacZΔneo was created by Cre-mediated deletion of the neomycin resistance cassette in Coch-LacZ (Fig. 4a in Rodriguez et al. 2004). It encodes a bicistronic transcript with the intact full-length Coch open reading frame and termination codon followed by a downstream internal ribosome entry site (IRES) and LacZ (Fig. 1a). Coch−/− has a deletion of exon 7 through part of exon 12: the deleted region encodes the two vWA domains and remaining cochlin protein downstream of the LCCL domain, including the stop codon (Fig. 1a). Mice were maintained by intercrossing on a mixed C57BL/6-cBrd/cBrd/Cr and CD1 background.
Fig. 1

a Recombinant Coch alleles. The stop codon at exon 12 is shown as an asterisk. b RT-PCR analysis. Coch amplification products were detected in both P19 Coch-LacZΔneo (lane 1) and Coch−/− (lane 2) inner ear RNA. Lane 3, Coch−/− RNA without reverse transcriptase. Lane 4, no RNA. Gapdh was amplified as a control. c Western blot analysis. Extracts of P19 wild type (+/+) and Coch−/− (−/−) whole inner ears were separated by SDS–PAGE and probed with an anti-cochlin antibody against the LCCL domain (Robertson et al. 2001). No residual cochlin products (expected size: 16 kDa) were detected in Coch−/− animals. The same membrane was stripped and reprobed for β-actin.

RT-PCR analysis

Inner ears were dissected from postnatal day 19 (P19) Coch−/− and Coch-LacZΔneo mice. Total RNA was isolated using the Qiagen RNeasy mini kit (Qiagen, USA). Eight whole inner ears were used to isolate total RNA (250 μg) for each genotype. Eight microliters of total RNA and random hexamers were used to generate cDNA (20 μl) by RT-PCR with the SuperScript First-Strand Synthesis System (Invitrogen, USA). Two microliters of cDNA were used for PCR with primers LCCL (5′-CCC ATT CCT GTC ACC TGC TTT AC-3′, 5′-AGA TGC TGG ACA CTG ACG CAT AC-3′) and Gapdh (5′-TGC TGA GTA TGT GGT GGA GTC TA-3′, 5′-AGT GGG AGT TGC TGT TGA AGT CG-3′). PCR reactions were performed in 50 μl for 30 cycles (94°C, 15 s; 60°C, 30 s; 68°C, 60 s). Amplification products (one-fifth of each final reaction volume) were separated in an agarose gel and visualized by ethidium bromide staining and ultraviolet transillumination. LCCL amplification products were subcloned and confirmed by sequence analysis.

Western blot analysis

P19 whole inner ears were mechanically disrupted and then lysed with 7M urea, 2M thiourea, 2% Triton X-100, 100 mM dithiothreitol, 1× Complete Protease Inhibitor Cocktail (Roche, Basel, Switzerland), and separated in 12.5% polyacrylamide–SDS gels. Proteins were transferred to polyvinylidene fluoride (PVDF) membranes (Millipore, USA) and detected with a 1:100 dilution of an antibody specific for the amino terminus of cochlin (Robertson et al. 2001). Bound antibodies were visualized using SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL, USA).

LacZ reporter expression analysis

Mouse inner ears were dissected and immediately processed for β-galactosidase detection with X-gal (Novagen, USA). At least three animals (six ears) of each genotype and age were evaluated. Specimens were fixed for 20 min at room temperature (1% formaldehyde, 0.2% glutaraldehyde and 0.02% Nonidet P-40 in phosphate-buffered saline (PBS)), washed twice (2 mM MgCl2 and 0.02% Nonidet P-40 in PBS) and stained overnight at 37°C (0.1 M MgCl2, 5 mM potassium ferricyanide and 1 mg/ml X-gal in PBS). For P10 and older mice, the cochlea was perfused by removing part of the bony capsule at the apex, followed by decalcification (0.25 M disodium ethylenediaminetetra acetate (EDTA) and 2% paraformaldehyde in PBS) for 2–3 days at 4°C. Specimens were then incubated overnight in 30% sucrose and embedded in Tissue-Tek OCT compound (Sakura, USA). Ten-micrometer cryosections were mounted on slides and visualized by light microscopy.

