signature=3bd29032edeb7086e197723d7d444350,Pan-cancer EMT-signature identifies RBM47 down-regulation...

Cell lines, cell culture and reagents

Colorectal cancer (SW480, SW620, Caco-2), breast cancer (MCF7, MDA-MB-231), and prostate cancer (LNCaP, PC-3) cell lines were maintained in Dulbecco’s Modified Eagles Medium (DMEM, Invitrogen) containing 10% fetal bovine serum (FBS, Invitrogen). The colorectal cancer cell lines HCT-15, HT29, LS174T, HCT116 and DLD-1 were maintained in McCoy’s 5 A Medium (Invitrogen) containing 10% FBS. All cells were cultivated in presence of 100 units/ml penicillin and 0.1 mg/ml streptomycin. SiRNAs (Ambion silencer siRNA: negative control (ID#4611) and RBM47 (ID#s29090) were transfected at a final concentration of 10 nM using HiPerfect transfection reagent (Qiagen). IL-6 (Immunotools) was dissolved in water and used at a final concentration of 20 ng/ml.

Conditional expression of SNAIL and SLUG alleles in cell pools

Stable DLD1/pRTR-SNAIL cell pools were described previouslySLUG cDNA was amplified by PCR from pcDNA-SLUG (a kind gift from Kou-Juey Wu, Institute of Biochemistry & Molecular Biology, National Yang-Ming University, Taiwan), verified by sequencing and cloned into the pRTR vector. Stable DLD1/pRTR cell pools were obtained transfection of DLD-1 cells with pRTR plasmids using FuGene reagent (Roche). After 24 hours, cells were transferred into media containing 4 µg/ml Puromycin for one week. Homogeneity of the derived cell pools was tested by addition of 100 ng/ml DOX for 48 hours and evaluation of GFP expression by fluorescence microscopy.

RNA isolation and quantitative real-time PCR (qPCR)

Total RNA was isolated using the Total RNA Isolation Kit (Roche) according to manufacturer’s instructions. cDNA was generated from 1 µg total RNA per sample using the Verso cDNA synthesis kit (Thermo scientific). Quantitative real-time PCR (qPCR) was performed by using the LightCycler 480 (Roche) and the Fast SYBR Green Master Mix (Applied Biosystems). Expression was normalized using detection of GAPDH using the ΔΔCt methodS4.

Chromatin immunoprecipitation (ChIP) assay

Cross-linking of cells was performed with 1% formaldehyde (Merck) and terminated after 5 minutes by addition of glycine at a final concentration of 0.125 M. Cells were harvested with SDS buffer (50 mM Tris pH 8.1, 0.5% SDS, 100 mM NaCl, 5 mM EDTA) and after pelleting resuspended in IP buffer (2 parts of SDS buffer and 1 part Triton dilution buffer (100 mM Tris-HCl pH 8.6, 100 mM NaCl, 5 mM EDTA, pH 8.0, 0.2% NaN3, 5.0% Triton X-100). Chromatin was sheered by 8 sonication cycles (HTU SONI 130, G. Heinemann) to generate DNA fragments with an average size of 700 bp for qChIP. Preclearing and incubation with polyclonal STAT3 antibody (sc-482, Santa Cruz), SNAIL antibody (#AF3639, R&D systems), or IgG control (#R-5506, Sigma or #AB-108-C, R&D systems) for 16 hours was performed as previously describedS5.

Western blot analysis and antibodies

Cell-lysates were collected in RIPA lysis buffer (50 mM Tris/HCl, pH 8.0, 250 mM NaCl, 1% NP40, 0.5% (w/v) sodium deoxycholate, 0.1% sodium dodecylsulfate, complete mini protease and phosphatase inhibitors (Roche). Lysates were sonicated and centrifuged for 15 min at 4 °C. Per lane 30–60 µg of whole cell lysate was separated on 10% SDS-acrylamide gels and transferred on Immobilon PVDF membranes (Millipore). For immunodetection membranes were incubated with antibodies listed in Supplemental Table S6. Signals from horse-radish-peroxidase (HRP) - coupled secondary antibodies were generated by enhanced chemiluminescence (Millipore) and recorded with a CCD camera (440CF imaging system, Eastman Kodak Co.). Intensities of protein expression signals were quantified using densitometric analysis with the Kodak Molecular Imaging Software v5.0.1.27.

Boyden-chamber invasion assay

To analyse invasion, cell inserts (8.0 µm pore size membrane; Corning) were first coated with Matrigel (BD Bioscience) at a dilution of 3.3 ng/ml in medium without serum. Subsequently, 5 × 104 cells, previously deprived of serum (0.1%) for 24 hours, were seeded on the Matrigel in the upper chamber in serum free medium. As chemo-attractant 20% FBS was placed in the lower chamber. After 48 hours, non-motile cells at the top of the filter were removed and the cells in the bottom chamber were fixed with methanol and stained with DAPI and counted using immunofluorescence microscopy. Results represent the average number of cells in five fields per membrane in triplicate inserts. Experiments were performed in triplicates.

