2.4. Sensorineural hearing loss
Hearing loss is a common disorder, the incidence of congenital hearing loss being estimated to be 1 in 1,000 births. Approximately equal numbers of cases are attributed to environmental and genetic factors (Morton 1991, Gorlin et al. 1995). If presbyacusis is excluded, 0.5-2% of the population may be said to be affected. The wide range is due to variability in the definition of normal hearing level (Spillmann 1994). About 70% of affected individuals have a non-syndromic form of hearing loss, and thus they do not have symptoms in other organs. Genetic defects cause some of the non-syndromic cases of deafness and all the syndromic ones. Non-syndromic deafness is usually sensorineural, whereas syndromic deafness may be conductive, sensorineural or mixed. Less than 1% of cases of deafness due to genetic defect have been estimated to be mitochondrially inherited, while about 99% are due to a nDNA defect (77% autosomal recessive, 22% autosomal dominant and 1% X-linked) (Morton 1991). Fifty-one loci for non-syndromic deafness have been identified, five of them in mitochondrial genes (Hereditary hearing loss homepage 2002). More than 32 genes contributing to syndromic deafness have been recognized (Resendes et al. 2001).
2.4.1. Hearing mechanisms
The hearing organ is located in the temporal bone, where the outer ear and middle ear are separated by the tympanic membrane. The ossicular chain is in contact with the cochlea via the oval window, and the cochlea forms a spiral canal with 2 turns that narrows towards the apex. There are three fluid-filled compartments in the cochlea, the scala vestibuli, scala media and scala tympani (Figure 3). The scala vestibuli, which is in contact with the oval window, and the scala tympani, in contact with the round window, are filled with perilymph, which has a high sodium concentration and a low potassium concentration, resembling normal extracellular fluid. The scala tympani and scala vestibuli are connected at the apex by the helicotrema. The scala media is filled with endolymph, which has a high potassium concentration and a low sodium concentration and is separated from the scala vestibuli by the Reissner membrane. The organ of Corti, including the sensory hair cells and supporting cells, sits on a basilar membrane, and the extracellular fluid within it, surrounding the body of outer hair cells, has sodium and potassium concentrations that resemble those in the perilymph, the potassium level being just slightly higher. The peripheral dendrites of the cochlear nerve are inside the spidal lamina and in contact with the inner hair cells. Stria vascularis forms the outer wall of the scala media. (Palva et al. 1991, Ulfendahl et al. 2000.)
Figure 3. Illustration of the cochlear duct. On the top and bottom are the scala vestibuli and scala tympani, fluid-filled canals containing perilymph. The scala media, which is separated from the scala vestibuli by Reissner’s membrane, is filled with endolymph. The relative movements of Reissner’s membrane and the basilar membrane lead to movements of the tectorial membrane, which cause deflection of the stereocilia in the inner hair cells (IHC) and outer hair cells (OHC).
Sound is converted to vibratory impulses in the tympanic membrane and transmitted via the ossicular chain in the middle ear to the oval window of the cochlea. This receives the impulses and transmits them to the perilymph of the scala vestibuli. Because of the pressure waves, the Reissner’s membrane bends towards the scala media. The pressure wave causes movements of the basilar membrane, and the outer hair cells attached to it brush against the tectorial membrane. The outer hair cells amplify the movements of the basilar membrane for the inner hair cells, which activate the auditory nerve. Thus the inner hair cells are the sensory cells and convert the impulses to the cochlear nerve. The sensitivity and selectivity for sound stimuli cannot be explained entirely by the pressure waves, but rather the outer hair cells serve as an active amplifier, increasing the sensitivity of the organ of Corti. (Palva et al. 1991, Ulfendahl et al. 2000.)
The concentrations of endolymph, the fluid covering the hair cells, are maintained at a high positive resting potential. This is essential for normal hair cell function, because when it is reduced to zero, the result is deafness (Steel et al. 1987). Thus diseases that alter the consistency of the endolymph lead to deafness (Steel 1999). The stria vascularis participates in maintaining the ion concentrations in the endolymph, and it has been proposed that potassium ions may be recycled within the cochlear duct from the outer hair cells to the supporting cells and then to the stria vascularis. Other routes have been proposed, but all of them have in common the fact that the stria vascularis pumps ions to the endolymph via Na-K-ATPase pumps, which is an energy-consuming process (Kikuchi et al. 1995, Spicer & Schulte 1998, Schulte & Steel 1994).
