A Mutation in Synaptojanin 2 Causes Progressive Hearing Loss in the ENU-Mutagenised Mouse Strain Mozart

Background Hearing impairment is the most common sensory impairment in humans, affecting 1∶1,000 births. We have identified an ENU generated mouse mutant, Mozart, with recessively inherited, non-syndromic progressive hearing loss caused by a mutation in the synaptojanin 2 (Synj2), a central regulatory enzyme in the phosphoinositide-signaling cascade. Methodology/Principal Findings The hearing loss in Mozart is caused by a p.Asn538Lys mutation in the catalytic domain of the inositol polyphosphate 5-phosphatase synaptojanin 2. Within the cochlea, Synj2 mRNA expression was detected in the inner and outer hair cells but not in the spiral ganglion. Synj2 N538K mutant protein showed loss of lipid phosphatase activity, and was unable to degrade phosphoinositide signaling molecules. Mutant Mozart mice (Synj2 N538K/N538K) exhibited progressive hearing loss and showed signs of hair cell degeneration as early as two weeks of age, with fusion of stereocilia followed by complete loss of hair bundles and ultimately loss of hair cells. No changes in vestibular or neurological function, or other clinical or behavioral manifestations were apparent. Conclusions/Significance Phosphoinositides are membrane associated signaling molecules that regulate many cellular processes including cell death, proliferation, actin polymerization and ion channel activity. These results reveal Synj2 as a critical regulator of hair cell survival that is essential for hair cell maintenance and hearing function.


Introduction
Deafness is an etiologically heterogeneous trait where the age of onset, severity and site of lesion can vary. It can be caused by genetic and/or environmental factors. Environmental risk factors include premature birth, bacterial and viral infections, exposure to loud noise and to ototoxic drugs (which include commonly used aminoglycoside antibiotics such as kanamycin, neomycin and gentamycin, as well as the anticancer drugs cisplatin and methotrexate). However, more than 60% of congenital non-syndromic deafness is caused by mutations in one of an estimated 200 or more ''deafness'' genes [1]. Inherited hearing losses are usually transmitted as monogenic disorders and most often in an autosomal recessive mode [2], the latter accounts for approximately 80% of non-syndromic inherited hearing loss. The identification of genes that cause deafness when mutated has significantly improved our knowledge about the auditory pathway and led to better genetic counseling and management of people affected by hearing loss. Despite this, there are still large gaps in our knowledge of genes involved in hearing loss and the molecular mechanisms of auditory function.
To identify novel genes that contribute to deafness we undertook a phenotypic driven approach in which mice at the Australian Phenomics Facility (APF) were ENU mutagenised and progeny were screened for recessively inherited hearing loss. In one of our strains, Mozart, we have identified the causative mutation to be in the Synj2 gene, which has not previously been associated with deafness. The causative mutation in Synj2 is located in a critical catalytic residue in the 5-phosphatase domain and results in loss of 5-phosphatase activity. In Mozart this mutation leads to gradual hearing loss accompanied by hair cell degeneration. This study identifies Synj2 as a critical regulator of hair cell survival that is essential for maintaining normal hearing.

