Integrin α8 and Pcdh15 act as a complex to regulate cilia biogenesis in sensory cells

Abstract

The way an organism perceives its surroundings depends on sensory systems and the highly specialized cilia present in the neurosensory cells. Here, we describe the existence of an integrin α8 (Itga8) and protocadherin-15a (Pcdh15a) ciliary complex in neuromast hair cells in a zebrafish model. Depletion of the complex via downregulation or loss-of-function mutation leads to a dysregulation of cilia biogenesis and endocytosis. At the molecular level, removal of the complex blocks the access of Rab8a into the cilia as well as normal recruitment of ciliary cargo by centriolar satellites. These defects can be reversed by the introduction of a constitutively active form of Rhoa, suggesting that Itga8-Pcdh15a complex mediates its effect through the activation of this small GTPase and probably by the regulation of actin cytoskeleton dynamics. Our data points to a novel mechanism involved in the regulation of sensory cilia development, with the corresponding implications for normal sensory function.

Keywords: Integrin α8; Pcdh15; Sensory cilia; Usher syndrome.

Fig. 1.

Downregulation or mutations of Itga8 or Pcdh15a inhibits ciliogenesis. (A–J) SR-SIM of 3 dpf (A–G) and 5 dpf (H–J) larvae immunostained for acetylated tubulin (ac tubulin, red) and counterstained with phalloidin (green). Control morpholino-injected animals (A), itga8 (Itga8 MO, B) or pcdh15a (15a MO, C) morphants, morphants co-injected with itga8 (8 MO+cRNA; D) or pcdh15a (15a MO+cRNA; E) cRNAs, Itga8 morphants co-injected with CA rhoab (+Rhoab; F) or rhoad (+Rhoad; G) cRNAs, WT (H), and orbiter (orb) strong (I) and orbiter weak mutants (J) are shown. Scale bars: 4 µm (A–G), 3.5 µm (H–J). (K,L) Quantification analysis of the kinociliary length. For each independent experiment, the average kinociliary length per neuromast was calculated and expressed as mean±s.e.m. (M) Scatter plot of individual kinociliary lengths for WT and orbiter mutants. (N) Frequency distribution analyses of kinociliary length in WT and orbiter mutants. (O) Cartoon of a neuromast (top view) showing the stained structures: hair bundle in green and kinocilia in red. **P<0.01; ***P<0.001; ns, not significant (one-way ANOVA followed by Dunnett's multiple comparisons test or two-tailed Student's t-test). At least five independent experiments were performed.

Fig. 2.

Itga8 and Pcdh15a proteins mediate Rhoa activation. Rhoa pulldown assays performed with total lysates from 1–2 dpf MOs (A,C) and 5 dpf pcdh15a mutants (B). (A) Animals were injected with control, specific Itga8 (Itga8 MO) or Pcdh15a (15a MO) MO or with the specific MO and the corresponding cRNA (+cRNA). (C) Itga8 MOs were also co-injected with the cRNAs for CA rhoab (Itga8 MO+b), CA rhoad (Itga8 MO+d) or both (Itga8 MO+b&d). Itga8 immunoblot (IB Itga8) from controls and the treated animals was performed in parallel with the pulldowns to confirm Rhoa activation in the absence of Itga8 protein expression. GTP-Rhoa, active Rhoa, red framed area; tRhoa, total Rhoa; His-Rhoa, recombinant his-tagged Rhoa. White asterisks denote the beads from the pulldown. Note that in some cases GTP-Rhoa is shown as a doublet including the recombinant protein (red asterisks). (D–G) GTP-Rhoa (from the red framed area) abundance was quantified, normalized to tRhoa and expressed as a percentage of that in controls (mean±s.e.m.). *P<0.05; **P<0.01; ***P<0.001; ns, not significant (one-way ANOVA followed by Dunnett's multiple comparisons test or two-tailed Student's t-test). At least three independent experiments were performed for each treatment.

Fig. 3.

Itga8 and Pcdh15a localize at the apical aspect of hair cells. Representative images of embryos/larvae immunostained for Pcdh15a (red), Itga8 (green) and acetylated tubulin (blue). Samples were counterstained with phalloidin (gray). (A–B′) Confocal microscopy analyses of a 1 dpf zebrafish inner ear (dorsal view, anterior to the left). (B,B′) Magnification of boxed area in A,A′. (C–H′) SR-SIM of 5 dpf neuromasts. (C–D′) WT. (D,D′) Magnification of boxed area in C,C′. (E–F′) orbiter strong; (G–H′) orbiter weak. Asterisks denote hair bundle immunostaining. The arrowhead in panel B′ denotes cilia immunostaining. (I,J) Cartoons of neuromasts (top views) showing colocalization (yellow) of Itga8 and Pcdh15a. In WT animals (I), both proteins colocalize at the hair bundle in a sub-set of cells and towards the tip of the kinocilia. In mutant animals (J), there is some colocalization at the hair cell bundle but also around it (cuticular plate). Scale bars: 25 µm (A), 8 µm (B), 4 μm (D), 7 µm (C,E–H). Three independent experiments were performed.

Fig. 4.

