Figures and data
Summary of GNAI-related functions proposed previously.
A) Apical HC differentiation from symmetry breaking to hair bundle development. The distribution of the GPSM2-GNAI complex at the bare zone and at stereocilia tips is indicated in orange. B) Defects observed with pertussis toxin (ptx) or when inactivating GNAI proteins. Defective off-center migration of the basal body and inverted OHC1-2s were only observed with ptx, respectively in cochlear explants (in vitro) and by expressing the ptx catalytic subunit (ptxA) in vivo. Mouse knock-out (KO)s of Gnai genes were to date only reported to affect hair bundle morphogenesis. Known GNAI regulators that produce similar defects when inactivated are indicated on top for each type of defect. DKO, double KO.
Individual GNAI proteins make different contributions to hair bundle development.
A-B) SEM images of representative OHCs (A) and IHCs (B) in 3-week old animals at the cochlear first turn. Gnai1neo, Gnai2del, and Gnai1neo; Gnai2del mutants show apparently normal hair bundle morphology in both HC types. In contrast, Gnai3neo and Gnai1neo; Gnai3neo mutants show numerous defects in both HC types, including truncated hair bundles in OHCs (arrow), as well as supernumerary rows of stunted (hollow arrowheads) or longer stereocilia of variable height (full arrowheads) in IHCs. In addition, in FoxG1-Cre; Gnai2del; Gnai3flox and Atoh1-Cre; LSL-myc:ptxA mutants, OHC1-2s are severely misoriented. C-F) Quantification of various hair bundle features in 3-week old IHCs at the cochlear first turn. Each mutant strain is compared to littermate controls (in black; exact genotypes detailed in Source Data file). At least 3 animals, 17 IHCs and 108 stereocilia are represented per condition, except for FoxG1-Cre; Gnai2del; Gnai3flox where we could only obtain a single animal due to postnatal lethality. Mann-:hitney 8 test; p<0.0001, p<0.001, p<0.01, p<0.05; non-significant p-values are indicated. Detailed cohort sizes and statistics can be found in the Source Data file. G) SEM images of representative OHCs showing a truncated hair bundle (arrowhead). Lengths for the left and right wing of the hair bundle were measured and plotted as paired values for the same OHC. A littermate control graph is only shown for Gnai3neo mutants (Gnai3neo/+ littermate controls). Littermate control graphs for the other mutants can be found in Fig. 2 Supp. 1G. p-values are for a F-test of variance of pooled left and right wing lengths compared to littermate controls. At least 3 animals and 88 OHCs are represented per genotype. Only Gnai3neo and Gnai1neo; Gnai3neo mutants show truncated hair bundles and a significant p value (p<0.05). Scale bars are 10μm (A) and 2μm (B, G).
Loss of GNAI3 leads to profound deafness at high frequencies.
A-D) ABR thresholds at 8, 16, 32, and 40 kHz for Gnai1neo (A), Gnai2del (B), Gnai1neo; Gnai2del (C), and Gnai3neo and Gnai1neo; Gnai3neo (D) mutants tested between P21 and P29. Boxplots are framed with 25-75% whisker boxes where exterior lines show the minimum and maximum values, the middle line represents the median, and + represent the mean. A plotted value of 100 dB indicates that animals did not respond to 90 dB stimuli. In (C), controls are a pool of Gnai1+/+; Gnai2del/+, Gnai1neo/+; Gnai2+/+ and Gnai1neo/+; Gnai2del/+ animals. N indicates the number of animals tested. Two-way ANOVA with Sidak’s multiple comparison. p<0.0001 ****, p<0.001 ***, p<0.01 **, p<0.05 *; non-significant p-values are indicated. P-values in orange were obtained comparing non-littermate animals and suggest possibly raised thresholds when GNAI1 is inactivated in addition to GNAI3 (see text). kHz, kilohertz, dB SPL, decibel sound pressure level.
Systematic immunolabeling of GNAI proteins in Gnai mutant strains.
A-F) Two different antibodies (scbt“GNAI3” and pt“GNAI2”) were used to label the auditory epithelium at P0-P3. Neither antibody is specific for its protein target, as scbt“GNAI3” is able to detect GNAI2 (E) and pt“GNAI2” is able to detect GNAI3 (C). Note how no apical GNAI signal is visible with either antibody in Gnai2; Gnai3 double mutants (F). When the identity of the GNAI protein detected is unambiguous based on the sample genotype, it is made explicit in orange. Scale bars are 10µm.
GNAI2 only partially rescues the loss of GNAI3 in individual postnatal hair cells.
