학술논문

Thymidylate synthase is functionally associated with ZEB1 and contributes to the epithelial-to-mesenchymal transition of cancer cells
Document Type
Report
Source
Journal of Pathology. June, 2017, Vol. 242 Issue 2, p221, 13 p.
Subject
Stem cells -- Analysis
Cancer cells -- Analysis
Health
Language
English
ISSN
0022-3417
Abstract
Byline: Aarif Siddiqui, Maria Eleni Vazakidou, Annemarie Schwab, Francesca Napoli, Cristina Fernandez-Molina, Ida Rapa, Marc P Stemmler, Marco Volante, Thomas Brabletz,Paolo Ceppi Keywords: thymidylate synthase; epithelial-to-mesenchymal transition; ZEB1; micro-RNAs; solid tumours Abstract Thymidylate synthase (TS) is a fundamental enzyme of nucleotide metabolism and one of the oldest anti-cancer targets. Beginning from the analysis of gene array data from the NCI-60 panel of cancer cell lines, we identified a significant correlation at both gene and protein level between TS and the markers of epithelial-to-mesenchymal transition (EMT), a developmental process that allows cancer cells to acquire features of aggressiveness, like motility and chemoresistance. TS levels were found to be significantly augmented in mesenchymal-like compared to epithelial-like cancer cells, to be regulated by EMT induction, and to negatively correlate with micro-RNAs (miRNAs) usually expressed in epithelial-like cells and known to actively suppress EMT. Transfection of EMT-suppressing miRNAs reduced TS levels, and a specific role for miR-375 in targeting the TS 3'-untranslated region was identified. A particularly relevant association was found between TS and the powerful EMT driver ZEB1, the shRNA-mediated knockdown of which up-regulated miR-375 and reduced TS cellular levels. The TS-ZEB1 association was confirmed in clinical specimens from lung tumours and in a genetic mouse model of pancreatic cancer with ZEB1 deletion. Interestingly, TS itself appeared to have a regulatory role in EMT in cancer cells, as TS knockdown could directly reduce the EMT phenotype, the migratory ability of cells, the expression of stem-like markers, and chemoresistance. Taken together, these data indicate that the TS enzyme is functionally linked with EMT and cancer differentiation, with several potential translational implications. Copyright [c] 2017 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd. CAPTION(S): Supplementary figure legends Appendix S1. Supplementary materials and methods Figure S1 Levels of expression of TS (TYMS) mRNA and protein in cancer cells from the NCI-60 panel. (A) Log-transformed TS (TYMS) mRNA expression levels in epithelial (green) and mesenchymal-like (red) cancer cell lines from the NCI-60 Novartis array dataset. (B) TYMS mRNA expression levels in NCI-60 cells divided as epithelial, low-mesenchymal, and high-mesenchymal status according to a VIM/CDH1 gene ratio. Data are from the NCI-60 Novartis array dataset. (C) TS protein levels (log2 scale) from a publicly available proteomics database (source: http://www.proteindb.org), with the cells lines from the NCI-60 panel grouped for EMT status. (D) TS levels in individual cells from the NCI-60 panel, as evaluated in C. Data are the average of three independent sources from the proteindb.org website, shown in E. P-values are two-tailed t-tests. Figure S2 Levels of expression of selected mRNAs and cell-doubling times in cancer cells from the NCI-60 panel. (A) mRNA expression levels of indicated genes in cancer cell lines from the NCI-60, divided by EMT status. TYMS=thymidylate synthase; RRM1=ribonucleoside-diphosphate reductase large subunit; PCNA=proliferating cell nuclear antigen; MIKI67=Ki-67; DHFR=dihydrofolate reductase, CCND1=cyclin D1. Data are from the Novartis array dataset. (B) Doubling-time data of NCI-60 cancer cells, divided by EMT status (doubling times are from the Developmental Therapeutic Program of the NCI; EMT status of the cells is according to ref 19). Figure S3 Immunofluorescence analysis of A549 cells and immunoblotting analyses of tissues and cell lines. (A) Immunofluorescence of A549 cells as in Figure TS levels are increased in cells with a mesenchymal-like phenotype. (A) Log-transformed TS (TYMS) mRNA expression levels in epithelial (green) and mesenchymal-like (red) cancer cell lines from the NCI-60 Stanford array dataset. Lines indicate median levels in both groups. (B) Correlation between E-cadherin (CDH1), ZEB1, and miR-200c and the TYMS gene. R-values are Pearson coefficients. (C) Western blot analysis of TS protein levels and EMT markers in cancer cells. [beta]-Actin was used as a loading control. (D) Densitometric quantification (ImageJ) of TS levels expressed as ratios with [beta]-actin in the western blots as in C. (E) Quantification of TS levels expressed as band area intensity in the western blots as in C, dividing for ZEB1 negativity or positivity. (F) Western blot analysis of TS and ZEB1 in a panel of non-small cell lung cancer (NSCLC) cell lines. (G) Western blot analysis of TS and EMT markers in A549 cells stably overexpressing a scrambled control or an shRNA targeting ZEB1. (H) qPCR analysis of EMT markers and TS in shZEB1 and control A549 cells. Data are[+ or -]SD, normalized to GAPDH, and expressed as fold change relative to control cells. P values are two-tailed t-tests. **p TS suppression by shRNA reduces EMT and the migratory phenotype. (A) Western blot analysis of TS and indicated EMT and CSCs markers in A549 cells infected with shRNA targeting two independent TS mRNA sequences, compared with scrambled-infected cells (PLKO). (B) Immunofluorescence of A549 cells stably overexpressing a scrambled control or an shRNA targeting TS, stained for E-cadherin and vimentin. The cells were stained concomitantly with the shZEB1 shown in the supplementary material, Figure S3A, with the same PLKO cells as a control; the field shown here for PLKO overlaps partially with that in Figure S3A (supplementary material). DAPI was used as a nuclear counterstain. (C) FACS plots of cells as in A 6 days after infection, stained with PI (Nicoletti). The percentages of subG1 nuclei are indicated. (D) Colony formation of cells as in A stained after 7 days with crystal violet and quantified in triplicate dishes. (E) Representative pictures and (F) quantification of wound-healing assay on A549 cells infected with non-targeting or with shTS overexpressing viruses. (G) Plots of relative wound densities of A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses. (H) Real-time analysis of cell death in A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses after treatment with cisplatin (CDDP) at 750A[micro]m dose. Cell death was monitored with the incorporation of a green dye by the Incucyte ZOOM system. (I) Western blot analysis on TS in A549 cells transfected with siRNA smart pool targeting TS, compared with non-targeting control cells. (J) FACS plots of cells as in I and stained with Aldefluor reagent, to quantify the percentage of CSCs. Gates were set using DEAB reagent; percentages indicate the proportion of ALDH1high cells. (K) Quantification of sphere number in cells treated as in I and plated in sphere-forming conditions in low-adherence plates. (L) Western blot analysis on TS, ZEB1, and E-cadherin in A549 cells infected with a vector overexpressing TS, compared with control-infected cells. (M) FACS plots of cells as in L and stained with Aldefluor reagent. P values are two-tailed t-tests. *p Figure S4 Assessment of DNA fragmentation and TS (TYMS) expression related to EMT (including GSE datasets). (A) Quantification of the percentage of subG1 nuclei in A549 and Calu-1 cells treated with 1 mm 5-FU, related to Figure A mesenchymal-like phenotype and higher TS levels are associated with increased 5-FU resistance. (A) Sensitivity of NCI-60 cancer cell lines to 5-FU, expressed as log-transformed values. Cells were divided into epithelial, mesenchymal low, and mesenchymal high, according to a pre-determined EMT gene ratio. Data are from DTP of the National Cancer Institute (NCI). (B) IC50 values of NCI-60 cancer cell lines relative to 5-FU. Data are from the GI50 database (NCI). (C) Representative images of breast (MDA-MB231, T47D), colorectal (HCT116, KM12), lung (A549, Calu-1), and ovarian (Ovcar5, Ovcar8) cancer cells treated with the indicated concentrations of the drug. (D) Dose-response