The ZEB1/miR-200c feedback loop regulates invasion via actin interacting proteins MYLK and TKS5

Abstract

Epithelial to mesenchymal transition (EMT) is a developmental process which is aberrantly activated during cancer invasion and metastasis. Elevated expression of EMT-inducers like ZEB1 enables tumor cells to detach from the primary tumor and invade into the surrounding tissue. The main antagonist of ZEB1 in controlling EMT is the microRNA-200 family that is reciprocally linked to ZEB1 in a double negative feedback loop. Here, we further elucidate how the ZEB1/miR-200 feedback loop controls invasion of tumor cells. The process of EMT is attended by major changes in the actin cytoskeleton. Via in silico screening of genes encoding for actin interacting proteins, we identified two novel targets of miR-200c - TKS5 and MYLK (MLCK). Co-expression of both genes with ZEB1 was observed in several cancer cell lines as well as in breast cancer patients and correlated with low miR-200c levels. Depletion of TKS5 or MYLK in breast cancer cells reduced their invasive potential and their ability to form invadopodia. Whereas TKS5 is known to be a major component, we could identify MYLK as a novel player in invadopodia formation. In summary, TKS5 and MYLK represent two mediators of invasive behavior of cancer cells that are regulated by the ZEB1/miR-200 feedback loop.

Keywords: MYLK (MLCK); ZEB1; epithelial to mesenchymal transition (EMT); invadopodia; miR-200.

Figure 1. Identification of potential cell invasion target genes of miR-200c

A. In silico screening of genes containing the term “actin interaction” in the functional annotation of nextprot database was merged to miR-200c targets predicted using the PITA target prediction program, resulting in 173 candidates. B. 18 of these 173 potential target genes harbor at least one miR-200c binding site that is conserved in vertebrates (TargetScanHuman6.2) and exhibit an expression pattern of <0.5 in the microarray analysis of MDA-MB-231 shZEB1/shCtrl. C. miRConnectL expression correlation of the 18 target genes with miR-200c (sPCC=summed Pearson correlation coefficient). D. Quantitative RT-PCR showing changes in expression of 8 candidate genes after transient overexpression of miR-200c in MDA-MB-231 in comparison to control miR (miR-ctrl). β-Actin was used for normalization.

Figure 2. miR-200c overexpression leads to downregulation of MYLK and TKS5

A. Immunoblots (left) showing expression levels of MYLK and TKS5 after transient overexpression of miR-200c in MDA-MB-231. β-Actin was used as loading control. Immunofluorescence images (right) showing that transient overexpression of miR-200c in MDA-MB-231 significantly reduced TKS5 expression and increased E-cadherin expression. Scale bar, 20 μm. B. Quantitative RT-PCR (left) and immunoblots (right) showing expression levels of MYLK, TKS5 and ZEB1 after transient overexpression of miR-200c in BT-549. β-Actin was used for normalization in qPCRs and as loading control in immunoblots. C. Gene expression levels of MYLK, TKS5 and ZEB1, measured by qRT-PCR in MCF7 upon transient inhibition of the miR-200c subgroup (miR-200b, miR-200c and miR-429). Upregulation of MYLK on protein level was confirmed by immunoblot analysis (right). β-Actin was used as loading control. D. Gene expression levels of miR-200c, MYLK, TKS5 and ZEB1 in MDA-MB-231 cells with stable short hairpin RNA (shRNA) mediated knockdown of ZEB1 (shZEB1) versus control (shCtrl) showing downregulation of MYLK and TKS5 upon miR-200c activation. A respective immunoblot is given using Tubulin as loading control. E. Luciferase reporter assay of MYLK- and TKS5- 3′UTR showing reduced activity upon transient miR-200c overexpression in MDA-MB-231 cells compared to control miRNAs. The ZEB1- 3′UTR reporter was included as a positive control. F. Mutated versions of the MYLK- and TKS5- 3′UTR reporters allow higher relative luciferase activity in MCF7 cells, in comparison to the wt constructs (left). After inhibition of the miR-200c subgroup using antagomiRs only the wt, but not the mutated reporters were significantly upregulated (right). The ZEB1- 3′UTR reporter was included as a positive control.

Figure 3. Expression correlation of MYLK, TKS5 and ZEB1 with miR-200c and/or E-cadherin in cancer cell lines and samples from breast cancer patients

A. Heat map and expression correlation analysis of MYLK, TKS5, ZEB1 and E-cadherin shown for 56 breast cancer cell lines of the CCLE panel (GSE36133). Pearson correlation coefficients (r) and P-values (two-tailed) computed are shown below. B. NCI60 cell lines were classified as epithelial, undefined and mesenchymal clusters (according to ref [6]). Expression levels of MYLK, TKS5 and ZEB1 in these groups show significant enrichment in the mesenchymal cluster, whereas miR-200c is strongly enriched in the epithelial cluster. C. Heat map showing expression levels of miR200c, E-cadherin, ZEB1, MYLK and TKS5 for 101-breast cancer patients (GSE19783). Pearson correlation coefficients (r) and P-values (two-tailed) computed are shown below.

Figure 4. Inhibition of MYLK and TKS5 attenuates cancer cell invasion

A. Quantitative RT-PCR and immunoblot analysis showing diminished expressing levels after transient siRNA-mediated knockdown of MYLK (left) and TKS5 (right) in MDA-MB-231. β-Actin was used as normalization gene and immunoblot loading control. B. Wounding assay for MDA-MB-231, transfected with si-MYLK, si-TKS5 or miR-200c. Quantification of three biological replicates shows a significant reduction in migration relative to control cells only after miR-200c overexpression, but not after si-MYLK or si-TKS5 treatment. Representative images are shown in Supplementary Figure S4A. C. Cultrex invasion assays for MDA-MB-231, transfected with si-MYLK, si-TKS5 or miR-200c. Quantification of three biological replicates shows a significant reduction in invasion relative to control cells after si-MYLK or si-TKS5 treatment as well as after miR-200c overexpression.

Figure 5. MYLK and TKS5 are essential for the establishment of invadopodia

MDA-MB-231, pre-treated with indicated siRNAs or miRNAs, were plated on fluorescein-gelatin coated coverslips. After 24 h cells were fixed, stained and visualized. Representative images are shown. Scale bar, 20 μm. Arrows indicate regions of matrix degradation. 10 random fields of view from three independent experiments were quantified (right) showing significantly reduced gelatin degradation upon transfection of si-MYLK and si-TKS5 to a similar extent as upon overexpression of miR-200c.

Figure 6. Analysis of MYLK in invadopodia formation

A. Immunofluorescence showing HA-tagged recombinant MYLK, Cortactin and F-actin (stained with Rhodamine phalloidin) in MDA-MB-231 cells on gelatin-coated coverslips. The three proteins co-localize in point shaped structures that are also obvious in the z axis (merge). Scale bar overview, 20 μm; inset, 10 μm. Arrows indicate points of co-localization. B. A representative image of an in situ proximity ligation assay (PLA) shows sites of interaction between MYLK-HA and Cortactin in a transfected MDA-MB231 cell indicated by red fluorescent spots. In the untransfected neighboring cells, no signal was observed. Empty vector transfected cells or incubation without antibody were used as negative controls (not shown). Scale bar, 10 μm.