A Synj2 Mutation is Responsible for Hearing Loss in the Mozart Mouse
The Mozart mouse was identified in a mouse ENU mutagenesis program in which G3 offspring of ENU-mutagenised mice were screened for recessively inherited hearing loss. To identify the causative mutation in the Mozart mouse, genome-wide microsatellite marker analysis on DNA from 20 affected F2 mice was completed. The results showed linkage to D17Mit113 on chromosome 17 with a LOD score of 4.5. Fine mapping of 69 affected mice using amplifluor SNP assays narrowed the candidate region to a 2.1Mb interval between rs13482843 and rs13482851. The coding regions of all 8 known protein coding genes within this region were sequenced and the only change found was in the Synj2 gene. The causative mutation in Mozart is a T to A nucleotide change at nucleotide 1641 (c.1641T.A) in the Synj2 mRNA (Ensembl transcript ENSMUST00000061091). This results in a p.Asn538Lys (N538K) substitution (UniProtKB/Swiss-Prot database Q9D2G5) in the highly conserved catalytic 5-phosphatase domain of the enzyme ( Fig. 1A and 1B). Previous studies have predicted that this Asn orientates a catalytic Asp by forming a hydrogen bond and mutation of this residue results in a complete loss in the ability of INPP5A to hydrolyze Ins(1,4,5)P 3 [21,22]. To investigate the effect of the N538K mutation on Synj2 phosphatase activity, GST-tagged wild-type and mutant Synj2 N538K 5phosphatase domains were expressed in E. coli, purified and analyzed for 5-phosphatase catalytic activity in hydrolyzing PtdIns(3,4,5)P 3 to form PtdIns(3,4)P 4 . While the wild-type Synj2 effectively hydrolysed PtdIns(3,4,5)P 3 generating PtdIns(3,4)P 2 , the Synj2 N538K construct exhibited no 5-phosphatase activity (Fig. 1C). Similarly, wild-type Synj2 degraded PtdIns(4,5)P 2 forming PtdIns(4)P, but the Synj2 N538K mutant construct exhibited no 5phosphatase activity at the same protein concentration (Fig. 1D). Therefore in Synj2, like other 5-phosphatase family members, mutation of the conserved Asp 538 residue results in loss of 5phosphatase catalytic activity.
Synj2 N538K/N538K Mice Show a Progressive Sensorineural Hearing Loss without Vestibular Dysfunction Auditory function was analyzed in Synj2 N538K/N538K mice by measuring auditory brainstem response (ABR). Synj2 N538K/N538K mice are born with normal hearing, but their hearing deteriorated by 8 weeks. By 12 weeks the mice were severely deaf. Fig. 2 shows the hearing profiles of Synj2 +/+ , Synj2 N538K/+ and Synj2 N538K/N538K mice at 4, 8, 12 and 24 weeks. A minimum of 20 mice were included in each group. Heterozygous mice showed a similar audiological profile as wild-type mice. No circling or head tossing/ tilting behavior indicative of a vestibular dysfunction was observed in Synj2 N538K/N538K mice. When we examined the hearing in F2 progeny from the C57BL/6 Synj2 N538K/N538K outcross with the CBA/H mapping strain we noted that a milder deafness phenotype was evident in 25% of the offspring. The hearing thresholds in these mice would be approximately 30dB lower than the C57BL/6 Synj2 N538K/N538K mice. This was not the case if Synj2 N538K/N538K mice were intercrossed with a BALB/c strain, suggesting the interaction of one or more modifier genes from the C57BL/6 background. Genotyping showed that the cadherin 23 Ahl variant (Cdh23 ahl ) is responsible for modifying the severity of the Synj2-associated hearing loss (results not shown).
Neurological and Behavioral Studies in Synj2 N538K/N538K Mice As mutations in genes regulating phosphoinositide signalling pathways are often associated with neurological features [12,23], we examined peripheral nerve conductance in Synj2 N538K/N538K mice. Fig. 3A shows the time from application of the sciatic stimulus pulse to the first and second muscle response in 6 Synj2 N538K/+ and 6 Synj2 N538K/N538K mice. Much of the variation in these data is likely to be due to a slight variation in the position of the stimulating electrode along the sciatic nerve. It is clear however, that there was no systematic variation in the latency between Synj2 N538K/+ and Synj2 N538K/N538K mice. Nerve conductance was also examined directly by determining the difference in latency of the first peak when changing the point of nerve stimulation by a known distance. This was analyzed in mice used previously in the nerve conductance test (numbers 4 to 12). The conduction velocities range from 16-23 meters/second across the 8 mice and no significant differences between Synj2 N538K/+ and Synj2 N538K/N538K mice were detected (Fig. 3B).
Primary behavioral screens of Synj2 +/+ and Synj2 N538K/N538K mice consisting of locomotor, light dark test, Y-maze, tail suspension, hot plate and marble burying tests, did not reveal any significant differences between wild-type and Synj2 N538K/N538K mice.
The mutant Mozart mice have a normal life span, are fertile and, apart from the hearing loss, do not appear to have any additional clinical or behavioural manifestations.