Interdependency of Itga8 and Pcdh15a protein expression and localization in hair cells. (A,B) Representative immunoblots from 3 dpf Pcdh15/Itga8 MOs (A) and 5 dpf pcdh15a mutants (B). (A) Expression of Itga8 and Pcdh15a was analyzed in controls, MOs (Itga8 MOs or 15a MOs) or MOs with the corresponding cRNA (+cRNA). (B) Expression of Itga8 and Pcdh15a was analyzed in WT and orbiter mutants. Red asterisks denote the full-length protein, black asterisks denote putative small protein variants. The blue asterisk denotes an additional variant not found in WT. Membranes were stripped and re-probed for actin as loading control. IB itga8, Itga8 immunoblot; IB 15a, Pcdh15a immunoblot. Three independent experiments were performed. (C–G′) Confocal images of 3 dpf controls (C,C′), MOs [Itga8 MOs or 15a MOs, (D,D′, F,F′) or MOs plus the corresponding cRNA (+cRNA; E,E′,G,G′)]. Larvae were immunostained for Pcdh15a (red), Itga8 (green) and acetylated tubulin (blue) and counterstained with phalloidin (gray). Arrowheads denote Itga8 and Pcdh15a colocalization at the tip of the cilia in controls and +cRNA MOs but not in MOs alone. Top right corner: number of neuromasts showing apical localization for the corresponding protein versus total number of neuromasts inspected. Two independent experiments were performed. Scale bars: 4.5 µm.

Fig. 5.

Itga8 and Pcdh15a interact in hair cells. (A–D) Representative immunoblots of co-IP studies from 3 dpf larvae showing Itga8–Pcdhs15 interaction. IP itga8: Itga8 IP; IP NGpS: IP with pre-immune guinea pig serum; IP 15a, Pcdh15a IP; IP NRS, IP with pre-immune rabbit serum; IB 15a, Pcdh15a immunoblot; IB Itga8, Itga8 immunoblot. Asterisks denote specific bands (red asterisks, full-length protein; black asterisks, small variants). Three independent experiments were performed. (E–H) Coronal sections of 5 dpf WT (E,H) orbiter (orb) strong (F) and weak (G) mutants. Positive PLA (red dots) (E and high magnification inset), demonstrate an in situ Itga8–Pcdh15a association in WT but not in pcdh15a mutants (F,G). (H) Negative control in which one of the primary antibodies (in this case anti-Itga8) was omitted. Sections were immunostained for acetylated tubulin (blue) and counterstained with phalloidin (green). (I) Cartoon of a neuromast (lateral view) showing the stained structures: hair bundle in green, kinocilia and cell bodies in blue and the Itga8–Pcdh15a complex as red dots. Ten animals from two independent experiments were inspected. Scale bar: 6 µm (for E–H), 10 µm (inset in E).

Fig. 6.

The Itga8–Pcdh15a complex regulates endocytosis/recycling in hair cells. (A–F,H–M) Confocal images of 3 dpf (A–F,H–J) and 5 dpf (K–M) larvae (labeled as in Fig. 1) incubated with FM1-43 for 30 min. Scale bars: 5.5 µm. (G,N) Quantification of total fluorescent intensity per neuromast. At least five independent experiments were performed for each treatment and total fluorescence intensity expressed as a percentage of that in control (mean±s.e.m.). (O) Cartoon of a neuromast (top view) showing the stained structures: hair cells loaded with FM1-43 are in red, and the hair cell bundle in green. **P<0.01; ***P<0.001; ns, not significant (one-way ANOVA followed by Dunnett's multiple comparisons test or two-tailed Student's t-test).

Fig. 7.

Lack of the Itga8–Pcdh15a complex activity results in ciliary cargo transport impairment. Confocal images and quantitative data (mean±s.e.m.) for Rab8a (A–H), Rabin8 (K–O) and centrin (Q–X) in control, Itga8 MOs, Pcdh15a MOs and orbiter mutants, and in rescued MOs (labeled as in Fig. 1). Scale bars: 6 µm. (I,J,P,Y,Z) Cartoons of neuromasts (top views) showing the corresponding staining: the hair cell bundle is in green, and Rab8a, Rabin8 or centrin is in red. (I) Control neuromast showing Rab8a ciliary staining. (J) Itga8- or Pcdh15a-deficient neuromast showing apical localization of Rab8a but no staining in the kinocilia. (P) Neuromasts showing apical distribution of Rabin8. (Y) Control neuromast showing basal body/transition zone localization of centrin with weak ciliary staining (pink). (Z) Itga8- or Pcdh15a-deficient neuromast showing centrin localization at the basal body/transition zone and also at the hair cell bundle (yellow). (AA) Rab8a graph. The presence of ciliary Rab8a was evaluated for each treatment in five independent experiments and expressed as percentage of that in control. (BB) Rabin8 graph. Neuromast apical fluorescence was quantified for each treatment in three independent experiments and is expressed as percentage of that in control. (CC) Centrin graph. The centrin punctate distribution correlating with the point of insertion of the kinocilium (basal body/centriole) was qualitatively assessed and the results expressed as percentages of that in control. Five independent experiments were performed. *P<0.05; **P<0.01; ***P<0.001; ns, not significant (one-way ANOVA followed by Dunnett's multiple comparisons test or two-tailed Student's t-test).

Fig. 8.

Itga8–Pcdh15a complex function. (A) Under normal conditions, Itga8b1 associates with Pcdh15a and activates Rhoa (1). By regulating acting dynamics, GTP-Rhoa, modulates the Rab8a (2) and centrin (3) ciliary distribution with the concomitant ciliary lengthening. Normal ciliary morphology and signaling lead to proper endocytic activity (4). (B) Absence of the Itga8b1–Pcdh15a complex from the cilia compartment results in Rhoa inactivation (5), leading to an accumulation of inactive Rab8a (6) at the ciliary base and centrin mislocalization (7). Cilia are shorter and endocytic activity (8) is impaired.