Foxn4 promotes gene expression required for the formation of multiple motile cilia

Abstract

Multiciliated cell (MCC) differentiation involves extensive organelle biogenesis required to extend hundreds of motile cilia. Key transcriptional regulators known to drive the gene expression required for this organelle biogenesis are activated by the related coiled-coil proteins Multicilin and Gemc1. Here we identify foxn4 as a new downstream target of Multicilin required for MCC differentiation in Xenopus skin. When Foxn4 activity is inhibited in Xenopus embryos, MCCs show transient ciliogenesis defects similar to those seen in mutants of Foxj1, a known key regulator of genes required for motile ciliation. RNAseq analysis indicates that Foxn4 co-activates some Foxj1 target genes strongly and many Foxj1 targets weakly. ChIPseq suggests that whereas Foxn4 and Foxj1 frequently bind to different targets at distal enhancers, they largely bind together at MCC gene promoters. Consistent with this co-regulation, cilia extension by MCCs is more severely compromised in foxn4 and foxj1 double mutants than in single mutants. In contrast to Foxj1, Foxn4 is not required to extend a single motile cilium by cells involved in left-right patterning. These results indicate that Foxn4 complements Foxj1 transcriptionally during MCC differentiation, thereby shaping the levels of gene expression required for the timely and complete biogenesis of multiple motile cilia.

Keywords: Cilia; Foxj1; Foxn4; Multiciliate cells; Xenopus laevis.

Fig. 1.

MCC differentiation is disabled in Foxn4 morphants. (A-C) Xenopus embryos were injected at the two-cell stage with a Foxn4 morpholino, followed with RNA encoding mRFP (blue) and Hysl1-GFP (green) to label membranes and centrioles, respectively. At stage 26, embryos were fixed and cilia (red) stained with an acetylated tubulin antibody. Shown are representative confocal images of the skin in a wild-type control (A) and Foxn4 morphant (B) embryo, along with the percentage of MCCs (±s.d.) displaying defective cilia based on scoring eight fields from different embryos. Scale bars: 10 μm. (C) Plot showing the average number of MCCs (based on cilia extension), outer cells (OC) and putative ionocytes (Inc) in a 0.13×0.13 mm field of the skin, based on data from eight or nine stage 26 embryos. Error bars indicate s.d. (D-G) Confocal images of the skin in control (D,F) or Foxn4 morphants (E,G) at stage 26, where basal bodies are marked by expression of Centrin4-GFP (green, D,E) or Chibby-GFP (green, F,G) and cell boundaries with mRFP. (D-G′) Below each image is a z-scan, where the apical surface and basal body positions in wild-type embryos are marked by the upper and lower dashed lines, respectively. (H,I) Confocal images of the skin in Foxn4 morphants, where cell boundaries are labeled with mRFP (blue), basal bodies with Centrin4-GFP (green), and cilia by acetylated tubulin immunostaining (red), fixed at stage 26 (H) and stage 30 (I), along with the percentage of MCCs (±s.d.) with undocked basal bodies or defective cilia extension obtained by scoring at least eight fields from different embryos. In wild-type controls, MCCs with these phenotypes were rarely observed. (J) Scatter plot of basal body number located within 1 μm of the apical surface in wild-type MCCs or those in Foxn4 morphants at the indicated stages. The central bar indicates the mean and error bars indicate s.d. ***P<0.001; n.s., not significant.

Fig. 2.

foxn4 Cas9/CRISPR phenotypes. (A-C) Representative confocal images of the skin in stage 26 wild-type control embryos (A) or embryos injected with Cas9 along with two independent gRNAs directed against foxn4 (B,C). Cell membranes (mRFP, blue), basal bodies (Centrin4-GFP, green) and cilia (acetylated tubulin, red) are labeled as in Fig. 1. (D,E) Representative confocal images of the skin of embryos at stage 26 or stage 30, following injection with Cas9 protein and Foxn4^gRNA1. Cell boundaries (blue), basal bodies (green) and cilia (red) are labeled as above. (A-E) The percentage of MCCs (±s.d.) with defective cilia or undocked basal bodies is indicated based on scoring at least 12 fields from different embryos (n≥50 cells). (F) Scatter plot of basal body number located within 1 μm of the apical surface in MCCs of wild-type controls (blue, Cont) and Cas9/Foxn4^gRNA1-injected embryos (red, G1) at the indicated stages. The central bar indicates the mean and error bars indicate s.d. ***P<0.001; n.s., not significant.

Fig. 3.

Similar functions of Foxj1 and Foxn4 downstream of Multicilin. (A,B) Representative confocal images of the skin of stage 30 wild-type control (A) or Cas9/Foxj1^gRNA1-injected (B) embryos, where cell membranes (mRFP, blue), basal bodies (Centrin4-GFP, green) and cilia (acetylated tubulin, red) are labeled. The percentage of MCCs with cilia extension defects is indicated (±s.d.) based on scoring 12 fields from different embryos (n≥50 total cells). (C) The average number of MCCs, outer cells and putative ionocytes within a 0.085×0.085 mm field based on data from ≥12 embryos. Error bars indicate s.d. *P<0.05. (D) The percentage of MCCs extending >50, 10-50, 1-10 and 0 cilia in control and Cas9/Foxj1^gRNA1-injected embryos based on scoring ≥147 cells from eight to ten embryos. Error bars indicate s.d. n.d., not detected. (E-G) Confocal images of the skin of wild-type (E), Cas9/Foxj1^gRNA1-injected (F) or Cas9/Foxn4^gRNA1-injected (G) embryos, further injected with RNAs encoding Multicilin-HGR, mRFP (blue) or Centrin4-GFP (green). At stage 11.5, Multicilin activity was induced by treatment with Dex, and embryos were fixed at stage 26 and stained for cilia (red). The percentage of MCCs with cilia extension defects is indicated (±s.d.). (H) Scatter plot showing basal body number located within 1 μm of the apical surface of MCCs induced by Multicilin, in wild-type control (Cont), Cas9/Foxj1^gRNA1-injected (Foxj1) or Cas9/Foxn4^gRNA1-injected (Foxn4) embryos based on scoring 12 fields from different embryos. Bars indicate the mean and s.d.