Degeneration of Hair Cell Structure and Stereocilia
Morphology in Synj2 N538K/N538K Mice Examination of outer and middle ear structures did not identify any visible structural changes in Synj2 N538K/N538K mice compared to wild-type mice. Gross structural changes in the inner ear were initially examined by hematoxylin and eosin (H&E) staining of cochlear sections. We observed a gradual loss of inner and outer hair cells in 8 and 12 week old mutant mice, starting in the basal turn of the cochlea and spreading towards the apical region, leading to a collapse of the Organ of Corti (Fig. 4). At the same time there was a reduction in the number of spiral ganglia cells and the remaining cells showed increased vacuolation. Synj2 N538K/+ GST-Synj2 wild-type or Synj2 N538K mutant 5-phosphatase domains or GST alone were purified from E. coli and assayed for PtdIns([ 32 P]3,4,5)P 3 5-phosphatase activity (Panel C). Lipid products were separated by thin layer chromatography (top left panel). The migration of PtdIns(3,4)P 2 and PtdIns(3,4,5)P 3 are indicated. The relative amount of recombinant protein added to each reaction was determined by immunoblotting with GST antibodies (lower left panel). The relative percentage of PtdIns(3,4,5)P 3 substrate hydrolyzed was determined by densitometry (right panel). Bars represent mean 6 SEM from 3 independent experiments. **p,0.001. Recombinant GST-Synj2 wild-type or Synj2 N538K mutant 5-phosphatase domains or GST alone were purified from E. coli and assayed for PtdIns(4,5)P 2 5-phosphatase activity (Panel D). Phosphate release (pmol/25 ml sample) was measured using a malachite green assay (left panel). Bars represent mean 6 SEM from 2 independent experiments. The relative amount of recombinant protein added to each reaction was determined by Western blotting with GST antibodies (right panel). **p,0.001. doi:10.1371/journal.pone.0017607.g001 mice resembled wild-type mice, showing no signs of hair cell degeneration or loss of spiral ganglion neurons (data not shown).
The process of degeneration of the organ of Corti around the time of the onset of hearing impairment was examined in more detail by ultrastructural analysis using scanning electron microscopy (SEM). Cochleae from 2, 4, 8 and 12 week old mice were analyzed (Fig. 5). Degeneration of outer hair cells and fusion of stereocilia was apparent as early as two weeks in the basal cochlear region ( Fig. 5C and 5D). At four weeks of age ( Fig. 5E, 5F, 5G, 5H) signs of degeneration were seen in all turns of the cochlea, although most evident at the basal turn (Fig. 5G, 5H). Both inner and outer hair cells showed fusion of the stereocilia, and basal outer hair cells appear sunken and withdrawn from neighboring supporting cells (Fig. 5H). At eight weeks of age, hair cell degeneration was more pronounced. In the basal cochlear turn, the majority of outer hair cell hair bundles were missing, and those that remained showed extensive stereocilia fusion (Fig. 5K). Some hair bundles were also missing from the outer hair cells of the mid cochlear turn (Fig. 5J). At 12 weeks of age, the basal region of the cochlea was completely void of outer hair cell hair bundles and inner hair cells showed extensive stereocilia fusion and loss ( Fig. 5O). Degeneration of apical and mid regions of the cochlea also progressed; the majority of hair cells exhibited fused stereocilia, and many outer hair cells lacked hair bundles (Fig. 5M, 5N). At each time-point investigated, degeneration of the outer hair cells appeared more advanced and severe than that of the inner hair cells of the same cochlear region.
No Loss of Nerve Fibers in Synj2 N538K/N538K Mice Hair cells are often the primary site of damage in sensorineural hearing loss, with the degeneration of cochlear nerve fibers a secondary consequence [24]. The loss of these fibers has been reported in a number of mouse models for hearing loss [25,26], and in mice exposed to harmful levels of noise [27]. To determine whether loss of cochlear nerve fibers underlies the hearing loss in Synj2 N538K/N538K mice, antibodies were used to detect neurofilament NF-L and neurofilament 200kD (NF-H) in cryosections of 12 week Synj2 +/+ and Synj2 N538K/N538K mice. NF-L expression was evident in fibers of the cochlear nerve in wild-type and Synj2 N538K/N538K mice, when compared to an isotype control (Fig. 6). No apparent differences in the levels of expression of NF-L or NF-H (data not shown), or in the number of fibers, were observed between Synj2 +/+ and Synj2 N538K/N538K mice by this method. Estimating neuronal fiber density or individually counting peripheral axons would however provide a more sensitive and accurate method to determine any discrete differences between genotypes.