A-B) GNAI (pt“GNAI2” antibody; see Fig. 4) and GPSM2 co-immunolabeling in P0 Gnai3neo (A) and Gnai1neo; Gnai3neo (B) animals at the cochlear base. Boxed regions are magnified on the right. In both mutants, incomplete GNAI patterns are observed at the bare zone (arrow) and stereocilia tips (arrowheads). Remaining GNAI signals are specifically GNAI2 in (B). C) Correlation plot of GNAI signal intensity at the bare zone and tips in half-OHCs at the P2 cochlear first turn. Presence or loss of GNAI is remarkably correlated spatially between bare zone and tips in the same half-HC. N=3 animals, n=37 OHCs, Pearson correlation with best fit (red line; plot for control littermates can be found in Fig. 5 Supp. 1A). D-E) GNAI (pt“GNAI2” antibody) immunolabeling in P6 Gnai1neo; Gnai3neo animals. Boxed IHC regions are magnified below. Loss of GNAI2 progresses with HC differentiation, with largely lost IHC signals at the P6 cochlear base (D) but partial rescue on one side of the cell at the P6 first turn (E), as observed at the cochlear base at P0 (A-B). F-G) ZO1 (apical junctions) and pericentrin (PCNT; basal body) immunolabeling in P8 OHCs. The position of the basal body was used to determine the vertex (middle) of the original hair bundle and to draw a radial line separating each OHC into two halves (F). The length of each hair bundle wing (y axis) is graphed in relation to the corresponding apical surface area (x axis) in the same half-OHC (G). Truncated OHC wings correlate with reduced apical membrane area on the same side. N=3 animals, n=58 OHCs, Pearson correlation with best fit (red line; plot for control littermates can be found in Fig. 5 Supp. 1C). A.U., arbitrary unit. Scale bars are 10µm.
Delayed bare zone expansion and severely dysmorphic hair bundles in absence of GNAI2 and GNAI3.
A-B) GNAI (scbt“GNAI3” antibody, see Fig. 4) and acetylated tubulin (AcTub) co-immunolabeling at the E17.5 cochlear base. Note how F-actin labeling (phalloidin) reveals a polarized bare zone marked by GNAI (arrows) in control but not in Gnai2; Gnai3 double mutants. C) GPSM2 and AcTub co-immunolabeling at the E17.5 cochlear base. In contrast to Gnai2; Gnai3 double mutants, FoxG1-Cre; DIO-ptxA mutants have a polarized bare zone (arrows) despite OHC1-2 adopting a reversed orientation (V brackets indicate orientation). GPSM2 marks the bare zone in controls and is reduced in mutants. D) Graphs of bare zone surface area in E17.5 HCs at the cochlear base. FoxG1-Cre; Gnai2del; Gnai3flox: controls (Gnai2del/+; Gnai3flox/+ and Gnai2del/+; Gnai3flox/flox) N=3 animals, n=19 IHC, 23 OHC1, 19 OHC2, 21 OHC3; mutants N=3, n=23 IHC, 24 OHC1, 23 OHC2, 24 OHC3. FoxG1-Cre; DIO-ptxA: controls (DIO-ptxA) N=3, n=18 IHC, 18 OHC1, 21 OHC2, 18 OHC3; mutants N=3, n= 21 IHC, 20 OHC1, 21 OHC2, 9 OHC3. E-G) Pericentrin (PCNT) and AcTub co-immunolabeling at P0 at the cochlear base. Unlike at E17.5 (A-B, D), most P0 Gnai2; Gnai3 double mutant HCs have a bare region (E-F, arrows). This bare region is unpolarized and its abnormal shape refects aberrant stereocilia distribution. In sharp contrast, ptxA mutants have normally-shaped hair bundles and bare zones despite OHC1-2 adopting a reversed orientation (G). H) Graphs of bare zone surface area in P0 HCs at the cochlear base. FoxG1-Cre; Gnai2del; Gnai3flox: controls (Gnai2del/+; Gnai3flox/+, Gnai2del/+; Gnai3flox/flox and FoxG1-Cre; Gnai2del/+; Gnai3flox/+) N=3, n= 19 IHC, 23 OHC1, 21 OHC2, 21 OHC3; mutant N=3, n= 21 IHC, 22 OHC1, 24 OHC2, 24 OHC3. FoxG1-Cre; DIO-ptxA: controls (DIO-ptxA) N=3, n=15 IHC, 23 OHC1, 18 OHC2, 15 OHC3; mutants N=3, n= 16 IHC, 24 OHC1, 18 OHC2, 19 OHC3. D, H: Mann-Whitney U test; p<0.0001 ****, p<0.001 ***, p<0.01 **, p<0.05 *; non-significant p-values are indicated. All ptxA samples are heterozygotes (R26DIO-ptxA/+). Scale bars are 10μm (A, C (OHC), E, G (OHC)) and 5μm (B,C (IHC), F, G (IHC)).
