Polymorphisms in microRNA binding sites of mucin genes as predictors of clinical outcome in colorectal cancer patients

Carcinogenesis, Jan 2017

Vymetalkova, Veronika, Pardini, Barbara, Rosa, Fabio, Jiraskova, Katerina, Di Gaetano, Cornelia, Bendova, Petra, Levy, Miloslav, Veskrnova, Veronika, Buchler, Tomas, Vodickova, Ludmila, et al.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:


Polymorphisms in microRNA binding sites of mucin genes as predictors of clinical outcome in colorectal cancer patients

Carcinogenesis Polymorphisms in microRNA binding sites of mucin genes as predictors of clinical outcome in colorectal cancer patients Veronika Vymetalkova 1 2 Barbara Pardin 0 Fabio Rosa 0 Katerina Jiraskova 1 2 Cornelia Di Gaetan 0 Petra Bendova 1 2 5 Miloslav Levy 4 Veronika Veskrnova 3 Tomas Buchler 3 Ludmila Vodickova 1 2 5 Alessio Naccarat 2 Pavel Vodicka 1 2 5 0 Hliucm , an Genetics Foundation, 10126 Turin , Italy 1 Ilnics,titute of Biology and Medical Genetics, First Faculty of Medicine, Charles University , 11000 Prague, Czech Repub 2 Department of Molecular Biology of Cancer, Institute of Experimental Medicine , 14200 Prague, Czech Repub 3 Dnedpartment of Oncology, Thomayer Hospital and First Faculty of Medicine, Charles University , 11000 Prague , Czech Republic 4 Dice,partment of Surgery, First Faculty of Medicine, Charles University and Thomayer University Hospital , 14200 Prague , Czech Republic a 5 By,iomedical Centre, Faculty of Medicine in Pilsen, Charles University in Prague , 30605 Pilsen, Czech Republ 6 ,Department of Medical Sciences, University of Turin , 10126 Turin, Ital Polymorphisms in microRNA (miRNA) binding sites may affect miRNA/target gene interaction, resulting in differential mRNA/protein expression and susceptibility to common diseases. Mucins have been identified as markers of adverse prognosis. We hypothesized that genetic variations in miRNA binding sites located in mucin genes may modulate signaling response and the maintenance of genomic stability ultimately affecting cancer susceptibility, efficacy of chemotherapy and survival. In this study, we analyzed the association of single nucleotide polymorphisms in predicted miRNA target sites (miRSNPs) of mucin genes with colorectal cancer (CRC) risk and clinical outcome. Thirteen miRSNPs in 9 genes were assessed in 1111 cases and 1469 controls. No strongly significant associations were observed in the case-control study. Patients carrying the CC genotype of rs886403 iMnUC21 displayed a shorter survival and higher recurrence risk when compared with TT carriers [overall survival (OS): hazard ratios (HR) 1.69; 95% confidence intervals (CI) 1.13-2P. 4=6 0;.01 and event-free survival (EFS): HR 1.99; 95% CI 1.38-2.84P; = 0.0002, respectively]. The observed associations were more striking after stratification for tumor site (in patients with colon cancer, OS: HR 2.63; 95% CI 1.69-4P.1<00; .0001 and EFS: HR 2.65; 95% CI 1.72-4.07; P < 0.0001). In contrast, rectal cancer cases carrying the CC genotype of rs472965M5UinC17 displayed a longer survival (OS: HR 0.27; 95% CI 0.14-0.54P; = 0.0002) than those with the most common genotype. To our knowledge, this is the first study investigating miRSNPs potentially affecting miRNA binding to mucin genes and revealing their impact on CRC susceptibility or patient's survival. Introduction metastasis itself is responsible for ~90% of all CRC-related deaths (1,2). The molecular basis of CRC pathogenesis and proOne of the most common malignancies, colorectal cancer (CRC), gression is complicated and poorly understood at present. accounts annually for almost half a million deaths worldwide. Mucinous colorectal carcinoma is generally defined as Mortality of cases results from uncontrolled metastatic- dis having greater than 50% of the tumor area with a mucinous ease, particularly in peritoneum, lymph nodes and liver. Tumor - Abbreviations [Prague (3 hospitals), Benesov, Brno, Liberec, Ples, Pribram, Usti nad Labem and Zlin] between September 2003 and October 2010. Two control groups, BMI body mass index whose samples were collected at the same time of cases recruitment, were