A reduction in Ptprq associated with specific features of the deafness phenotype of the miR-96 mutant mouse diminuendo

miR-96 is a microRNA, a non-coding RNA gene which regulates a wide array of downstream genes. The miR-96 mouse mutant diminuendo exhibits deafness and arrested hair cell functional and morphological differentiation. We have previously shown that several genes are markedly downregulated in the diminuendo organ of Corti; one of these is Ptprq, a gene known to be important for maturation and maintenance of hair cells. In order to study the contribution that downregulation of Ptprq makes to the diminuendo phenotype, we carried out microarrays, scanning electron microscopy and single hair cell electrophysiology to compare diminuendo mutants (heterozygous and homozygous) with mice homozygous for a functional null allele of Ptprq. In terms of both morphology and electrophysiology, the auditory phenotype of mice lacking Ptprq resembles that of diminuendo heterozygotes, while diminuendo homozygotes are more severely affected. A comparison of transcriptomes indicates there is a broad similarity between diminuendo homozygotes and Ptprq-null mice. The reduction in Ptprq observed in diminuendo mice appears to be a major contributor to the morphological, transcriptional and electrophysiological phenotype, but does not account for the complete diminuendo phenotype.


Introduction
The diminuendo mutant mouse exhibits deafness and arrested development of the cochlear sensory hair cells. Inner and outer hair cells (IHCs and OHCs) in homozygotes appear immature at young stages and later degenerate; by postnatal day (P)28, very few OHCs are visible. Heterozygotes display poor hearing at P15 followed by rapid deterioration until they lack any response to sound. Their hair cells also appear immature, but are still present at P28 (Lewis et al., 2009;Kuhn et al., 2011). In humans, two MIR96 mutations have been found to be associated with dominantly-inherited progressive hearing loss (Mencia et al., 2009).
The diminuendo causative mutation is a single nucleotide change in the seed region of the microRNA miR-96, which disrupts its function by both loss of its proper targets and gain of novel ones (Lewis et al., 2009). miR-96 is a regulator controlling a wide array of genes, as seen in the microarrays comparing wildtypes and homozygotes (Lewis et al., 2009). However, a few genes were markedly downregulated in diminuendo homozygotes, including Ptprq, which is downregulated to~50% of the wildtype level (Lewis et al., 2009).
Ptprq is a phosphatidylinositol phosphatase, which in the mouse is expressed specifically in the hair bundles of cochlear and vestibular hair cells at early postnatal stages, although expression in the OHCs is transient. In mature hair cells, the staining is restricted to the base of the hair bundle (Goodyear et al., 2003). Three isoforms have been identified in chick hair cells, one expressed at immature stages and two in mature hair cells (Nayak et al., 2011). Ptprq localisation at the base of the stereocilia is actively maintained by Myo6, and is thought to be important for hair bundle maturation and maintenance (Goodyear et al., 2003;Sakaguchi et al., 2008). Mice homozygous for a catalytically inactive form of Ptprq (Ptprq-CAT-KO), which has a shortened intracellular domain but a wildtype C-terminus, do not respond to sound. It is likely to be a functional null; no Ptprq protein is detectable in the hair cells of homozygous Ptprq-CAT-KO mice (Goodyear et al., 2003). At early postnatal stages hair cells are present but the hair bundles are disorganised, and smaller and rounder in shape. At 3 months old hair cells are absent from the basal turn of the organ of Corti. The shaft connectors of vestibular hair cells are absent from Ptprq-CAT-KO homozygotes, as are vestibular evoked potentials (Goodyear et al., 2012). miR-96 is a master regulator, controlling multiple genes and coordinating maturation of the hair cells (Kuhn et al., 2011), and its absence would be expected to affect a wide range of genes, although the pathways it regulates are not yet known. Understanding how much the downregulation of Ptprq contributes to the diminuendo phenotype is important for understanding the overall action of miR-96. We therefore undertook a detailed morphological examination of the hair cells, electrophysiology and transcriptome of diminuendo homozygotes and heterozygotes, Ptprq-CAT-KO homozygotes and controls, to ascertain what contribution the downregulation of Ptprq makes to the diminuendo phenotype.

Mice
Mouse studies were carried out in accordance with UK Home Office regulations and the UK Animals (Scientific Procedures) Act of 1986 (ASPA) under a UK Home Office licence, and the study was approved by the Wellcome Trust Sanger Institute's Ethical Review Committee. Mice were culled using methods approved under this licence to minimize any possibility of suffering.
Two genetically modified mouse lines were used; the diminuendo line and the Ptprq-CAT-KO line. The diminuendo phenotype is the result of an N-ethyl-N-nitrosourea (ENU) mutagenesis screen (Lewis et al., 2009); the line was generated and has been maintained on a C3HeB/FeJ background. The Ptprq-CAT-KO allele is a targeted knock-out in which a 1061 bp fragment of Ptprq, which contains two of the exons coding for the catalytic domain, is replaced by a neomycin resistance cassette. 129/Sv stem cells were used, and the mouse line was established and has been maintained on a mixed C57BL/6J/129/Sv background (Goodyear et al., 2003). Fifty-eight animals were used in this study in total.