Loss of GNAI2 and GNAI3 provokes hair cell eccentricity defects absent in ptxA mutants.
A-B) Graphs of HC eccentricity representing the position of the basal body as a ratio of the radius (BB/r, top diagram). Data cover E17.5 mid and base and P0 apex, mid and base cochlear positions for each HC type. HCs were considered symmetrical when their eccentricity ratio was lower than 0.25 (red zone). Only FoxG1-Cre; Gnai2del; Gnai3flox mutants harbor symmetrical HCs. Their proportion is indicated in the bar graphs on the right (A). Overall, the proportion of symmetrical cells tends to decrease in maturing OHCs but remains higher in IHCs. At least 3 animals and 39 cells per HC type are represented for each stage, cochlear position and genotype (detailed in the Source Data file). Controls for FoxG1-Cre; Gnai2del/del; Gnai3flox/flox are Gnai2del/+; Gnai3flox/+, Gnai2del/+; Gnai3flox/flox, FoxG1-Cre; Gnai2del/+; Gnai3flox/+ and FoxG1-Cre; Gnai2del/+; Gnai3flox/flox. Controls for FoxG1-Cre; DIO-ptxA are DIO-ptxA heterozygotes. Mann-Whitney U test: p<0.0001 ****, p<0.001 ***, p<0.01 **, p<0.05 *; non-significant p-values are indicated. Detailed cohort sizes and statistics can be found in Source Data file.OHCs but remains higher in IHCs. At least 3 animals and 39 cells per HC type are represented for each stage, cochlear position and genotype (detailed in the Source Data file). Controls for FoxG1-Cre; Gnai2del/del; Gnai3flox/flox are Gnai2del/+; Gnai3flox/+, Gnai2del/+; Gnai3flox/flox, FoxG1-Cre; Gnai2del/+; Gnai3flox/+ and FoxG1-Cre; Gnai2del/+; Gnai3flox/flox. Controls for FoxG1-Cre; DIO-ptxA are DIO-ptxA heterozygotes. Mann-Whitney U test: p<0.0001 ****, p<0.001 ***, p<0.01 **, p<0.05 *; non-significant p-values are indicated. Detailed cohort sizes and statistics can be found in Source Data file.
Loss of GNAI2 and GNAI3 recapitulates hair cell orientation defects observed in ptxA mutants.
A-D) Circular histograms show HC orientation (α) based on the position of the basal body (purple dot in top diagram) at the stage and cochlear position indicated. 0° is towards the cochlear base and 90° is lateral. Only HCs with an eccentricity above 0.25 are represented (see Fig. 7). Histograms show frequency distribution (10° bins) and red radial lines and arcs respectively indicate the circular mean and the circular mean deviation. In control cochleae, HCs are tightly oriented laterally (90°) except for OHC3 that show a slight bias towards the cochlear apex (180°). As reported previously, ptxA expression induces a reversal of OHC1 and OHC2, and more imprecise lateral orientation of OHC3. This phenotype is progressively recapitulated in time in Gnai2; Gnai3 double mutants (compare least mature E17.5 mid (A) and most mature P0 base (D)). Interestingly, IHCs also show severe misorientation in Gnai2; Gnai3 double mutants, unlike in ptxA mutants. n, HC number in N=3-4 animals. Controls for FoxG1-Cre; Gnai2del/del; Gnai3flox/flox are Gnai2del/+; Gnai3flox/+, Gnai2del/+; Gnai3flox/flox, FoxG1-Cre; Gnai2del/+; Gnai3flox/+ and FoxG1-Cre; Gnai2del/+; Gnai3flox/flox. Histograms for littermate controls of FoxG1-Cre; DIO-ptxA mutants, as well as data at the P0 cochlear mid position, can be found in Fig. 8 Supp. 1. Detailed cohorts and results can be found in the Source Data file.
Summary and model: three roles validated in vivo for GNAI proteins during hair cell polarized morphogenesis.