2.4.2. Nuclear DNA mutations causing SNHI
A defect leading to sensorineural hearing impairment will either be in the cochlea or in the auditory nerve. Many of the genes responsible for hearing have been shown to be expressed in the cochlea, and form of cochlear deafness can be caused by defects in the hair cells, non-sensory cells or tectorial membrane. The defect in some forms of deafness is unknown.
Deafness caused by a hair cell defect. Three myosin genes encoding structural proteins in hair cells have been found to be important for structural integrity (Friedman et al. 1999a). MYO7A is involved in Usher syndrome type 1B, suggesting that similar macromolecular interactions are required for the functioning of the ear and eye (Liu et al. 1998). Other motor proteins in hair cells are MYO15 and MYO6, which are both expressed only in the hair cells in the inner ear (Hasson et al. 1995, Liang et al. 1999). Three mutations in MYO15 causing profound congenital deafness have been recognised (Friedman et al. 1995, Wang et al. 1998), and it has been suggested that the pathomechanism may be a defect in actin polymerization, in view of the observation of particularly short stereocilia in mutant mice (Liang et al. 1999). Mutations in the MYO6 gene were first identified in a deaf mouse mutant (Avraham et al. 1997), but a missense mutation in MYO6 has been also detected in an Italian family with a dominant form of progressive hearing loss (Melchionda et al. 2001). Mutations in the harmonin gene and cadherin-23 gene, respectively, have been recognised in patients with forms of Usher syndrome, i.e. USH1C (Bitner-Glindzicz et al. 2000, Verpy et al. 2000) and USH1D (Bryda et al. 1997, Yonezawa et al. 1999, Bolz et al. 2001, Bork et al. 2001). Harmonin contains a PDZ domain, which is a central organizer of high-order supramolecular complexes located at specific emplacements in the plasma membrane. These domains can form homomeric or heteromeric structures and bind to transmembrane proteins, ion channels or transporters, and to actin or actin-binding proteins. Hence they cluster and coordinate the activity of various plasma membrane proteins and provide a bridge between these and the cortical cytoskeleton (Fanning & Anderson 1999, Garner et al. 2000, Sheng & Pak 2000). Interestingly, myosin VIIA and harmonin are present in the whole cell body and the stereocilia, and there may be interaction between them (Verpy et al. 2000, Petit 2001). Several potassium channel proteins, including KCNQ4 and KCNE1, are crucial to potassium recycling (Steel 1999). Other nuclear-encoded proteins present in hair cells and mutations detected in those genes in patients with deafness are the vesicle trafficking protein otoferlin (Yasunaga et al. 1999, 2000) and transcription factor POU4F3 (see section 3.4.4.; Willot et al. 1995, Vahava et al. 1998).
Deafness caused by non-sensory cell defects. Mutations causing deafness have been found in genes encoding gap junction subunits, i.e. connexins, particularly connexin26 (Kelsell et al. 1997), connexin30 (Grifa et al. 1999) and connexin31 (Xia et al. 1998). Mutations in gap junctions and ion channels affect the ion homeostasis in the endolymph. Connexin26 has been found to be responsible for autosomal recessive forms of non-syndromic deafness, and mutational screening has revealed a high prevalence of defects in this gene in deaf populations. Connexin26 accounts for 30-60% of the autosomal recessive forms of isolated deafness in Europe and the United States (Estivill et al. 1998a, Lench et al. 1998, Green et al. 1999, Murgia et al. 1999, Gabriel et al. 2001) and about 50 mutations have been reported (for reviews, see Prasad et al. 2000, Rabionet et al. 2000, Kelsell et al. 2001). Connexin26 has four transmembrane domains, forming a channel which exhibits a greater permeability for positively charged ions or molecules than for negatively charged ones (Cao et al. 1998). The pathogenesis of connexin26 mutations remains unknown. It has been suggested that Pendred syndrome may be the most common form of syndromic deafness, although its prevalence is not known exactly (Fraser 1965). The enlargement of the vestibular aqueduct is the most common anomaly in patients with this syndrome (Phelps et al. 1998), and mutations have been found in the gene encoding pendrin, a chloride and iodide transporter (Scott et al. 1999). The detection of pathogenic mutations in genes encoding proteins such as claudin-14, a tight junction component that separates the fluid compartments of the inner ear, has shown the importance of maintaining the concentrations of these fluids (Wilcox et al. 2001). Other recognised nuclear genes classified into this group are ones that encode proteins such as the extracellular matrix component cochlin (Robertson et al. 1998), the transcriptional coactivator EYA4 (Verhoeven et al. 2000) and the transcription factor POU3F4 (de Kok et al. 1996).