Scanning electron microscopy
Scanning electron microscopy was carried out on diminuendo homozygote, heterozygote and wildtype and Ptprq-CAT-KO homozygote and wildtype mouse inner ears at P4 (four mice of each genotype, of either sex) as described in Chen et al. (2013), using the OTOTO method (Hunter-Duvar, 1978). Images were taken using an Hitachi SE-4800 microscope at 5 kV under a standard magnification (9 80) to show the whole organ of Corti and a higher magnification (9 15 k) for IHCs and the outermost row of OHCs at 10% intervals along the length of the cochlear duct from 10 to 90% of the distance from the base to apex (Fig. 1). No obvious differences of the size of the organ of Corti were observed during dissection of the different genotypes, and all mice examined had the normal number of cochlear turns. Cartoons of hair cells typical of each point along the organ of Corti and genotype were drawn based on the micrographs, to summarise their appearance from a similar angle.

Microarray
Organ of Corti dissection and RNA extraction were carried out as described in Lewis et al. (2009), using male Ptprq-CAT-KO homozygous (n = 6) and wildtype (n = 6) P4 littermates collected within a 2-h time window to control for circadian changes. Total RNA (500 ng) for each sample was amplified and purified using the Illumina TotalPrep-96 RNA Amplification kit (Ambion, UK), according to the manufacturer's instructions. Biotin-labelled cRNA was then normalized to a concentration of 150 ng/uL and 1500 ng was hybridised to Illumina MouseWG-6 v2 beadarrays (Illumina, CA, USA) for 16 h (overnight) at 58°C. Following hybridisation, beadarrays were washed and stained with streptavidin-cy3 (GE Healthcare, UK). Beadarrays were then scanned using the Beadarray reader and image data was then processed using Genome Studio software (Illumina). The data were normalised using a quantile normalisation, making the assumption that the overall intensity distributions of the arrays should be comparable (Bolstad, 2001;Bolstad et al., 2003). Normalised data was then analysed using the Bioconductor LUMI and LIMMA packages (Gentleman et al., 2004;Smyth, 2004;Smyth et al., 2005). The P-value was adjusted for multiple tests as described in Benjamini & Hochberg (1995), and the P-value cut-off was set to 0.05. Probe data from wildtype mice was compared to probe data from homozygote mutant mice; the only difference between each pair of littermates was the genotype. Microarray data are available in the ArrayExpress database (www.ebi.ac.uk/arrayex press) under accession number E-MTAB-1785. Further analysis was carried out using the R statistical software package (Team, 2012) and the GSEA (Gene Set Enrichment Analysis) software (Mootha et al., 2003;Subramanian et al., 2005) on ranked microarray gene lists, using gene pathway sets downloaded from mSigDB (http:// www.broadinstitute.org/gsea/msigdb/index.jsp, August 2013), Reactome (http://www.reactome.org/, July 2013) and Pathway Commons (http://www.pathwaycommons.org, July 2013). GSEA does not accept duplicate gene IDs, so if duplicates were present in the input gene list the highest absolute value was used.

cDNA creation and RTPCR
Each pair of RNA samples (homozygote and wildtype Ptprq-CAT-KO P4 littermates of either sex; fourteen mice used in total) were normalised to the same concentration before treatment with DNAse 1 (Sigma; cat.no. AMP-D1). cDNA was created using the Superscript II Reverse Transcriptase kit (Invitrogen; cat. no. 11904-018). Method of measuring the length of the cochlear duct to make comparisons in corresponding positions. In order to compare hair cell bundles in corresponding positions along cochlear ducts between wildtype and mutant mice or between different mutant mice, the images were taken under a standard magnification (9 80) to show the whole organ of Corti. The cochlear duct was divided into ten parts of 10% from base to apex. A higher magnification (9 15 k) was used for analysis of IHCs and the outermost row of OHCs at every 10% interval along the length of the cochlear duct. Scale bar, 300 lm.
Primer/probe sets and the Taqman Master Mix were purchased from Applied Biosystems (Master Mix: 4364340; Hprt1: Mm01318747_g1; Jag1: Mm01270190_m1; Calb2: Mm00801461_m1; Grxcr1: Mm01217856_m1; Hsd17b7: Mm00501703_m1. Primer/ probe sets for Chrna1, Chrna9 and Otof were manually designed using a tool supplied by Applied Biosystems). Hprt1 was used as the internal control, and Jag1, which is expressed in supporting cells (Morrison et al., 1999;Zine et al., 2000;Lewis et al., 2009), was used to control for sensory tissue content. Pairs were only used if their Jag1 expression levels did not differ significantly (where significance was determined to be P < 0.05). Quantitative PCR was run on an ABI7000 machine (Applied Biosystems). Four technical replicates were carried out per sample, and three wildtype-homozygote pairs were tested for each probe. The 2 ÀDDCt test was used to determine relative expression levels for each probe (Livak & Schmittgen, 2001); the calibrator was the wildtype littermate threshold for the same probe and the internal control was Hprt. Student's t-test was used for results which were both parametric and homoscedastic, Welch's t-test for parametric, heteroscedastic data, the Wilcoxon test for non-parametric, homoscedastic data and the Brunner-Munzel test for results which were neither parametric nor homoscedastic. All significance tests were non-directional, and a significance level of a < 0.05 was used. The F-test and Fligner-Killeen test were used to estimate homoscedasticity, and the Shapiro-Wilks test for normality. The Bonferroni test was used to adjust P-values to take multiple testing into account. Excel (Microsoft, Seattle, USA) and R (Team, 2012) were used to carry out the calculations and statistical tests.

