Defective Gpsm2/Gαi3 signalling disrupts stereocilia development and growth cone actin dynamics in Chudley-McCullough syndrome

Abstract

Mutations in GPSM2 cause Chudley-McCullough syndrome (CMCS), an autosomal recessive neurological disorder characterized by early-onset sensorineural deafness and brain anomalies. Here, we show that mutation of the mouse orthologue of GPSM2 affects actin-rich stereocilia elongation in auditory and vestibular hair cells, causing deafness and balance defects. The G-protein subunit Gα_i3, a well-documented partner of Gpsm2, participates in the elongation process, and its absence also causes hearing deficits. We show that Gpsm2 defines an ∼200 nm nanodomain at the tips of stereocilia and this localization requires the presence of Gα_i3, myosin 15 and whirlin. Using single-molecule tracking, we report that loss of Gpsm2 leads to decreased outgrowth and a disruption of actin dynamics in neuronal growth cones. Our results elucidate the aetiology of CMCS and highlight a new molecular role for Gpsm2/Gα_i3 in the regulation of actin dynamics in epithelial and neuronal tissues.

Figure 1. Gpsm2 and Gα_i3 are dynamically expressed at the tips of stereocilia.

(a–c) Surface view of whole mounts of rat cochlear sensory epithelium at E17.5 (a) and P7 (b,c) illustrating Gpsm2 (a,b, green) and Gα_i3 (c, green) labelling in the actin-rich hair bundle labelled by phalloidin (Ph, magenta). (a) At E17.5, Gpsm2 localizes at the tips of stereocilia at the onset of hair bundle growth (yellow arrows), but also in an asymmetrical crescent in a distal region of the apical membrane of HCs (green asterisks). (b,c) By P7, both proteins accumulate at tips of stereocilia (green), strongly in IHC (yellow arrows) and more weakly in OHC (green arrows). Arrow: inner hair cell (IHC). Bracket: outer hair cell. Scale bars (a–c), 4 μm. (d) Gpsm2 (green) is localized at tips of P8 stereocilia of rat vestibular HC bundles. Ph: phalloidin. Scale bar, 2 μm. (e) At P15, confocal imaging reveals the accumulation of Gα_i3 protein at the tip of individual stereocilium. Scale bar 2 μm. (f–h) At P15, STED super-resolution imaging of the Gpsm2-expression domain at an individual stereocilium tip (f). Gpsm2 accumulated at tips of IHC stereocilia (green), above the F-actin labelling (magenta). (f, right panel, g) Acquisition of single plane images in two perpendicular axis as illustrated on the schematic in h reveals the cap-like structure of the Gpsm2 nanodomain. Scale bars, 2 μm. (i) Triple STED labelling reveals two mostly overlapping nanodomains at stereocilia tips with Gpsm2 (green) and Eps8 (magenta) above the F-actin signal (Ph, white). Left images: individual channels for Gpsm2 (top) and Eps8 (middle). The bottom image illustrates the phalloidin channel (grey) with two-colour binary representation of Gpsm2 (green) and Eps8 (magenta), with the overlapping domain (plain white). Scale bar, 2 μm. (j) Isolated long (j) and short (j, inset) stereocilia illustrate the accumulation of the two proteins in the long stereocilium only. Scale bar, 1 μm. (k) Intensity profiles of phalloidin, Gpsm2 and Eps8 from (j) (orange line across tip domain). Immunostainings repeated more than six times.

Figure 2. Gpsm2 and Gnai3 mutations inhibit stereocilia elongation.

