WNT16b promotes the proliferation and self-renewal of human limbal epithelial stem/progenitor cells via activating the calcium/calcineurin A/NFATC2 pathway

Abstract

Our previous finding revealed that WNT16b promoted the proliferation of human limbal epithelial stem cells (hLESCs) through a β-catenin independent pathway. Here, we aimed to explore its underlying molecular mechanism and evaluate its potential in the treatment of limbal stem cell deficiency (LSCD). Based on the findings of mRNA-sequencing, the expression of key molecules in WNT/calcineurin A/NFATC2 signalling pathway was investigated in WNT16b-co-incubated hLESCs and control hLESCs. An epithelial wound healing model was established on Wnt16b-KO mice to confirm the regulatory effect of WNT16b in vivo. The therapeutic potential of WNT16b-co-incubated hLESCs was also evaluated in mice with LSCD. Our findings showed that WNT16b bound with Frizzled7, promoted the release of Ca²⁺ and activated calcineurin A and NFATC2. With the translocation of NFATC2 into cell nucleus and the activation of HDAC3, WDR5 and GCN5L2, the expression of H3K4me3, H3K14ac and H3K27ac in the promoter regions of FoxM1 and c-MYC increased, which led to hLESC proliferation. The effect of the WNT16b/calcium/calcineurin A/NFATC2 pathway on LESC homeostasis maintenance and corneal epithelial repair was confirmed in Wnt16b-KO mice. Moreover, WNT16b-coincubated hLESCs could reconstruct a stable ocular surface and inhibit corneal neovascularization in mice with LSCD. In conclusion, WNT16b enhances the proliferation and maintains the stemness of hLESCs by activating the non-canonical calcium/calcineurin A/NFATC2 pathway in vitro and in vivo, and accelerates corneal epithelial wound healing.

FIGURE 1

The non‐canonical WNT‐Ca²⁺ pathway is involved in human limbal epithelial stem cell (hLESC) proliferation induced by WNT16b. The mRNA‐seq assay showed that WNT/Ca²⁺ pathway‐related genes were significantly enriched after treatment with WNT16b, as shown by Kyoto Encyclopedia of Genes and Genomes (KEGG) (A), heatmap (B) and Gene Ontology (GO) (C) enrichment analyses. The fluorescence intensity of calcium was significantly higher in WNT16b‐treated hLESCs (p = 0.006) (D). Western blotting showed increased expression of pPLCβ3 after WNT16b treatment (p = 0.003), although the level of PLCβ3 was unchanged (E), confirming the mobilization of intracellular calcium. ** p < 0.01.

FIGURE 2

WNT16b activates the Ca²⁺/calcineurin A/NFATC2 signalling pathway. The mRNA levels of the genes encoding the catalytic subunits of calcineurin, PPP3CA, PPP3CB and PPP3CC and the isoforms of NFAT (NFATC1 and NFATC2) were significantly increased in WNT16b‐treated human limbal epithelial stem cells (hLESCs) (PPP3CA: p < 0.001; PPP3CB: p = 0.003; PPP3CC: p = 0.002; NFATC1: p = 0.003; NFATC2: p = 0.001) (A). Nevertheless, WNT16b treatment did not affect the expression of NLK (p = 0.29), PRKCA (p = 0.63), PPP3R1 (p = 0.20) or NFATC3 (p = 0.65). A reduction of only the mRNA level of NFATC4 was found (p = 0.004), and the expression of CaMK2A was not detected. Although the expression of CDC42 increased at the mRNA level (p = 0.002), it did not show a significant change at the protein level (p = 0.16) (B). The protein expression of other key molecules in the PKC/CaMKII pathway, such as p‐PKC pan, pCaMKII, was also unchanged (p = 0.84, 0.81, respectively). In contrast, WNT16b treatment significantly increased the protein levels of calcineurin A (p = 0.045) and NFATC2 (p = 0.008) (C) and enhanced the phosphatase activity of calcineurin A (p = 0.002) (D). However, it had no effect on NFATC1 (p = 0.15). Moreover, activated NFATC2 translocated from the cytoplasm into the cell nuclei after WNT16b treatment, as evidenced by Western blotting (E) and immunofluorescence staining (F). Scale bars: 50 μm. *p<0.05, ** p < 0.01, ***p < 0.001.