Single hair cell electrophysiology
Outer hair cells from the apical coil of diminuendo homozygote, heterozygote and wildtype mice of either sex were studied in acutely dissected organs of Corti from P1 to P6, where the day of birth is P0. Twenty-four mice were used in total. Animals were killed by cervical dislocation in accordance with UK Home Office regulations. Cochleae were dissected as previously described (Johnson et al., 2012) in normal extracellular solution (in mM): NaCl, 135; KCl, 5.8; CaCl 2 , 1.3; MgCl 2 , 0.9; NaH 2 PO 4 , 0.7; D-glucose, 5.6; and Hepes-NaOH, 10. Sodium pyruvate (2 mM), MEM amino acids solution (50 9, without L-glutamine) and MEM vitamins solution (100 9) were added from concentrates (Fisher Scientific, UK). The pH was adjusted to 7.5 (osmolality~308 mmol/kg). All animals were genotyped as previously described (Lewis et al., 2009).
Mechanoelectrical transducer (MET) currents were elicited by stimulating the hair bundles of OHCs using a fluid jet from a pipette (tip diameter 8-10 lm) driven by a piezoelectric disc (Johnson et al., , 2012. The pipette tip of the fluid jet was positioned near to the bundles to elicit a maximal transducer current. Mechanical stimuli were applied as force-steps or 50-Hz sinusoids (filtered at 0.25 kHz, 8-pole Bessel) with driving voltages of AE 40 V. MET currents were recorded with a patch pipette solution containing 1 mM EGTA as the calcium buffer, which was previously assessed using perforated patch recordings (Johnson et al., 2008). The effect of endolymphatic Ca 2+ concentration [reported to be between 0.02 and 0.04 mM Ca 2+ -Bosher & Warren, 1978;Salt et al., 1989)] was examined by superfusing the hair bundle with 0.04 mM Ca 2+ (buffered with 4 mM HEDTA) using a multi-barrelled pipette positioned closed to the patched cells. During the experiments in which different extracellular Ca 2+ concentrations were used (0.04 mM or 1.3 mM), the fluid jet pipette was also filled with the same solution.
Statistical comparisons of means were made by one-way ANOVA followed by the Tukey multiple comparison test with alpha = 0.05. Mean values are quoted AE SEM.

Hair bundle structure -OHCs
In the wildtype mice at P4, there was a gradient in maturity of OHCs from cochlear apex towards base with apical hair cell bundles showing a slightly more immature, rounded arrangement, a few short microvilli remaining within the cleft of the bundles, and narrower angles between the two arms of the V-shaped array. Basal OHCs showed a wide-angled array of stereocilia, three rows of graded height organised in straight lines with few if any excess microvilli remaining in the cleft of the V-shaped array (Figs 2A and B and 3).
All the OHCs in diminuendo homozygous mice looked immature compared with wildtype OHCs at the corresponding positions. Mutant OHCs from the base up to the 80% position showed a similar appearance, without an obvious developmental gradient (Fig. 3). In contrast, above the 80% position a developmental gradient was observed, suggesting that the development of mutant OHCs may have stalled at a set point in differentiation. The hair bundles of mutant OHCs showed a more rounded shape than those of wildtype mice and where there was evidence of a V-shaped array, the angle was narrower than that of wildtype mice. These mutant OHCs had six to eight rows of stereocilia in contrast to three rows in wildtype mice, and displayed retention of microvilli in the cleft of the bundle, with some clearly ectopic stereocilia making some bundles look more like an O-shape than a V-shape ( Heterozygote (Dmdo/+) OHCs showed an intermediate phenotype in all notable respects. For example, they did show a developmental gradient from apex towards base, where the hair bundles had a V shape and wider angle between two arms than diminuendo homozygous mutants, but even at the base they were narrower than wildtype bundles. There were a few ectopic stereocilia at 80%, similar to diminuendo homozygous mutants, but not at 30%, where they more closely resembled wildtypes (Figs 2C and D and 3).
