Systematic Selection And Primary Identification Of Differential Expression Genes In The Crosstalk Of Colorectal Cancer Cells And Microenvironment During Epithelial Mesenchymal Transition Process

Based on statistical data of American Cancer Society(ACS),one in four American death toll dies of cancer every year.The morbidity and mortality of cancer are staying at a high level in worldwide.The aspects which induce the high morbidity and mortality of cancer are cancer distant metastasis and chemotherapeutic resistance.Since the mechanisms of neoplasm development are more diversified,cancer therapy become personalized.Epithelial mesenchymal transition(EMT)is such a process that epithelial cells lose their epithelial characteristics and obtain mesenchymal ones under the regulation of inner and outer aspects.EMT process plays important roles in embryonic development and wound healing.Studies in these years show that the EMT process also takes part in the cancer development.Cells undergoing EMT will lose cell polarity and obtain invasive and migrating abilities.On the molecular level,EMT-characteristic cells are down-regulated epithelial biomarkers,such as E-cadherin and occludin,and up-regulated mesenchymal biomarkers,such as,N-cadherin,vimentin,fibronectin.Meantime,cells undergoing EMT prove to have the same properties as those of stem cells,which will benefit to metastasis and drug resistance.Complicated signaling pathways mediated by cytokines regulate the EMT process,such as EGF,VEGF,HGF,IGF,FGF,Rho,PI3K,Wnt,etc.Transcriptional factors,such as Snail,ZEB,Twist,are activated by signal pathways,and directly or indirectly affect the expression level of E-cadherin.Tumor microenvironment is suggested to play a critical role in EMT process and metastasis.Tumor budding is characterized as single tumor cell or a cluster of tumor cells(up to five)located at the invasive front of tumor,employing invasive and migratory abilities.Tumor budding is recognized as EMT-characteristic tumor cells which employed both epithelial and mesenchymal phenotypes.Previous study of our lab showed that tumor budding expressed both epithelial and mesenchymal biomarkers,which is an ideal model to study EMT in vivo.The purpose of this project is to select potent EMT biomarkers during the crosstalk regulation of tumor cells and the microenvironment,and laid the foundation of figuring out the mechanisms of EMT.1 Potent EMT markers selection in tumor budding and its microenvironment.We accurately captured 500-600 tumor buddings and the stromal components at the invasive front of three cases of colorectal cancer specimens using laser capture microdissection after the pathological screening was executed to rule out the tumor budding-rare specimens.Moreover,we captured intestinal epithelial cells,cancer cells in the center of primary tumor,and the stromal components of both of them as controls.We lately employed high-throughout screening and systematical bioinformatics to select and analyze potent EMT genomic markers.Comparing to cancer cells in the center of primary tumor,we found that there were 494 differentially expressed genes in tumor budding,including 432 up-regulated genes and 62 down-regulated genes(P<0.05;fold change ?2&? 0.5).Then we constructed EMT regulation network based on public data.And we calculated the Pearson Correlation Coefficient between 494 alternative genes and verified EMT biomarkers in the regulatory network,and mapped 494 genes in the ready-built EMT regulatory network based on protein-protein interaction(PPI).Forty-nine genes which had interaction with verified EMT biomarkers based on online PPI network were dug up and recognized as potent EMT biomarkers.To figure out how microenvironment affects EMT and cancer metastasis,we executed data mining on the surrounding stroma of tumor budding.Statistical analysis found that 3495 genes differentially expressed in stromal components compared to the surrounding stroma of colorectal cancer cells in cancer center area,including 1752 up-regulated genes and 1743 down-regulated genes.Since we lost one case of the stroma of tumor budding and the cDNA sample was non-renewable,we executed no statistical analysis in the stroma of tumor budding.We calculated the Pearson correlation coefficient between 3495 alternative genes and the verified EMT biomarkers in the regulatory network,and mapped these genes in the ready-built EMT regulatory