Wbp2 is required for normal glutamatergic synapses in the cochlea and is crucial for hearing

Abstract

WBP2 encodes the WW domain-binding protein 2 that acts as a transcriptional coactivator for estrogen receptor α (ESR1) and progesterone receptor (PGR). We reported that the loss of Wbp2 expression leads to progressive high-frequency hearing loss in mouse, as well as in two deaf children, each carrying two different variants in the WBP2 gene. The earliest abnormality we detect in Wbp2-deficient mice is a primary defect at inner hair cell afferent synapses. This study defines a new gene involved in the molecular pathway linking hearing impairment to hormonal signalling and provides new therapeutic targets.

Keywords: glutamate excitotoxicity; hearing impairment; hormonal signalling; ribbon synapses; transcriptional coactivator.

Figure 1. Wbp2 mutation and Wbp2 expression in the cochlea

1. A

   Diagram showing the design of the mutated Wbp2 allele. A promoterless cassette including LacZ and neo genes was inserted in the second intron of the Wbp2 gene flanked by FRT sites (green triangles). LoxP sites (red triangles) flank the critical exon (exon 2) of the Wbp2 gene (exons in yellow).

2. B

   Western blot showing no detectable Wbp2 protein in 4‐week‐old mutant brain compared to wt littermate controls; 5 μg of the protein lysate was subjected to 10% SDS–PAGE. β‐tubulin was used as a loading control. Wbp2 l and Wbp2 s refer to the long and short isoforms, respectively.

3. C

   Quantitative real‐time PCR showing severe knockdown of Wbp2 transcription in 4‐week‐old mutant (n = 3) inner ears and eyes, compared to wt littermate controls (n = 3). Heterozygotes show intermediate levels. Hprt was used as a control and levels are normalised to wt levels. Data plotted as mean ± SD. Two‐tailed t‐test: Wbp2 ear: het *P = 0.03, hom ***P = 0.000000033; Wbp2 eye: het *P = 0.01, hom ***P = 5.35537E‐13.

4. D–F

   X‐gal staining of Wbp2 hets at P14 showing Wbp2 expression (blue) in all the main cochlear structures: the stria vascularis (black arrowhead in D), spiral prominence (empty arrowhead in D), Reissner's membrane (arrow in D), strong expression in the spiral ganglion cells (arrowheads in E) and in IHCs and OHCs in the organ of Corti (arrowheads in F). Scale bars: (D), 50 μm; (E, F), 20 μm. ihc: inner hair cells; ohc: outer hair cells. No X‐gal staining is observed in wt controls (not shown). The X‐gal reaction is always cytoplasmic.

Source data are available online for this figure.

Figure 2. Auditory responses of Wbp2‐deficient mice

1. A–E

   Mean ABR thresholds (± SD) for clicks and tone pips are plotted for wt (green), het (blue) and hom (red) mice at ages (A) P14 (wt, n = 3; het, n = 8; hom, n = 6); (B) 4 weeks (wt, n = 38; het, n = 26; hom, n = 37); (C) 14 weeks (wt, n = 10; hom, n = 14); (D) 28 weeks (wt, n = 15; het, n = 5; hom, n = 25); (E) 44 weeks (wt, n = 9; het, n = 2; hom, n = 11). Grey symbols and lines indicate thresholds of individual mutants. In (B), open symbols represent thresholds under urethane anaesthesia (see Materials and Methods), showing no difference compared with ketamine/xylazine used for all other thresholds.

2. F

   Mean thresholds for mutants aged 2 weeks (yellow), 4 weeks (purple), 14 weeks (cyan), 28 weeks (grey) and 44 weeks (black).

3. G

   Averaged click‐evoked ABR waveforms from 4‐week‐old wt (n = 23, green) and mutants (n = 34, red), at 50‐dB sensation level (SL) (left panel). SP and ABR wave 1 (W1) are indicated by grey areas. Expanded averaged SP and ABR W1 waveform patterns for 10‐ to 60‐dB SL in 10‐dB increments are plotted in green (wt; middle panel) and red (mutants; right panel), to illustrate the growth of SP and W1 with stimulus level.

4. H, I

   Mean 2f1‐f2 DPOAE thresholds (± SD) are plotted for wt (green), heterozygote (blue) and homozygous (red) mice aged 4 weeks (H: wt, n = 5; hom, n = 5) or 21 weeks (I: het, n = 3; hom, n = 5), as a function of f2 frequency.