The Ptprq-CAT-KO homozygous OHCs showed abnormal hair bundles at P4, and we found that they appeared to share more similarities with diminuendo heterozygous OHCs than with diminuendo homozygous OHCs. First, both diminuendo heterozygote and Ptprq-CAT-KO homozygote OHCs displayed a developmental gradient from apex towards base, which seemed to be stalled in diminuendo homozygotes (Fig. 3). Second, the gross morphology of OHC bundles in diminuendo heterozygous and Ptprq-CAT-KO homozygous mice was not as rounded as diminuendo homozygous mutants ( there were still extra rows of stereocilia/microvilli in diminuendo heterozygotes and Ptprq-CAT-KO homozygotes, but they appeared to be fewer than in diminuendo homozygotes ( Fig. 2D and F; compare to Fig. 2H). Typical OHC bundles for each genotype are summarised and drawn based on micrographs for a more direct comparison as shown in Fig. 3. The kinocilium was present in all genotypes including wildtype along the length of the cochlear duct, so was not useful as a measure of differentiation.

Hair bundle structure -IHCs
Most wildtype IHCs had a well-defined wide V-shape, although at the 90% position near the apex, the bundles were more rounded than further towards the base of the cochlear duct ( Fig. 4A and B). IHC stereocilia had grown thicker, with the second tallest row having the thickest shafts and forming wedge-shaped (tented) apical ends, followed by the tallest row, and the innermost stereocilia were the thinnest as reported previously (Erven et al., 2002;Holme et al., 2002). Persistent microvilli were observed within the cleft of the bundle along the length of the organ of Corti. The stereocilia from the same row were almost the same height as each other, so hair bundles had a clear staircase arrangement (Figs 4A,B and 5).
IHCs of diminuendo homozygous mutants differed from those of wildtype mice mainly by not showing thickening of their stereocilia, such that there was little distinction between stereocilia rows save the height. There was also less distinction between stereocilia and microvilli than in IHCs of wildtype mice ( Fig. 4G and H). A minority of homozygote inner hair cells in the basal half had hair bundles consisting of bunched stereocilia with no height gradation. No microvilli were observed in these hair cells (Fig. 6). Diminuendo heterozygote IHCs showed an intermediate appearance but generally were much more similar to wildtype IHCs than to diminuendo homozygote IHCs ( Fig. 4C and D). Heterozygote IHCs showed some thickening of stereocilia, unlike the homozygote IHCs. However, there was a slight developmental delay observed towards the base of the cochlea (Fig. 5).
The Ptprq-CAT-KO homozygous IHCs displayed a phenotype intermediate between diminuendo homozygotes and wildtypes. For example, both diminuendo heterozygote and Ptprq-CAT-KO homozygote IHCs bore extra rows of stereocilia/microvilli, but not as many as diminuendo homozygote IHCs ( Fig. 4D and F; compare to Fig. 4H). Thickening of IHC stereocilia was seen in Ptprq-CAT-KO homozygous and diminuendo heterozygous mice and the arrangement of the stereocilia in the same row was neater than that of the diminuendo homozygous mice ( Fig. 4C and E; compare to Fig. 4G). However, we also observed some Ptprq-CAT-KO inner hair cells resembling the diminuendo homozygote hair cells with bunched stereocilia (as shown in Fig. 6C). Typical inner hair cell bundles for each genotype are summarised and drawn based on micrographs for a more direct comparison in Fig. 5. As with OHCs, the kinocilium was observed in all genotypes and at all positions in the cochlea.

Summary of hair cell appearance
In summary, we observed developmentally immature morphology in diminuendo heterozygous, homozygous and Ptprq-CAT-KO homozygous mice. When comparing the development of OHC hair bundles in terms of gross morphology, number of stereocilia rows and form and arrangement of stereocilia, diminuendo heterozygous and Ptprq-CAT-KO homozygous mice are more similar to each other and the morphological changes observed in these two genotypes seem not as severe as those seen in diminuendo homozygous mice. However, Ptprq-CAT-KO homozygote and diminuendo homozygote IHCs both appear markedly more affected than diminuendo heterozygote IHCs, which closely resemble wildtype IHCs (Fig. 5). The relationships can be summarised as follows: OHCs: Dmdo/Dmdo < Ptprq/Ptprq = Dmdo/+ < WT IHCs: Dmdo/Dmdo < Ptprq/Ptprq < Dmdo/+ < WT These findings suggest that the downregulation of Ptprq is likely to be an important contributor to the hair cell phenotype observed in diminuendo mice, but it is not sufficient to explain all of it.