Deafness caused by a tectorial membrane anomaly. The tectorial membrane is an extracellular gel-like matrix, the proper structure of which is required for normal hearing. Mutations in genes encoding α-tectorin, collagen 11α2 and otogelin all lead to defects in the ultrastructure of the membrane (McGuirt et al. 1999, Legan et al. 2000, Simmler et al. 2000).
Deafness of unknown cellular origin. Mitochondrial mutations causing deafness can be classified into this group (see section 2.4.3). Type II transmembrane serine proteases represent an emerging class of cell surface proteolytic enzymes which have been identified relatively recently and have not yet been functionally characterized. Point mutations in the gene have been identified in patients with severe or profound congenital hearing loss (Ben-Yosef et al. 2001, Masmoudi et al. 2001). Diaphanous-1 is involved in cell polarization and cytokinesis (Castrillon & Wasserman 1994, Kohmo et al. 1996, Evangelista et al. 1997) and has been found in a dominantly transmitted non-syndromic form in a large family in Costa Rica (Leon et al. 1981).
2.4.3. Deafness as a mitochondrial disorder
Hearing impairment as a mitochondrial disease is well-recognized and occurs either in a non-syndromic form (Prezant et al. 1993, Oshima et al. 1999) or as part of a syndrome (DiMauro & Bonilla 1997). Mitochondrial deafness is characteristically sensorineural, bilateral, usually progressive and inherited maternally. The age of onset of mitochondrial sensorineural deafness is usually in early adulthood, thus differing from the other types of sensorineural deafness, which are more often congenital.
The pathological mtDNA mutations associated with SNHI can be heteroplasmic or homoplasmic. Mutations have been detected mostly in tRNA genes, but also in rRNA genes. The majority of mtDNA mutations cause syndromic deafness, although non-syndromic forms have been recognized. Hearing loss as a feature of mitochondrial disease was first recognized in patients with the Kearns-Sayre, MELAS and MERRF syndromes. Approximately 70% of patients with these multi-system disorders have SNHI (Gold & Rapin 1994, Uimonen et al. 2001) and 7.4% of adult patients with SNHI who use a hearing aid have been found to harbour the 3243A>G mutation (Majamaa et al. 1998), which is the most common mtDNA point mutation in patients with SNHI (Goto et al. 1990). Hearing loss is often associated with diabetes mellitus and symptoms can occur without neurological symptoms in patients with the 3243A>G mutation or a large deletion (Reardon et al. 1992, van den Ouweland et al. 1992, Ballinger et al. 1992b). The association between SNHI and diabetes mellitus has been confirmed among diabetic patients (Oka et al. 1993, Alcolado et al. 1994). Hearing loss usually develops after the onset of diabetes, so that where 1.3-5.7% of non-insulin dependent diabetic patients in Japan, for example, harboured the 3243A>G mutation, 61% of these also had hearing loss (Kadowaki et al. 1994, 1995, Oka et al. 1995). A mutation 14709T>C in tRNAGlu has been reported in patients with maternally inherited deafness and diabetes mellitus (Vialettes et al. 1997).
Several mtDNA mutations have been found to cause non-syndromic SNHI, including 1555A>G (Prezant et al. 1993), 7445T>C (Reid et al. 1994a), 7472insC (Verhoeven et al. 1999) and 7511T>C (Sue et al. 1998). The mutation 7445A>G was initially described as causing non-syndromic deafness, but it was subsequently found to be associated with the skin condition palmoplantar keratoderma, with additional deafness in at least some of the cases (Reid et al. 1994a, Fischel-Ghodsian et al. 1995, Sevior et al. 1998). There are two case reports of the mtDNA mutations 1555A>G and 1095T>C in two families with deafness and Parkinson’s disease, the inheritance pattern of these conditions being consistent with maternal inheritance (Shoffner 1996, Thyagarajan et al. 1997). These examples appear to complicate the classification of mtDNA mutations into those that causing either a syndromic or a non-syndromic form of deafness.