Figure EV1. Supplementary physiology

1. A–C

   SP amplitude (thick lines) and latency (thin lines) are plotted as a function of dB SL, for wt (green, n = 23) and mutant (red, n = 34) responses to clicks (A), 12‐kHz tones (B) and 24‐kHz tones (C). Values were pooled across stimulus level for statistical analysis. For each stimulus, SP amplitude was significantly reduced in mutants. Click: mean wt SP = 0.61 μV, mean mutant SP = 0.28 μV, t‐test P = 0.0000323. 12 kHz: mean wt SP = 0.47 μV, mean mutant SP = 0.25 μV, t‐test P = 0.00257. 24 kHz: median wt SP = 0.26 μV, median mutant SP = 0.13 μV, Mann–Whitney rank‐sum test U = 6.000 P = 0.003. For each stimulus, SP latency was not changed. Click: mean wt = 1.21 ms, mean mutant = 1.26 ms, t‐test P = 0.208. 12 kHz: mean wt = 1.61 ms, mean mutant = 1.68 ms, t‐test P = 0.154. 24 kHz; median wt = 1.55 ms, median mutant = 1.54 ms, Mann–Whitney rank‐sum test U = 30.500 P = 0.401.

2. D–F

   Mean ABR wave 1 amplitude (± SD) is plotted as a function for dB SL for wt (green, n = 23) and mutant (red, n = 34) responses to clicks (D), 12‐kHz tones (E) and 24‐kHz tones (F). For each stimulus, W1 amplitude was reduced in mutants; Kruskal–Wallis one‐way ANOVA on ranks; click, H = 591.579, P < 0.001; 12 kHz, H = 631.535, P < 0.001; 24 kHz, H = 524.426, P < 0.001.

3. G–I

   Mean ABR wave 1 latency (±SD) is plotted as a function for dB SL for wt (green, n = 23) and mutant (red, n = 34) responses to clicks (G), 12‐kHz tones (H) and 24‐kHz tones (I). For each stimulus, positive peak P1 latency was increased in mutants; Kruskal–Wallis one‐way ANOVA on ranks; click, H = 606.088, P < 0.001; 12 kHz, H = 612.422, P < 0.001; 24 kHz, H = 497.363, P < 0.001.

4. J

   Endocochlear potential was recorded in Hom and Wt, and no significant difference in the values was observed (t‐test, wt controls: 119.5 ± 5.2 mV, n = 5, range 113.0–124.7; mutants: 116.8 ± 4.6 mV, n = 6, range 109.7–121.7). P = 0.39526.

Figure EV2. Normal appearance of hair cell bundles and normal pre‐synaptic function in Wbp2‐deficient mice

1. SEM showing normal appearance of IHC and OHC hair bundles in 6‐week‐old Wbp2‐deficient mice compared to wt controls, illustrated for the 24‐kHz (40% of the length of the cochlear duct from the base) and 9‐kHz (80% position) best frequency regions. There was no sign of excess degeneration up to 30 weeks in mutants compared to littermate controls. Scale bars: upper row, 20 μm; middle row, 3 μm; bottom row, 1 μm.

2. Saturating mechanoelectrical transducer (MET) currents recorded from a P7 control and a Wbp2‐mutant IHC by applying voltage steps from −121 mV to +99 mV in 20‐mV increments (holding potential −81 mV). For clarity, only two voltage steps are shown. During the voltage steps, hair bundles were displaced by applying 50‐Hz sinusoidal force stimuli (the driver voltage, DV, to the fluid jet is shown above the traces). Negative deflections of the DV are inhibitory. The arrows indicate the closure of the transducer channels, that is disappearance of the resting current, during inhibitory bundle displacements. Dashed lines indicate the holding current, which is the current at the holding potential of −81 mV.

3. Peak‐to‐peak current–voltage curves obtained from 11 controls and 5 Wbp2‐mutant IHCs (P7–P8).

4. I_{C a} and ΔC _m responses from adult (P19–P33) control and Wbp2‐mutant IHCs from high‐frequency region. Recordings were obtained in response to 50‐ms voltage steps, in 10‐mV increments, from −81 mV. For clarity, only average maximal responses are shown (control: n = 7; Wbp2 mutant: n = 8).

5. Average peak I_{C a}‐voltage (left axis) and ΔC _m‐voltage (right axis) curves from control and Wbp2‐mutant IHCs (control: n = 7; Wbp2 mutant: n = 8).