Microarray and qRTPCR
None of the probes showed a significant difference in expression levels between wildtypes and Ptprq-CAT-KO homozygotes after significance was adjusted to allow for multiple tests. However, there were a number of interesting genes among those with an unadjusted P < 0.01 (Table 1). These include Calb2, Hsd17b7, Chrna1, Chrna9 and Otof, which showed altered expression in the diminuendo microarray and further quantitative PCR, as we have previously reported (Lewis et al., 2009; see also Kuhn et al., 2011), and Grxcr1, which was of interest as the gene mutated in the pirouette mutant (Odeh et al., 2010). All six genes were downregulated in the Ptprq-CAT-KO microarray (Table 1), while in diminuendo homozygotes Calb2, Chrna1, Chrna9 and Hsd17b7 were downregulated (Lewis et al., 2009) and Otof was upregulated (Kuhn et al., 2011). These six genes were chosen for validation by quantitative RTPCR in the present study. . This gradient is still present in diminuendo heterozygote mice (centre left), but appears delayed; at each stage depicted, the hair cells resemble the next most apical stage of the wildtype. A similar difference can be observed between Ptprq-CAT-KO homozygote OHCs (centre right) and diminuendo heterozygote OHCs, with Ptprq-CAT-KO OHCs at 30% resembling diminuendo heterozygote OHCs at 50%, which resemble wildtype OHCs at 70%, for example. No Ptprq-CAT-KO OHCs were preserved for observation at 10%. Diminuendo homozygotes (right) display a much more extreme phenotype, with hair cell development appearing to stall at~50%, and hair cells more apical than that appearing very immature. qRTPCR of these six genes showed that Grxcr1, Otof and Hsd17b7 were all significantly downregulated in Ptprq-CAT-KO homozygotes compared to wildtype littermates [littermates Grxcr1, P = 1.72 9 10 À7 (Student's t-test); Otof, P = 4.52 9 10 À5 (Welch's t-test); Hsd17b7, P = 7.52 9 10 À9 (Student's t-test)] ( (Fig. 7), confirming the microarray results. The expression levels of Chrna1 in Ptprq-CAT-KO homozygotes relative to wildtype littermates varied widely between the three pairs used, and Chrna9 and Calb2 were not significantly affected [Chrna9, P = 1 (Wilcoxon t-test); Calb2, P = 0.15 (Welch's t-test)]. Expression of Grxcr1 was not significantly altered in the diminuendo microarray and Otof is upregulated in diminuendo homozygotes (Lewis et al., 2009;Kuhn et al., 2011), so in this case only the Hsd17b7 downregulation resembles the diminuendo homozygote phenotype.
We also compared the microarray data on a larger scale, using first a Pearson correlation comparing the probe log proportional changes across both microarrays (13 542 of the probes were present in both microarrays, out of 21 229 in the Ptprq microarray and 25 044 in the diminuendo array). This resulted in a Pearson coefficient of 0.31, which is a comparatively low correlation (Fig. 8A). When considering only probes with an unadjusted P-value < 0.1 in both arrays (146 probes, of 1012 probes with P < 0.1 in the Ptprq microarray and 2506 probes with P < 0.1 in the diminuendo microarray), however, the coefficient was 0.62 (Fig. 8B). The same data (probes with unadjusted P < 0.1 from both microarrays) were used to create a heatmap visualisation of the probe proportional changes from both arrays (Fig. 8C). While the magnitude of regulation differed between diminuendo homozygotes and Ptprq-CAT-KO homozygotes, the scattergraph and heatmap show that most of the genes were regulated similarly; those upregulated in diminuendo homozygotes were upregulated in Ptprq-CAT-KO homozygotes (green in the heatmap) and those downregulated in diminuendo homozygotes were downregulated in Ptprq-CAT-KO homozygotes (pink in the heatmap). The magnitude of up-or downregulation compared to wildtype littermates was noticeably greater in the diminuendo array than in the Ptprq-CAT-KO array (indicated by the darker colours). Pathway analysis was carried out using GSEA with ranked lists of genes from both arrays; again, an unadjusted P < 0.1 was used. GSEA takes a list of genes with a ranking (in this case, the log of the proportional change) and a set of known pathways and tests to see whether any of the pathways are over-represented by the genes at the top or the bottom of the ranked list. Overall, we found more pathways significantly over-represented in diminuendo dysregulated genes (P < 0.05) but, after removing uninformative pathways (such as 'system development' and 'system process'), several pathways were significantly enriched in both Ptprq and diminuendo dysregulated genes ( Fig. 9A and B, Table 2). The shared pathways related to the neurological system and to cell proliferation. No canonical pathways (as defined by mSigDB) were shared, but there were multiple transcription factors implicated in the transcriptional phenotype of both mutants (Fig. 9C).
The broad pattern of the transcriptome of Ptprq-CAT-KO homozygotes resembled that of diminuendo homozygotes but was not as severe. In addition, not all the genes affected were regulated in the same direction. The downregulation of Ptprq may contribute to but cannot explain all the transcriptome effects seen in diminuendo mice.