curves and IC50 calculations of cells as in C treated for 48h with the indicated concentrations of 5-FU. Numbers indicate IC50 values[+ or -]SD. Percentages of dead cells were calculated with FACS dead/live cells gating in quadruplicate conditions. (E) FACS plots of Ovcar8 and Ovcar5 cells treated with 1mm 5-FU and stained after 48h with PI (Nicoletti). Percentages of the subG1 fraction (gated) are indicated. (F) Quantification of the percentage of the subG1 fraction in cells treated as in E. (G) Western blot of E-cadherin, ZEB1, and TS in A549 cells treated for 48h with 10ng/A[micro]l TGF-[beta]. (H) Percentage of viable cells in A549 cells untreated and treated with 10ng/A[micro]l TGF-[beta] and with 5-FU or control for the indicated time-points. All non-treated controls were given DMSO concentrations equivalent to the highest dose of 5-FU administered. NT=non-treated. P values are two-tailed t-tests. *p Figure S5 The effects of the TS 3'-UTR sequence on mRNA stability and unbiased correlation analysis of miRNAs negatively correlated with TYMS expression. (A) Left: FACS plots showing the red fluorescence of A549 cells overexpressing a dsRed expression plasmid fused with the TS 3'-UTR (after the stop codon), compared with the control dsRed vector. Right: representative white-light and red fluorescence pictures of the cells. (B) Unbiased correlation analysis of the miRNAs most negatively correlated with the TYMS gene in the panel of NCI-60 cancer cells, according to two independent databases (miRConnectL and miRConnectQ). Data are from the miRConnect.org website. Figure S6 The abundance of miR-375 in HCT116 cells and colon cancer cell lines and the interaction of miR-375 with 5-FU on A549 cell viability. (A) Absolute quantification of miR-375 in HCT116 cells by the standard curve method. The average number of (mature) miR-375 molecules calculated per HCT116 cell[+ or -]SD is indicated. (B) Relative levels of miR-375 in the colon cancer subpanel of the NCI-60 cells (data from Cellminer.com). (C) Viability of A549 cells overexpressing the indicated pre-miRNAs and treated with the indicated concentrations of 5-FU. (D) Viability of A549 cells overexpressing TS or an empty vector, transfected with pre-miR-375 and pre-miR-control and treated with the indicated concentration of 5-FU. Assays were performed with MTS reagent and values normalized to untreated controls. P-values are two-tailed t-tests. Figure S7 Assessment of DNA fragmentation, cell proliferation, ALDH1 status, sphere growth, and the effects of shTS on miRNA levels. (A) Quantification of subG1 cells (expressed as percentages) in cells treated as in Figure TS suppression by shRNA reduces EMT and the migratory phenotype. (A) Western blot analysis of TS and indicated EMT and CSCs markers in A549 cells infected with shRNA targeting two independent TS mRNA sequences, compared with scrambled-infected cells (PLKO). (B) Immunofluorescence of A549 cells stably overexpressing a scrambled control or an shRNA targeting TS, stained for E-cadherin and vimentin. The cells were stained concomitantly with the shZEB1 shown in the supplementary material, Figure S3A, with the same PLKO cells as a control; the field shown here for PLKO overlaps partially with that in Figure S3A (supplementary material). DAPI was used as a nuclear counterstain. (C) FACS plots of cells as in A 6 days after infection, stained with PI (Nicoletti). The percentages of subG1 nuclei are indicated. (D) Colony formation of cells as in A stained after 7 days with crystal violet and quantified in triplicate dishes. (E) Representative pictures and (F) quantification of wound-healing assay on A549 cells infected with non-targeting or with shTS overexpressing viruses. (G) Plots of relative wound densities of A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses. (H) Real-time analysis of cell death in A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses after treatment with cisplatin (CDDP) at 750A[micro]m dose. Cell death was monitored with the incorporation of a green dye by the Incucyte ZOOM system. (I) Western blot analysis on TS in A549 cells transfected with