6. Average ΔC _m from 10 control and 11 Wbp2‐mutant IHCs in response to voltage steps from 2 ms to 100 ms (to around −11 mV) showing mainly the RRP and the initial recruitment of the SRP for the 100‐ms step.

7. Average ΔC _m from 12 control and 11 Wbp2‐mutant IHCs in response to voltage steps from 100 ms to 2 s (to around −11 mV) showing the SRP.

Data information: Data are shown as mean ± SD.

Figure EV3. Histology of organ of Corti

Semi‐thin sections stained with toluidine blue show no cochlear abnormalities and no obvious loss of spiral ganglion cells in Wbp2‐deficient mice compared to controls at 4 weeks. Scale bar: 20 μm. SG: spiral ganglion cells.

Figure 3. Analysis of the human variants in WBP2

1. Pure tone audiogram recorded from one of the probands, showing severe to profound bilateral asymmetric hearing loss.

2. Capillary sequence traces from fathers, mothers and probands, showing the heterozygosity of the parents for the two separate variants and the compound heterozygosity for each proband.

3. ConSeq (http://conseq.tau.ac.il/) analysis of the residues; the locations of the three human variants (p.Ala160Thr, p.Met163Leu and p.Ala224Val; boxed in black) are average to highly conserved. “b” and “e” indicate buried and exposed residues (according to the neural network algorithm), and “f” and “s” indicate predicted functional and predicted structural residues. The GRAM and WW binding domains (WW1 and WW2) are marked by blue and green boxes. The GRAM domain is thought to be an intracellular protein‐ or lipid‐binding signalling domain and may play an important role in membrane‐associated processes.

4. Alignment of the protein sequence from a range of vertebrates. The human variants are boxed in black.

Figure 4. Analysis of the Wbp2 mouse isoforms in the brain and in the cochlea

1. Splice forms of WBP2, numbered according to the Ensembl numbering scheme. The GRAM domain is marked in blue, the WW domains in green and the locations of the variants in black (the p.Ala160Thr and p.Met163Leu variants are too close to show separately in this view).

2. Agarose gel trace of cDNA obtained from mouse wt P28 inner ear (IE), P28 brain, adult and P4 organ of Corti (OC). Results show the expression of two Wbp2 isoforms in the brain at P28 and very faint band for the short isoform together with a strong band for the long isoform in the inner ear at P28. If we look at just the organ of Corti (adult and P4), we observe a strong band for the long isoform in the P4 OC and an almost undetectable band for the short isoform, which was not even picked up by sequencing (see C). Wbp2 l: long isoform (550 bp); Wbp2 s: short isoform (480 bp).

3. Cartoon illustrating the results from the sequencing of mouse wt cDNA performed at P4 and P28. While in the brain we detect both Wbp2 isoforms and in the organ of Corti only the long isoform, in the whole inner ear sample we detect the presence of the long isoform with a small band for the short one.

4. Western blot showing the predominant presence of the Wbp2 long isoform in the cochlea at P28, with a weak trace of the short isoform showing up only when a higher concentration of protein lysate (20 μg) is loaded. Both isoforms are absent in the Wbp2‐deficient mouse. Gapdh was used as a loading control. Wbp2 l: long isoform; Wbp2 s: short isoform.

Source data are available online for this figure.

Figure 5. Afferent innervation in Wbp2‐deficient mice

1. At P14, afferent terminals below IHCs are slightly swollen in the mutants (neurofilament labelling in green, arrowheads to compare; CtBP2 labels ribbons and IHC nuclei in red). Scale bars, 10 μm. At P28, neurofilament/CtBP2 labelling in the organ of Corti of 4‐week‐old Wbp2‐deficient mice and littermate controls shows severe swelling of IHC afferent terminals in the mutants, especially in the 24‐kHz region (yellow arrowheads). The pre‐synaptic ribbons do not look as well aligned to the terminals in the mutants (white arrows). At this stage, we also observe swelling of OHC afferent terminals in the 24‐kHz region (empty arrowheads). Scale bars, 5 μm. ihc: IHC nucleus; p: pillar side; m: modiolar side.

2. Counts of pre‐synaptic ribbons per IHC in the 8‐, 18‐ and 24‐kHz regions, showing no difference between mutants and controls at P28.

3. TEM of the organ of Corti performed at P28 showing swollen afferent terminals below inner and outer hair cells (arrowheads for comparisons between mutants and controls). Scale bars, 5 μm.