Hair cell transducer current is impaired in diminuendo mutant mice
We measured MET currents from OHCs of diminuendo mice in order to compare them with those previously recorded from cells of (B) This is the more common diminuendo homozygote phenotype. The diminuendo homozygote hair cell resembles the wildtype but appears less mature, with still-thin stereocilia and more microvilli. (C) In contrast, a minority of homozygote inner hair cells lacked the staircase organisation and microvilli entirely. We observed no hair bundles intermediate between these two phenotypes. Scale bars, 300 lm. Ptprq-mutant mice, and they showed a reduced amplitude (Goodyear et al., 2003). Transducer currents from P6 apical-coil OHCs in diminuendo control (+/+), heterozygous (Dmdo/+) and homozygous mutant (Dmdo/Dmdo) mice were elicited by alternating inhibitory and excitatory bundle displacements using 50-Hz sinusoidal force stimuli (Johnson et al., , 2012. Upon moving the bundles in the excitatory direction (i.e. towards the taller stereocilia) and at negative membrane potentials, an inward MET current could be elicited in OHCs from all genotypes ( Fig. 10A-C), suggesting that the tip links were present and functional. The maximal MET current in control OHCs (À1941 AE 79 pA at À122 mV, n = 7) was significantly larger than that in heterozygous (À1581 AE 68 pA, n = 6, one-way ANOVA, Tukey post-test -P < 0.01) and homozygous (À273 AE 78 pA n = 5, one-way ANOVA, Tukey post-test -P < 0.0001) mutant cells (overall significance -F 2,15 = 123.1, P < 0.0001). Any resting current flowing through open MET channels in the absence of mechanical stimulation was reduced when bundles were moved in the inhibitory direction (i.e. away from the taller stereocilia) in all control and mutant OHCs ( Fig. 10A-C,  arrows). However, the open probability of MET channels at rest in 1.3-mM Ca 2+ was significantly larger in control (0.055 AE 0.004 at À122 mV, n = 7) than in heterozygous (0.026 AE 0.005 pA, n = 6, one-way ANOVA, Tukey post-test -P < 0.001) and homozygous (0.004 AE 0.003 pA, n = 5, one-way ANOVA, Tukey post-test -P < 0.0001) mutant cells (overall significance -F 2,15 = 33.5, P < 0.0001). Because the MET current reverses near 0 mV List of genes from the microarray with differences in expression between Pprq-CAT-KO homozygotes and wildtype littermates (unadjusted P < 0.01). Four genes are known to be involved in deafness (PTPRQ, OTOF, GRXCR1 and CHRNA9). , it became outward when excitatory bundle stimulation was applied during voltage steps positive to its reversal potential ( Fig. 10A-C). At positive potentials (+ 98 mV), the larger resting transducer current ( Fig. 10A-C, arrowheads) is likely to be due to an increased open probability of the transducer channel resulting from a reduced driving force for Ca 2+ influx (Crawford et al., 1989;Johnson et al., 2011). Note that the residual outward K + current observed at + 98 mV, just before the beginning of the MET current, was largely reduced in heterozygous and homozygous mutant OHCs, which is consistent with a failure in the normal development of the basolateral membrane currents in these diminuendo mutant cells (Kuhn et al., 2011). We asked whether the Ca 2+ sensitivity of the MET current was affected in diminuendo mutant OHCs by locally superfusing their hair bundle with a solution containing an endolymph-like concentration of Ca 2+ (0.04 mM; Bosher & Warren, 1978). Lowering the extracellular Ca 2+ concentration is known to increase both the maximum MET current and its fraction activated at rest. Calcium is a permeant blocker of the transducer channel (Ricci & Fettiplace, 1998;Marcotti et al., 2005), so the increased current amplitude in low Ca 2+ is caused by the partial relief of this block. Moreover, extracellular Ca 2+ causes adaptation and as such closes some transducer channels. Therefore, reducing Ca 2+ influx into the transducer channel, by either depolarizing hair cells to near the Ca 2+ equilibrium potential (as shown in Fig. 10A-C) or lowering the extracellular concentration, causes an increased open probability of the channel (Crawford et al., 1991). In diminuendo mice, both phenomena were observed because decreasing the Ca 2+ concentration from 1.3 to 0.04 mM increased the resting (Fig. 10D-F) and maximal (Fig. 10D, E and G) MET current in control and homozygous OHCs. The above results indicate that the size of the MET current, but not its Ca 2+ sensitivity, is affected by mutation in miR-96. miR-96 has recently been shown to regulate the progression of the physiological differentiation of cochlear hair cells as in its absence the development of the basolateral biophysical properties is arrested just after the day of birth (Kuhn et al., 2011). We tested whether miR-96 was also able to control the normal developmental maturation of the mechanoelectrical transduction apparatus in hair cells by comparing the properties of the MET current in heterozygous and homozygous diminuendo OHCs to that of control cells. Hair bundles from OHCs were stimulated using alternating inhibitory and excitatory step mechanical stimuli instead of sinusoids, which allows the measurement of MET channel adaptation in addition to current size. In control P6 OHCs, excitatory bundle movements with non-saturating stimuli elicited rapid inward currents at a holding potential of À84 mV (Fig. 11A, arrow in left panel). Inhibitory hair bundle stimulation shuts off the small fraction of current flowing at rest (see also Fig. 10A) and, at the offset of large inhibitory steps, a transient rebound (downward dip; Fig. 11A, arrowhead in left panel) was observed. All these manifestations of MET current adaptation are similar to those described in control auditory hair cells from lower vertebrates and mice (Crawford et al., 1989;Kennedy et al., 2005). Slipping and rebound adaptation were largely reduced or absent in heterozygous and homozygous diminuendo P6 OHCs (Fig. 11A, middle and right panels). We then compared the MET current responses in diminuendo mutant OHCs (Fig. 11A) with those recorded in control cells during early postnatal development (P1-6; Fig. 11B shows examples at P2, P3 and P4). We found that both the amplitude (Fig. 11C) and the extent (Fig. 11D) of adaptation during excitatory stimulation (Fig. 11A, arrow in left panel) of the MET current in P6 OHCs from diminuendo homozygous mutant were similar to those recorded in P1-P2 control cells (Fig. 11C and D,respectively;red). An intermediate phenotype was seen for heterozygous diminuendo OHCs ( Fig. 11C and D; blue).
