RUNX1 promotes tumour metastasis by activating the Wnt/β-catenin signalling pathway and EMT in colorectal cancer

A total of 73 patients who underwent radical operation for CRC at Nanfang Hospital of Southern Medical University were included in the study after obtaining informed consent. A diagnosis of CRC was histopathologically confirmed for each patient sample. Cancer tissues and matched normal tissues were stored at -80C° for quantitative real-time PCR (qRT-PCR) and Western blotting analyses. Additionally, none of the patients had received radiotherapy or chemotherapy before surgery. The protocols used in this study were approved by Nanfang Hospital’s Protection of Human Subjects Committee.

Human embryonic kidney cell line (293 T), human normal colon epithelial cell line (FHC) and six human CRC cell lines (SW480, SW620, RKO, HCT116 HT29 and LOVO) were purchased from the Cell Bank of Type Culture Collection (CBTCC, Chinese Academy of Sciences, Shanghai, China) and were cultured in Dulbecco’s modified Eagle medium (DMEM) (Gibco, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS; Gibco, Carlsbad, CA). Cells were maintained at 37 °C in a humidified 5% CO₂ atmosphere.

RUNX1 overexpression and knockdown were performed using a lentiviral packaging system. To construct overexpressing exogenous and RNA-interfered endogenous RUNX1 cell lines, full-length RUNX1 (NM_001754) was cloned into the expression vector pLenti-EF1a-EGFP-P2A-Puro-CMV (Obio Technology, Shanghai, China) and transfected into HCT116 and RKO cell lines according to the manufacturer’s instructions. Knockdown of endogenous RUNX1 was mediated by designed shRNAs (Cyagen, Guangzhou, China) that were transfected into SW480 and RKO cell lines according to the manufacturer’s instructions. The RUNX1 shRNA sequence was sense 5′-CCAGGTTGCAAGATTTAAT-3′ and 5′-GGCAGAAACTAGATGATCA-3′, and the scramble sequence was sense 5′-CCTAAGGTTAAGTCGCCCTCG-3′. Transduced cells were selected in medium containing puromycin (#EZ2811D376, BioFroxx, Germany) (2 μg/ml) and maintained in medium containing puromycin (1 μg/ml).

Total RNA was extracted from cells or tissues with TRIzol reagent (TaKaRa, China). qRT-PCR was performed using the PrimeScript RT Reagent Kit (#RR035A, TaKaRa, China) and SYBR Premix Ex Taq (#RR820A, TaKaRa, Dalian, China) following the manufacturer’s instructions. Our results were normalized to the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or U6. The specific primers used are listed in Additional file 5: Table S1. The qRT-PCR results were analyzed to obtain the Ct values of the amplified products, and data were analyzed by the 2-ΔΔCt method.

We performed Western blotting according to the methods of a previous study. Protein lysates were prepared, subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto polyvinylidene difluoride (PVDF) membranes and blotted according to standard methods using the following antibodies: RUNX1 (#4336, CST), KIT (#3074, CST), CD44 (#3570, CST), cyclin D1 (#2978, CST), c-Jun (#9165, CST), LEF1 (#2230, CST), Met (#8198, CST), c-Myc (#5605, CST), TCF1/TCF7 (#2230, CST), β-catenin (#8480, CST), E-cadherin (#3195, CST), N-cadherin (#13116, CST), Vimentin (#5741, CST), TCF8/ZEB1 (#3396, CST), AXIN1 (#2087, CST) and GAPDH (60004–1-Ig, Proteintech).

Immunohistochemistry was performed following the manufacturer’s instructions (PV-6001, ZSGB-BIO, Beijing, China) using the E-cadherin antibody (60335–1-Ig, Proteintech), N-cadherin (66219–1-Ig, Proteintech), and Vimentin (60330–1-Ig, Proteintech). One independent pathologist used software ImageJ to calculate gray values for pathological scoring.

Approximately 2 × 10⁵ cells were suspended in 100 μl serum-free medium and seeded in the upper 8-μm-pore Transwell chambers (3422, Corning, USA), and the lower chambers were filled with 500 μL of 20% FBS medium. For the cell invasion assay, the Transwell chambers were Matrigel-coated (354234, Corning, USA). Then, the cells were incubated at 37 °C for 24–48 h to allow for migration or invasion. For quantification, the cells were fixed with 4% paraformaldehyde for 20 min, stained with hematoxylin for 20 min at room temperature, and counted in five randomly chosen fields (× 200) under a microscope.

Cells were seeded on six-well culture plates at a density of approximately 1 × 10⁶ cells per well and incubated for 24 h (80–90% confluence). Scratch wounds were produced by a 10 μl plastic pipette tip, and then cells were cultured in DMEM with 2% FBS. Wound margins were photographed, and migration was monitored after 48 h of wound formation. Cell motility was quantified by measuring the distance between the advancing margins of cells in three randomly selected microscopic fields (× 200) at each time point.