FIGURE 3

WNT16b binds to FZD7 to activate the downstream calcineurin A/NFATC2 signalling pathway. A coimmunoprecipitation (Co‐IP) assay (A) showed that WNT16b directly bound to FZD7. Blocking FZD7 with FZ7‐21 not only prohibited the effect of WNT16b on human limbal epithelial stem cell (hLESC) proliferation (B, C) and the expression of stemness biomarkers (D) (cell density: p < 0.001; EdU⁺ cell: p = 0.003; Ki67⁺ cell: p = 0.02; ΔNp63α⁺ cell: p = 0.02) but also downregulated the levels of calcineurin A (p = 0.04) and NFATC2 (p = 0.03) (E) and inhibited intracellular Ca²⁺ release (F). Only the expression of p‐PLCβ3 was not affected (p = 0.2). Scale bars: 50 μm. A~E: *p < 0.05, ** p < 0.01, ***p < 0.001. *in (F) stands for the comparison between WNT16b group and control, # stands for the comparison between WNT16b group and WNT16b+Fz7‐21 group. * and #: p < 0.05, ** and ##: p < 0.01, *** and ###: p < 0.001.

FIGURE 4

Inhibition of the calcineurin A/NFATC2 signalling pathway attenuates the proliferation of human limbal epithelial stem cells (hLESCs) induced by WNT16b. EdU assays (A, B) showed that both 5 μM FK506 (A) and 1 μM VIVIT (B) attenuated hLESC proliferation induced by WNT16b (p = 0.02 and 0.001, respectively), which was further confirmed by a decrease in cell density (p = 0.002 and 0.005, respectively; C). RT–PCR (D, E) showed that the expression of PPP3CA (p < 0.001), PPP3CB (p = 0.004), PPP3CC (p < 0.001) and NFATC2 (p < 0.001) decreased with FK506 treatment (D), whereas VIVIT caused a reduced level of only NFATC2 (p < 0.001; E). Accordingly, Western blotting (F) revealed that FK506 inhibited the expression of both calcineurin A and NFATC2 (p = 0.001 and 0.007, respectively), while VIVIT only prohibited the expression of NFATC2 (calcineurin A: p = 0.15, NFATC2: p = 0.004). In addition, the increased expression of ΔNp63α and Ki67 in WNT16b‐treated hLESCs was suppressed by treatment with either FK506 (p = 0.05 and 0.02, respectively; G) or VIVIT (p = 0.02 and 0.007, respectively; H). Scale bars: 50 μm (B, H); 100 μm (A, G). *p < 0.05, ** p < 0.01, ***p < 0.001.

FIGURE 5

WNT16b modulates the epigenetic modifications on c‐MYC and FoxM1. Schematic diagram (A) showed the locations of the NFATC2 binding sites in the c‐MYC and FoxM1 promoters. The sequences at the most frequent binding sites were used for the ChIP–qPCR assay. WNT16b induced an enrichment of NFATC2 at the promoters of c‐MYC and FoxM1 (p < 0.001 and p = 0.034, respectively; B). The inhibition of NFATC2 with VIVIT treatment prohibited the expression of FoxM1 and c‐MYC (C). Coimmunoprecipitation (Co‐IP) assays (D‐F) showed upregulation of GCN5L2 (p = 0.005; D) and WDR5 (p = 0.01; F) and downregulation of HDAC3 (p = 0.03; D) after WNT16b treatment. Nevertheless, HDAC1, HDAC2 and EZH2 did not have significant changes compared with the control group (E). Elevated levels of H3K27ac, H3K14ac and H3K4me3 were also found in the nuclei of WNT16b‐treated hLESCs (G). ChIP–qPCR (H–J) revealed the enrichment of H3K4me3 (J), H3K14ac (I) and H3K27ac (H) at the promoter loci of FoxM1 (H3K4me3: p = 0.021, H3K14ac: p = 0.006; H3K27ac: p = 0.04) and the enrichment of H3K4me3 in the promoter regions of c‐MYC (p = 0.002; J). *p < 0.05, ** p < 0.01, ***p < 0.001