In summary, we have shown that the biophysical properties of the MET currents in diminuendo homozygous OHCs are arrested at a very early stage of postnatal development. We also see a strong similarity between the reduced amplitude of the MET current in diminuendo heterozygotes with that previously reported in OHCs from Ptprq-null mutant mice (Goodyear et al., 2003). However, both slipping and rebound adaptation are absent or largely reduced in diminuendo heterozygotes while they were still present in Ptprq-null mutants (Goodyear et al., 2003).

Discussion
Homozygous diminuendo mice at P4 lack the normal gradient of hair bundle development of OHCs between 0 and 80% of the cochlear length while above that in the apex the hair cells are still less mature than those at the equivalent position in wildtype littermates. This suggests developmental retardation followed by stalling at a Genes with unadjusted P < 0.1 were used, compared to four different gene setsgenes annotated with GO Biological Processes, genes playing a role in Reactome pathways, genes associated with Pathway Commons pathways and genes regulated by transcription factor-associated regulatory sites. The number of significantly enriched pathways (P < 0.05) is followed by the total number of gene sets considered in brackets. In order for a gene set to be considered, at least ten of its genes had to be present in the list compiled from the microarray. certain point in differentiation. The hair cells appear unable to differentiate any further. Heterozygous diminuendo mice show a delay in differentiation of OHCs at P4 but there is no stalling of development at this stage; the gradient of maturity is present throughout the organ of Corti. We have previously shown that, at 4 weeks old, diminuendo heterozygote OHCs still display significantly narrower bundle span than wildtypes while supporting cells have wider apices, and hair cell density remains the same (see Supporting Information Fig. s1p and q in Lewis et al., 2009). The OHCs of P4 Ptprq-CAT-KO homozygous mice also appear to exhibit delayed differentiation, similar to diminuendo heterozygotes but with a greater delay (Fig. 3). The IHCs of diminuendo heterozygotes are broadly similar to wildtypes, but both diminuendo homozygote and Ptprq-CAT-KO homozygote IHCs have disorganised stereocilia bundles. However, Ptprq-CAT-KO IHCs display more hallmarks of maturity than diminuendo homozygote IHCs, such as well-organised stereocilia rows and thickening of stereocilia tips. The overall phenotype of Ptprq-CAT-KO hair cells bears a striking resemblance to that of diminuendo heterozygous mice and is not as severe as that of diminuendo homozygous mice, suggesting that the reduction in Ptprq activity is a major contributor to the diminuendo stereocilia morphological phenotype. This observation also fits with part of the electrophysiological phenotype, where heterozygous diminuendo mice resemble homozygous Ptprq-null mice (Goodyear et al., 2003), with a reduced MET current amplitude compared to controls. In diminuendo homozygotes the phenotype is more severe, as normal development of the MET current amplitude appears to be arrested just after birth (P1-2), consistent with the proposed role of miR-96 in regulating the progression of the physiological and morphological differentiation of mammalian cochlear hair cells (Kuhn et al., 2011). Furthermore, the downregulation of Ptprq expression in the diminuendo mutants is not likely to be responsible for the adaptation defects seen in these mice, because adaptation in Ptprq mutants is near to normal (Goodyear et al., 2003). The transcriptome analyses, on the other hand, show less of a similarity between Ptprq-CAT-KO and diminuendo. This is not unexpected -Ptprq is not a regulatory factor but performs a specific function in the hair cell, while miR-96, which is a regulator, would be expected to affect the expression of many genes. However, when just the genes with a P-value < 0.1 are considered, there are similarities between Ptprq-CAT-KO and diminuendo, with many of the genes affected in Ptprq-CAT-KO homozygotes being similarly regulated in diminuendo homozygotes, as seen in the heatmap (Fig. 8C) and reflected in the correlation coefficient, r = 0.62 (Fig. 8B). An initial exploration of pathways affected shows that some of the pathways and transcription factors implicated by the transcriptional data are common to both mutants. Of the three genes whose expression changes were confirmed by qRTPCR, however, only downregulation of Hsd17b7 is seen in both Ptprq-CAT-KO and diminuendo homozygotes. Grxcr1 is not significantly affected in diminuendo homozygotes, and Otof is upregulated in diminuendo homozygotes and downregulated in Ptprq-CAT-KO homozygotes (Lewis et al., 2009;Kuhn et al., 2011). In addition, many of the other major molecular changes observed in diminuendo homozygotes are not reproduced in Ptprq-CAT-KO homozygotes, for example the downregulation of Ocm. It is possible that Hsd17b7 is downregulated in both mutants because of a specific effect for which lack of Ptprq is responsible. Hsd17b7 (hydroxysteroid (17-beta) dehydrogenase 7) is involved