For immunofluorescence, cells were seeded on cover slips. Overexpression or shRNA knockdown treatment was performed after 24 h. After the indicated treatment, the cells were cultured for 48 h, fixed with 4% formaldehyde for 10 min at room temperature, and washed three times with wash buffer (0.02% Tween 20/PBS). Then, the cells were permeabilized with 0.5% Triton X-100/PBS for 10 min at room temperature. The cells were washed three times with wash buffer (5 min each time) and then incubated with 1.5% bovine serum albumin (BSA)/phosphate-buffered saline (PBS) solution (blocking solution) for 30 min at room temperature. The E-cadherin antibody (60335–1-Ig, Proteintech), N-cadherin (66219–1-Ig, Proteintech), Vimentin (60330–1-Ig, Proteintech), and β-catenin (66379–1-Ig, Proteintech) were incubated in blocking solution at 4 °C overnight. Rhodamine phalloidin (1:1500, Cytoskeleton, PHDR1), which was used for detecting F-actin, was incubated in blocking solution at room temperature in the dark for 60 min. After washing, the cells were incubated with Alexa594-conjugated secondary antibodies for 60 min at room temperature in a protected environment in the dark (Life Technologies, A-21235, 1:500 in blocking buffer) followed by counterstaining with DAPI (Thermo Fisher). Samples were mounted with ProLong Gold antifade reagent (Life Technologies) and imaged on a confocal microscope (× 200).

ChIP assays were performed using a kit (#17–10085, Merck, German), and all the experimental procedures were performed according to the manufacturer’s instructions. Briefly, cells (1 × 10⁷) in a 15-cm culture dish were treated with 1% formaldehyde to cross-link chromatin-associated proteins to DNA. The cell lysates were subjected to ultrasound for 9–10 sets of 10-s pulses at 40% output to shear the DNA into fragments between 200 and 1000 bps. Equal cell lysates were incubated with 10 μl of DYKDDDDK Tag (bound to the same epitope as the Sigma Anti-Flag M2 antibody) (#14793, CST), anti-IgG antibody (Merck) as a negative control and anti-RNA polymerase II as a positive control. All the above chromatin supernatants were incubated with 20 μL magnetic protein A/G beads overnight at 4 °C with rotation. On the second day, the protein-DNA complexes were reversed and purified for pure DNA. The human c-Kit binding sites were amplified with qRT-PCR and PCR. The specific primers used are listed in Additional file 6: Table S2.

Female athymic 4 to 5-week-old BALB/c nude mice were purchased from the Laboratory Animal Services Centre of Guangdong Province and were maintained in a specific pathogen-free facility. For the orthotopic injection metastatic mouse model assay, the nude mice of the experimental group were anesthetized with 1% pentobarbital, and 5 × 10⁶ cells were injected under the ileocecal serosa. The general growth conditions in the nude mice were observed after surgery to determine whether there were signs of cachexia, such as wasting and arched backs. Then, 40 days after the operation, all mice were sacrificed, and the intestines, liver, spleen and lungs were removed for photographing. After formalin fixation, paraffin sections were generated and stained with hematoxylin and eosin (H&E).

The SPSS 22.0 (IBM; Chicago, IL, USA) and Microsoft Excel 2016 (Microsoft, Redmond, WA, USA) statistical analysis software programs were used for statistical analysis of the experimental data. The significance of differences between groups was estimated by Student’s t-test. Additionally, multiple group comparisons were analyzed with one-way ANOVA. Bivariate correlations between study variables were calculated by Spearman’s rank correlation coefficients. * P < 0.05, ** P < 0.01, and *** P < 0.001 were considered significant.

An analysis of The Cancer Genome Atlas (TCGA) database using the online Gene Expression Profiling Interactive Analysis (GEPIA) tool (http://​gepia.​cancer-pku.​cn/​) indicated that RUNX1 is highly expressed in most types of digestive system tumours (Fig. 1a). To screen CRC gene expression profiles, we selected two qualified gene expression microarray datasets (TCGA and GSE106582) (Fig. 1b) and determined that RUNX1 was differentially expressed in colorectal tumour tissues compared to its expression in non-neoplastic tissues (p < 0.001) in two independent samples and in paired samples (Fig. 1c). An analysis of the correlation between CRC-related gene set enrichment and RUNX1 expression showed statistically significant (NES = 1.54, p = 0.006) results in TCGA database, and Gene Set Enrichment Analysis (GSEA) (http://​software.​broadinstitute.​org/​gsea/​index.​jsp) suggested a significant correlation between RUNX1 expression and these genes (Fig. 1d). Notably, Kaplan-Meier survival analysis revealed that patients with high RUNX1 expression levels had poorer disease-free survival and overall survival than patients with low RUNX1 expression levels. Similar results were found when the GEPIA web server was used to analyze TCGA database (Fig. 1e) as well as in the PROGgeneV2 - Pan Cancer Prognostics Database using the GEO database (http://​genomics.​jefferson.​edu/​proggene/​) (Additional file 1: Figure S1A and B).

