The interplay between Wnt and mTOR signaling modulates ciliogenesis in human retinal epithelial cells

Abstract

The primary cilium is a microtubule-based organelle essential for various cellular functions, particularly signal transduction. While the role of cilia in regulating signaling pathways has been extensively studied, the impact of signaling pathways on cilia formation remains less well understood. Wnt signals are critical modulators of cell fate. In this study, we investigate how modulating Wnt signaling affects cilia formation in human retinal pigment epithelial (hTERT-RPE1) cells. Our findings show that enhancement of Wnt/LRP6 signaling before serum starvation delays ciliogenesis. Cells with high baseline Wnt activity exhibited distal appendage dysregulation, failure to remove CP110-CEP97 from mother centrioles, and reduced Rab8-vesicle docking, which are critical events for cilia membrane establishment and axoneme extension. Additionally, these cells displayed reduced autophagic flux, increased mTOR kinase activity, and elevated OFD1 levels at centriolar satellites. Importantly, mTOR inhibition rescued ciliogenesis in cells with elevated Wnt activity, underscoring the interplay between these signaling pathways. Our data also indicate that insufficient Wnt signaling activation disrupts ciliogenesis, emphasizing the need for precisely regulated Wnt levels.

Fig 1. Changes in baseline Wnt activity impair cilia formation in RPE1 cells.

(A, B) Quantification of ciliation of RPE1 cells treated with control (siLUC) and LRP6 (A) or β-catenin (B, CTNNB1) siRNA and serum starved for 24 h prior to fixation and staining with γ-tubulin (basal body marker) and ARL13b (cilia membrane marker). The bar graph indicates mean ± S.D. from three independent experiments. (A) siLUC, n = 561; siLRP6, n = 339; (B) siLUC, n = 312; siCTNNB1, n = 647. (C) Experimental scheme for cilia induction and treatment of cells with Wnt agonist. (D) Western blot analysis of saponin lysed RPE1 cells treated with Co-CM or Wnt3a-CM (as depicted in C) and serum-starved for 16 h. Actin served as a loading control. (E) RPE1 cells were treated as described in (C) and subjected to immunofluorescence analysis 16 h after serum starvation. The bar graph indicates the percentages (mean ± S.D.) of ciliated cells treated with Co-CM and Wnt3a-CM from three independent experiments. Serum starved cells without treatment (serum free) were used as a control. The images show representative ciliated and non-ciliated cells with enlarged views of the centrosomal area. Basal bodies (γ-tubulin, red), ciliary membrane (ARL13b, green) and DNA (DAPI staining, blue) are shown as single channels or merged. Scale bar: 5 μm. (F) RPE1 cells were treated as described in (C) and stained with Rb and phospho-Rb (pRb; S807/S811) antibodies 16 h after serum-starvation. Representative images and quantifications of the percentage of non-proliferating cells (pRb negative) are shown. Co-CM, n = 638; Wnt3a-CM, n = 876. The bar graph indicates the mean ± S.D. from three independent experiments. Scale bar: 10 μm. (G) Quantification of ciliation in RPE1 cells treated with buffer control (n = 561) or purified surrogate Wnt agonist (n = 639) after 16 h of serum starvation. The bar graph indicates the mean ± S.D. from three independent experiments. (H) Quantification of ciliation after 16 h of serum starvation in RPE1 7xTGC cells expressing DKK1 alone or in combination with WNT1. Empty plasmids were used as a control. The bar graph indicates the mean ± S.D. from three independent experiments. Control, n = 559; DKK1, n = 597; Wnt1, n = 553; Wnt1 + DKK1, n = 612. P values are based on Student t test. The data underlying the graphs and blots in this figure can be found in the S1 Data and S1 Raw Images files.

Fig 2. Increasing baseline Wnt activity delays cilia formation independently of TCF7.

