Identification Of The Differentially Expressed Molecules In MSI-H/S Colorectal Cancer And Their Effects On Immunotherapy

Colorectal cancer (CRC) is one of the most common solid tumors in the worldand becomes the one of major cause of cancer-related death. Though higher responserate have been achieved with the improvement in pharmaceutical strategies over thepast decades, the relatively high resistance to systemic chemotherapy and the geneticheterogeneity of colorectal cancer have lead to poor prognosis and unfavorableclinical outcomes. Almost50%of the patients died as the disease progresses.Therefore, it is necessary to develop novel strategies for the treatment of metastaticcolorectal cancer. Personalized medicine, which refers to the tailoring of medicaltreatment to the individual characteristics of each patient, has been verified to beeffective. Examples provided by Herceptin in breast cancer and Gleevec in chronicmyeloid leukaemia (CML) have long carried the mantle for individualized therapy.Microsatellite instability (MSI) status is a potential prognostic factor in CRC,patients with high frequency MSI (MSI-H) have unique characteristics compared withmicrosatellite stability (MSS) cancers, such as proximal anatomic location and severeinflammatory cell infiltration. Several studies have also indicated that patients withMSI-H achieve better prognosis compared with MSS CRCs. In our previous study，tumor samples obtained at baseline from101patients, who were selected from arandom phase II clinical trial of sequential administration of OXA+5-Fu and dendriticcell (DC) vaccine in treatment of metastatic colorectal cancer, were suitable for MSIstatus analysis. Of them60patients received sequential administration ofOXA+5-Fu/LV and DC vaccine (Group A),41received OXA+5-Fu/LV (Group B)only. There were18MSI-H patients, of them,10MSI-H in group A and8MSI-H ingroup B. The results indicated that patients with MSI-H in Group A achieved80%(8/10) objective clinical response, while2patients achieved25%(2/8) objectiveclinical response in Group B.However, the underlying mechanisms remain to be unidentified. Our researchwas intended to identify the biological features of CRCs with MSI-H, so as to elucidate the underlying mechanisms for the improved prognosis in CRCs withMSI-H. Our major findings are summarized in four parts as followed:PartⅠ: Establishment of the colorectal cancer cell models with MSI-H and MSSin vitro.Five colorectal cancer cells with MSI-H（LS174t, LS180, HCT15, HCT116,SW48) were obtained for isolating their corresponding MSS cells according to tumorcell heterogeneity via cloning. The multiple cell clones of each colorectal cancer cellline were obtained by serial dilutions and plating in96-microwell plates. The numberof cell clones of LS174t, LS180, HCT15, HCT16, SW48was71,81,25,65,37,respectively. Genomic DNA was extracted from all separate cell clones. MSI at agiven microsatellite locus was detected by comparison of the allele patterns ofmultiple separate cell clones of each cell line. The microsatellite markers included theBAT25, BAT26, D2S123, D17S250, D5S346. The numbers of different allele patternsfor each microsatellite marker in a given cell line was scored. If different alleles of agiven microsatellite marker were detected in the separate cell clones of a given cellline, the locus was then scored as MSI-positive, and if no different alleles weredetected, then it was MSS. When two or more of the five microsatellite markersshowed MSI, based on the NCI criteria, MSI-high was scored. We isolated multiplecolorectal cancer cells with MSS status, the number of MSS cells in LS174t, LS180,HCT15, HCT16, SW48was21,11,12,10,10and named them LS174t-MSS,LS180-MSS, HCT15-MSS, HCT16-MSS, SW48-MSS, respectively. The rest of celllines were