Furthermore, to determine whether RUNX1 protein expression is also increased in CRC, qRT-PCR and Western blotting were performed in 61 pairs of fresh samples and in 12 pairs of fresh samples, including cancers and matched adjacent normal tissues or independent normal tissues (Fig. 1f and g). In this analysis, the RUNX1 levels were significantly upregulated in most CRC tissues. In the patient clinical information statistics table (Additional file 7: Table S3), we found that patients with high expression of RUNX1 were worse in AJCC stage, T stage and M stage, when the patients were divided into high and low RUNX1 expression groups by the median expression of RUNX1 mRNA. Western blotting and qRT-PCR analysis revealed varying levels of RUNX1 expression in six CRC cell lines: SW480, SW620, HT29, HCT116, RKO and LoVo. RUNX1 expression was highest in SW480 and SW620 cells; lowest in HCT116, HT29 and LoVo cells; and moderately expressed in RKO cells (Additional file 1: Figure S1C).

To investigate the biological function of RUNX1 in CRC cells in vitro and in vivo, we established stable RUNX1-overexpressing CRC cell lines with HCT116 and RKO cells and RUNX1-silenced cell lines using a lentiviral vector carrying two specific shRNAs in SW480 and RKO cells. Forty-eight hours after transfection, qRT-PCR and Western blotting analysis showed a significant increase in RUNX1 mRNA and protein expression in HCT116-RUNX1 and RKO-RUNX1 cells and a significant decrease in RUNX1 expression in SW480-shRUNX1_1, SW480-shRUNX1_2, RKO-shRUNX1_1 and RKO-shRUNX1_2 cells compared to the expression levels in control cells (Fig. 2a).

Markedly faster wound closure was observed in HCT116-RUNX1 and RKO-RUNX1 cells, but SW480-shRUNX1_1, SW480-shRUNX1_2, RKO-shRUNX1_1 and RKO-shRUNX1_2 cells exhibited slower wound closure, as revealed by a wound healing assay (Fig. 2c and Additional file 2: Figure S2B). Moreover, the results from Transwell assays revealed that overexpression of RUNX1 significantly increased the migration and invasion abilities of CRC cells. Next, our results showed that the migration and invasion abilities were significantly decreased in cells with RUNX1 downregulation (Fig. 2b and Additional file 2: Figure S2A).

Furthermore, we also achieved consistent results for in vivo metastasis assays. As shown by our results, tumours formed by cells overexpressing RUNX1 showed stronger metastatic ability, and more tumours were formed in the spleen (Additional file 2: Figure S2C) in the nude mice cecum tumour orthotopic model transfected with RUNX1-overexpressing CRC HCT116-RUNX1 cells than in the control group transfected with the HCT116-vector (Fig. 2d).

GSEA analysis using the TCGA and GSE17538 datasets showed that RUNX1 expression is associated with regulation of the actin cytoskeleton and focal adhesion (Fig. 3a). We retrieved genes with more than 100 RUNX1 transcription factor binding sites from the Gene Transcription Regulation Database (GTRD) (http://​gtrd.​biouml.​org/​) and used the results to conduct a Gene Ontology (GO) enrichment analysis using DAVID Bioinformatics Resources 6.8 (https://​david.​ncifcrf.​gov/​). The GO-Cellular Components enrichment analysis results showed gene enrichment in the cellular components of the cytoskeleton, cell junction, actin cytoskeleton and cell-cell junction (Fig. 3b). Next, immunofluorescence labelling of F-actin was performed using phalloidin. Immunofluorescence analysis showed that the cell lines with upregulated RUNX1 were characterized by a spindle-shaped morphology and the rearrangement of F-actin fibres (Fig. 3c). The spindle shape change and loss of cell-cell contact are characteristics of the cell morphology changes that occur during EMT. Through mRNA expression correlation analysis on website GEPIA, we found that the expression of RUNX1 is positively correlated with multiple molecules of the EMT and is statistically significant, such as N-cadherin (CDH2), Vimentin (VIM), Snail and ZEB1 (Additional file 3: Figure S3A). Western blotting was used to analyze EMT markers induced by RUNX1 in the HCT116, SW480 and RKO cell lines. E-cadherin was downregulated in HCT116-RUNX1 cells and upregulated in SW480-shRUNX1 cells compared to its expression in control cells. The upregulation of ZEB1, N-cadherin and vimentin was also induced by RUNX1 in HCT116-RUNX1 cells and inhibited in SW480-shRUNX1 cells (Fig. 3d). Consistent with these results, the immunofluorescence assay results revealed decreased expression of E-cadherin, whereas the expression levels of N-cadherin and vimentin were decreased in HCT116-RUNX1 cells, and the opposite result was observed in SW480-shRUNX1_2 cells (Fig. 3e). Consistent with these findings, decreased E-cadherin and increased N-cadherin and vimentin expression levels were shown by immunohistochemistry in situ in the RUNX1-overexpressing cecum tumours of nude mice (Fig. 3f). Through mRNA expression correlation analysis on website GEPIA, we found that the expression of RUNX1 is positively correlated with multiple molecules of the MMP family and is statistically significant, such as MMP3, MMP7, MMP9 and MMP14 (Additional file 3: Figure S3C). Subsequently, we also used qPCR experiments to detect elevated expression of the above molecules in the HCT116 cell line overexpressing RUNX1 (Additional file 4: Figure S4A). These results strongly suggest that RUNX1 is involved in enhancing the metastatic capacity of CRC.