network based on PPI.Sixty-six genes were figured out to have interaction with verified EMT biomarkers on protein level,implying that these 66 genes might directly or indirectly be involved in the EMT program and metastatic dramas.In order to further explore the mechanisms of EMT development,we executed in-depth data mining with tumor budding data.We defined intestinal epithelial cell,cancer cells in the center of primary tumor and tumor budding as three processing stages of cancer development.We constructed the transcriptional regulation networks of three stages based on gene expression value.We got the gene union of intestinal epithelial cells,cancer cells in the center of primary tumor and tumor buddings after comparing these three groups in pairs.Then we calculated the correlation coefficient between these genes in each group,and came to the topological overlap,which indicated the connection degree between one gene and other genes.We classified genes who shared the same expression trend based on the topological overlap using average linkage hierarchical clustering.Then we got the gene models of intestinal epithelial cells,cancer cells in the center of primary tumor and tumor buddings which the performance category of genes was exhibited by evolutionary tree.Comparing gene models of tumor buddings to other two groups,we obtained a series of gene sets which made the tumor budding cells unique.Then we filtrated 5 gene models in the gene models of tumor budding(PC1_var>0.90,density>0.7).Inflammatory cytokines,metabolic products and epigenetic regulators were all involved in the EMT program;and these mechanisms all played critical roles in cancer therapy-resistance.What is more,cancer stem cells(CSCs)in the primary tumor were not inhibited for most of the antagonists or inhibitors which could lead to a non-complete treatment or neoplastic recurrence.To figure this out,we selected a bunch of potent biomarkers from tumor budding and its stromal microenvironment,concentrating on EMT-related inflammatory,epigenetic,metabolic and sternness regulators,and genes that probably affected drug-resistance in the EMT program.We presented functional annotation on the differentially expressed genes in tumor budding and the stroma individually,and selected genes possessing functions in inflammation,epigenetics,metabolism,cancer sternness and drug resistance based on KEGG database.Then we calculated the correlation index between genes based on gene-gene interaction in online database.Finally,we selected genes based on three parameters,including Betweenness Centrality,Degree and Clustering Coefficient.We selected 56 inflammatory factors,80 epigenetic regulators,95 metabolic mediators,87 cancer sternness-related genes,and 17 drug resistance-related genes.2 Identification of potent and classic EMT biomarkers in colorectal cancers using Nanostring nCounter systemWe elaborately picked out 268 genes,including embraced 202 potent new EMT biomarkers selected from tumor budding and its stromal environment through bioinformatics based on microarray and 66 classic EMT biomarkers(as control)for identification.We intensively laser-captured single or clusters of tumor budding cells and the surrounding stromal components at the invasive front of 4 cases of colorectal cancer specimens and detected expression of 268 genes via nCounter system.We normalized gene expression level of tumor buddings and the surrounding stroma based on the fluorescence signal value(housekeeping genes were GAPDH and ABCF1),and executed t-test and ratio calculation.As a conclusion,220 out of 268 genes(82%)shared the same expression trend in tumor budding data,and 222 out of 268 genes(83%)shared the same expression trend in the microenvironment of tumor buddings.We surprisingly found that there were 18 genes differentially expressed both in tumor budding and the microenvironment.Among these genes,IL6,GAB1,VANGL1 were up-regulated both in tumor budding and the microenvironment,implying crosstalk between tumor cells and the microenvironment.We concluded a cross-regulation network to show the interaction of epithelial tumor budding cell and its microenvironment during EMT program.EMT program depended on the activation of transcriptional factors,such as SNAIL,TWIST and ZEB family members following the signal transmission from WNT,TGF?,PI3K/MAPK,and growth factors pathways,etc.,and the signalized interactions among them as we