Identifying the genes of hearing, deafness, and dysequilibrium

Hearing loss affects more than 25 million Americans and costs over 50 billion dollars each year, surpassing the combined financial impact of multiple sclerosis, stroke, epilepsy, spinal injury, Huntington’s, and Parkinson’s disease (1). Inherited deafness affects one child in 2,000, and equal numbers of children are born with a significant loss of hearing from other causes (2). In the general population, losses acquired through trauma and disease are even more frequent. A representative sample of 2,000 people would include over 300 who have a significant hearing impairment (3). More than 25% of people suffer from such a condition by the age of 65 and nearly 50% by the age of 80 (2). Infections and diseases of the external and middle ear are treatable, but the majority of significant hearing loss is permanent “nerve deafness”, a misnomer for conditions that typically result from loss of hair cells (4). Like the rods and cones in the retina, hair cells in the inner ear convert physical stimuli into neural signals. Our sense of hearing originates from 16,000 hair cells in each cochlea (5). The rapid and critical reflex stabilization of our visual gaze and our sensitivity to head rotation depend on the hair cells in the inner ear’s three semicircular canals. Hair cells in the two other end organs of the vestibular portion of the inner ear lie beneath masses of calcium carbonate crystals and provide part of our sensitivity to gravity.

Purely scientific interests provide justification for efforts to understand the mechanisms of the inner ear. The cochlea is sensitive to vibrations so minute that they approach the diameter of an atom, and it achieves temporal resolution on the level of 10 μs (1). It does this while providing excellent resolution of sound frequency across broad bandwidths and deep dynamic range. Yet, for many years the inner ear was effectively “off limits” to protein identification because its small number of detector cells located deep within the temporal bone was unsuited for conventional biochemistry. A few genes were identified via biochemical methods, but the efforts were limited to abundantly expressed genes (6, 7). Now, methods for amplifying small quantities of nucleic acids permit small samples from the sensory transducers in the ear to be investigated effectively. An article in a previous issue of the Proceedings illustrates this point. Heller et al. (8) have identified 120 clones in a cDNA library from auditory epithelia that code for 12 genes that are ear-specific or highly expressed in the ear. Three of the genes had been reported in the ear previously. Two are related to known forms of inherited nonsyndromic deafness. Two are highly expressed in cell types that are unique to the ear, and five others encode collagen isoforms. Such results are heartening in an area of sensory research in which only a limited number of genes had been identified.

The type of approach that Heller et al. (8) have taken should continue to provide rapid identification of genes in the ears of chickens and other species. Important progress has been made before this in the identification of mutant genes that are responsible for inherited forms of deafness, but those genes have come to light one at a time (G. Van Camp and R. J. H. Smith, http://dnalab-www.uia.ac.be/dnalab/hhh/). Not surprisingly the first success came in identifying the cause of a syndromic form of inherited deafness, in which hearing impairment is accompanied by other phenotypes. In 1990, a mutation in the COL4A5 collagen gene was identified as the cause of Alport syndrome (9). In 1992, Waardenburg’s syndrome type 1 was identified with a mutation in a homologue of the Pax-3 gene (10). In 1993, a mutation in mitochondrial rRNA was identified with x-linked nonsyndromic and antibiotic-induced deafness (11). In 1995, the first genes in which mutations result in autosomal recessive forms of deafness were identified in mice. Analysis of Snell’s waltzer and shaker-1 demonstrated mutations in the genes for unconventional myosins 6 and 7, respectively (12, 13). Approximately 60 mutant loci in mice are candidates for involvement in the auditory system (14). Mapping of loci for deafness and balance dysfunction in mice already has speeded the identification of several genes that cause inherited deafness in humans. Such information should continue to be of great value in the analysis of the genes discovered in a range of species. In the case of shaker-1, the defects mapped to a site in the mouse genome that is homologous to a region on human chromosome 11 near the locus for Usher syndrome 1B. Based on that indication, reverse transcription-PCR analysis of affected families was undertaken and it showed that the human mutations were indeed in the gene for myosin 7a (15). These findings and the recent discovery that the phenotypes of shaker-2 mice and the human nonsyndromic deafness DFNB3 arise from mutations in myosin 15 (16, 17) have focused attention on the presence of unconventional myosins in hair cells themselves (18). These motor proteins are essential for maintenance of hearing and nonredundant in their actions, but their actual roles in the ear remain to be determined.