(a–c) SEM of cochlear inner hair cells (IHC) from controls (a), Gpsm2 (b) or Gnai3 cKOs (c) in P5 mouse. The kinocilium is indicated with yellow stars. In Gpsm2 (b) and Gnai3 (c) cKOs, the staircase pattern is almost absent with stereocilia of similar length and width. Lateral links between adjacent stereocilia are preserved (magenta arrows, insets). Scale bars, 1 μm. (d,e) Quantifications at P5 show reduced length of IHC tallest stereocilia and supernumerary stereocilia in both Gpsm2 and Gnai3 cKOs. Quantifications are presented as whisker box plots (min/max), *P<0.05; ***P<0.001 with one-way ANOVA (post hoc Bonferonni’s test). (f–h) SEM of basal cochlear IHC from controls (f), Gpsm2 (g) or Gnai3 cKOs (h) in P21 mouse. In Gpsm2 cKOs, stereocilia are short, typically with more than four rows and severely reduced staircase pattern. A similar, but weaker phenotype was observed in Gnai3 cKOs, with occasional longer stereocilia in a bundle with overall shorter stereocilia (h, magenta brackets). Scale bars, 1 μm. (i,j) Quantifications at P21 are consistent with the above illustrations with a severely reduced length of IHC stereocilia and supernumerary stereocilia in Gpsm2 cKOs, and a similar but milder phenotype in Gnai3 cKOs. Quantifications are presented as whisker box plots (min/max), *P<0.05; ***P<0.001 with one-way ANOVA (post hoc Bonferonni’s test). (k) 3D rendering of the surface of a control (top) and a Pou4f3-Gpsm2 mutant (bottom) at P8. Mutants exhibit supernumerary rows of abnormally short stereocilia, sometimes split or fragmented (yellow asterisks). Scale bars, 4 μm. (l) SEM of postnatal cochlear explants treated or not for 8 days (DIV8) with 100 ng ml⁻¹ PTX. In the PTX-treated samples (bottom panels), the stereocilia are shorter than in controls (upper pannels), with similar widths. Scale bars, 1 μm. (m) Quantification of the length of the tallest stereocilia cultures treated with PTX (blue line=mean). n=34 (control) and 76 stereocilia (PTX). Cultures repeated three times. ***P<0.0001 with unpaired Student’s t-test.

Figure 3. Gpsm2 and Gnai3 mutations affect cochlear and vestibular function.

(a,b) Hearing tests on 4-week-old mice reveal severe threshold increases in Gpsm2 cKO (a), compared with high frequency loss only in Gnai3 cKOs (b). Arrow in a indicates ABR thresholds exceeding the maximum testable intensity. Mean±s.d. click-evoked ABR (click-ABR) and tone-burst-evoked ABR (f-ABR). Mean threshold values (in dB SPL) of click-ABR of control mice are shown above corresponding bars. ***P<0.001 (Grey shaded area: P<0.05) by two-way ANOVA (post hoc Bonferonni's multi comparisons test). f-ABR: Control (Ctr) and cKO, n=8 ears from eight mice click-ABR: Control and cKOs: n=16 ears from eight mice. (c,d) Left panel: Gpsm2 cKOs (green traces) display increased circling activity in a representative open-field during the first 30 s and at the end of the track (10 min) compared with control littermates, whereas Gnai3 cKOs (magenta traces) are unaffected (right panel). (e,f) Gpsm2 cKOs mice cover more distance and rotate more than Gnai3 cKOs mice (each circle is an individual mouse). Open white circles are controls. (g) Top: heat map of force swim test occupancy for control and Gpsm2 cKOs. Bottom: during the 2 min test, Gpsm2 cKO mice showed less immobility and more body axis rotation compared with controls. (h) SEM of the surface view of the macula of the utricle of P11 mice in control (left) and Gpsm2 cKO (right). Stereocilia elongation in the mutant is dramatically reduced compared with control. Scale bars, 1 μm.

Figure 4. Gpsm2 and Gα_i3 localization at stereocilia tips is dependent on myosin 15 and whirlin.

(a–d) Immunocytochemistry for Gpsm2 (a,b) or Gα_i3 (c,d) and staining with phalloidin (Ph, magenta) shows protein localization at the tips of stereocilia in IHCs from shaker 2 (sh2/+, a,c) and whirler (wi/+, b,d) control mice at P8. Homozygous mutations (sh2/sh2 and wi/wi) lead to a loss of Gpsm2 and Gα_i3 stereocilia tip staining. The immunostainings were repeated four times. (e,f) Immunocytochemistry for myosin 15 (green) and staining with phalloidin (Ph, magenta) shows protein localization at tips of stereocilia in IHCs from controls of Gpsm2 (e) and Gnai3 (f) cKO P8 mice. Myosin 15 protein is still present at the tip of shortened stereocilia of both cKOs, though in reduced amounts. (g,h) Immunocytochemistry for whirlin (green) and staining with phalloidin (Ph, magenta) shows protein localization at tips of stereocilia in IHCs from controls of Gpsm2 (g) and Gnai3 (h) P8 mutant mice. Gpsm2 and Gnai3 mutation leads to a strong reduction of whirlin stereocilia staining. Scale bars, 4 μm. The immunostainings (e–h) were repeated six times and imaged blindly.