Auditory brainstem response analysis

Auditory brainstem response (ABR) thresholds were measured as previously described (Griffith et al. 2002). Mice were anesthetized by intraperitoneal injection (0.375 g/kg body weight) of tribromoethanol (Sigma, USA). ABR thresholds were measured in a sound chamber using Intelligent Hearing System software (IHS, USA). Click and pure-tone (8, 16, 32 kHz) response thresholds were measured for both ears. Averages of 500 responses were recorded for descending 5-dB stimulus steps to determine the threshold. ABR thresholds of better-hearing ears were compared between genotypes by one-tailed Student’s t-tests.


Inner ear expression of Coch

Previous Northern and Western blot analyses failed to detect any Coch mRNA or protein, respectively, in Coch−/− inner ears (Rodriguez et al. 2004). We performed an RT-PCR analysis to increase the sensitivity for detecting residual Coch mRNA. We identified products corresponding to the LCCL domain from both positive control (homozygous Coch-LacZΔneo) and Coch−/− mouse inner ears (Fig. 1b, lanes 1 and 2), although the amount of amplification product was lower in Coch−/− mice.

To determine if any residual cochlin is expressed in Coch−/− inner ears, we performed a Western blot analysis with an anti-cochlin antibody directed specifically against the LCCL domain (Robertson et al. 2001). No immunoreactivity was detected in Coch−/− inner ear tissue (Fig. 1c), showing that there is little, if any, LCCL peptide expressed.

Spatiotemporal expression of Coch-LacZΔneo

We characterized the temporal and spatial patterns of Coch mRNA expression in the inner ear by X-gal-based detection of Coch-LacZΔneo expression (Fig. 2). Coch promoter-directed β-galactosidase expression was seen throughout the developing and mature inner ear. X-gal staining was stronger in the vestibule, compared to the cochlea, through P5. Staining was stronger in the cochlea at later time points, although the overall staining was much weaker (Fig. 2a). LacZ expression was seen in the tympanic membrane (Fig. 2b), but nowhere else in the middle or outer ear. In the cochlea, staining was observed in the spiral limbus, spiral ligament and cells lining the scala vestibuli (Figs. 2c, d). Whereas the spiral ligament in the middle turn (Fig. 2d) is only faintly stained, likely due to perfusion artifact, it was consistently stained in the apical turn of the cochlear duct. Due to technical difficulty achieving uniform X-gal perfusion throughout the length of the cochlear duct, we were unable to determine if there is a tonotopic gradient of Coch-LacZΔneo expression. In the vestibule, staining was seen in nonsensory epithelial cells in the utricle, saccule and semicircular canals, but not within sensory epithelial (hair) cells (data not shown). There was no staining observed in the stroma of the crista ampullaris (Fig. 2d). The spatial and temporal expression patterns of LacZ reporter are consistent with previous reports of in situ hybridization and immunohistochemical analyses (Robertson et al. 2001).
Fig. 2

LacZ reporter expression. aCoch-LacZΔneo (upper row) and wild type (lower row) inner ears stained with X-gal. A time course study from embryonic day 15 to postnatal day 30 shows staining in the cochlea and vestibular labyrinth in Coch-LacZΔneo mice. bCoch-LacZΔneo (upper panel) and wild type (lower panel) tympanic membranes at P0 (left) and P5 (right) stained with X-gal. TM tympanic membrane; arrowhead malleus. c Whole-mount organs of Corti. The spiral limbus of Coch-LacZΔneo, but not wild type, organ of Corti was stained with X-gal. Left panel, wild type. Right panel, Coch-LacZΔneo. IHC inner hair cell; OHC outer hair cell; Slb spiral limbus. d Ten-micrometer sections of inner ears. The spiral limbus and cells lining the scala vestibuli were stained in Coch-LacZΔneo inner ears (middle panel) but not in wild-type inner ears (left panel). In the ampulla, the sensory epithelium and the stroma of the crista were not stained. Slb spiral limbus; Slg spiral ligament; SV stria vascularis; RM Reissner’s membrane; TM tectorial membrane; ScV scala vestibuli; OC organ of Corti.