identified in the diagram.Genes,which only differentially expressed in the stroma of tumor budding such as PTPRM,SMURF2,SMAD4,RUNX1T1,SRC,GRB were individually the mediators in these pathways and were theoretically recognized as potent mediators of EMT.Genes,which only differentially expressed in tumor budding such as CAV1/2,ITGA,TCF/LEF,KIT also participated in the regulation of EMT,in theoretically.However,it was worth noting that genes,such as PTPRF,FUS,DCN,LTBP1,INHBA,SMAD2/3,ARHGAP5,IL6,PARD6A,WAS,ITGB,ROHA,MET,IGF2,KRAS,MMP1 which all differentially expressed both in tumor budding and its microenvironment as we identified,might affect the transition from epithelial cells to mesenchymal cells,in theory.In-depth researches on these genes will benefit to figure out the mechanism of EMT.To study the relation between potent EMT and metastasis biomarkers and tumor therapy,we collected 131 drugs from cancer therapy.We executed enrichment analysis and hypergeometric distribution test on selected genes and drugs,and executed fisher exact test on target binding event which were less than five.We found that genes,such as ACTA2,BGN,EHD2,FN1,LGALS1,THBS2,IGFBP7,MMP2,OLFML3,had target binding to clinical drug GSK269962A(P<0.0001);genes,such as BGN,CDH1,CLCA1,TBXA2R,MET,PDGFRA,had target binding to axitinib(P = 0.017);genes,MEF2C and TBXA2R,had target binding to nilotinib P = 0.036).3 The identification of potent EMT biomarkers in colorectal cancer cell lines.We picked out potent EMT biomarker MMP1 and LGALS1 for further verification.We constructed EMT model in HT29 and SW480 colorectal cell lines using TGF? and TNFa.We found that MMP1 and LGALS1 was upregulated in the EMT models of colorectal cell lines,which implied that MMP1 and LGALS1 might take part in EMT development in colorectal cancer.4 The correlation analysis of potent EMT biomarkers with classic EMT biomarkers in colorectal tissues and prognosis analysis.To figure out the relationship of MMP1 and LGALS1 and EMT,we analyzed the expression correlations between candidate genes and classic EMT biomarkers,including CDH1,CDH2,VIM,SNAI1,TWIST 1,using tumor budding data of 7 cases of patients-derived colorectal cancer specimens.We found that LGALS1 was strongly and positively correlated with ZEB1(r = 0.821;P = 0.023);MMP1 was strongly and positively correlated with CDH2(r = 0.891;P = 0.007).We eliminated the missing data of candidate gene expression value.Then we retained 83 cases to analyze the prognostic significance of MMP1 on colorectal cancer patients,and 82 cases for LGALS1,We found that LGALS1,MMP1 were up-regulated in colorectal cancer,compareing to normal tissue(P<0.05).Surprisingly,we found that LGALS1(P = 0.032)and MMP1(P = 0.006)were related with the survival rate of colorectal cancer patients.LGALS1 and MMP1 could be recoganized as prognosis biomarkers of colorectal cancer patients.From the above results,we draw the following conclusions:(1)We screened and got 106 potential EMT biomarkers in tumor budding and its microenvironment via microarray expression profile,systematically bioinformatical data mining and nCounter identification experiments.And we analyzed the crosstalk between tumor cell and microenvironment during EMT process through the construction of EMT regulation network.Meantime,we predicted 15 potent EMT biomarkers which probably had target binding with 3 cancer durgs.This was the first time to study the crosstalk regulation of tumor cells and the microenvironment on EMT process using patient-derived colorectal cancer specimens,and the huge amounts of data laid fundations for in-depth analysis on EMT process.(2)Using colorectal cancer cell lines model of EMT induced by classic EMT inducer TGF? and TNF?,we found that potent EMT biomarker MMP1 and LGALS1 were upregulated in the EMT models of colorectal cancer cells,which implied the potential correlation between MMP1 and LGALS1 and EMT in colorectal cancer.(3)Through multi-stage validation experiments in colorectal cancer specimens,we found that LGALS1 and MMP1 were positively correlated with classic EMT biomarkers in expression level,and these two genes were respectively prognosis biomarkers of colorectal cancer patients.Overexpression LGALS1 and MMP1 predicted worse clinical outcomes of patients.Moreover,LGALS1 and cancer drug GSK269962A had potential combination effect,which implied that LGALS1 might be a promising target for cancer therapy.