Wound healing assay

Cells were cultured until they reached complete confluence. Mitomycin C [10 ng/ml] was added two hours before generating a scratch using a pipette tip. After washing twice with HBSS to remove Mitomycin C and detached cells, medium was added. Images were captured on an Axiovert Observer Z.1 microscope connected to an AxioCam MRm camera using the Axiovision software (Zeiss) at the respective time-points. Experiments were performed in triplicates.

Metastases formation in a xenograft mouse model

DLD-1 cells stably expressing Luc2 were generated as described previously6/0.2 ml Luciferase tagged cells were injected into the lateral tail vein of NOD/SCID mice using 25-gauge needles. In weekly intervals anesthetized mice were injected intraperitoneal with D-luciferin (150 mg/kg) and imaged 10 minutes after injection using the IVIS Illumina System (Caliper Life Sciences). The acquisition time was 2 minutes. 8 weeks after tail vein injection, mice were sacrificed and examined for lung metastases. For H&E stainings, lungs were fixed with 4% paraformaldehyde and 3 μm sections were stained with haematoxylin and eosin. All studies involving mice were conducted with approval by the local Animal Experimentation Committee (Regierung of Oberbayern). All experiments were performed in accordance with relevant guidelines and regulations.

Clinical samples and immunohistochemistry

RBM47 expression was evaluated using formalin-fixed, paraffin-embedded (FFPE) colon cancer samples of 86 patients who underwent surgical tumor resection at the Ludwig-Maximilians University of Munich (LMU) between 1994 and 2005. Follow-up data were recorded by the tumor registry Munich. All tumors were located on the right side of the colon. Half of the patients had colon cancers with synchronous liver metastases, where metastasis was diagnosed by clinical imaging or liver biopsy. Controls consisted of colon cancer patients without distant metastases at the time of diagnosis and with a disease-free survival of at least 5 years after primary surgical resection. The samples of cases and controls were matched by tumor grade (according to WHO 2000), T-classification (according to TNM Classification of Malignant Tumors 2009), and tumor localization, resulting in 43 matched pairs. Tissue microarrays (TMAs) were generated with 6 representative 1 mm cores of each case. 5 µm TMA sections were prepared, deparaffinized, and stained with anti-RBM47 (Abcam ab167164) rabbit monoclonal antibody on a Ventana Benchmark XT Autostainer with UltraView Universal DAB and alkaline phosphatase detection kits (Ventana Medical Systems). RBM47 expression was scored semi-quantitatively, ranging from complete absence (score 0), weak (score 1), moderate (score 2), or strong expression (score 3). The difference in RBM47 expression between patients without distant metastases was analyzed using the Chi-square test. The study was performed with permission of the ethics committee of the Medical Faculty of the LMU. All analyses were performed in accordance with relevant guidelines and regulations.

Analysis of expression data from public databases

TCGA expression and clinical data was obtained from the TCGA data portal (tcga-data.nci.nih.gov/tcga/tcgaDownload.jsp)https://genome-cancer.ucsc.edu)http://www.broadinstitute.org/ccle/home)www.synapse.org), #syn2623706. Expression and clinical data of GSE39582, GSE37892, GSE24551, GSE17536, and GSE33113 datasets was downloaded from NCBI GEO (www.ncbi.nlm.nih.gov/geo).

Statistical analysis

For CCLE analyses, cell lines from colorectal, breast, lung, bladder, and pancreatic cancer were first grouped according to their epithelial/mesenchymal characteristics: epithelial-like cell lines (expression of CDH1 more than 10-fold higher than expression of VIM) and mesenchymal-like cell lines (expression of VIM more than 10-fold higher than expression of CDH1). Next, the difference in expression of all mRNAs between epithelial-like and mesenchymal-like cancer cell lines was analyzed by a Student’s t-test (p  0.1) and negatively correlate with expression of the mesenchymal marker VIM (r  0.1). Data was analyzed and results visualized using the Multiple Experiment Viewer packagehttps://omics.pnl.gov/software/venn-diagram-plotter). Expression of mRNA signatures was calculated as average of z-score normalized expression of all mRNAs from the signature. The statistics for Kaplan-Meier survival curves was calculated by log-rank test. For binary classification of cases (high/low expression), receiver operated characteristics (ROC) curve analysis was used to determine optimal cut-off values using the Cut-off Finder tool (http://molpath.charite.de/cutoff/assign.jsp). The classification of TCGA COAD and GSE39582 samples into epithelial- or mesenchymal-associated groups in Fig. 2a and b was performed using k-means clustering (Pearson correlation, number of clusters: 2, maximum iterations: 50). The association of RBM47 expression with nodal status or metastasis in TCGA data was calculated by Student’s t-test. The association of RBM47 protein expression with metastasis was calculated using the chi-square test. qPCR and invasion data is expressed as mean ± SD. Differences were analyzed by a two tailed Student’s t-test. Differences in total photon flux in xenograft experiments were determined by a paired two tailed Student’s t-test. Calculations were performed using Prism5 (Graph Pad Software Inc.) and p-values ≤ 0.05 were considered as significant. * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001.