Synj2 is Expressed in the Hair Cells, but not the Spiral Ganglion of the Inner Ear As Synj2 mutant mice exhibit deafness, in situ hybridization was performed to determine the site of Synj2 expression within the cochlea (Fig. 7). In wild-type Synj2 +/+ adult mice, Synj2 mRNA expression was detected in the inner and outer hair cells ( Fig. 7A and 7B), but was not observed in the spiral ganglion. In Synj2 N538K/N538K mutants, Synj2 expression was detected at 4 weeks of age, but was not observed at 8 and 12 weeks following the degeneration of hair cells and collapse of the organ of Corti ( Fig. 7C and 7D). Expression of the highly related paralog enzyme, Synj1, was investigated in 12 week old mice. In contrast to Synj2, expression of Synj1 was not detected in hair cells, but was highly  Synj1 expression was still evident in remaining spiral ganglion cells in the Synj2 N538K/N538K mouse, despite obvious degeneration of the spiral ganglion neurons (Fig. 7G). No differences in Synj1 and Synj2 expression levels or patterns were observed when heterozygotes and wild-type mice were compared. We conclude that Synj2, but not Synj1, is expressed in the hair cells of the inner ear.

Discussion
This study has identified a novel regulator of hair cell survival, which is essential for normal hearing, as the phosphoinositide signal regulating enzyme, Synj2. We have identified a mouse mutant with a non-syndromic, recessively inherited, progressive hearing loss due to an ENU-induced mutation in the Synj2 gene. Synj2 N538K/N538K mice exhibit a rapidly progressive hearing loss and at 12 weeks of age are severely deaf. The ABR hearing thresholds in heterozygous Synj2 N538K/+ mice are similar to those in wild-type littermates and the hearing loss in Mozart is therefore recessively inherited. Synj2 N538K/N538K mice do not exhibit circling or head tossing behavior, and respond normally in the simple trunk curl test. We therefore conclude that they do not have vestibular dysfunction. Scanning electron microscopy of cochleae from postnatal day 4 (P4) Synj2 N538K/N538K mice suggests that hair cells have developed normally (data not shown). However, they show signs of degeneration as early as two weeks of age. Initial stages of degeneration include fusion of stereocilia followed by complete loss of hair bundles and ultimately loss of hair cells, which is most profound at the basal cochlear region. Outer hair cells begin to degenerate earlier than inner hair cells and show the most profound degeneration. As the neurofilament numbers are normal in Synj2 N538K/N538K mice, it is likely that loss of such fibers is not the underlying cause of hearing loss in these mice. It therefore appears that Synj2 is essential for hair cell survival and normal hearing.
Synj2 is a member of the inositol polyphosphate 5-phosphatase family, of which ten mammalian enzymes have been characterized, many of which exhibit diverse roles in physiological function. Synj1 and Synj2 are two closely related 5-phosphatases [10,28], that contain an N-terminal catalytic domain (Sac1) involved in hydrolysis of additional phosphoinositides [29] followed by a central 5-phosphatase domain that hydrolyses the 5-position phosphate from the inositol ring of PtdIns(4,5)P 2 , PtdIns(3,4,5)P 3 and Ins(1,4,5)P 3 (Fig. 1). Mice with mutations in Synj1 die shortly after birth due to neurological complications [14] associated with the accumulation of clathrin-coated vesicles in the nerve terminals as a consequence of failure to degrade PtdIns(4,5)P 2 . Deafness or inner ear abnormalities have not been reported in Synj1 mice and Synj1 is not expressed at detectable levels in adult mouse hair cells. However, expression of Synj1 has been previously reported in adult zebrafish hair-cell transciptome using amplified RNA [30], and more recently in developing hair cells of zebrafish larval by in situ hybridization analysis [31], and synaptosomes from 2 week old chicken cochleae using microarray and western blot analysis [32]. Synj1 has been implicated in hair cell function in zebrafish. Synj1deficient nrc and comet zebrafish mutants revealed impaired vestibular-ocular defects with normal startle acoustic response [31,33,34]. Examination of the Synj1 Q296X mutant zebrafish neuromast hair cells showed blebbing near synaptic ribbons. It was shown that this is a consequence of an imbalance between exo-and endocytosis and dependent on Cav1.3 calcium channel activity. Electrophysiological investigations revealed decreased number of readily releasable vesicles and far fewer reserve pool vesicles, as well as defective phase locking of afferent activity. Synj1 was proposed to play a critical role in facilitating vesicle recycling by affecting the number of vesicles released and the timing of release [31]. The absence of detectable Synj1 expression in hair cell in 12 week old mice in this study could reflect the limited sensitivity of the technique utilized and/or indicate a restricted temporal expression of Synj1 in mouse hair cells. Synj2 is a much less characterized enzyme, and our Synj2 N538K/N538K mice are the only reported mouse strain with a mutation in the Synj2 gene (the gene affected in a mouse mentioned in [35] appears to be in Synj1 and not in Synj2). Therefore our study represents the first association of Synj2 with a disease phenotype.