The phenotypic variability in diseases caused by mtDNA mutations remains unclear. It is not uncommon for the same mutation to cause severe multiorganic disease in one person and a mild phenotype in another member of the family but have no effect at all on a third member. MtDNA mutations are often heteroplasmic, and the quantity of mutant mtDNA differs from one individual to another. This variability, coupled with tissue-specific differences in the threshold and varied dependence on oxidative metabolism, is part of the explanation for the varied clinical phenotypes. In addition, the heteroplasmic level of mtDNA mutation varies between tissues and even between single cells. This may help to clarify the pathomechanisms of the heteroplasmic mtDNA mutations which usually cause syndromic deafness. Two homoplasmic mitochondrial DNA disorders, LHON and non-syndromic deafness, have nevertheless been found to involve similar differences in clinical severity (Newman et al. 1991, Prezant et al. 1993, Chinnery et al. 2000).
Environmental factors such as the use of aminoglycosides can modulate phenotypic expression (Prezant et al. 1993, Fischel-Ghodsian et al. 1997). The mechanism of toxicity of aminoglycosides in the presence of the 1555 mutation is fairly well understood, but unfortunately no other environmental factor has been identified, so that it has proved impossible to build any common pathophysiological model (Fischel-Ghodsian 2000). The mtDNA haplotype and nDNA may influence the phenotype, so that the penetrance of the mutation 7445A>G, for example, is thought to be increased in a certain haplotype (Fischel-Ghodsian 1999). However, some members of families sharing the same mitochondrial haplotype have normal hearing, while others have severe hearing loss (Prezant et al. 1993, Estivill et al. 1998b). A nuclear factor has been suggested as a cause of phenotypic variability in the 1555A>G mutation (Bykhovskaya et al. 2000). Based on the present data, sequence changes in cochlea-specific subunits of mitochondrial ribosomes or RNA-processing proteins have been suggested as interacting abnormally with the mitochondrial defect, leading to insufficient oxidative phosphorylation or loss of a secondary function in the cochlea (Fischel-Ghodsian 2000).
2.4.3.1. Mutations in the 12S rRNA gene
The first mtDNA mutation to be identified as causing maternally inherited non-syndromic hearing loss was 1555A>G in 12SrRNA (Prezant et al. 1993), and since deafness from this cause was first described in patients with a history of taking aminoglycoside antibiotics, an ototoxic mechanism with increased aminoglycoside sensitivity was suggested. Since then deafness has been described as occurring without aminoclygosides (Estivill et al. 1998b). As the phenotypes vary within families, and some of family members may have completely normal hearing (Prezant et al. 1993), other factors probably contribute to deafness in the presence of this predisposing mutation. A genome-wide search for modifying nuclear genes has suggested that a chromosome 8 locus may be involved, but no genes have been detected so far (Bykhovskaya et al. 1998, Bykhovskaya 2000, 2001). The 1555A>G mutation has been found in different mtDNA haplotypes, indicating that it has arisen several times (Fischel-Ghodsian et al. 1993, Prezant et al. 1993, Matthijs et al. 1996, Pandya et al. 1997, Usami et al. 1997, Estivill et al. 1998b, Casano et al. 1998). It is been shown to be homoplasmic in all but one case (El-Schahawi et al. 1997), and the phenotype has been non-syndromic in all except two distinct cases. An association of the mutation with PD or with cardiomyopathy has been reported, although no causal relationship has yet been proved (Shoffner et al. 1999, Santorelli et al. 1999). Other organs, including vestibular system, have been found to be functionally normal (Braverman et al. 1996). The age of onset varies, being mostly during infancy in an Arab-Israeli pedigree and two Italian ones (Braverman et al. 1996, Casano et al. 1998) and during childhood or even adulthood in Spanish families (Estivill et al. 1998b). A marked variation in the frequency of the mutation has been detected between populations, with detection in 19 out of 70 Spanish families with SNHI representing a surprisingly high frequency, even though this may partly be explained by selection bias (Estivill et al. 1998b, Torroni et al. 1999). In Japan, 3% of 319 unrelated SNHI patients and 10% of 140 cochlear implantation patients harboured the 1555A>G mutation, the corresponding prevalences among patients with a history of the use of aminoglycosides being 33% and 59% (Usami et al. 2000).
