Epithelial to mesenchymal transition influences fibroblast phenotype in colorectal cancer by altering miR-200 levels in extracellular vesicles

Abstract

Colorectal cancer (CRC) with a mesenchymal gene expression signature has the greatest propensity for distant metastasis and is characterised by the accumulation of cancer-associated fibroblasts in the stroma. We investigated whether the epithelial to mesenchymal transition status of CRC cells influences fibroblast phenotype, with a focus on the transfer of extracellular vesicles (EVs), as a controlled means of cell-cell communication. Epithelial CRC EVs suppressed TGF-β-driven myofibroblast differentiation, whereas mesenchymal CRC EVs did not. This was driven by miR-200 (miR-200a/b/c, -141), which was enriched in epithelial CRC EVs and transferred to recipient fibroblasts. Ectopic miR-200 expression or ZEB1 knockdown, in fibroblasts, similarly suppressed myofibroblast differentiation. Supporting these findings, there was a strong negative correlation between miR-200 and myofibroblastic markers in a cohort of CRC patients in the TCGA dataset. This was replicated in mice, by co-injecting epithelial or mesenchymal CRC cells with fibroblasts and analysing stromal markers of myofibroblastic phenotype. Fibroblasts from epithelial tumours contained more miR-200 and expressed less ACTA2 and FN1 than those from mesenchymal tumours. As such, these data provide a new mechanism for the development of fibroblast heterogeneity in CRC, through EV-mediated transfer of miRNAs, and provide an explanation as to why CRC tumours with greater metastatic potential are CAF rich.

Keywords: MiR-200; Zeb1; cancer-associated fibroblast; colorectal cancer; epithelial to mesenchymal transition; extracellular vesicle; stroma.

FIGURE 1

EMT status of CRC models. (a) Assessment of EMT marker expression in a spectrum of epithelial and mesenchymal CRC cell lines by western blotting. (b) EMT marker expression in the SW480 MET model (SW480‐ZKD cells), where HCT116 cells are shown as a positive control. Low and high exposures for E‐cadherin in both panels. E‐epithelial and M‐mesenchymal. Representative of three separate experiments.

FIGURE 2

Characterisation and transfer of CRC EVs. (a) Endosomal (Alix, TSG101) and tetraspanin (CD63, CD81) marker expression in HCT116 cells and EVs. The mitochondrial protein, cytochrome C, was used as a negative EV marker. Representative of three separate experiments. (b) Transmission electron micrograph of HCT116 EVs at 100,000× magnification. Scale bar represents 100 nm. Representative of three EV preparations. (c) Nanoparticle tracking analysis of HCT116 EVs from five separate videos, each 90s duration. (d) Visualisation of DiO‐labelled HCT116 EVs within MRC5 fibroblasts after 24 h conditioning. Phase contrast and fluorescence images are shown. Scale bar represents 50 μm. Representative of three experiments. (e) Detection of DiO‐labelled HCT116 EVs in MRC5 fibroblasts by flow cytometry. Representative histograms for control (DiO‐labelled medium) and EV‐conditioned fibroblasts from three experiments.

FIGURE 3

Relative abundance of miRNAs in epithelial and mesenchymal CRC cells and EVs by qPCR array. (a) CRC cells. (b) CRC EVs. CT values were normalised to the geometric mean of all values for that sample. Combined mean 2^(‐ΔΔCT) values from three biological replicates for epithelial cells or EVs (DLD‐1, HCT116 and SW620) were compared with mesenchymal (SW480) cells or EVs for each miRNA, to generate fold changes. MiRNAs for which 2^(‐ΔΔCT) values were less than 0.1 in all samples were excluded. Fold changes for each miRNA are represented on a blue‐white‐red (low‐median‐high) colour scale. Fold changes for miR‐200 family members are highlighted in yellow.

FIGURE 4

MiR‐200 levels in epithelial and mesenchymal CRC cells and EVs. (a) CRC cells. (b) CRC EVs. (c) SW480‐ZKD cells (SW480‐MET model). (d) SW480‐ZKD EVs. MiRNA levels were normalised to miR‐423‐5p, calculated from the triplicate of CT values, using the ΔΔCT method, and expressed relative to SW480 or SW480 control cells or EVs, which were assigned the value 1. Statistical significance was determined by two‐tailed unpaired t‐test (*p < 0.05; **p < 0.01; ***p < 0.001). In (a) and (b), statistical significance is shown for DLD‐1, HCT116 and SW620, compared to SW480 (from top to bottom, respectively). Values plotted are the means of three technical replicates from three experiments.