4. Neurofilament/CtBP2 labelling shows swollen and retracting terminals (white arrows), especially in the 24‐kHz regions in the mutants at 8 weeks (arrows). Scale bars, 10 μm.

5. Counts of pre‐synaptic ribbons in the 8‐, 18‐ and 24‐kHz regions, showing no difference in their number per IHC in the mutants compared to littermate controls at 8 weeks.

Data information: All data are shown as mean ± SD and statistically analysed by two‐tailed Student's t‐test. n = 35 hair cells. Synaptic count at 4 weeks: wild type: 24 kHz 16.9 ± 2.4, 16 kHz 13.67 ± 2, 9 kHz 13.0 ± 1.8; mutants: 24 kHz 18.17 ± 1.18 (P = 0.27), 16 kHz 15.89 ± 1.8 (P = 0.26), 9 kHz 15 ± 2 (P = 0.07). Synaptic count at 8 weeks: wild type: 24 kHz: 15.35 ± 3.75, 18 kHz 17.73 ± 1.79, 8 kHz: 11.25 ± 0.92; mutants: 24 kHz: 15.78 ± 3.46 (P = 0.9), 18 kHz 15.78 ± 0.17 (P = 0.26), 8 kHz: 12.3 ± 2.26 (P = 0.6) Source data are available online for this figure.

Figure 6. GluR2/3 expression and synaptic defects in Wbp2‐deficient mice

1. A

   We arrayed synapse images aligned by ascending size of the post‐synaptic site after GluR2/3 and CtBP2 labelling. This is a composite image made of several synapses taken from a single IHC from a single wt and a single hom, representing double labelling experiments performed on 3 mutants and 3 controls. Synapses in the mutants show abnormal morphology and smaller green patches, suggesting reduced expression of the GluR2/3 AMPA receptor subunits. Scale bar (shown on the bottom right), 1 μm.

2. B–G

   TEM images of IHC ribbon synapses of wt (B, arrowheads for synaptic vesicles) and homs (C–G) at 4 weeks of age, showing a representative array of synaptic phenotypes in the 24‐kHz cochlear region of mutants. While in (C) and (D) the ribbons look slightly abnormal in size with misplaced synaptic vesicles (arrowheads in C), we also observe orphan post‐synaptic densities surrounded by floating synaptic vesicles (arrowheads in E); ribbons with synaptic membranes (arrow in F) that have detached from the IHC membrane (the arrowhead labels the original position of the synapse before detachment) and are floating in the swollen nerve terminal (F); ribbons (arrow in G) that are detached from their densities (arrowhead in G). Scale bars: (B–E), 200 nm; (F, G), 500 nm. nt: nerve terminal; psd: post‐synaptic density; snt: swollen nerve terminal.

Source data are available online for this figure.

Figure 7. The Wbp2 molecular pathway

1. Diagram showing the Wbp2 molecular pathway, including its downstream targets and their functional relationship. The blue arrows, light blue lines and green lines link data from the literature (in vivo and in vitro); the orange squares and red arrows indicate up‐ or down‐regulation shown in our experimental observations, as reported in this study.

2. Quantitative real‐time PCR showing reduced mRNA levels for Esr1, Esr2 and Pgr and up‐regulation of Shank3 and Psd‐95 in cochleae of 4‐week‐old Wbp2‐deficient mice compared to littermate controls (n = 3 for each genotype). Hprt is used as a relative control. *P = 0.03 for Psd‐95; **P = 0.007 for Shank3; *P = 0.03 for Esr2; **P = 0.0016 for Esr1; *P = 0.037 for Pgr.

3. Synapses from one mutant and one control IHC at 4 weeks of age after Psd‐95 and CtBP2 labelling, showing stronger Psd‐95 expression in the mutants compared to controls, representing double labelling experiments performed on 3 mutants and 3 controls. Scale bar, 10 μm.

4. Quantification of Psd95 fluorescence in IHC synapses, representing expression in the apical (9‐kHz best frequency region of the cochlea) and basal (24‐kHz best frequency region of the cochlea) regions at 4 weeks of age. Data from 2 wt and 2 homs were analysed (16 synapses per cochlear region per mouse). AU, arbitrary units. Wt: 24 kHz 22.41 ± 10.70, 9 kHz 11.045 ± 2.128; mutants: 24 kHz 66.73 ± 13.70, P = 0.069; 9 kHz: 26.02 ± 5.79 P = 0.075.

Data information: Data are shown as mean ± SD and were statistically analysed by two‐tailed Student's t‐test.Source data are available online for this figure.