Dishevelled stabilization by the ciliopathy protein Rpgrip1l is essential for planar cell polarity

Abstract

Cilia are at the core of planar polarity cellular events in many systems. However, the molecular mechanisms by which they influence the polarization process are unclear. Here, we identify the function of the ciliopathy protein Rpgrip1l in planar polarity. In the mouse cochlea and in the zebrafish floor plate, Rpgrip1l was required for positioning the basal body along the planar polarity axis. Rpgrip1l was also essential for stabilizing dishevelled at the cilium base in the zebrafish floor plate and in mammalian renal cells. In rescue experiments, we showed that in the zebrafish floor plate the function of Rpgrip1l in planar polarity was mediated by dishevelled stabilization. In cultured cells, Rpgrip1l participated in a complex with inversin and nephrocystin-4, two ciliopathy proteins known to target dishevelled to the proteasome, and, in this complex, Rpgrip1l prevented dishevelled degradation. We thus uncover a ciliopathy protein complex that finely tunes dishevelled levels, thereby modulating planar cell polarity processes.

Figure 1.

Ftm mouse mutants show planar polarity defects in the organ of Corti. (A) Immunofluorescence (IF) on an E18.5 wild-type cochlea showing Rpgrip1l accumulation at the transition zone of the cilium (green arrow), between the basal body (red arrow) and the axoneme. (B) Schematic drawing of a wild-type sensory hair cell, showing the relative positions of the stereociliary bundle (red), the kinocilium basal body (green), and the daughter centriole (yellow). The green numbers (1–6) correspond to the six quadrants used to monitor the position of the kinocilium in J. The proximo-distal axis of the cochlea (inner-outer axis, PCP axis) is indicated. (C) Diagram illustrating the mean length/width (LW) ratio of flat-mounted cochlear ducts of E18.5 controls: (Ftm^+/+ or Ftm^+/−, n = 20) and Ftm^−/− (n = 18) fetuses. Double asterisk indicates that the difference in mean LW ratio between control and Ftm^−/− cochleae is significant (Student’s t-test, α < 0.0005). (D–G) Scanning electron microscopy on dissected cochleae of E18.5 control (D and F) and Ftm^−/− (E and G) fetuses. The position of the row of inner hair cells (IHC) and of the three rows of outer hair cells (OHC1-3) is indicated in D. (H and I) Phalloidin staining of F-actin on E18.5 control (H) and Ftm^−/− (I) cochleae. (J) Diagrams showing the position of the centrioles in different cell quadrants illustrated in B. The normal position in wild-type cells is in quadrant 3. 123 control cells (three fetuses from three experiments) and 84 Ftm^−/− cells (three fetuses from three experiments) were analyzed. The difference between control and Ftm^−/− cells in the distribution of the centrioles in the different cell quadrants is significant (Khi2 test: α < 0.001). (K–N1) Detection of acetylated α-tubulin (K–L1) or γ-tubulin (M–N1) and F-actin in E18.5 control (K, K1, M, and M1) or Ftm^−/− (L, L1, N, and N1) cochleae. K, L, M, and N: IHCs; K1, L1, M1, and N1: OHCs. In selected cells (circled), white arrows point to the position of cilia (K–L1) or centrioles (M–N1). In all the pictures, proximal is to the top. Bars, 5 µm. Antibodies are indicated and color coded. All pictures and measurements were performed in the basal 25% of the cochlea.

Figure 2.