Figure 5. Gpsm2 and Gα_i3 depend upon myosin 15 to reach stereocilia tips but not for their targeting at HC apical membrane.

(a,b) Immunocytochemistry for Gpsm2 (a) or Gα_i3 (b) shows both protein localization at the tips of stereocilia (green asterisks), and at the apical membrane of HC as a crescent-shape (yellow dashed-line) from controls of shaker 2 (sh2/+). Homozygous mutation (sh2/sh2) leads to a loss of Gpsm2 and Gα_i3 stereocilia tip staining, whereas the apical crescent is maintained. (c) Immunocytochemistry for Gpsm2 (green) shows protein localization at the tips of stereocilia in IHCs from controls of Gnai3, but absent from Gnai3 cKO. (d) Reciprocally, immunocytochemistry for Gα_i3 (green) shows protein localization at the tips of stereocilia in IHCs from controls of Gpsm2, but absent from Gpsm2 cKO. Scale bars, 4 μm. The immunostainings were repeated four times.

Figure 6. New protein complex between Gpsm2 and whirlin.

(a–c) COS-7 cells transiently transfected with GFP–Myo15, DsRed-whirlin, myc-Gpsm2^FL, untagged Gα_i3 or DsRed as indicated. (a) Co-expression of myosin 15 and whirlin leads to the formation of actin-rich filopodial structures with Gpsm2 accumulating at the tip with the two proteins. Absence of whirlin (DsRed only) markedly reduces the colocalization at the tips. (b) Gα_i3 also accumulates at the tip of filopodia with myosin 15 and whirlin. Transfections were repeated more than three times. Scale bars, 4 μm. (c) Percentage of filopodia tips (±s.e.m.) displaying colocalization of Gpsm2, myosin 15 and whirlin in the two contexts. n=number of filopodia tips from three independent experiments. ***P<0.001 with Mann–Whitney test. (d) Schematic representation of Gpsm2^FL and whirlin proteins. Four variants identified in CMCS patients, including Gpsm2^R318RfsX8 are indicated. (e) Whirlin co-immunoprecipitates with myc-Gpsm2^FL but not with non-immune IgG. (f) The interaction is maintained, though reduced, with the Gpsm2^R318RfsX8 (Gpsm2^R318R) variant. Representative images of n>3 independent experiments. (g) A GST-pull-down assay indicates that the unstructured region of whirlin between aa 672 and aa 810 binds to Gpsm2^R318RfsX8. Representative image of n=3 independent experiments. (h,i) Increasing amounts of myc-Gpsm2^FL-encoding cDNA over constant amounts of whirlin leads to a net increase in whirlin expression levels, whereas increasing doses of whirlin had no significant effect on Gpsm2 levels. Five independent experiments presented as whisker box plots (min/max) combined with dot plot (blue bars representing the mean values) ***P<0.001, **P<0.01, *P<0.05 (for h,i) with one simple t-test or one-way ANOVA (post-hoc Bonferonni’s test). NS, not significant. (j) Gpsm2^FL significantly increased the length of filopodia generated by eGFP–myosin 15 and DsRed-whirlin (control) in COS-7 cells. Constructs bearing CMCS mutations led to shorter filopodia, compared with control or to Gpsm2^FL. Data are presented as whisker box plots (min/max) from three independent experiments (n=220, 207, 345, 228 and 188 filopodia, in the order indicated on the histogram’s x axis) (blue bars represent the mean values). ***P<0.001, **P<0.01 with Kruskal–Wallis test (post-hoc Dunn's Multiple Comparison Test). NS, not significant.

Figure 7. Gpsm2 mutation leads to hypoplasia of the corpus callosum and affects growth cone outgrowth.