The mutation in Synj2, p.N538K, leads to significant loss of catalytic activity towards two of its substrates, PtdIns(3,4,5)P 3 and PtdIns(4,5)P 2 . The Synj2 Asn 538 residue is highly conserved in species as divergent as Drosophila and yeast. Whisstock et al mutated the corresponding asparagine to an alanine in the mouse 43 kDa inositol polyphosphate 5-phosphatase (encoded by the Inpp5a gene) and showed that this residue is essential for catalytic hydrolysis of Ins(1,4,5)P 3 [22].
Phosphoinositides are components of eukaryotic cell membranes that act as signaling molecules, regulating many cellular processes including actin polymerization, apoptosis and vesicular membrane trafficking [14,36,37,38,39,40,41,42]. Mutations in genes affecting phosphoinositide metabolism can therefore be expected to have important consequences for cell function and survival. Mozart's relatively mild phenotype is somewhat unexpected, especially since the p.Asn538Lys Synj2 mutation appears to significantly ablate 5-phosphatase activity in enzyme assays. Although we are unable to yet explain the mild phenotype of the Synj2 N538K/N538K mice, we can suggest several possible explanations: first, the mutated Synj2 N538K enzyme may have enough residual phosphatase expression to allow normal cell function, except in the hair cells where our in situ results suggest a higher level of Synj2 activity is present and may be required for cell function; second, Synj2 has several important domains that function independently from the 5-phosphatase activity. These functions are retained in our mutant mouse and the lack of Synj2 5-phosphatase activity could be largely compensated for in mice (apart from within the cochlea) by other 5-phosphatases [43]; third, Synj2 N538K protein may bind its substrate PtdIns(4,5)P 2 and/or PtdIns(3,4,5)P 3 , but not release it, thereby affecting an inner ear specific function, or fourth, low levels of Synj1, undetectable by in situ hybridisation, may compensate for the lack of Synj2 5-phosphatase activity in the mutant in a temporal fashion. In support of this hypothesis the detection of phosphoinositide biosensors, AKT/PH-GST or PLCd1PH-GST, by immunohistochemistry in tissue sections from wild-type versus Synj2 N538K/N538K mice did not identify any detectable difference in their distribution or levels.