Rpgrip1l depletion in zebrafish embryos leads to CE defects, hydrocephaly, and impaired laterality. (A) IF on embryos injected with an RNA coding for a tagged form of Rpgrip1l (Rpgrip1l-Myc), showing its localization at the transition zone of the cilium in the neural tube (nt), floor plate (fp), and notochord (nc). Enlarged views of selected cilia are presented in insets on the right. (B–D) Rpgrip1l morpholino-injected embryos (Mo-1) at the 12-s stage, representative of the three phenotypic classes: unaffected (B, indistinguishable from uninjected controls, body gap angle [bga] around 50°), class I (C, mild, 60° < bga < 90°, kinked notochord, and class II (D, severe: 90° < bga < 120°, malformed somites and kinked, broader notochord). (E–G) dorsal views of representative embryos of the three classes, showing the kinked notochord (nc) and malformed somites (s) in class I/II embryos. (H and I) 5-dpf control (H) and rpgrip1l morphant (I) larvae, showing the shortened and abnormally curved trunk in morphants. (J and K) Lateral views of 48-hpf control (J) and rpgrip1l morphant (K) embryos showing hydrocephaly in morphants (K, arrowheads). (L–O) Laterality defects in rpgrip1l morphants. southpaw (spw) expression patterns in the lateral mesoderm (arrows), visualized by ISH, fall into four different classes: left, right, bilateral, and absent. (P) Diagrams illustrating the axis elongation phenotype. For each class (color coded), the percentage of embryos is indicated by the length of the horizontal bars. The number of embryos scored is 236 for uninjected (ni), 159 for 1 mM Mo-Rpgrip1l, 408 for 0.4 mM Mo-Rpgrip1l, and 171 for 0.4 mM Mo-Rpgrip1l + hRPGRIP1L RNA. Double asterisk indicates that the difference in the distribution of embryos in the different phenotypic classes is significant (Khi2 test: α < 0.001). The morphant phenotype is dependent on the dose of morpholino (0.4 vs. 0.8 mM; 1 mM corresponds to 1 pmol/embryo) and is partially rescued by co-injection of human RPGRIP1L mRNA (100 pg). (Q) Diagram illustrating the percentage of hydrocephalic embryos. Number of embryos scored: 32 for uninjected, 32 for hRPGRIP1L RNA, 25 for 0.8 mM Mo-1 + GFP RNA, and 22 for 0.8 mM Mo-1 + hRPGRIP1L RNA. **, α < 0.001, Khi2 test. (R) Diagram indicating the number of control and morphant embryos in each of the laterality classes illustrated in L–O: left (L), right (R), bilateral (B), and no expression (N). Number of embryos scored: 17 for uninjected, 79 for 0.4 mM Mo-1. **, α < 0.001, Khi2 test. In Q and R, the data shown are from a single representative experiment out of two (Q) or four (R) repeats, respectively. Bars: (A) 10 µm; (A, insets) 2.5 µm; (B–O) 250 µm.

Figure 3.

Rpgrip1l is necessary for posterior positioning of the basal body in floor plate cells. (A–F) Lateral confocal sections of 18-s stage control (A, D, and G) and rpgrip1l morphant (B, C, E, F, H, and I) embryos stained with antibodies to acetylated α-tubulin (A–C), γ-tubulin, and GFP. The embryos have been injected at the 1-cell stage with an mbGFP RNA in order to visualize cell membranes. G, H, and I are higher magnifications of the regions boxed in D, E, and F, respectively. (J and K) Diagrams illustrating the percentage of floor plate cells with a basal body in posterior (P), medial (M), or anterior (A) position in 18-s stage control embryos (J) or after injection of 4 ng/µl Dsh-GFP RNA (K), 0.4 mM Mo-Rpgrip1l (J and K), or co-injected with both (K). The diagram in J corresponds to 2 independent injection experiments with a total of 7 controls (142 cells analyzed) and 14 rpgrip1l morphants (172 cells analyzed). The diagram in K corresponds to 2 independent injection experiments with a total of 8 controls (130 cells analyzed), 11 rpgrip1l morphants (183 cells analyzed), and 11 morphants co-injected with Dsh-GFP mRNA (168 cells). The double asterisks in J and K mean that the differences in the distributions of the three phenotypic classes between the compared experimental conditions are statistically significant (Khi2 test: α < 0.001). In A–I, anterior is to the left and dorsal is up. In D–I, a single confocal section is shown. Bars, 10 µm in all pictures.

Figure 4.

Rpgrip1l stabilizes dishevelled at the cilium base. (A) Diagram illustrating the functional interaction between Rpgrip1l and dishevelled in axis elongation. For each lane, the injected morpholinos are indicated on the right. n = 37 control, 29 Mo-Rpgrip1l–injected, 37 Mo-3Dvl–injected, and 24 co-injected embryos. The differences in the repartition in classes between batches of embryos injected with one morpholino only and the batch of co-injected embryos are significant (α < 0.001, Khi2 test). (B) Western blots illustrating the amounts of Dvl2-Myc protein after injection of 15 pg Dvl2-Myc RNA per embryo at the 1-cell stage with or without co-injecting Mo-Rpgrip1l (0.8 mM). The diagrams illustrate the relative amounts of Dvl2-Myc in control and Mo-Rpgrip1l–injected embryos at the 80% epiboly and at the 12-s stages, in six independent experiments. Dvl amounts are in arbitrary units and the average amounts in control embryos are arbitrarily set at 1.0. (C) Diagram illustrating the axis elongation phenotype in embryos injected with Mo-Rpgrip1l and/or with Dsh-GFP RNA (12 ng/µl, corresponding to 12 pg/embryo). n = 23 control, 23 Dsh-GFP injected, 24 Mo-Rpgrip1l–injected, and 58 doubly injected embryos. Dsh-GFP mRNA (12 pg/embryo) significantly rescues the morphant phenotype (α < 0.001, Khi2 test). (D and E) IF with an anti-GFP antibody to reveal the GFP tagged Dsh protein and with an anti-acetylated α-tubulin antibody (Ac Tub) to label cilia in embryos injected with Dsh-GFP (10 pg/embryo) with (E) or without (D, controls) Mo-Rpgrip1l (0.8 mM). (F and G) View of the floor plate after IF with anti-GFP, anti-βgal, and anti–Ac-Tub antibodies in 18-s stage embryos injected with DshGFP (10 pg/embryo) and nlsLacZ RNAs (60 pg/embryo) with (G) or without (F, controls) Mo-Rpgrip1l. anti-βgal staining in nuclei indicates that the corresponding cells have received injected RNA. (H) IF with anti–Ac-Tub, anti-GFP, and anti–γ-Tub antibodies in embryos injected with Dsh-GFP RNA alone. (I) IF with anti-GFP, anti-Myc, and anti–γ-Tub antibodies in embryos injected with Dsh-GFP (10 pg/embryo) and Rpgrip1l-Myc (7 pg/embryo) RNAs. Bars, 10 µm.