Successes continued in 1997 with evidence supporting the conclusion that mutations in connexin-26, a gap junction protein, are responsible for a major form of nonsyndromic recessive deafness, DFNB1, and for a more limited form of nonsyndromic dominant hearing loss, DFNA3 (19, 20). In the same year, nonsyndromic dominant deafness DFNA1 was identified with a mutation in the human homologue of the Drosophila gene diaphanous. That gene encodes a protein ligand for profilin, which is a target for regulation of actin polymerization by Rho GTPases. Those facts suggest a possible role for the diaphanous product in the formation of or in dynamic changes that may occur within the cuticular plate or the stereocilia bundle, actin-rich cytoskeletal elements that are unique to hair cells (21). Also in 1997, mutations in the human α-tectorin gene were shown to cause nonsyndromic dominant hearing impairment in one Belgian (DFNA12) and one Austrian (DFNA8) family (22). In 1998, mutations in the transcription factor POU4F3, which is known to be highly expressed in hair cells of mice, were shown to cause a nonsyndromic dominant form of progressive hearing loss (DFNA15) (23). The occurrence of a hearing loss with some indications of possible vestibular dysfunction is consistent with the finding that mice that are homozygous for targeted deletion of this transcription factor are deaf and have vestibular dysfunction from birth (23). The absence of a mutant phenotype in heterozygous mice remains to be explained, however, in view of the dominant action of the mutant gene in humans.

The molecular mechanisms that govern inner ear development, function, response to injury, and aging should be revealed by contributions from such studies. In situ hybridization already has provided important insights into the role of neurotrophic signaling in vivo through studies of the development of the auditory and balance systems (25) as have similar studies of the BMPs (26). Phenotypic analyses of mice with targeted deletions have been informative in the case of the genes for Int2, Brn 3.1, and the neurotrophins and their receptors among others (24, 27, 28). Analysis of zebra fish mutants that have effects in the ear also will be informative although identifying the mutations in specific genes is difficult at present (29, 30). Cloning in expression libraries by screening with antibodies to the elements of the ear also will continue to be informative.

The value of the information reported by Heller et al. (8) derives from its specificity and is illustrated by considering four of the genes. Chicken COCH5B2 is expressed in spindle-shaped cells along the path of the auditory neurites. It also is expressed in muscle spindle organs. This gene is an ortholog of a gene that is mutated in three families affected by, DFNA9, a nonsyndromic autosomal dominant hearing loss that is accompanied by vestibular dysfunction. In the chicken ear, the gene is expressed in a location that is consistent with pathology of affected human ears. The exact reason why a defect in this gene is sufficient to cause deafness and balance dysfunction is not yet known, but the observation that it is expressed in both the ear and the muscle spindle should be informative. Although those two sites are both associated with mechanically sensitive receptors there would have been little reason to suspect that they share specific elements of molecular structure. Determining the details of that expression pattern should help to explain the mechanism by which the mutation of COCH5B2 causes hearing and balance dysfunctions. The cells of the tegmentum vasculosum of the cochlea and the cells in the endothelial lining of the cardiac ventricles both express the gene for otokeratin at high levels. Heller et al. (8) hypothesize that this may indicate that the cells of the tegmentum vasculosum have specific adaptations of their cytoskeletons for resisting deformation by mechanical stress. The hypothesis would not have been predicted on the basis of prior knowledge. The expression pattern also suggests that otokeratin is a good marker for cells of the tegmentum vasculosum. Chicken connexin-31 is expressed in most epithelial cells of the cochlea but not in hair cells. The current assessment of its expression pattern appears consistent with ear specificity. It is unclear whether connexin-31 is an ortholog of mammalian connexin-26, the form mutated in DFNB1 and DFNA3, an ortholog of mammalian connexin-30, or a unique form. Its structure raises questions that may help to improve understanding of the roles played by the different domains of connexin molecules in deafness and in normal function. Homogenin is a new member of the gelsolin family of proteins. It is expressed at very high levels in homogene cells that are unique to the ear. Homogene cells form part of the chicken cochlea’s lining. The tectorial membrane is attached to these cells. On the basis of the expression pattern of this gelsolin family member, Heller et al. (8) have examined the cytoskeletons of homogene cells, finding that they stain intensely for actin filaments. It is known that the tectorial membrane moves in relation to the sensory bundles of the hair cells and the basilar membrane with each cycle of acoustic stimulation. Therefore, Heller et al. hypothesize that the actin cytoskeleton in the homogene cells may adjust tension in the tectorial membrane and thereby modulate the performance of the cochlea.