(A) Immunofluorescence images showing the localization of TCF7 (green) in Co-CM and Wnt3a-CM treated RPE1 cells as depicted in 1C. γ-tubulin (red) and DAPI (blue) serve as markers for centrosomes and nuclei, respectively. Scale bar: 5 μm. (B, C) RPE1 cells incubated with control (siLUC) or TCF7 siRNAs were treated with Co-CM or Wnt3a-CM as depicted in 1C and analyzed 16 h after serum starvation. (B) Western blot analysis of saponin-lysed RPE1 cells using DVL2 and TCF7 antibodies. Actin served as a loading control. The asterisk marks the phosphorylated form of DVL2. (C) Quantification of the percentage of ciliated cells based on γ-tubulin and ARL13b staining. The bar graph indicates the mean ± S.D. of three independent experiments. siLUC + Co-CM, n = 338; siLUC + Wnt3a-CM, n = 325; siTCF7 + Co-CM, n = 336; siTCF7 + Wnt3a-CM, n = 322. (D) Quantification of ciliary length from (C). The box/dot plots show quantification of ciliary length from three independent experiments. siLUC + Co-CM, n = 104; siLUC + Wnt3a-CM, n = 93; siTCF7 + Co-CM, n = 107; siTCF7 + Wnt3a-CM, n = 100. P values are based on Student t test. The data underlying the graphs and blots in this figure can be found in the S1 Data and S1 Raw Images files.

Fig 3. Wnt signaling decreases the recruitment of RAB8a to basal bodies.

(A) RPE1 cells were serum starved for 16 h in the presence of solvent control (DMSO), BIO or CHIR99021. Cells were processed for immunofluorescence analysis using γ-tubulin and ARL13b antibodies. DNA was stained with DAPI. The bar graph shows the percentage of ciliated cells (mean ± S.D.) of three independent experiments. DMSO, n = 449; BIO, n = 497; CHIR99021, n = 391. (B) RPE1 cells stably expressing GFP-RAB8a were treated with CHIR99021 as described in (A). GFP-RAB8a was detected by direct fluorescence. The graph shows the percentage of non-ciliated cells with GFP-RAB8a docked at centrosomes (stained with γ-tubulin antibodies). DNA was stained with DAPI. The mean ± S.D. of three independent experiments is shown. Representative images with enlarged centrosomal areas (as indicated) show merged γ-tubulin and GFP-RAB8a (top panel), γ-tubulin (red, middle panel) and GFP-RAB8a (green, lower panel) in control (DMSO) and CHIR99021-treated cells. DMSO, n = 561; CHIR99021, n = 396. Scale bar: 5 µm. (C) Quantification of ciliary length from (A). Representative images show basal body (γ-tubulin, red) and ciliary membrane (ARLB13b, green) merged signals. The box/dot plots show quantification of ciliary length from three independent experiments. Control, n = 99; BIO, n = 68; CHIR99021, n = 99. Scale bar: 5 μm. (D) RPE1 cells stably expressing GFP-RAB8a were treated with Co-CM and Wnt3a-CM as depicted in 1C and fixed for GFP-RAB8a localization analysis as described in (B). Quantifications and representative images are shown. Enlarged images show merged γ-tubulin (red) and GFP-RAB8a (green) signals (top panel). DNA was stained with DAPI. Co-CM, n = 216; Wnt3a-CM, n = 306. Scale bar: 5 μm. (E) Cartoons representing daughter and mother centrioles with distal (DAs) and subdistal (SDAs) appendages and components analyzed. RAB8a-positive vesicles (yellow) dock at the mother centriole at initial steps of ciliogenesis prior to extension of the axoneme (green). Created using Adobe Illustrator CS3. (F) RPE1 cells were treated as depicted in 1C and stained for ODF2 (SDA component, green) and γ-tubulin (red). The box/dot plots show quantification of fluorescence intensities of ODF2 at the centrosomes from three independent experiments. Representative images are shown on the right. DNA was stained with DAPI (blue). Co-CM, n = 84; Wnt3a-CM, n = 97. Scale bar: 10 μm. (G, H) RPE1 cells were treated with DMSO control or CHIR99021 and serum starved for 16 h prior to immunofluorescence analysis for CEP164 and CHIBBY. The box/dot plots show quantification of fluorescence intensities of CEP164 (G) and CHIBBY (H) at the centrosomes from three independent experiments. Representative images with enlargements of the centrosomal area are shown on the right. DNA was stained with DAPI (blue). (G) Control, n = 110; CHIR99021, n = 111. (H) Control, n = 91; CHIR99021, n = 98. Scale bar: 10 μm. P values are based on Student t test. A.U., arbitrary units. The data underlying the graphs in this figure can be found in the S1 Data file.

Fig 4. Wnt signaling activation impairs the removal of the cilia inhibitory proteins CP110 and OFD1.