siRNA smart pool targeting TS, compared with non-targeting control cells. (J) FACS plots of cells as in I and stained with Aldefluor reagent, to quantify the percentage of CSCs. Gates were set using DEAB reagent; percentages indicate the proportion of ALDH1high cells. (K) Quantification of sphere number in cells treated as in I and plated in sphere-forming conditions in low-adherence plates. (L) Western blot analysis on TS, ZEB1, and E-cadherin in A549 cells infected with a vector overexpressing TS, compared with control-infected cells. (M) FACS plots of cells as in L and stained with Aldefluor reagent. P values are two-tailed t-tests. *p TS suppression by shRNA reduces EMT and the migratory phenotype. (A) Western blot analysis of TS and indicated EMT and CSCs markers in A549 cells infected with shRNA targeting two independent TS mRNA sequences, compared with scrambled-infected cells (PLKO). (B) Immunofluorescence of A549 cells stably overexpressing a scrambled control or an shRNA targeting TS, stained for E-cadherin and vimentin. The cells were stained concomitantly with the shZEB1 shown in the supplementary material, Figure S3A, with the same PLKO cells as a control; the field shown here for PLKO overlaps partially with that in Figure S3A (supplementary material). DAPI was used as a nuclear counterstain. (C) FACS plots of cells as in A 6 days after infection, stained with PI (Nicoletti). The percentages of subG1 nuclei are indicated. (D) Colony formation of cells as in A stained after 7 days with crystal violet and quantified in triplicate dishes. (E) Representative pictures and (F) quantification of wound-healing assay on A549 cells infected with non-targeting or with shTS overexpressing viruses. (G) Plots of relative wound densities of A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses. (H) Real-time analysis of cell death in A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses after treatment with cisplatin (CDDP) at 750A[micro]m dose. Cell death was monitored with the incorporation of a green dye by the Incucyte ZOOM system. (I) Western blot analysis on TS in A549 cells transfected with siRNA smart pool targeting TS, compared with non-targeting control cells. (J) FACS plots of cells as in I and stained with Aldefluor reagent, to quantify the percentage of CSCs. Gates were set using DEAB reagent; percentages indicate the proportion of ALDH1high cells. (K) Quantification of sphere number in cells treated as in I and plated in sphere-forming conditions in low-adherence plates. (L) Western blot analysis on TS, ZEB1, and E-cadherin in A549 cells infected with a vector overexpressing TS, compared with control-infected cells. (M) FACS plots of cells as in L and stained with Aldefluor reagent. P values are two-tailed t-tests. *p TS suppression by shRNA reduces EMT and the migratory phenotype. (A) Western blot analysis of TS and indicated EMT and CSCs markers in A549 cells infected with shRNA targeting two independent TS mRNA sequences, compared with scrambled-infected cells (PLKO). (B) Immunofluorescence of A549 cells stably overexpressing a scrambled control or an shRNA targeting TS, stained for E-cadherin and vimentin. The cells were stained concomitantly with the shZEB1 shown in the supplementary material, Figure S3A, with the same PLKO cells as a control; the field shown here for PLKO overlaps partially with that in Figure S3A (supplementary material). DAPI was used as a nuclear counterstain. (C) FACS plots of cells as in A 6 days after infection, stained with PI (Nicoletti). The percentages of subG1 nuclei are indicated. (D) Colony formation of cells as in A stained after 7 days with crystal violet and quantified in triplicate dishes. (E) Representative pictures and (F) quantification of wound-healing assay on A549 cells infected with non-targeting or with shTS overexpressing viruses. (G) Plots of relative wound densities of A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses. (H) Real-time analysis of cell death in A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses after treatment with cisplatin (CDDP) at 750A[micro]m dose. Cell death was monitored with the