GNAI proteins are required for the centrifugal migration of the basal body, and thus HC symmetry breaking (a). This activity is distinct from defining binary HC orientation downstream of GPR156 (b), as Gpr156 mutant HCs have normal basal body positioning and normal hair bundles. Finally, GNAI partners with GPSM2 to shape the hair bundle and elongate stereocilia (c). The specific identity and the dosage of the GNAI proteins required differ for each role. GNAI3 is the principal architect of hair bundle development (c), with GNAI2 playing an important but progressively waning role. In this role, high amounts of GNAI3/GNAI2 are required. A lower dose of any GNAI protein is sufficient for proper HC orientation (b), and a still lower dose can ensure symmetry breaking (a) since this function is intact in the ptxA model. We speculate that heterotrimeric signaling underpins basal body migration (a), as it does HC orientation (b). A regulator working with GNAI proteins for symmetry breaking remains to be identified.
A-D) New mouse strains generated in this study are Gnai2del (A), Gnai3flox (B) and R26DIO-ptxA (C; see Methods for details). R26LSL-myc:ptxA was published previously and is illustrated for comparison as it is used in the study as well (D). E-F) Loss of GNAO has no obvious impact on apical HC morphology or HC orientation. Representative confocal images of phalloidin-stained Gnao1 constitutive mutants at P21 (E; Gnao1neo) and scanning electron microscopy images of conditional Gnao1 inactivation in the Gnai1; Gnai3 double mutant background (F; Gnao1flox). In F, top panels compare control (left) and conditional Gnao1 inactivation (right) when GNAI1 and GNAI3 functions are preserved. Bottom panels compare control (left) and conditional Gnao1 inactivation (right) when GNAI1 and GNAI3 are inactivated. Note how HC-specific inactivation of GNAO using the Atoh1-Cre driver does not obviously enhance defects observed in Gnai1; Gnai3 double mutants (F, bottom). G) Graphs showing the length of the left and right wing of 3 week-old OHC hair bundles paired by cell. Graphs for control littermates that were omitted in Fig. 2G are added here (left) next to repeated mutant graphs (right). p-values are for a F-test of variance of pooled left and right wing lengths compared to littermate controls. At least 3 animals and 88 OHCs are represented per genotype. Only Gnai3neo (Fig. 2G) and Gnai1neo; Gnai3neo mutants show truncated hair bundles and a significant p value (p<0.05).
A-B) ABR thresholds at 8, 16, 32 kHz for constitutive Gnao1 (A) and conditional Atoh1-Cre; Gnao1flox/flox (B) mutants tested between P20 and P25. Boxplots are framed with 25-75% whisker boxes where exterior lines show the minimum and maximum values, the middle line represents the median, and + represent the mean. N indicates the number of animals tested. In (B), because the original animal cohort was established to test whether the loss of GNAO could enhance defects in Gnai1; Gnai3 double mutants (see Fig. 2 Supp. 1E-F), controls include the following genotypes: Atoh1-Cre; Gnao1flox/+, Gnao1flox/flox, Gnai1neo/+; Gnai3neo/+, Gnai1neo/+, Atoh1-Cre; Gnao1flox/+; Gnai1neo/neo; Gnai3neo/+ and Gnai1neo/neo; Gnai3neo/+ where other alleles are wild-type. Mutants include the following: Atoh1-Cre; Gnao1flox/flox; Gnai3neo/+ and Atoh1-Cre; Gnao1flox/flox; Gnai1neo/neo; Gnai3neo/+ where other alleles are wild-type. Cohort details can be found in the Source Data file. Two-way ANOVA with Sidak’s multiple comparison. non-significant p-values are indicated. kHz, kilohertz, dB SPL, decibel sound pressure level.
GNAI protein distribution is undisturbed in Gnao1 mutants and no evidence for specific GNAO enrichment at the hair cell apex. A-B) GNAI (scbt “GNAI3” antibody; see Fig. 4 and Methods) immunolabeling in P5 Gnao1 constitutive (A) and Atoh1-Cre; Gnao1flox conditional (B) mutants at the cochlear base. GNAI and GPSM2 (B) distribution at the bare zone and stereocilia tips is unchanged in absence of GNAO. C) GNAO immunolabeling at the P3 cochlear mid produces signals that are unspecific since they appear unchanged in conditional Gnao1 mutants. Scale bars are 10μm.
GNAI2 fully rescues loss of GNAI3 at embryonic stages and is still detected in low amounts at stereocilia tips in adult hair cells lacking GNAI3.