It is known that inositol phosphates and phosphoinositides do play an important role in hearing. PtdIns(4,5)P 2 is hydrolyzed by phospholipase C to generate Ins(1,4,5)P 3 which mobilizes intracellular calcium. Ins(1,4,5)P 3 and ATP release in cochlear cells leads to the propagation of regenerative Ca 2+ waves, intracellular free Ca 2+ that spreads through cells supporting the sensory cells [6,44,45,46]. This intracellular signaling affects important cellular processes [47], some of which might be essential for correct sound processing [46]. A number of connexin 26 mutations have been reported in people with hearing loss, causing impaired Ins(1,4,5)P 3 connexin hemichannel permeability [7,9], therefore demonstrating that this is a process essential for propagation of Ca 2+ waves and normal cochlear function [8]. Mechanisms for sensing damage to sensory hair cells from acoustic trauma have also been linked to the availability of functional Ins(1,4,5)P 3 stores in supporting cells [48]. Hirono et al showed that PtdIns(4,5)P 2 itself plays an essential role in hair cell transduction and adaptation in frogs [49]. Mice with mutations in the transmembrane phosphatidylinositol phosphatase gene Ptprq [50] develop abnormal hair cell stereocilia, most likely from altered actin organization, or turnover, as well as changes to membrane trafficking due to changes in phosphoinositide levels in hair cells [51]. Phosphoinositides are also involved in mediating ototoxicity of aminoglycoside antibiotics [52,53] through a process involving formation of reactive oxygen species [54]. PtdIns(3,4,5)P 3 has been associated with a decline of survival capacity in ageing outer hair cells [55]. PtdIns(4,5)P 2 has been shown to play a significant role in regulating ion channel activity, including potassium channels such  as KCNQ [56]. Notably mutations in the potassium channel gene, KCNQ4, which is expressed in the hair cells of the inner ear, underlie DFNA2, a subtype of autosomal dominant progressive, high-frequency hearing loss. Mice with loss of function of KCNQ4 exhibit normal vestibular function, but have progressive hearing loss associated with loss of the outer hair cells, reminiscent of Synj2 N538K/N538K Mozart mice [57]. Therefore it is tempting to speculate that loss of Synj2 function in degrading PtdIns(4,5)P 2 leads to altered KCNQ channel activity and hair cell degeneration.
Mutations in other lipid phosphatases that regulate phosphoinositide signaling molecules have been associated with abnormal neurological function. Recently Inpp4a, an enzyme that degrades PtdIns(3,4)P 2 , the lipid product of 5-phosphatase hydrolysis of PtdIns(3,4,5)P 3 , was identified as a suppressor of neuroexcitatory cell death [58]. Pale tremor mice have light fur pigmentation, severe tremors and abnormal gait due to a mutation in the Fig4 gene, coding for a PtdIns(3,5)P 2 5-phosphatase [59]. In humans, mutations in FIG4 cause Charcot-Marie-Tooth type 4J neuropathy [59]. We therefore investigated if Synj2 N538K/N538K mice had more subtle neurological features by measuring nerve conductance in the sciatic nerve, as well as conducting a primary behavioral screen. However, homozygous mutant mice did not show any statistically significant differences from Synj2 N538K/+ littermates.
Mouse models of disease are often excellent models of human conditions. Analysis of human DNA samples to determine if mutations in the SYNJ2 gene cause hearing loss in humans is in progress.
Although studies are beginning to provide an insight into the role of some phosphoinositides in hearing loss, the specific mechanisms of how the p.Asn538Lys mutation in Synj2 N538K/N538K mice causes a progressive, non-syndromic hearing loss, and the role of phosphoinositides in this process is not yet understood. Further characterization of our novel Synj2 N538K/N538K mouse model will provide new insights into the functional and structural features of the auditory system, and the role of Synj2 and phosphoinositides in the auditory process.

Materials and Methods
The Mozart Mouse The Mozart mouse was generated at the Australian Phenomics Facility (APF) as part of a mouse ENU mutagenesis program (http:// www.australianphenomics.org.au/services.html). Male C57BL/6 mice were treated weekly for 3 weeks with 100 mg/kg N-ethyl-Nnitrosourea (ENU). To identify recessive phenotypes, G2 siblings were mated to generate consanguineous G3 offspring homozygous for ENU mutations. Recessive inheritance was validated in the Mozart strain by backcrossing deaf mice onto wild-type C57BL/6 mice followed by brother-sister matings, to confirm that approximately 25% of the offspring had a hearing loss. All mouse procedures were approved by the Royal Children's Hospital Animal Ethics Committee, RCH AEEC #A488 and #A585.

Hearing Testing of Mice
G3 offspring of ENU-mutagenised mice were initially screened for hearing loss using a clickbox that elicits a Preyer reflex or startle response in hearing mice. The clickbox produces an 18.9 kHz burst of 106 dB SPL at a distance of 10 cm (Institute of Hearing Research, Nottingham, UK). It provided a convenient, fast, low-cost and thus suitable high-throughput phenotypic screen. Mice that failed the initial clickbox hearing test had a more detailed assessment of possible hearing loss using an evoked auditory brainstem response (ABR) test (AEP, Bio-logic Systems Corp.). Subdermal active, reference and ground electrodes were placed at the vertex, ventrolateral to the left ear and the abdomen, respectively, of the anaesthetized mouse. Specific auditory stimulus in the form of broadband clicks was delivered in a range of decibel sound pressure levels (dB SPL) and the ABR recorded. Subsequent offspring were then screened by ABR to obtain a hearing profile for each genotype. Mice were also screened for the presence of vestibular dysfunction, identified by the display of hyperactivity that manifests either as circling, head tossing/tilting and/or star gazing behaviour.

