Fig 2: CEP97-Dyrk1a modulates centriole disengagement. (A) The subcellular localization of CEP97 to deuterosomes. The mRNAs of CEP97-HA and GFP-Ccdc78 (deuterosome marker) were injected into two ventral blastomeres at the eight-cell stage. Embryos were fixed at stage 19 and transverse sectioned, followed by immunofluorescence with HA probe (red), anti-GFP (cyan), and anti-centrin (magenta) antibodies. Images were generated by 3D-SIM. Scale bar, 1 µm. (B) Defective centriole disengagement upon knockdown of CEP97 and Dyrk1a. The indicated combinations of the mRNAs and MOs were injected into one blastomere of eight-cell stage embryos and fixed at stage 23. Cent-RFP marks basal bodies (red). GFP-Ccdc78 marks deuterosomes (cyan). Cilia are visualized by acetylated tubulin antibody staining. Images in white dotted square are magnified to the right. Rightmost images are 3D projections. Scale bars, large images = 20 µm; small images = 5 µm. Con, control. (C) Quantification of completed centriole disengagement in MCCs in B. Cent-RFP expressing MCCs in a field were considered for quantification. Graph is mean percentage ± SD of MCCs per image. Image numbers for quantification, n = 30; embryos per group from three independent experiments for each condition, n = 25; ***, P < 0.001; ****, P < 0.0001; one-way ANOVA.

Fig 3: CEP97 and Dyrk1a are required for ciliation in MCCs, and CEP97 is required in GRP. (A and C) A cocktail of the indicated mRNAs and MOs was injected into one ventral blastomere of four-cell stage embryos and fixed at stage 27. Acetylated tubulin staining (magenta) marks cilia. Mem-GFP (green) was used as a tracer. Scale bars, 50 µm. Con, control. (B) Quantification of relative acetylated tubulin intensity in A (image numbers for analysis, n = 30; embryos per group from three independent experiments for each condition, n = 16). ****, P < 0.0001; one-way ANOVA. Error bars indicate ±SD. (D) Quantification of relative acetylated tubulin intensity in C (image number for analysis, n = 25; embryos per group from three independent experiments, n = 14). ****, P < 0.0001; one-way ANOVA. Error bars indicate ±SD. (E and F) Quantification of the total percentage of MCCs in A and C. No significant difference is observed in the percentage of total MCCs in CEP97 and Dyrk1a morphants. Image numbers for analysis, n = 25; one-way ANOVA. (G) TEM of multicilia in control, CEP97, and Dyrk1a morphant MCCs. Cross-sectioned TEM images do not show any structural defects in axonemes on CEP97 and Dyrk1a knockdown. Scale bars, 100 nm. (H) CEP97 knockdown, but not Dyrk1a knockdown, decreases the length of GRP cilia. A cocktail of the indicated mRNAs and MOs was injected into two dorsal blastomeres of eight-cell stage embryos and fixed at stage 16. Acetylated tubulin staining (magenta) marks GRP cilia. Mem-GFP (green) was used as a tracer. Scale bars, 50 µm. (I) Quantification of GRP cilia length in H. Measured cilia number per group, n > 800; dissected GRPs per group from three independent experiments, n = 20. ****, P < 0.0001; one-way ANOVA. Error bars indicate ±SD. A.U., arbitrary units.

Fig 4: CEP97 forms a complex with Dyrk1a and Plk1. (A) The subcellular colocalization of CEP97 and Plk1 in migrating MCCs. Embryos injected with the indicated mRNAs were fixed at stage 19 and transversely sectioned. CEP97-mCherry, magenta; Plk1-GFP, green. DAPI was used to stain nuclei. Scale bars, 5 µm. (B) The phosphorylation of CEP97 is required for CEP97-Plk1 interaction. The embryos were injected with the indicated mRNAs, and reciprocal co-IPs were performed. (C) CEP97 mutant lacking aa 541–565 does not associate with Plk1. Reciprocal co-IPs were performed with embryo lysates as indicated. (D) Schematic diagram of Xenopus Plk1 deletion mutants. (E and F) PB domain is required for CEP97-Plk1 binding. CEP97 WT (E), but not CEP97 SA (F), associates with WT Plk1 and PB alone. The indicated mRNAs were injected into one-cell stage embryos, and reciprocal co-IPs were performed. (G) CEP97 mediates Dyrk1a-Plk1 interaction. Embryos injected with the indicated MOs or mRNA were lysed at stage 16 for co-IPs. (H) Quantification of co-IP protein bands in G. Co-IP protein band density was normalized to input. Graph was generated from three independent reciprocal co-IPs. ***, P < 0.001; one-way ANOVA. Data are mean ± SD. (I) The His region of Dyrk1a is necessary for Dyrk1a–Plk1 interaction. The embryo lysates expressing the indicated exogenous proteins were used to perform co-IPs. Numbers marked on left side of Western blot images represent protein molecular weight (kilodaltons).

Fig 5: Human Separase expression restores defective centriole disengagement in CEP97 and Dyrk1a morphant MCCs. (A) The embryos injected with the indicated mRNAs and MOs were fixed at stage 23, and then immunofluorescence was performed. Cent-RFP (red) marks basal bodies. GFP-Ccdc78 (cyan) marks deuterosomes. Multicilia are visualized by antiacetylated tubulin antibody. Images in dotted square are magnified to the right. Rightmost images are 3D projections. Scale bars, large images = 20 µm; small images = 5 µm. Con, control; sepa, Separase. (B) Quantification of the completed centriole disengagement in A. Cent-RFP expressing MCCs in a field were considered for quantification. Bar graphs display the mean percentage of disengaged MCCs ± SD. Number of field images for measurement per group, n = 26; embryos per group, n = 18; ****, P < 0.0001; ***, P < 0.0001; one-way ANOVA. (C) Working model. The phosphorylation of CEP97 by Dyrk1a modulates the affinity of CEP97 for Plk1, which leads to mature centriole disengagement, basal body migration, and docking during ciliogenesis in MCCs. Centrioles in red, cilia in orange, actin meshwork in light green, and deuterosome in green.