named LS174t-MSI-H, LS180-MSI-H, HCT15-MSI-H, HCT16-MSI-H,SW48-MSI-H, respectively. Finally we established the colorectal cancer cell modelswith MSI-H and MSS in vitro. We detected the different morphological features inMSI-H/S cell models, especially in LS180cell model. We also detected the differentsensitivity of the established colorectal cancer cell models to5-Fu and CPT, andfound that MSI-H cells were more resistant to5-Fu and more sensitive to CPT, inconsistence with previous reports. PartⅡ: Identification of differentially expressed molecules in MSI-H/S colorectalcancer cell models.We profiled gene expression in3pairs of MSI-H/S colorectal cancer cell models,consisting of LS180-MSI-H, LS174t-MSI-H, HCT116-MSI-H, and LS180-MSS,LS174t-MSS, HCT116-MSS, and explored the molecular variation in MSI-H andMSS cells through Whole-Genome gene array (Affymetrix). We identified2216,3446,989genes in LS180, LS174t, HCT116cell model, respectively, the expressionof which were significantly different between MSI-H and MSS groups. Mostly relatedpathway about the1456up-regulated genes in LS180-MSI-H is cytokine-cytokinereceptor interaction, in LS174t-MSI-H is polyadenylation of mRNA pathway, and inHCT116MSI-H is cell cycle pathway. Cluster analysis of the up-regulated genes inthree MSI-H cells showed that as few as7genes was up-regulated in MSI-H,including Fas, Versican, Ezrin, translocated promoter region (TPR), chromosome14open reading frame139(C14orf139), similar to ankyrin repeat domain20family,member A1(LOC727770) and chromosome5open reading frame24(C5orf24).We investigated the protein variation in MSI-H and MSS cells via HumanCytokine Antibody Array (RayBiotech). We found many up-regulated cytokines inMSI-H cells compared with MSS, such as elevated IL-8, TIMP-1, TIMP-2, BMP-4,BMP-5, AR, GH, GDF-15, IGFBP-4, CCL28, CCL20, CXCL10, CXCL5and so on inLS180-MSI-H cells. The mRNA levels of IL-8, TIMP-2, GDF-15, CXCL5andCCL20were consistently upregualated in LS180-MSI-H cells with gene array; withIGFBP-2and IGFBP-3upregulated in LS174t-MSI-H cells.FACS analysis of Fas expression in three MSI-H/S models indicated48.3%Faspositive cells in MSI-H as compared with29.9%in MSS in LS180cell model;80.4%in MSI-H as compared with70.2%MSS in LS174t cell model, and68.7%MSI-H ascompared with48.6%MSS in HCT116cell model.We also found up-regulated tumor associated antigen genes in MSI-H cellswithin three MSI-H/S models, including cancer/testis antigen (CTAG), melanomaantigen family A (MAGE-A) and P antigen family. CTAG-1(NY-ESO-1)，CTAG2(NY-ESO-2) and MAGE-A4were increased in LS180-MSI-H cells by qPCR and FACS analysis. QPCR analysis also revealed up-regulated MAGE-3and MAGE-12in LS174t-MSI-H cells, as well as up-regulated Versican in all three MSI-H/S cellmodels.FACS analysis of the expression of NY-ESO-1，NY-ESO-2，MAGE-A4in10colorectal cancer cell lines (7MSI-H,3MSS), in which MSI status have been perviousascertained, showed that the average expression of NY-ESO-1， NY-ESO-2，MAGE-A4is higher in MSI-H than MSS cells.Part Ⅲ: Potential mechannism analysis for better prognosis of MSI-H CRCWe measured Fas-agonist induced apoptosis in MSI-H cells by FACS analysis,and found that the percentage of Annexin and PI double positive cells is higher inMSI-H than MSS cells. The cell viability is lower in MSI-H than MSS cells via cellcounting CCK8-assay. Thus, the elevated expression of Fas in MSI-H may enhancethe sensitivity to CTLs.To explore whether DCs pulsed with antigen specific peptides were capable ofinducing peptide-specific CTLs in vitro. PBMCs from HLA-A2positive volunteerwere separated by density gradient centrifugation using Ficoll-Hypaque. IsolatedPBMCs were plated (1×107cells/ml per well) into a6-well plate in RPMI1640medium supplemented10%fetal calf serum (FCS). After2hours