Heller et al. (8) achieved their results by first harvesting auditory epithelia from 250 chicken embryos via microdissection to prepare a RNA source for the library that would be enriched for messages expressed in the sensory epithelium. In fact, calbindin and β-tectorin are the only genes identified in the study that are specifically expressed in the sensory epithelium, but those genes were in 30 of the 120 clones selected. The rationale for enrichment is convincing, however. Such procedures are likely to provide a bigger payoff in the future. To generate an ear-specific probe, subtractive amplification was performed by using poly(A)+ RNA from auditory epithelia as the tester and by using a mixture of poly(A)+ RNA from retina, brain, and liver as the driver. The probe was then used to select 196 clones for additional investigation. The payoff of this method was substantial. One can expect even more interesting genes to be identified via refinements of this type of probe generation protocol. PCR-based subtractions of this type should be amenable to different forms of fine-tuning and allow great improvements in this critical step of the library screening process. In situ hybridization on frozen sections of embryonic chicken ears was used for the second stage of the selection of the 120 clones that encoded genes specifically or highly expressed in the ear. Anatomy can provide clear indications of the potential importance of the genes expressed in the ear, and elsewhere in the body, but the step may work best for highly expressed genes. Heller et al. (8) suggest that they will be refining the screening probe and will subtract out the cDNAs that have been identified here as they seek to identify rare mRNAs from the auditory organ.

Despite the small database of known genes, the chicken ear is an opportune choice for efforts to identify genes expressed in the ear, in part because its development has been thoroughly studied. Also, unlike mammals, chickens are able to recover hearing and balance sensitivity after experiencing a loss of hair cells. Damaged ears of chickens generate replacement cells that become innervated and provide significant recovery of function within a few weeks (31). Identifying the genes that are specific to these ears may well reveal control elements for that repair process. Such efforts also can be expected to provide useful markers for the currently recognized cell types within the sensory epithelia of vertebrate ears. Results also may provide specific markers to identify as yet unrecognized subtypes of cells, such as potential stem cells within regenerative epithelia and markers for stages of differentiation in the lineages that lead to replacement cells. It is to be expected that screening for ear specific genes may also reveal elements of the signaling systems that control development of the ear.

The chicken ear is thus far unique in the degree to which cell life span differs between the auditory and vestibular epithelia. In the absence of trauma, the cells of the chicken’s auditory epithelium are mitotically quiescent, perhaps remaining quiescent for the life of the organism. In contrast, the cells in vestibular epithelia that are within 500 μm of those auditory cells turnover rapidly, so that cell life spans average less than 1 month (32). Identification of genes that are differentially expressed in those divisions of the chicken’s ear should provide insights into the processes that limit the life spans of these cells. In contrast, the hair cells in human ears are assumed to have life spans of a century or more. The hair cells we were born with are believed to provide our senses of hearing and equilibrium throughout life. Of course, hair cell losses do occur thereby contributing to a propensity for falls in the elderly and interfering with our ability to communicate by spoken language. Studies that are beginning to define the molecular elements of the ear and the processes that they control appear likely to someday provide the knowledge that will allow treatment, protection, and understanding of this organ.