Switch-like enhancement of epithelial-mesenchymal transition by YAP through feedback regulation of WT1 and Rho-family GTPases

Abstract

Collective cell migration occurs in many patho-physiological states, including wound healing and invasive cancer growth. The integrity of the expanding epithelial sheets depends on extracellular cues, including cell-cell and cell-matrix interactions. We show that the nano-scale topography of the extracellular matrix underlying epithelial cell layers can strongly affect the speed and morphology of the fronts of the expanding sheet, triggering partial and complete epithelial-mesenchymal transitions (EMTs). We further demonstrate that this behavior depends on the mechano-sensitivity of the transcription regulator YAP and two new YAP-mediated cross-regulating feedback mechanisms: Wilms Tumor-1-YAP-mediated downregulation of E-cadherin, loosening cell-cell contacts, and YAP-TRIO-Merlin mediated regulation of Rho GTPase family proteins, enhancing cell migration. These YAP-dependent feedback loops result in a switch-like change in the signaling and the expression of EMT-related markers, leading to a robust enhancement in invasive cell spread, which may lead to a worsened clinical outcome in renal and other cancers.

Fig. 1

Anisotropic texture of the mechanical environment increases EMT. a Schematic depiction of the edge areas of the expanding epithelial sheets, including formation of the finger-like protrusions (FLPs) led by “leading” cells and occasional separation of the tip cells (“dissemination”). It also defines the distance metric for the cell migration analysis in this and subsequent figures. The inset shows the electron micrograph of the NRA. b Cell trajectories colored to reflect the cell speed values in an expanding sheet on flat surfaces (top) and NRA (bottom). The colors of each dot indicate the mean speed values of the corresponding cells (left). The trajectories were tracked for 6 h with the speed measured every 20 min (right). c Migration speed of individual cells on flat and NRA substrata at different initial distances from the edge of the sheet (corresponding to values of d in panel a). Each dot represents the average speed of an individual cell. Dashed lines indicate the averaged speed of isolated individual cells on a flat surface (red) and NRA (blue) (each number of independently analyzed cells, n, is indicated, # = statistical significance of speed on the marginal region vs. the most submarginal region of cells on a flat surface (red) and NRA (blue), ^##P < 5 × 10⁻⁴ and ^###P < 5 × 10⁻⁶, * = statistical significance of the speed of cells on a flat surface vs. NRA, and *P < 0.05, **P < 5 × 10⁻⁴, and ***P < 5 × 10⁻⁶). d Disseminations from epithelial sheets per unit length of epithelial sheets on flat substrata and NRA for 18 h (n = 4 biologically independent experiments, * = statistical significance of numbers of dissemination on flat surfaces vs. NRA. ***P < 5 × 10⁻⁶). e The expression of Snail in the cells on flat surfaces and NRA analyzed by immunoblotting (n = 3 biologically independent samples, * = statistical significance of Snail expression in cells on flat vs. NRA substrata, *P < 0.05). f Dissemination from epithelial sheets on flat substrata and NRA in the presence of TGFβ (n = 4 biologically independent samples, * = statistical significance of the number of disseminations with control vs. TGFβ on the flat substratum (red) and on NRA (blue), flat substrata vs. NRA with TGFβ (black), *P < 0.05 and **P < 1 × 10⁻²). All error bars are S.E.M and statistical significance was determined by two-sided Student’s t-test

Fig. 2

Switch-like regulation of YAP in expanding epithelial layers loosening cell–cell adhesion. a Cells on flat substrata and NRA analyzed by immunoblotting using YAP and phosphorylated YAP antibodies. b Cell migration speed of individual YAP^KD and YAP^OE cells in cell sheets as a function of the distance from the sheet edge on NRA (each number of independently analyzed cells, n, is indicated, * = statistical significance of speed of control vs. YAP^KD cells on NRA (green) and of control vs. YAP^OE cells on NRA (purple), *P < 0.05, **P < 5 × 10⁻⁴, and ***P < 5 × 10⁻⁶). c Immunofluorescence staining for YAP and active β-catenin in epithelial cell sheets on flat substrata and NRA. Translocation of YAP and active β-catenin into nuclei observed in marginal zones and FLPs of sheets expanding on NRA (brown boxes) and mostly cytoplasmic YAP and active β-catenin localization in submarginal cells on NRA (red boxes) and on flat substrata. The samples were analyzed after 8 h from the initiation of sheet expansion. d Fractions of nuclei displaying different intensities of YAP and active β-catenin staining as a function of the distance from the sheet edge determined at 8 h after initiation of sheet expansion (the edge is coincident with the concave regions at the bases of FLPs in panel c). e Velocity correlation length in y-direction, parallel to the direction of epithelial expansion for the control and YAP^KD cells on flat substrata and NRA (each number of independently analyzed cells, n, is indicated, * = statistical significance of correlation length, *P < 0.05, **P < 0.01, and n.s = no significance). f Cell migration speed of individual cells in cell sheets as a function of the distance from the sheet edge on flat substrata and NRA, in the presence of an E-cadherin functional blocking antibody. Dashed line with a square marker indicates the average cell migration speed of isolated control cells and solid line with a circle marker corresponds to that of isolated YAP^KD cells (each number of independently analyzed cells, n, is indicated, # = statistical significance of speed values in the marginal region vs. the most submarginal region of YAP^KD cells with drugs, ^###P < 5 × 10⁻⁶. * = statistical significance of control vs. YAP^KD cells with drugs (green) and of YAP^KD cells with vs. without drugs (black), *P < 0.05, **P < 5 × 10⁻⁴, and ***P < 5 × 10⁻⁶). g Control, YAP^KD, and YAP^OE cells were immunoblotted using active β-catenin and E-cadherin antibodies. All error bars are S.E.M and statistical significance was determined by two-sided Student’s t-test