Mapping and Mutation Detection
Deaf Mozart mice were outcrossed with the CBA/H mapping strain and the Mozart/CBA/H offspring were then crossed to each other to produce affected F2 generation mice for homozygosity mapping. Genomic DNA (isolated by phenol/chloroform extraction from tail biopsies) from 20 affected mice were analyzed by genome wide scans using 120 microsatellite markers (AGRF, Australia). Fine mapping was performed using Amplifluor-based SNP assays (APF, Australia) on 69 additional affected mice. For statistical analysis we used the normal approximation to the binomial test for proportions of homozygous C57BL/6 genotypes (hearing loss mutants), to map the deafness loci. In regions of linkage, the proportion of homozygous C57BL/6 genotypes would approach 1.0 whilst at unlinked loci this proportion would be 0.25. Linkage intervals were examined for known or putative deafness genes using the UCSC genome browser. PCR primers were designed to amplify and sequence the coding regions, intron-exon boundaries and the 59 and 39 untranslated regions of candidate genes. Primer sequences are available upon request. Sequencing was performed using BigDye terminators V3.1 (ABI).

Site-directed Mutagenesis and 5-Phosphatase Enzyme Assay
Synj2 or Synj2 N538K 5-phosphatase domains were sub-cloned into the pGEX-KG vector in-frame with the GST tag. Expression of recombinant GST-tagged proteins was induced in E.coli by incubation of 50 ml cultures of cells with 0.1 mM IPTG for 3 hours at 30uC. Cell pellets were suspended in 5 ml GST extraction buffer (50 mM HEPES (pH 7.5), 150 mM NaCl, 1 mM EDTA, 10 mM MgCl 2 ) + 1% Triton X-100, 1 mM PMSF, 0.2 mg/ml aprotinin, 0.2 mg/ml leupeptin, 1 mM benzamidine and incubated at 4uC for 90 minutes with gentle agitation. Lysates were centrifuged at 13,000 g for 10 minutes then the supernatants incubated with 100 ml pre-washed glutathione Sepharose TM 4B beads overnight at 4uC with rotation. Samples were washed 6 times in ice-cold GST extraction buffer then PtdIns(3,4,5)P 3 or PtdIns(4,5)P 2 5-phosphatase assays were performed on the affinity precipitates. For PtdIns(3,4,5)P 3 5phosphatase assays, 50 ml PtdIns([ 32 P]3,4,5)P 3 prepared as described [60] was added to the affinity precipitates together with 4 ml 20 x kinase buffer (400 mM HEPES (pH 7.5), 100 mM MgCl 2 , 20 mM EGTA). Reactions were incubated at 37uC for 20 minutes and the lipids extracted and analyzed by TLC as previously described [60]. For PtdIns(4,5)P 2 5-phosphatase assays, 5 mg PtdIns(4,5)P 2 (Echelon Biosciences, Salt Lake City, UT) was added to 40 ml of the affinity precipitates together with 55 ml 1 x kinase buffer. Reactions were incubated at 37uC for 30 minutes then 20 ml supernatants and 80 ml malachite green (Echelon Biosciences) were added to a 96 well plate in duplicate, incubated at room temperature for 15 minutess and the absorbance measured at 655 nm. Phosphate release was quantitated according to the manufacturer's instructions.

Cochleae Isolation
Adult mice were anaesthetized with isoflurane and culled by cervical dislocation according to the National Health and Medical Research Council Australian code of practice for the care and use of animals for scientific purposes. Intact cochleae were surgically removed using a posterolateral approach, as described previously [61]. Brains and other tissues were dissected from mice when needed for further analysis.