FIGURE 6

WNT16b promotes corneal wound healing in Wnt16b‐KO mice. The expression of P63α and ABCG2 in the limbal epithelium (A–B) was significantly decreased in Wnt16b‐KO mice. Moreover, both qRT–PCR (C) and Western blotting (D) showed downregulated expression of FZD7, calcineurin A, NFATC2, c‐MYC and FoxM1 (qRT–PCR: Ppp3ca: p = 0.002, Nfatc2: p = 0.02, c‐Myc: p = 0.006, Foxm1: p = 0.01; WB: calcineurin A: p = 0.007, NFATC2: p = 0.01, FZD7: p = 0.004; FoxM1: p = 0.0001, c‐MYC: p = 0.009). Evaluation of the epithelial defect sizes showed that Wnt16b‐KO mice had a slower epithelial healing rate than WT mice (24 h: p = 0.007, 36 h: p = 0.002, 48 h: p = 0.021). Topical application of WNT16b significantly accelerated epithelial healing in both WT mice (36 h: p = 0.005, 48 h: p = 0.016) and Wnt16b‐KO mice (24 h: p = 0.02, 36 h: p = 0.005, 48 h: p = 0.022) (E). KO: WNT16b‐knockout mice; WT: wild‐type mice. *,# and $ stand for the subgroup comparison between WT and KO, KO and KO+WNT16b, and WT and WT+WNT16b, respectively. *,#,$: p < 0.05, **,##, $$: p < 0.01, ***: p < 0.001.

FIGURE 7

Evaluation of ocular surface reconstruction in total limbal stem cell deficiency (LSCD) eyes treated with or without human limbal epithelial stem cells (hLESCs) transplantation. Representative slitlamp images of the ocular surface on 0, 2, 4 and 10 weeks (A) revealed a complete epithelium recovery after the treatment of WNT16b‐hLESCs (n = 6) and Cntl‐hLESCs (n = 6) for 10 weeks. Fluorescein staining was present in eyes without hLESCs transplantation (n = 6). Among three groups, the lowest neovascularization score was found in WNT16b‐hLESCs‐treated group at 10 weeks after the treatment (p = 0.006 and 0.042; B). The corneal opacity scores of WNT16b‐hLESCs‐ and Cntl‐hLESCs‐treated groups were similar, both of which were significantly lower than that in Non‐hLESCs group (p = 0.026 and 0.017; C). The corneal epithelial cells in WNT16b‐hLESCs‐treated group were recovered to 4–5 layers but slightly irregularly arranged as evidenced by histopathological examination (D). Moreover, the stroma oedema completely vanished and little neovascularization was observed in WNT16b‐hLESCs‐treated group. Compared to WNT16b‐hLESCs‐treated group, stromal neovascularization was still found in Cntl‐hLESCs‐treated group. Immunofluorescence staining of K14 and α‐SMA (E) showed that human derived K14⁺ cells were identified on the corneal surface of WNT16b‐hLESCs and Cntl‐hLESCs‐treated groups. No human K14⁺ cells were found in either LSCD eyes without hLESCs transplantation or normal mice eyes. The expression level of α‐SMA in WNT16b‐hLESCs and Cntl‐hLESCs‐treated groups was significantly lower than that in Non‐LESCs treated group. *The comparison between Cntl‐hLESCs and Non‐hLESCs‐treated groups. *p < 0.05, **p < 0.01. #, the comparison between WNT16b‐hLESCs‐ and Non‐hLESCs‐treated groups. #p<0.05，##p<0.01.

FIGURE 8

Schematic diagram of the molecular mechanism of the WNT16b/FZD7/calcineurin A/NFATC2 pathway in WNT16b‐treated hLESCs. With binding to FZD7, WNT16b causes the phosphorylation of PLCβ3, mobilizes the release of Ca²⁺ into the cytoplasm and activates calcineurin A. NFATC2 is dephosphorylated by calcineurin A and translocates into the cell nucleus and then binds to the promotors of c‐MYC and FoxM1. With the recruitment of GCN5L2, WDR5 and HDAC3, the enrichment of H3K27ac, H3K14ac and H3K4me3 occurs at the promoter area of FoxM1 and/or c‐MYC, which leads to the proliferation of hLESCs.