At the molecular level, nucleotide 1555 in the 12SrRNA gene of human mtDNA is equivalent to position 1491 in the 16SrRNA of E.coli. The mutant nucleotide 1555A>G forms a novel bp with C at position 1494, creating a new binding site for aminoglycosides and facilitating aminoglycoside sensitivity (Hamasaki & Rando 1997). These drugs are known to exert their antibacterial effects at the level of the decoding site of the small ribosomal subunit, causing miscoding or premature termination of translation (Davis & Davis 1968, Chamber & Sande 1996). Both symptomatic or asymptomatic members of a family carrying 1555A>G in their lymphoblastoid cell lines exhibited a specific and significant decrease in growth rate relative to controls in the presence of the aminoglycosides paromycin and neomycin (Guan et al. 1996), and they also showed a decrease in the rate of mitochondrial protein synthesis and respiration, suggesting that the mutation was responsible for the biochemical defects associated with the deafness phenotype. The fact that the decreases were greater in the cell lines derived from symptomatic individuals suggests an additional effect of other factors, e.g. nDNA (Guan et al. 2000, Bykhovskaya et al. 2000). Experiments in which mitochondria harbouring 1555A>G have been transferred to mtDNA-less cells have revealed that the growth rate is similar in asymptomatic and symptomatic cybrids, providing evidence for the assumption that the nuclear background determines the phenotypic manifestation. Although the 1555A>G mutation is clearly the primary factor in the development of deafness, it is not, however, sufficient to produce the clinical phenotype (Guan et al. 2001).
2.4.3.2. Mutations in the tRNASer(UCN) gene
The 7445A>G mutation in the 3’ end of tRNASer(UCN) has been described in at least three unrelated families in Scotland, New Zealand and Japan (Reid et al. 1994a, Fischel-Ghodsian et al. 1995, Sevior et al. 1998). Deafness is combined with palmoplantar keratoderma in two families (Sevior et al. 1998). Clinical re-evaluation of the New Zealand and Scottish pedigrees also revealed thickening of the epidermis on the palms of the hands and soles of the feet in many persons (van Camp & Smith 2000). The penetrance of the mutation is variable, with only few family members in the Scottish pedigree affected while the penetrance was clearly higher in the New Zealand and Japanese pedigrees. Thus the mtDNA mutation by itself does not appear to be sufficient to cause hearing loss, but requires additional genetic or environmental factors. It has been suggested that the difference in appearance may be due to a difference in the mitochondrial haplotype (Fischel-Ghodsian 1999). Three additional sequence changes in complex I were identified in New Zealand pedigree, two of which were considered to be secondary to LHON mutations (Fischel-Ghodsian et al. 1995) and were absent in the Scottish pedigree (Reid et al. 1994b). The 7445A>G mutation does not alter the structure of the tRNA, but rather affects the rate of processing of its precursor, causing approximately a 60-70% reduction in the tRNA level and a decrease in the rate of mitochondrial protein synthesis of light-chain mRNAs (ND6), as detected in lymphoblastoid cell lines from patients (Guan et al. 1997).
7472insC, the second pathogenic mutation in tRNASer(UCN) associated with SNHI, has been reported as being heteroplasmic in a Sicilian family (Tiranti et al. 1995) and in a large Dutch family (Verhoeven et al. 1999). In both families the hearing loss is non-syndromic or combined with ataxia, dysarthria or myoclonus. 7472insC has been reported to be homoplasmic in two families with myoclonic epilepsy, ataxia or cognitive impairment in addition to deafness (Jaksch et al. 1998a, 1998b, Schuelke et al. 1998). In the Sicilian family a full neurological syndrome was observed in patients with a mutant heteroplasmy exceeding 95% (Tiranti et al. 1995), whereas in the Dutch family the only symptom was deafness, the only person with additional neurological symptoms having 99% mutant mtDNA (Verhoeven et al. 1999). The differences in clinical phenotypes suggest the influence of other factors.