Hematoxylin and Eosin (H&E) Staining
Cochleae were isolated from 4, 8 and 12 week Synj2 +/+ , Synj2 N538K/+ and Synj2 N538K/N538K mice. Tissues were fixed in 4% paraformaldehyde (PFA) for 1-1.5 hours at room temperature with rotation, washed 3 x 5 minutes in TBS, and then transferred to 10% EDTA for 4-7 days at 4uC to decalcify. Decalcified cochleae were washed through a series of increasing sucrose solutions and incubated overnight at 4uC in 30% sucrose/PBS, then transferred to OCT (Sakura, USA) and left overnight at 4uC to allow the OCT to fully penetrate the tissue. Cochleae were embedded in OCT, sectioned (10 mm) using a Leica cryostat and placed onto Superfrost Plus microscope slides (Thermo Scientific, USA). Sections were dried at room temperature and stored at 220uC until required. Sections were then subjected to standard H&E staining.

Immunohistochemistry
Cochleae were prepared as described above for H&E staining. Cryosections were thawed at room temperature for a minimum of 30 minutes, washed 3 x 10 minutes in PBS and then incubated with 0.3% Triton X-100 for 20 minutes to permeabilise the tissue. Slides were then briefly washed in PBS and incubated with Image-iT FX Signal Enhancer (Invitrogen) for 30 minutes in a humidified chamber at room temperature to reduce background staining. Slides were washed a further 3 x 10 minutes with PBS, and preblocked in Blocking Solution (30% goat serum (GS), 2% BSA and 0.3% Triton X-100 in PBS) for 30 minutes before incubation with primary antibody in 1% GS, 0.1% Triton X-100 in PBS overnight at 4uC. Slides were washed for 3 x 10 minutes in PBS and incubated for a further 2 hours in the dark with an Alexa Fluor 594-conjugated goat anti rabbit IgG antibody or a 488-conjugated donkey anti mouse IgG antibody (both 1:1500; Molecular Probes) in 1% GS/PBS. Slides were washed an additional 3 x 10 minutes and mounted with ProlongH Gold antifade reagent with DAPI (Invitrogen). Slides incubated with primary antibodies omitted the incubation step with anti-GST antibody. Fluorescence was analysed using a Leica TCS SP2 laser scanning confocal microscope, and images generated with Leica Confocal Software (Leica Microsystems).

Scanning Electron Microscopy (SEM)
For SEM analysis, inner ears were isolated, the round and oval windows were cleared and the cochlear apex was pierced to allow thorough perfusion of the fixative. Inner ears were fixed in 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) with 3 mM CaCl 2 for 3 hours at room temperature. Samples were then washed in PBS and further dissected by removing the bony shell, stria vascularis, Reissner's and tectorial membranes, to reveal the cochlear sensory epithelium. Four, 8 and 12 week cochlear specimens were processed with the OTOTO method (osmium tetroxide/thiocarbohydrazide) adapted from Hunter-Duvar et al [62], then dehydrated in ethanol, critical point dried (CPD 20, Bal-Tec), mounted on stubs using conductive silver paint and viewed using a Hitachi FE S-4800 Scanning Electron microscope. Cochleae from 2 week old mice, and some additional 12 week mice, were coated with osmium tetroxide for 2.5 hours, washed, dehydrated in ethanol, critical point dried (CPD 030, Bal-Tec), mounted on stubs using silver paint, gold sputter coated (Edwards S150B sputter coater, ,10 nm) and viewed using a Philips XL30 FE Scanning electron microscope.

Nerve Conductance Tests
Peripheral nerve conduction was assessed in two ways by electromyography (EMG). The sciatic nerve was stimulated and the time it took for the compound muscle action potential to be generated was measured. These same recordings were also investigated for a second peak, indicating that sciatic stimulation activated Ia sensory fibers in the nerve and evoked a muscle action potential via their monosynaptic connections with sciatic motor neurons in the spinal cord. The second method of examining nerve conduction was to determine nerve conduction velocity directly by taking the difference in the latency of the first peak when changing the point of sciatic nerve stimulation by a known distance. By subtracting the latencies we cancel out any delays in the generation of the EMG response caused by neuromuscular transmission and generation of the muscle action potential, and are left with the difference in latency caused by the change in distance. This is an accurate way to determine if there are differences in conduction velocity in different groups of mice, although the velocities measured by this ''whole nerve'' stimulation approach differ from some single fibre measurements.