of incubation,non-adherent cells as T-cell-enriched fractions were removed and placed in a new6-well plate. Adherent cells were cultured in RPMI1640medium supplemented with10%FCS,500U/ml GM-CSF and40ng/ml IL-4. On day6, DCs were collected,stimulated with10μg/ml synthesized peptides, NY-ESO-1(HLA-A2restricted),NY-ESO-2(HLA-A2restricted) and MAGE-A4(HLA-A2restricted) for48hoursand then washed twice.2×105peptide-pulsed DCs were co-cultured with2×106autologous T-cell-enriched non-adherent cells in1ml RPMI1640mediumsupplemented with10%FCS. The cells were further re-stimulated with freshpeptide-pulsed DCs every7days. On day3after second stimulation,20U/ml rhIL-2was supplemented. Media were changed every3days with half fresh media inpresence of rhIL-2and expanded as necessary. On day7after the last stimulation,cytotoxicity assays were performed using a standard CFSE-PI assay. The results showed that CTLs induced by NY-ESO-1, NY-ESO-2and MAGE-A4pulsed DCsexhibited a stronger peptide-specific response against LS180-MSI-H as comparedwith LS180-MSS and T2cells. On the contrary, no specific lysis was observed in Tcells co-cultured with OVA peptide pulsed DCs or untreated DC. In ELISPOT assay,LS180-MSI-H cells，LS180-MSS cells were treated with mitomycin and used asstimulators，with T lymphocytes as responders. The results showed that T cellsinduced by DC pulsed with NY-ESO-1, NY-ESO-2and MAGE-A4elicited moreIFN-γ positive blots upon in vitro restimulation of LS180-MSI-H than LS180-MSScells or OVA.To investigate whether DCs pulsed with antigen specific peptides were capableof inducing peptide-specific CTLs in vivo. HLA-A2/Kbtransgenic mice,6–8weeksold, were subcutaneously immunized three times at a week’s interval with1×106BMDCs pulsed with10μg/ml NY-ESO-1, NY-ESO-2and MAGE-A4or untreatedBMDCs. Splenocytes from immunized mice were re-stimulated withMitomycin-treated LS180-MSI-H, Mitomycin-treated LS180-MSS or OVA peptide.Elispot results showed that more IFN-γ secreting cells were induced in splenocytesgenerated from mice immunized with NY-ESO-1, NY-ESO-2and MAGE-A4pulsedBMDCs restimulated with LS180-MSI-H, than LS180-MSS cells or OVA. Weinvestigated whether peptide-specific CTL could be induced by the immunization ofBMDCs pulsed with NY-ESO-1, NY-ESO-2and MAGE-A4. More potentpeptide-specific lysis of LS180-MSI-H were observed in splenocytes from miceimmunized with DCs pulsed with NY-ESO-1, NY-ESO-2and MAGEA4peptide thanlysis of LS180-MSS or T2cells, while effector cells from mice immunized with OVApulsed BMDCs or untreated BMDCs showed almost no killing of target cells. Overall,these results indicated that immunization of NY-ESO-1, NY-ESO-2and MAGE-A4pulsed BMDCs could induce peptide specific CTLs in vivo.In order to analyze whether the splenocytes from immunized mice indeed havethe ability of tumor rejection, we created an in vivo model of adoptive transfer inC57BL/6nu/numice bearing human tumors. The human colorectal cancer cell linesLS180-MSI-H (HLA-A2positive) were tested for progressive tumor growth in nude mice. C57BL/6nu/numice were challenged subcutaneously with2×106LS180-MSI-Htumor cells in the flank area. The mice were injected intravenously with1×108splenocytes per animal on day5after tumor bearing. This adoptive transfer wasperformed one time, splenocytes induced by NY-ESO-1or NY-ESO-2peptide pulsedBMDCs in HLA-A2/Kbtransgenic mice were able to suppress LS180-MSI-H growthin nude mice. All the control mice developed palpable tumor11days after tumorchallenge. All mice in control groups died between day37and day50after tumorchallenge. In this two groups,1,2of the eight animals were tumor free ever since,respectively, while no significant protection or improvement in survival was observedin control groups. The experiments demonstrated