Figure 5.

RPGRIP1L stabilizes dishevelled at the cilium base in cultured mammalian renal cells. (A) Co-immunostaining of Dvl2 (top) or Dvl3 (middle) with γ-tubulin on control and RPGRIP1L-KD MDCK cells. The bottom panel shows z-section of coimmunostaining of Dvl3 and γ-tubulin in the three cell lines. The histogram compares the percentage of cells with Dvl2 or Dvl3 staining and their accumulation in the pericentriolar region versus the cytoplasm (800–900 cells analyzed). (B) Quantitative RT-PCR analysis of Dvl2 and Dvl3 mRNA levels in control and RPGRIP1L-KD MDCK cells. (C) Western blot analysis and quantification of endogenous Dvl2 and Dvl3 in control and RPGRIP1L-KD MDCK cells. (D) Coimmunostaining of inversin with γ-tubulin on control and RPGRIP1L-KD MDCK cells. The right panel shows z-section of the stainings in both cell lines. (E) Immunostaining of NPHP4-V5 on control and RPGRIP1L-KD MDCK cells transduced with the NPHP4-V5 expression construct (Burcklé et al., 2011). In A, D, and E nuclei were stained with Hoechst. Bars: (C, D, z-sections A or C) 10 µm; (A, top and middle panels) 2 µm.

Figure 6.

Rpgrip1l forms a complex with dishevelled, inversin, and nephrocystin-4. (A) Western blot analysis and quantification of endogenous levels of Dvl3 in control and RPGRIP1L-depleted MDCK cells. The reduction observed in RPGRIP1L-KD was blocked by the proteasome inhibitor clasto-lactacystin β-lactone (cL). Western blot was reprobed for α-tubulin as a loading control (n = 3). (B) Western blot analysis and quantification of Dvl2-Myc expression levels. HEK293T cells were transiently transfected with a Dvl2-Myc expression construct alone or together with either RPGRIP1L-Myc or Inv-Flag expression constructs (n = 4). *, P < 0.03; **, P < 0.002; Mann Whitney test. (C and D) Dvl2-Myc and RPGRIP1L-Flag were coexpressed with inv-GFP (C) or V5-tagged NPHP4 (D, NPHP4-V5) in HEK293T cells. After immunoprecipitation with anti-GFP (C) or anti-V5 (D), both Dvl2-Myc and RPGRIP1L-Flag were specifically detected in immunoprecipitates (C and D, left panels). Expression level of transfected proteins in cell lysates was confirmed by immunoblotting with appropriate antibodies (C and D, right panels). (E) Western blot analysis and quantification of Dvl2-Myc expression levels. HEK293T cells were transiently transfected with a Dvl2-Myc expression construct alone or together with RPGRIP1L-Flag or RPGRIP1L-Flag with patient mutation (RPGRIP1L-T615P-Flag or RPGRIP1L-T677I-Flag). While RPGRIP1L-Flag construct stabilized Dvl2-Myc expression, constructs with patient mutation had a reduced efficiency (left panels). Then HEK293T cells were transiently transfected with a Dvl2-Myc expression construct together with either RPGRIP1L-Flag or RPGRIP1L-Flag with patient mutation and NPHP4-V5 (middle panels) or inv-GFP (right panels). Dvl2-Myc was not as efficiently protected from degradation when coexpressed with RPGRIP1L-Flag patient mutation constructs than the wild-type RPGRIP1L construct (n = 3).