Fig. 3

Similar subcellular localization of WT1 with YAP. a Immunofluorescence staining for WT1 in epithelial cell sheets on flat substrata and NRA. Translocation of WT1 into nuclei was observed in marginal zones and FLPs of sheets expanding on NRA (brown boxes). Submarginal cells on NRA (red boxes) and on flat substrata showed YAP in the cytoplasm. The samples were fixed after 8 h to remove stencils. b Control and YAP^KD cells and their nuclear fractions were analyzed using immunoblotting using YAP and WT1 antibodies. c Immunofluorescence staining for WT1 in cells and d immunoblotting of YAP and WT1 abundance in the nuclei in different density–epithelial cell sheets showing density-dependent translocation of WT1 into the nucleus

Fig. 4

YAP regulates E-cadherin through WT1 in epithelial layers on NRA. a Immunofluorescence staining for WT1 in YAP^KD cell sheets (top), and for YAP in WT1^KD cell sheets cultured on NRA. The samples were fixed after 8 h to remove stencils. b Co-IP analysis using the YAP antibody, followed by immunoblotting using the WT1 antibody. c Chromatin immunoprecipitation (ChIP) analysis of the WT1-YAP complex binding at the E-cadherin promoter (n = 3). We extracted cross-linked chromatin and immunoprecipitated it using antibodies against YAP, WT1, IgG (negative control), and RNAPII or histone H3 (see Supplementary Fig. 12a) antibodies (positive controls). The immunoprecipitated chromatin and input genomic DNA were used for amplification of the E-cadherin promoter, CTGF promoter (known to be regulated by YAP but not WT1, and used as another control), and GAPDH promoter (n = 4 biologically independent samples, * = statistical significance of PCR products from each sample vs. IgG, *P < 0.05 and **P < 0.01). d E-cadherin expressions of control and WT1^KD cells analyzed by immunoblotting with WT1 and E-cadherin antibodies (n = 3 biologically independent samples, * = statistical significance of E-cadherin expression of control and WT1^KD cells, ***P < 5 × 10⁻³). e Cell migration speed of individual cells in control and WT1^KD epithelial cells in the marginal region of the sheets in the presence of an E-cadherin blocking antibody (each number of independently analyzed cells, n, is indicated, * = statistical significance, n.s = no significance, and ***P < 5 × 10⁻³). f Dissemination of cells in control and WT1^KD epithelial sheets on NRA in the presence of an E-cadherin blocking antibody (n = 4 biologically independent experiments, * = statistical significance, *P < 0.05, ***P < 5 × 10⁻³, and n.s = no significance). g Immunoblotting of phosphorylated YAP and total YAP in cell sheets cultured in the presence of different concentrations of E-cadherin blocking antibody (n = 3 biologically independent samples). h Schematic of the YAP-mediated cell dissemination triggered by mechanical cues stemming from NRA, operating through E-cadherin control by YAP-WT1 complexes and leading to cell dissemination in epithelial cell sheets on NRA. All error bars are S.E.M and statistical significance was determined by two-sided Student’s t-test

Fig. 5

Cell migration-enhancing effect of YAP via Rac1. a Rac1 activity assay using cells cultured on flat surfaces and NRA (n = 3 biologically independent samples, * = statistical significance of the ratio of active Rac1 to Rac1 on flat surfaces vs. NRA. *P < 0.05). b Control, YAP^KD, and YAP^OE cells were immunoblotted to evaluate Rac1 activity via pull-down assay, expression of TRIO, and phosphorylation of PAK and Merlin. c Cell migration speed of individual cells in YAP^KD (left) and YAP^OE (right) cells in the marginal region of epithelial cell sheets on NRA in the presence of Rac1 inhibitor, NSC23766 and a TRIO inhibitor, ITX3 (each number of independently analyzed cells, n, is indicated, * = statistical significance of control vs. YAP^KD and control vs. YAP^OE cells with drugs, n.s = no significance, *P < 0.05, **P < 1 × 10⁻², and ***P < 5 × 10⁻³). All error bars are S.E.M and statistical significance was determined by two-sided Student’s t-test