that immunization of mice with theNY-ESO-1or NY-ESO-2pulsed DCs induces potent protective immune responseagainst LS180-MSI-H tumor cells.Part Ⅳ: Identification of cancer associated antigen expression on CRCTumor samples obtained from186patients with colorectal cancer regardless oftreatments, who were selected from four hospitals, from2005to2009.63MSI-H and123MSS were identified. In this study, the primary tumor lesion was resected bysurgery and IHC was performed using NY-ESO-1or NY-ESO-2antibodies.4-μmsections of paraffin-embedded tissue were used for immunostaining NY-ESO-1(1:250,Bioss), NY-ESO-2(1:700, Bioss) and slides were then deparaffinized and rehydrated.Endogenous peroxidase was blocked with3%hydrogen peroxide in methanol for20min. Antigen retrieval was achieved with heating in a pressure cooker using10mmol/L of sodium-citrate buffer (pH6.0). Sections were incubated with20%goatblood serum at room temperature for blocking nonspecific binding. After washingwith PBS, sections were incubated with primary antibody at37℃for two hours, andthen incubated with secondary antibody (HRP-R/M, EnVision, Dako) at37℃for30mins. All antibodies were visualized using3,3-diaminobenzidine as chromagen(DAB). Negative controls omitted the primary antibody.NY-ESO-1+expression were detected as positive in36from63MSI-H patients(57.14%), as compared with53negative from123MSS patients (43.09%).NY-ESO-2+expression were detected as positive in39from63MSI-H patients (61.90%), significantly higher than MSS patients who achieved38.21%(47/123)(P=0.0022).We analyzed the3-year survival rate of patients with NY-ESO-1and NY-ESO-2protein expression, according to MSI status. The results showed that3-year survivalrate of MSI-H patients with NY-ESO-1positive was74.18%, compared with55.71%survival in patients with NY-ESO-1negative (P=0.2133).The results also indicated that3-year survival rate of MSI-H patients withNY-ESO-2positive was80.96%, significantly higher than in patients with NY-ESO-2negative46.05%(95%CI,0.08-0.62; P=0.0109). Kaplan-Meier Analysis also showedthat3-year survival rate of MSI-H patients with NY-ESO-2positive was significantlyhigher than the patients with NY-ESO-2negative (P=0.0044).These results showed that the expression of NY-ESO-2in MSI-H patients wassignificantly higher than MSS, and3-year survival rate of MSI-H patients withNY-ESO-2positive was also significantly higher than that of patients with NY-ESO-2negative. The results suggested that MSI-H CRCs had a better prognosis, mayresulting from the expression of NY-ESO-2protein, which promotes theimmunogenicity of MSI-H colorectal cancer cells.In conclusion, we successfully established the colorectal cancer cell models withMSI-H and MSS in vitro, and identified the differences in drug sensitivity and MSIstatus between MSI-H/S cell types. We also identified and characterized thedifferentially expressed molecules in MSI-H/S colorectal cancer cell models throughWhole-Genome gene array, showing that Fas as well as some tumor antigen associategenes are up-regulated in MSI-H cells within the MSI-H/S cell models.MSI-H cells with tumor antigen expression (NY-ESO-1, NY-ESO-2, MAGE-A4)could elite more effective immune response than MSS. Fas protein up-regulated inMSI-H tumor cells may enhance the sensitivity to effector cells. The expression ofNY-ESO-2in MSI-H colorectal cancer was significantly higher than in MSS, and3-year survival rate of MSI-H patients with NY-ESO-2positive was also significantlyhigher than NY-ESO-2negative patients. The results indicated patients with MSI-H colorectal cancer could promote its immunogenicity through the expression ofNY-ESO-2protein, enhance the sensitivity to effector cells by upregulation of Fasprotein, all these effects would contribute to its improved prognosis.