(A) RPE1 cells were treated as depicted in 1C and stained for CP110 (green), ODF2 (red) and γ-tubulin (gray). The graph shows the percentage of mother centrioles (marked by ODF2) with or without CP110. Mean ± S.D. of three independent experiments are shown. Representative images with enlargements of the centrosomal area as depicted are shown on the right. Co-CM, n = 455; Wnt3a-CM, n = 390. DNA was stained with DAPI (blue). Scale bar: 5 μm. (B) RPE1 cells were treated as depicted in 1C and stained for OFD1 (green) and γ-tubulin (red) after 16 h of serum starvation. Representative images show OFD1 localization in Co-CM or Wnt3a-CM treated cells. DNA was stained with DAPI (blue). Scale bar: 20 μm. (C) Quantification of the cytoplasmic centriolar satellite OFD1 signal of (B) in arbitrary units. Co-CM, n = 480; Wnt3a-CM, n = 430. (D) RPE1 cells incubated with control (siLUC) or CP110 siRNAs were treated with Co-CM or Wnt3a-CM as depicted in 1C and analyzed by immunofluorescence 16 h after serum starvation. The bar graph indicates the percentage of ciliated cells (mean ± S.D.) from three independent experiments. siLUC + Co-CM, n = 436; siLUC + Wnt3a-CM, n = 393; siCP110 + Co-CM, n = 235; siCP110 + Wnt3a-CM, n = 338. (E) Representative images of (D). Cells were stained with acetylated tubulin (Ace-tubulin, gray), ODF2 (mother centriole marker, red), CP110 (green). Enlargements are shown on the right. Scale bar: 5 μm. (F) RPE1 cells incubated with control (siLUC) or OFD1 siRNAs were treated with Co-CM or Wnt3a-CM as depicted in 1C and analyzed by immunofluorescence 16 h after serum starvation. The bar graph shows the percentage of ciliated cells (mean ± S.D.) from three independent experiments. siLUC + Co-CM, n = 690; siLUC + Wnt3a-CM, n = 677; siOFD1 + Co-CM, n = 521; siOFD1 + Wnt3a-CM, n = 509. (G) Western blot analysis of (F) showing the levels of phospho-LRP6 (p-LRP6), OFD1 (marked by an asterisk) and TCF7 in the indicated samples serum starved for 16 h. Actin served as a loading control. P values are based on Student t test. The data underlying the graphs and blots in this figure can be found in the S1 Data and S1 Raw Images files.

Fig 5. Inhibition of mTOR signaling rescues cilia formation in Wnt activated cells.

(A, B) RPE1 cells stably expressing mCherry-GFP-LC3 were treated with Co-CM and Wnt3a-CM as depicted in 1C and fixed 16 h after serum starvation for direct fluorescence analysis of mCherry and GFP signals. Representative images (A) and quantification of the number of GFP⁺mCherry⁺ foci (yellow dots, autophagosomes) (B) from three independent experiments are shown. Co-CM, n = 123; Wnt3a-CM, n = 113. Scale bar: 10 µm. (C) RPE1 cells were treated as depicted in 1C in the presence or absence of rapamycin and analyzed by western blot 16 h after serum starvation using phospho-S6K (T389) and S6K antibodies. Actin served as a loading control. DMSO was used as a solvent control. (D) RPE1 cells were treated as depicted in 1C in the presence of solvent control (DMSO) or rapamycin and analyzed by immunofluorescence 16 h after serum starvation. The bar graph shows the percentage of ciliated cells from three independent experiments. Co-CM+DMSO, n = 394; Co-CM+Rapamycin, n = 361; Wnt3a-CM+DMSO, n = 468; Wnt3a-CM+Rapamycin, n = 372. (E) RPE1 cells were treated with solvent control (DMSO) or CHIR99021 in the presence or absence of rapamycin as indicated and serum starved for 16 h. The percentage of ciliated cells was quantified based on ARL13b (cilia membrane marker) and γ-tubulin (basal body marker). The graph shows the mean ± S.D. from three independent experiments. DMSO, n = 484; DMSO+rapamycin, n = 588; CHIR99021 + DMSO, n = 478; CHIR99021 + Rapamycin, n = 476. P values are based on Student t test. (F) Model of how Wnt signaling affects ciliogenesis. Basal Wnt signaling promotes cilia formation, whereas Wnt hyperactivation prior to ciliogenesis delays this process by increasing mTORC1 activity and impairing the removal of OFD1 from centriolar satellites. M, mother centriole; D, daughter centriole. The cilia illustration was created using ChatGPT and Adobe Illustrator CS3. The data underlying the graphs and blots in this figure can be found in the S1 Data and S1 Raw Images files.