A) Control correlation plot for Gnai1; Gnai3 double mutants in Fig. 5C. GNAI signal intensity at the bare zone and stereocilia tips in control half-HCs at the P2 cochlear base. N=3 animals, n=40 OHCs, Pearson correlation with best fit (red line). The graph shows relatively uniform GNAI enrichment in each sub-cellular compartment, unlike in Gnai1; Gnai3 double mutants (Fig. 5C). B) GNAI (pt “GNAI2” antibody; see Fig. 4) and GPSM2 co-immunolabeling at the E18.5 5 cochlear base. As sole remaining GNAI, GNAI2 can still encompass the entire bare zone in Gnai1; Gnai3 double mutants and no obvious hair bundle defects are observed, unlike at P0 (Fig. 5A-B). The boxed region is magnified to the right. C) Control correlation plot for Gnai1; Gnai3 double mutants in Fig. 5G. The position of the basal body was used to determine the vertex (middle) of the hair bundle and to draw a radial line separating each OHC into two halves. The length of each hair bundle wing (y axis) is graphed in relation to the corresponding apical surface area (x axis) in the same half-OHC. The plot shows relatively uniform wing lengths and half apical areas in littermate controls, unlike in Gnai1; Gnai3 double mutants (Fig. 5G). N=3 animals, n=52 OHCs, Pearson correlation with best fit (red line). D-E) GNAI (pt “GNAI2” antibody; see Fig. 4) immunolabeling in P28 adults. As sole remaining GNAI, GNAI2 can still be detected at low levels at stereocilia tips in Gnai1; Gnai3 double mutants at the cochlear base, mid and apex (arrowheads). E) Higher magnifications views of a single OHC1 (left) and IHC (right). A.U., arbitrary unit. Scale bars are 10µm (B, D) and 5 µm (E).
Validation of the Gnai3flox and DIO-ptxA mouse strains.
A) GNAI (pt “GNAI2” antibody; see Fig. 4) immunolabeling at the P0 cochlear base. Note how in absence of Cre recombinase, Gnai3flox/flox HCs have normal hair bundles and normal GNAI enrichment at the bare zone and stereocilia tips. In contrast, breeding in Atoh1-Cre produces incomplete GNAI enrichment (arrows; compare with Fig. 4D) and shortened stereocilia typical of constitutive Gnai3 mutants. Boxed IHC regions are magnified below. B-D) Comparison of the two ptxA strains used in this study (see also Fig. 2 Supp. 1C-D). B) PtxA expression with the Atoh1-Cre driver produces comparable P4 apical HC defects in the LSL-myc:ptxA and DIO-ptxA strains. “Escaper” OHC1 that are not inverted in orientation (arrows) likely do not express Cre, or did not undergo recombination. C) When Atoh1-Cre induces ptxA expression in post-mitotic HCs, the number of escaper OHC1 is higher in the DIO-ptxA strain. This is likely because lox-based genomic deletions (LSL) are favored over genomic inversions (DIO; see Fig. 2 Supp. 1C-D). D) In contrast, an earlier Cre driver (FoxG1-Cre) shows virtually no escapers in either strain. N and n indicate the number of animals and OHC1 analyzed, respectively. All ptxA samples in the study are heterozygotes (R26LSL-myc:ptxA/+ or R26DIO-ptxA/+). Scale bars are 10µm (A, B), 20μm (D).
Loss of GNAI2 and GNAI3 recapitulates hair cell orientation defects observed in ptxA mutants.
A) P0 mid-cochlea histograms that complement P0 apex and P0 base histograms in Fig. 8C-D. B-F) Histograms including littermate controls for FoxG1-Cre; DIO-ptxA mutants in Fig. 8 (B-C, D, F), and P0 mid-cochlear position for FoxG1-Cre; DIO-ptxA mutants and controls (E). Note that all mutant histograms except at the P0 cochlear mid (A, E) are repeated from Fig. 8. Circular histograms show HC orientation (α) based on the position of the basal body (purple dot in top diagram) at the stage and cochlear position indicated. 0° is towards the cochlear base and 90° is lateral/abneural. Only HCs with an eccentricity above 0.25 are represented (see Fig. 7). Histograms show frequency distribution (10° bins) and red radial lines and arcs respectively indicate the circular mean and the circular mean deviation. n, HC number in N=3-4 animals. Controls for FoxG1-Cre; Gnai2del/del; Gnai3flox/ flox are Gnai2del/+; Gnai3flox/+, Gnai2del/+; Gnai3flox/flox, FoxG1-Cre; Gnai2del/+; Gnai3flox/+ and FoxG1-Cre; Gnai2del/+; Gnai3flox/flox. Controls for FoxG1-Cre; DIO-ptxA are DIO-ptxA heterozygotes. Detailed cohorts and results can be found in the Source Data file.
Mouse strain details.