incorporation of a green dye by the Incucyte ZOOM system. (I) Western blot analysis on TS in A549 cells transfected with siRNA smart pool targeting TS, compared with non-targeting control cells. (J) FACS plots of cells as in I and stained with Aldefluor reagent, to quantify the percentage of CSCs. Gates were set using DEAB reagent; percentages indicate the proportion of ALDH1high cells. (K) Quantification of sphere number in cells treated as in I and plated in sphere-forming conditions in low-adherence plates. (L) Western blot analysis on TS, ZEB1, and E-cadherin in A549 cells infected with a vector overexpressing TS, compared with control-infected cells. (M) FACS plots of cells as in L and stained with Aldefluor reagent. P values are two-tailed t-tests. *p TS suppression by shRNA reduces EMT and the migratory phenotype. (A) Western blot analysis of TS and indicated EMT and CSCs markers in A549 cells infected with shRNA targeting two independent TS mRNA sequences, compared with scrambled-infected cells (PLKO). (B) Immunofluorescence of A549 cells stably overexpressing a scrambled control or an shRNA targeting TS, stained for E-cadherin and vimentin. The cells were stained concomitantly with the shZEB1 shown in the supplementary material, Figure S3A, with the same PLKO cells as a control; the field shown here for PLKO overlaps partially with that in Figure S3A (supplementary material). DAPI was used as a nuclear counterstain. (C) FACS plots of cells as in A 6 days after infection, stained with PI (Nicoletti). The percentages of subG1 nuclei are indicated. (D) Colony formation of cells as in A stained after 7 days with crystal violet and quantified in triplicate dishes. (E) Representative pictures and (F) quantification of wound-healing assay on A549 cells infected with non-targeting or with shTS overexpressing viruses. (G) Plots of relative wound densities of A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses. (H) Real-time analysis of cell death in A549 cells infected with non-targeting or with shTS and shZEB1 overexpressing viruses after treatment with cisplatin (CDDP) at 750A[micro]m dose. Cell death was monitored with the incorporation of a green dye by the Incucyte ZOOM system. (I) Western blot analysis on TS in A549 cells transfected with siRNA smart pool targeting TS, compared with non-targeting control cells. (J) FACS plots of cells as in I and stained with Aldefluor reagent, to quantify the percentage of CSCs. Gates were set using DEAB reagent; percentages indicate the proportion of ALDH1high cells. (K) Quantification of sphere number in cells treated as in I and plated in sphere-forming conditions in low-adherence plates. (L) Western blot analysis on TS, ZEB1, and E-cadherin in A549 cells infected with a vector overexpressing TS, compared with control-infected cells. (M) FACS plots of cells as in L and stained with Aldefluor reagent. P values are two-tailed t-tests. *p Figure S8 Characteristics of the 61 NSCLC patients and their TS expression levels according to grade of tumour differentiation. (A) Characteristics of the 61 NSCLC patients shown in Figure TS protein levels are associated with ZEB1 and E-cadherin in NSCLC tissues. (A) Quantification of TS expression levels in tissues from 61 NSCLC patients. Patients were divided into two groups according to ZEB1 IHC positivity or negativity. (B) Quantification of TS expression levels in tissues from NSCLC patients. Patients were divided into two groups according to E-cadherin IHC negativity (either negative or cytoplasmic diffuse staining pattern) or positivity (membrane pattern of staining). Patients were further divided into adenocarcinoma (ADC) and squamous cell carcinoma (SSC). The P values are from a Mann-Whitney U-test. (C) Representative microscopic images of an NSCLC case with high (#1) and low (#2) TS staining, with corresponding staining of E-cadherin and ZEB1. (D) Survival curves of NSCLC patients divided by high and low expression of TS (TYMS gene) and E-cadherin (CDH1 gene) from a transcriptomic database and divided for histological diagnosis (ADC and SSC). The P values indicated are log-rank. ns=non-significant.