Auditory function of Coch−/− mice

ABR thresholds for click and pure-tone stimuli (8, 16, 32 kHz) were measured in Coch−/− and wild-type mice at P19, 3 months and 5 months of age. There was no significant difference in ABR thresholds of Coch−/− and wild-type mice at P19 (Fig. 3a) or at 3 months of age (Fig. 3b). At 5 months of age Coch−/− had ABR thresholds slightly better than the wild-type mice (Fig. 3c). There was no difference in ABR thresholds between males and females (data not shown).
Fig. 3

Auditory function of Coch−/− mice. a ABR thresholds at P19. There were no significant differences between Coch−/− (n=6; black bars) and wild type (n=6; white bars) responses to either click or pure tone stimuli at any frequency (P > 0.10). b ABR thresholds at 3 months of age. There were no significant differences between Coch−/− (n=5; black bars) and wild type (n=5; white bars) mice (P > 0.10). c ABR thresholds at 5 months of age. Coch−/− (n=5; black bars) mice had slightly better hearing than wild type (n=5; white bars) mice (P > 0.10).


Cochlin is the major protein component of bovine inner ears (Ikezono et al. 2001) and it has been thought that this abundance implies a critical role for hearing. Nonetheless, our study shows that a deficiency of cochlin does not affect hearing in young mice. Although we were able to detect Coch RNA transcripts by RT-PCR analysis, our results suggest there is reduced transcription, increased decay (perhaps nonsense-mediated), or both (Fig. 1b). Moreover, there were no detectable bands in a Northern blot (Rodriguez et al. 2004) or a Western blot analysis (Fig. 1c) of Coch−/− inner ears. Since cochlin comprises approximately 70% of bovine whole inner ear protein (Ikezono et al. 2001), we conclude that the amount of cochlin protein must be profoundly decreased in Coch−/− mice and that Coch might not be necessary for normal auditory function. Furthermore, we observed no abnormal circling, waltzing, head shaking, or other overt behaviors suggestive of vestibular dysfunction in the mutant mice. However, we cannot rule out a subtle vestibular abnormality since we did not perform specific tests of vestibular function.

A 16-kDa short isoform (cochlin-tomoprotein) corresponding to the LCCL domain is abundant in inner ear perilymph fluid but not in inner ear tissues of humans or cows (Ikezono et al. 2004), raising the possibility of a distinct and critical role for this isoform. Alternatively, cochlin-tomoprotein may represent a proteolytic cleavage by-product that is not incorporated into the extracellular matrix and is free to diffuse into perilymph. Although, it is possible that sufficient levels of cochlin-tomoprotein with an intact LCCL domain may exist in the perilymph of Coch−/− inner ears to preserve normal auditory function, the volume of cochlear perilymph in mice is only about 0.6 μl (Thorne et al. 1999) and we were unable to assess the amount of perilymph retained in our inner ear extracts. However, proteomic analysis of Coch−/− inner ear tissue detected no peptides containing the LCCL domain, whereas, cochlin was found to be the most abundant peptide (14% contained LCCL fragments) in wild-type controls (unpublished observation, NGR and CCM).

We tested mice up to 5 months of age in order to exclude the confounding, nonnormal ABR threshold elevations caused by age-related hearing loss alleles segregating from the 129/SvJ and C57BL/6 background strains (Szymko-Bennett et al. 2003). Since the onset of DFNA9 hearing loss occurs from the second to fifth decades of life (de Kok et al. 1999; Eavey et al. 2000; Fransen et al. 1999, 2001; Kamarinos et al. 2001; Nagy et al. 2004; Robertson et al. 1998; Usami et al. 2003; Verhagen et al. 2000), a potential progressive hearing loss phenotype might start after 5 months of age in Coch−/− mice. Analysis at later time points on a different strain background (e.g., CBA) could reveal a hearing loss phenotype. There is also a possibility that the hearing loss phenotype cannot be observed due to inter-species differences between human and mice. Nonetheless, since heterozygosity for COCH mutations causes DFNA9, the lack of a detectable phenotype in homozygous Coch−/− mice is consistent with a gain of function or dominant negative mutation of COCH in DFNA9. A knock-in mouse model is needed to determine whether it is a dominant negative effect, a gain of function, or both that lead to hearing loss and vestibular dysfunction in DFNA9.


We thank Tom Friedman and Rob Morell for helpful discussion and critical review of the manuscript, and Rachel McNamara for excellent technical assistance This work was supported by NIDCD/NIH intramural research fund Z01-DC-000060-03 (to AJG) and NIH grant DC03402 (to CCM). TM was supported in part by a Japan Society for the Promotion of Science Research Fellowship for Japanese biomedical and behavioral researchers at the National Institutes of Health.

Copyright information

© Springer-Verlag 2005