The 7472Cins extends the highly conserved TψC arm of the tRNA from five bp to six. The size of this loop has been reported to be crucial for its function (Gu et al. 1996b). The level of tRNASer(UCN) harbouring the insertion is lower in cells, but protein synthesis is not significantly affected (Toompuu et al. 1999).
Non-syndromic deafness has been described in association with homoplasmic 7510T>C and heteroplasmic 7511T>C in tRNASer (Hutchin et al. 1999, Friedman et al.1999b, Sue et al. 1999). These mutations, and also 7512T>C, which causes deafness, progressive myoclonic epilepsy and ataxia (Jaksch et al. 1998b), are located in the acceptor stem of the tRNA molecule, and a mutation in this area is known to disrupt the highly conserved secondary structure.
2.4.4. Presbyacusis
Presbyacusis, age-related hearing loss, occurs in a moderate form (>25dB loss) in 92% of the population over 60 years of age in the UK, and in a severe form (>45dB loss) in 31% (Davis 1995). In a longitudinal study, 97% of subjects with moderate hearing loss experienced a decline in hearing level over time, a 3db decline every ten years in persons under 55 years of age and 9dB in those over 55 years (Davis et al. 1991).
The cochlea has been found to be the site of age-related hearing loss. Decreased sensitivity of the hair cells has been found in patients with presbyacusis relative to younger controls matched for hearing loss (Lehnhardt 1984), and significantly prolonged latencies from the sound stimulus to generation of the peaks have been found when studying evoked auditory responses (Soucek & Michaels 1990), both of these observations suggesting a cochlea lesion. The location of the cell types is less evident, although the outer hair cells seem most likely to be responsible (Wright et al. 1987, Jennings & Jones 2001) on the basis of electrocochleography (Soueck & Michaels 1990), otoacoustic emissions (Bonfils et al. 1988, Nieschalk & Hustert 1996) and histology (Wright 1982, Schuknecht & Gacek 1993).
The cochlea relies on about 100-200 genes for its normal functioning (Morton 1991), and a genetic explanation for presbyacusis has been sought. There is familial aggregation in presbyacusis, this being most notable in women (Gates et al. 1999). Adult hearing loss (Ahl) genes, which code for a form of collagen, have been found to be involved in hearing loss in mice (Johnson et al. 1997), and interestingly, the phenotype of age-related hearing loss caused by a mutation in Ahl gene in mice has been found to be modified by mtDNA mutations (Johnson et al. 2000, Johnson et al. 2001), indicating that mtDNA mutations at least participate in age-related hearing loss. A defect in the gene for the transcription factor POU4F3, which is only expressed in the cochlea hair cells and is necessary for their terminal differentation and trophic support, has been found in a Jewish family who suffered from progressive age-related hearing loss (Willot et al. 1995).
The audiological findings in deafness caused by mtDNA mutations are similar to those in presbyacusis. The hearing threshold is compromised only at high frequencies in the early phase, but the condition eventually leads to a decrease at all frequencies. Similar histological findings, damage to hair cells and spiral ganglion cell degeneration, were observed in old mice with age-related hearing loss (Li & Hultcranz 1994, Spongr et al. 1997, Willot & Erway 1998), supporting the hypothesis of mtDNA mutations causing presbyacusis. The molecular basis of presbyacusis is unknown, but acquired heteroplasmic mtDNA mutations are thought to be associated with it. MtDNA mutations causing loss of oxidative phosphorylation activity seem to play an important role in ageing and degenerative diseases (Golden & Melov 2001), and since lifelong maintenance of the cochlea is critical, it is not unlikely that mitochondrial mutations in the auditory system could lead to presbyacusis. A study of the mtDNA-encoded cytochrome oxidase II gene in the spiral ganglion and membranous labyrinth from the archival temporal bones of five patients with presbyacusis revealed that at least some people with presbyacusis have a significant load of mtDNA mutations in their auditory tissue (Fischel-Ghodsian et al. 1997). Great individual variability was also noted in the quantity and cellular location of the mutations. Although the study included only five patients, it still provides the first evidence of mtDNA mutations causing presbyacusis. In addition, deletions in mtDNA have been found in archival temporal bones of patients with presbyacusis (Bai et al. 1997).