(a) Hematoxylin staining of coronal sections from P6 Gpsm2^Emx1 control (upper panel) and cKO brains (lower panel) at a caudal levels. Note the abscence of corpus callossum (CC) in the cKO mouse (inset). Scale bar, 1 mm, n=4 independent experiments. (b,c) In cultured hippocampal neurons at DIV2, N-cadherin-dependent outgrowth was reduced in Gpsm2 cKOs, but not in Gnai3 cKOs (b). The reduction in outgrowth was maintained on a laminin substrate in Gpsm2 cKOs (c). Data from three to six independent experiments are presented as whisker box plots (min/max) (n=number of growth cones). ***P<0.001 with unpaired Student’s t-test. (d) Images from three time points (0, 15 and 30 min) of a time-lapse movie from 2 DIV control and Gpsm2 cKO hippocampal neurons plated on N-cadherin-coated glass and showing the difference in distance covered (dotted lines and double-headed arrows). Scale bar=10 μm. (e) Quantification of the average number of pauses (±s.e.m.) during a 30 min period of growth cones from control (n=31 neurons) compared with Gpsm2 cKO (n=51 neurons), from four independent experiments. The Gpsm2 cKO growth cones pause more than controls. ***P<0.001, *P<0.05 with unpaired Student’s t-test. (f) Cumulated distance covered by control and Gpsm2 cKO growth cones over a 30 min period. Some pauses are indicated with blue arrows.

Figure 8. Gpsm2 co-immunoprecipitates whirlin in brain lysates and increases actin polymerization.

(a,b) Representative DIV2 growth cones from a control (a) and a Gpsm2 cKO (b) with the outlined peripheral region where sptPALM data were collected (yellow) and the corresponding individual actin-mEOS2 trajectories (3 min recording). Note the overall more-confined behaviour of the actin molecules in the cKO. Scale bars, 5 μm. (c) Distribution of the actin-mEOS2 trajectory length shows a median trajectory length of nine frames for both genotypes (blue line). (d) Representative mean squared displacement (MSD) over time for each of the three types of actin-mEOS2 behaviours with their corresponding α values. The plain curves represent fits to the function MSD=4Dt^α, where D is a diffusion coefficient and α is a power law exponent. (e) Repartition of the α values of actin-mEOS2 molecules in control and Gpsm2 cKO neurons on N-cadherin substrate. Values from 11 (control) and 12 (cKO) growth cones from three separate experiments (±s.e.m., n=1344 trajectories for control and 1121 for mutant). ***P<0.001, **P<0.01, *P<0.05 with an unpaired Student’s t-test or Mann Whitney test when a normality test failed. NS, not significant. (f) Immunoprecipitation of Gpsm2 together with whirlin using anti-Gpsm2 serum. Membranes were immunoblotted with the antibodies indicated on the left. The experiment was replicated twice. (g) DIV3 hippocampal neurons electroporated with eGFP–myosin 15, DsRed-whirlin and myc-Gpsm2 show enrichment of all three proteins at the tips of filopodia (arrowheads). Filopodia are outlined with dotted lines. The LUT was modified (left, Orange hot) to better visualise the accumulation. n=5 independent experiments. Scale bar, 4 μm. (h) Actin assay shows that the combination of Gpsm2^FL and Gα_i3 expression shift the F/G-actin ratio, whereas the Gpsm2^R318RfsX8 mutation decreases this shift. Dot plot from five biological repeats (black bar represent mean values). **P<0.01 with Unpaired Student’s t-test, **P<0.01 with one sample t-test. (i) Actin assays on cultures of neurons show a shift in the F/G-actin ratio, suggesting a decrease in actin polymerization in the Gpsm2 cKOs. Dot plot from five biological repeats (black bar represent mean values). **P<0.01 with one sample t-test.

Figure 9. Gpsm2 mutations affect stereocilia elongation and neuronal outgrowth by regulating actin dynamics at tip complexes.

Mechanistic model for Gpsm2-dependent stereocilia elongation and neuronal outgrowth. Gpsm2 accumulates at the tip complex of both structures via the myosin 15 motor protein in stereocilia and a comparable motor protein in filopodia. Gpsm2-dependent macromolecular protein complexes modulate actin dynamics at the tip of stereocilia (growing end) or the leading edge of the growth cone, participating respectively in the elongation and motility of the two structures. See text for details.