FIGURE 5

(a) Schematic of RNA pull down experimental set‐up. Nascent RNA was labelled in donor cells or cell‐free control media, from which EVs were isolated and transferred to MRC5 fibroblasts. Labelled RNA was captured from recipient fibroblasts and probed for miR‐200. (b) MiR‐200 levels in 5EU labelled RNA from recipient fibroblasts. MiRNA levels were normalised to miR‐423‐5p, calculated from the triplicate of CT values, using the ΔΔCT method, and expressed relative to MRC5 cells conditioned with MRC5 (self) EVs, which were assigned the value 1. Values plotted are means of three technical replicates from three experiments. Statistical significance was determined by two‐tailed unpaired t‐test (***p < 0.001).

FIGURE 6

Effects of EVs and miR‐200 on myofibroblast differentiation. Protein expression of ZEB1, α‐SMA and fibronectin by western blotting in MRC5 fibroblasts (with or without TGF‐β stimulation): (a) Conditioning with CRC EVs (epithelial or mesenchymal) at a concentration of 1.5 × 10⁹ EVs/ml (quantified using Nanosight); (b) transfection with miR‐200b/c; (c) transfection with ZEB1 siRNA. For Panels (a–c), band intensities are relative to β‐actin and normalised to the first lane of the blot which was given the value 1. Values plotted are from three independent experiments. (d) Immunocytochemistry of MRC5 and NCFs (control and ZEB1 siRNA‐transfected) before and after TGF‐β stimulation. Scale bars represent 100 μm. Staining intensity per cell (relative to control‐transfected/ TGF‐β unstimulated fibroblasts) is shown on the right. The graph represents individual data points from three independent experiments. Statistical significance (a‐d) was determined by two‐tailed unpaired t‐test (*p < 0.05; **p < 0.01; ***p < 0.001).

FIGURE 7

Correlation between EV‐miRNAs and CRC‐related epithelial and stromal genes. (a) Matrix constructed from 304 human CRC samples with matched miRNA and gene expression logCPMs. Unbiased hierarchical clustering of gene (mRNA) and miRNA expression according to correlation. Red to blue colour scale represents r from +1 to ‐1. (b) Protein expression by western blotting in CRC cells and fibroblasts (with or without TGF‐β) to demonstrate the expression of classical CAF markers across these cell types. (c) Protein expression by western blotting for additional CAF markers in MRC5 fibroblasts conditioned with DLD‐1 (epithelial), SW480 (mesenchymal) and SW480‐ZKD (epithelial) EVs, with or without TGF‐β stimulation. E‐cadherin 130 kDa; vimentin 55 kDa; α‐ SMA 42 kDa; fibronectin 220 kDa; periostin 93 kDa; PDGFR β 190 kDa; paladin 96 kDa;  S100A4 12 kDa; β ‐actin 42 kDa. Lysates obtained to prepare Figure 6a were used in Figure 7c.

FIGURE 8

The role of CRC EMT in determining fibroblast phenotype in vivo. (a) Scheme of in vivo experimental setup. (b) Haematoxylin and eosin staining of sections from SW480 control and SW480‐ ZKD CRC xenografts. Scale bars represent 100 μm (200×; top panels) and 50 μm (400x; bottom panels). Representative of three xenografts from three animals. (c) MiR‐200 levels and (d) ACTA2 (α‐SMA) and FN1 (fibronectin) mRNA levels in PKH+ve cells (fibroblasts) from SW480 control and ZKD xenografts. Normalised to SW480 control xenografts, which were assigned the value 1. Statistical significance was determined by two‐tailed unpaired t‐test (**p < 0.01; ***p < 0.001). Values plotted are the means of three technical replicates from nine pooled xenografts. (e) ZEB1 and (f) α‐SMA immunohistochemical staining of sections from SW480 control and SW480‐ZKD xenografts. Scale bar represents 100 μm. Representative of three xenografts.