Fig. 4.

RNAseq analysis of foxn4 and foxj1 mutant phenotypes. (A,B) The foxn4 and foxj1 sequences targeted by the Foxn4 and Foxj1 gRNAs, on chromosomes Chr1l and Ch9_10L, respectively. The sequence and location of the gRNAs are indicated, along with the coding frame, and location of the forkhead domain (shaded in red). Beneath are 20 randomly chosen sequence reads that map to these regions in an RNAseq analysis of progenitors from embryos injected with Cas9 protein and Foxn4^gRNA1 (A) or Foxj1^gRNA1 (B). (C) The total wild-type and mutant sequence reads observed at the two homeologs of the foxn4 and foxj1 genes in replicate RNAseq analysis of embryos injected with Cas9 protein and Foxn4^gRNA1, Foxj1^gRNA1, or the Foxn4 morpholino as a control. (D) Scatter plot of genes based on a log₂-fold change in expression (P<0.05) in RNAseq analysis of progenitors induced to undergo MCC differentiation with Multicilin, in the presence of a Foxn4 morpholino, or Cas9/Foxn4^gRNA1. Points in red are genes where Foxn4 binds directly within 1 kb of the TSS, based on ChIPseq analysis. (E-G) Scatter plots of genes based on log₂-fold change in expression (P<0.05) in RNAseq analysis of progenitors induced to under MCC differentiation by Multicilin, in the presence and absence of E2f4ΔCT to disable the EDM complex (F,G), with Cas9/Foxj1^gRNA1 to mutate foxj1 (E,F) or with Cas9/Foxn4^gRNA1 to mutate foxn4 (E,G). All genes changes with P<0.05 are indicated in gray, MCC core genes defined in Quigley and Kintner (2016 preprint) are in blue, and genes associated with centriole biogenesis are in red (Ma et al., 2014).

Fig. 5.

Foxn4 binding in MCC promoters. ChIPseq analysis was carried out on a GFP-tagged form of Foxn4 expressed in epithelial progenitors along with Multicilin to induce MCC differentiation. Sequence reads obtained in this analysis were compared with that previously obtained for Rfx2 (Chung et al., 2014), E2f4 (Ma et al., 2014) or Foxj1 and various chromatin marks (Quigley and Kintner, 2016 preprint). (A) Genome browser screenshot of the rfx2 promoter along with tag counts obtained in ChIPseq analysis of Foxn4, Rfx2, Foxj1 and chromatin marks as indicated. (B) The binding of different MCC transcription factors to the promoters of 950 genes that are markedly upregulated during MCC differentiation, based on previous extensive RNAseq analyses of skin progenitors (Quigley and Kintner, 2016 preprint). (C) Tag counts of Foxn4 and Foxj1 ChIPseq at all sites (gray) and all core MCC promoters (red) bound by either.

Fig. 6.

MCC differentiation in foxn4 and foxj1 double mutants. (A-D) Shown are representative confocal images of the skin in embryos that were injected with Cas9/Foxj1^gRNA1 (B), Cas9/Foxn4^gRNA1 (C) or both (D), or left uninjected as control (A). Cell membranes are labeled with mRFP (blue), basal bodies with Centrin4-GFP (green) and cilia with acetylated tubulin antibody (red). (A-D′) z-sections, with dashed lines indicating the apical location of docked basal bodies in control MCCs. Scale bars: 10 μm. (E) The frequency of all MCCs (scored based on centriole expansion, docked or undocked) and outer cells in the skin of embryos injected with Cas9 and gRNAs as indicated. No values differ significantly from control embryos. (F) MCCs in embryos injected with Cas9 and gRNAs as indicated were scored based on wild-type cilia extension (as in A), reduced cilia (as in B) or absent cilia (as in D). (E,F) Data are based on >19 randomly chosen fields (0.13×0.13 mm) from different embryos. n.d., not detected. Error bars indicate s.d.

Fig. 7.

GRP cilia formation in foxn4 and foxj1 mutant embryos. (A-C) Embryos were injected with Foxj1^gRNA1 or Foxn4^gRNA1 along with Cas9 protein, followed by RNA encoding mRFP (blue) or Cep152-GFP (green) to label membranes and centrioles, respectively. At stage 17/18, the dorsal tissue was dissected away to expose the GRP, fixed, stained for cilia (red) and imaged by confocal microscopy. Typical confocal images are shown of GRP cells in control (A), Cas9/Foxj1^gRNA1-injected (B) or Cas9/Foxn4^gRNA1-injected (C) embryos. Anterior is up. (D) Scatter plot showing length of GRP cilia in cells from individual embryos injected as indicated. Mean and s.d. are indicated. (E) Cilia positioning in cells along the anteroposterior axis in the GRP was measured in embryos injected as indicated. Each bar shows data from individual GRPs based on data obtained from 10-30 cells, with the percentage of cells with cilia located in the anterior, middle or posterior third indicated. Most GRP cells have a posteriorly localized cilium in control (69±11%) and Cas9/Foxn4^gRNA1-injected (69±9%) embryos, but not in Cas9/Foxj1^gRNA1-injected embryos (29±9%; P=9×10⁻⁷).