in cholesterol biosynthesis , a heterozygous and (C) a homozygous mutant P6 apical-coil diminuendo OHC by applying sinusoidal force stimuli of 50 Hz to the hair bundles at À122 mV and + 98 mV. The driver voltage (DV) signal of AE 40 V to the fluid jet is shown above the traces (negative deflections of the DV are inhibitory). The holding potential was À82 mV. Extracellular Ca 2+ concentration was 1.3 mM. The arrows and arrowheads indicate the closure of the transducer currents (i.e. resting current) elicited during inhibitory bundle displacements at hyperpolarized and depolarized membrane potentials, respectively. Note that the resting current increases with membrane depolarization. Dashed lines indicate the holding current, which is the current at the holding membrane potential. (D and E) Comparison of transducer currents recorded from (D) a control and (E) a homozygous mutant P6 diminuendo OHC in the presence of 1.3 mM Ca 2+ (black/red) and endolymphatic-like Ca 2+ concentration (0.04 mM; grey/pink lines) at À122 mV. (F and G) Resting current (F) and (G) peak transducer current at a membrane potential of À122 mV recorded in OHCs from the three genotypes in the presence of 1.3 mM (black/blue/red) and 0.04 mM (grey/pale blue/pink) extracellular Ca 2+ . and is required for neuronal survival early in development; mice homozygous for a disrupted allele of Hsd17b7 die at embryonic day 10.5, with increased apoptosis observed in the neural tube, dorsal root ganglia and trigeminal nerve. They also display cardiovascular defects (Shehu et al., 2008;Jokela et al., 2010). Mice with an ENU-induced mutation in Hsd17b7 display disrupted Hedgehog signalling. The mutation is hypomorphic; it leads to most of the Hsd17b7 transcripts being truncated due to a mis-splice, but low levels of wildtype transcript are still present (Stottmann et al., 2011). Hedgehog signalling is required for inner ear development (Liu et al., 2002;Riccomagno et al., 2002), but any more direct link between Ptprq and Hsd17b7 remains unclear. The changes in expression of Grxcr1 in Ptprq-CAT-KO homozygotes, and of Otof in both diminuendo and Ptprq-CAT-KO homozygotes, may be only indicative of a dysfunctional hair cell.
In conclusion, many of the morphological and electrophysiological features of diminuendo heterozygotes are reproduced in Ptprq-CAT-KO homozygotes, and in some features the transcriptome of Ptprq-CAT-KO homozygotes resembles that of diminuendo homozygotes. It is particularly interesting that in the P4 diminuendo cochlea there is very little difference in expression of Ptprq between heterozygotes and homozygotes; Ptprq is downregulated to~62% of the wildtype level in heterozygotes and $ 56% of the wildtype level in homozygotes (Lewis et al., 2009), which would suggest that diminuendo heterozygotes and homozygotes should both resemble Ptprq heterozygotes, which is not the case; Ptprq-CAT-KO heterozygotes have no hair cell defects (Goodyear et al., 2003). The reduction in Ptprq seen in diminuendo mice appears to be a major contributor to the phenotype, particularly the apparent delay and eventual stalling of hair cell development observed by scanning electron microscopy and the delayed development of the MET current, but it does not account for all of the observations. The more severe phenotypes in the diminuendo homozygous mice indicate that other genes besides Ptprq are contributing to the diminuendo homozygous phenotype, and these genes may not be affected in diminuendo heterozygous or Ptprq-null mice.
A B C D Fig. 11. The development of transducer currents in diminuendo mutant OHCs is prematurely arrested. (A) Driver voltages to the fluid jet (top panels) and transducer currents recorded at À84 mV from a control, a heterozygous and a homozygous diminuendo mutant P6 OHC. Positive driver voltages (excitatory direction) elicited inward (negative) transducer currents that declined or adapted over time only in control OHCs (arrow). A small transducer current was present at rest (before t = 0) and inhibitory bundle displacement turned this off. Upon termination of the inhibitory stimulus, the transducer current in control OHCs showed evidence of rebound adaptation (arrowhead). (B) Example of transducer currents in developing control OHCs (P2, P3 and P4). Note that both signs of current adaptation (arrow and arrowhead) are first detected at P4. (C) Maximal transducer current recorded at different postnatal ages in control OHCs (black) and that of the heterozygous (blue) and homozygous (red) diminuendo mutant cells at P6. The number of OHCs tested is shown above each data point. (D) Extent of adaptation for small bundle deflection towards the excitatory directions (see arrows in panel A, left panel) of the transducer current at different postnatal ages in control OHCs (black) and that of the heterozygous (blue) and homozygous (red) diminuendo mutant cells at P6. The number of OHCs showing MET current adaptation over the total number of cells tested is shown above each data point. Note that (C) the size and (D) the extent of adaptation recorded in P6 homozygous diminuendo OHCs is similar to that measured in P1-2 control cells.