Fig. 6

Cross-regulation of YAP, Rac1, and Merlin in epithelial layers on NRA. a Immunofluorescence staining for Merlin in epithelial cell sheets on a flat surface and NRA. In the marginal zone and FLPs of sheets on NRA, Merlin was primarily localized in the cytosol (brown boxes), whereas it was localized in the cell–cell contacts of submarginal cells cultured on NRA and flat substrata (red boxes). The samples were fixed after 8 h to remove stencils. b Cells on flat surfaces and NRA were used for Co-IP analysis with the AMOT antibody for investigating physical bounding between AMOT and a target protein, Merlin, and the precipitates were immunoblotted using Merlin antibody. c Lysates from control and YAP^KD cells were used for co-immunoprecipitation (Co-IP) study using the AMOT antibody, analyzed by immunoblotting with Merlin antibody. d Control and Merlin^KD cells were immunoblotted for evaluation of Rac1 activity via pull-down assay, YAP expression, and its phosphorylation. (n = 3 biologically independent samples, * = statistical significance of the ratio of active Rac1 to Rac1 on control vs. Merlin^KD cells. ***P < 5 × 10⁻³). e Cell migration speed of individual cells in Merlin^KD epithelial cell sheets as a function of the distance from the sheet edge in cell sheets cultured on NRA in the presence of a Rac1 inhibitor, NSC23766. (each number of independently analyzed cells, n, is indicated, # = statistical significance of speed values in the marginal region vs. the most submarginal region of Merlin^KD cells, ^###P < 5 × 10⁻⁶. * = statistical significance of control vs. Merlin^KD cells without drugs (blue) and of Merlin^KD cells with vs. without drugs (black), *P < 0.05 and **P < 5 × 10⁻³). f Dissemination of cells in control and Merlin^KD epithelial sheets on NRA. All error bars are S.E.M. (n = 4 biologically independent experiments, * = statistical significance of the number of disseminations of cells from control vs. Merlin^KD epithelial sheets, *P < 0.05). g Schematic of YAP-mediated signaling cascades operating via Rac1 and Merlin, controlling cell speed in cells undergoing partial and complete EMT. All error bars are S.E.M and statistical significance was determined by two-sided Student’s t-test

Fig. 7

Double-feedback loop with YAP activated on NRA regulating EMT. a Simulation result showing that feedback mechanisms lead to emergence of two stable states of high and low YAP activity, indicating epithelial (low Rac1, high E-cadherin, and low YAP) and partial EMT (high Rac1, low E-cadherin, and high YAP) states depending on Rac1 and E-cadherin activity in this system. b Cell migration speed on NRAs with different rigidity values, suggesting that rigidity can regulate the basal rate of YAP activation at different initial distances from the edge of the sheet (all error bars are S.E.M, each number of independently analyzed cells, n, is indicated, * = statistical significance of the speed of cells on NRA in which the rigidities are 10 MPa vs. 1 GPa, *P < 5 × 10⁻², **P < 1 × 10⁻², and ***P < 5 × 10⁻³, all two-sided Student’s t-test). c Simulated bimodal distribution of YAP activity as a function of the distance from the sheet edge on NRA substrata of different rigidity values. d Rigidity-dependent YAP localization in nuclei of cells cultured on NRA. Immunofluorescence staining of YAP showing nuclear localization at different distances from the sheet edge on NRAs with different rigidity values (top). Fractions of nuclei displaying different intensities of YAP staining as a function of the distance from the sheet edge on NRA having different rigidity (bottom) (see details in Supplementary discussion and Supplementary Fig. 17). The samples were fixed after 6 h to remove stencils and induce epithelial expansion. e Schematic description of the regulation of EMT by YAP-mediated topographically induced mechanical input, showing full EMT in marginal cells exposed to cell-free areas at the fronts of FLPs, partial EMT in extensive marginal areas, and epithelial organization (no EMT) in the areas most distant from the edge

Fig. 8

Potential role of the proposed mechanisms in the renal cancer. a Difference of signaling of renal cell carcinoma, ACHN, related to EMT, cell–cell adhesion, cell migration between disseminated cells, and cells remaining within the monolayer after dissemination elicited by mechanical cues. b Analysis of patient data, sequenced tumors (451 samples) of renal clear cell carcinoma, (TCGA, provisional) via cbioportal. Oncoprints show alterations in YAP1 and WT1 across a set of kidney renal clear cell carcinoma. Z-scores for mRNA expression (RNA Seq V2RSEM) and protein expression were 2.326 for 99% confidentiality (one tail). c Disease-free survival Kaplan–Meier estimate of upregulated YAP1 (left) and YAP1/WT1 (right) (see details in the “Methods” section)