GSEA analysis using both TCGA and the GSE17538 database showed that RUNX1 expression is associated with the Wnt signalling pathway (Fig. 4a). Next, we found a positive correlation between the expression of RUNX1 and CTNNB1, also called β-catenin, in the gene expression microarray dataset (GSE17538) (Fig. 4b) and in CRC cell lines (Fig. 4c). Through mRNA expression correlation analysis on website GEPIA, we found that the expression of RUNX1 is positively correlated with multiple molecules of the Wnt/β-catenin signalling pathway and is statistically significant, such as CTNNB1, WNT5a, CD44 and LEF1 (Additional file 3: Figure S3B). Furthermore, we verified the expression of β-catenin by Western blot and immunofluorescence analysis in lentivirus-infected cell lines, and the results showed that β-catenin expression increased after RUNX1 was overexpressed and decreased after RUNX1 was downregulated (Fig. 4d and g). Western blot analysis showed that β-catenin was upregulated in the nucleus with RUNX1 overexpression and downregulated in the nucleus with RUNX1 knockdown (Fig. 4e). A co-immunoprecipitation assay indicated that RUNX1 directly interacts with β-catenin in SW480 cells (Fig. 4f). We suggest that RUNX1 and β-catenin possibly enter the nucleus together to perform their functions. We used Western blotting to measure downstream targets, such as Met, C-Jun, C-Myc, TCF1/7, CD44, cyclin D1 and LEF1 (Fig. 4h). Our results showed that RUNX1 increased the expression of proteins, including Met, C-Jun, C-Myc, TCF1/7, CD44, cyclin D1 and LEF1, that are associated with the Wnt/β-catenin signalling pathway.

Using genes with more than 100 RUNX1 transcription factor binding sites, a GO-Biological Process enrichment analysis was performed using DAVID Bioinformatics Resources 6.8. The three most significant biological process involved positive regulation of GTPase activity, intracellular signal transduction and signal transduction (Fig. 5a). Through a Wayne diagram using the website (http://​bioinfogp.​cnb.​csic.​es/​tools/​venny/​index.​html), we determined the intersection of these three gene sets, which contained eight molecules (Fig. 5b), including KIT, PLCB1, BCR, ARHGEF7, KALRN, MY09B, ABR and CHN2. Using the protein interaction analysis website STRING (https://​version-10-5.​string-db.​org/​cgi/​input.​pl), we found that KIT may be a bridge between RUNX1 and the Wnt/β-catenin signalling pathway (Fig. 5c). Moreover, Western blot results showed that the expression of KIT increased and decreased along with the expression of RUNX1 (Fig. 5d). After siRNA was used to interfere with KIT expression, the expression of several proteins, including C-Myc and TCF1, that activate the Wnt/β-catenin signalling pathway was decreased (Fig. 5e) RUNX1 transcription factor binding sites have been reported in the promoter and enhancer regions of KIT [26]. Therefore, we used ChIP-qPCR and ChIP-PCR to examine the binding sites at + 700 bp and + 30 kb from the transcription start site KIT, and the results showed that RUNX1 can bind at those two sites (Fig. 5f).

XAV-939, a Wnt/β-catenin signalling pathway specific inhibitor, promotes β-catenin degradation. Moreover, using Transwell assays, we found that when cells were treated with XAV-939, the migration and invasion ability of HCT116 and RKO RUNX1-overexpressing cell lines was decreased (Fig. 6a). At the same time, XAV-939 can inhibit the expression changes of EMT-related molecules induced by RUNX1 (Fig. 6b). We used qRT-PCR and Western blotting to examine RUNX1 expression after treatment with siRNA of β-catenin for 48 h, and the results showed that the expression of RUNX1 decreased when the β-catenin expression was disturbed. (Fig. 6c).