CRC colorectal cancer included in the study. The first group consisted of 688 hospital-based in-di EFS event-free survival viduals admitted to five of the above-mentioned gastroenterological de-part HRs hazard ratios ments that had negative colonoscopy results for malignancy or idiopathic miRNA microRNA bowel diseases (Control group 1). The reasons for undergoing the colo-nos OS overall survival copy were: (1) positive fecal occult blood test, (2) hemorrhoids, (3) abdominal ORs odds ratios pain of unknown origin and (4) macroscopic bleeding. The second group of controls consisted of 781 healthy blood donor volunteers (Control group 2) collected from a blood donor centre in Prague. All individuals were-sub differentiation by histologic examination. The incidence of jected to standard examinations to verify the health status for blood d-ona mucinous carcinoma is about 12% among CRC patients, ranging tion and were cancer-free at the time of the sampling. Details of CRC cases from 6 to 19% in different studies 3(). The increased incidence and controls have been reported previously18(). occurs in proximal colon and among younger patients, as com- All subjects were informed and provided written consent to part-ici pared to nonmucinous adenocarcinoma. Mucinous CRCs have pate in the study and to approve the use of their biological samples for been found to have a higher Duke stage at diagnosis and, con-se genetic analyses, according to the Helsinki declaration. The local Ethics quently, a lower survival4)(. Committee approved the design of the study. Study subjects provided Mucins are large extracellular glycoproteins produced by information on their lifestyle habits, body mass index (BMI), diabetes epithelial cells, which are heavily glycosylated with complex and family/personal history of cancer, using a structured questionnaire to determine demographic characteristics and potential risk factors for CRC. oligosaccharides 5( ,6 ). The core proteins for human mucins (MUC1-MUC8, MUC12, MUC13, MUC15-17 and MUC19-21) have already been identified7(). Many mucins are abnormally Follow up of patients expressed and aberrantly glycosylated in adenocarcinomas. In Eight hundred sixty-six CRC cases and a second group consisting of 232 general, increased levels of mucin genes have been associated CRC patients recruited later on were monitored and followed up until 31st with increased cancer risk, tumor invasion, and a poor patient March 2013. For all subjects, clinical data at the time of diagnosis, inc-lud outcome (8–11). Previous studies have found that an upregul-a ing location of the tumor, UICC (International Union Against Cancer) tion of MUC1 and MUC5AC and a down-regulation oMfUC2 are tumor-node metastasis stage system, grade and adjuvant chemotherapy involved in the development and progression of CRC3(,12,13). treatment were collected, along with information about distant met-asta Recently, the overexpression oMfUC20 was considered as a pre- sis, relapse and date of death2(1). In our patients dataset, a subgroup of dictor of poor prognosis in CRC 1( 0). In CRC, the overexpression patients (n = 47) was diagnosed with mucinous cancer. of transmembrane mucins suggests their role in signaling cell Four hundred and eleven CRC cases received a 5-FU-based adjuvant growth and survival. Mucins have thus been identified as ma-rk reeitghimerena Masayfiorsrte-gliimneenp,odsetloivpeerraetdivaes athbeorlaupsy. inTfhuesiotnheorfap5y-FUwa(4s2b5 amsegd/m2)on ers of adverse prognosis 6(). and leucovorin (10 mg/m2) for five days every four weeks six times or a sim Interestingly, recent studies have unraveled a role of mi-cro plified De Gramont regimen which consisted of a 2-h intravenous (i.v.) inf-u RNAs (miRNAs) also in the regulation of various mucins1(4). sion of leucovorin (200 mg/m2), then a 5-FU i.v. bolus (400 mg/m2) followed MiRNAs are short (20–22 nucleotide) non-coding RNAs that r-eg by a 46-h 5-FU continuous i.v. infusion (2400–3000  mg/m2). Four hundred ulate gene expression by binding mainly to the′3untranslated forty subjects did not receive any adjuvant chemotherapy after surgery. regions (3′ UTR) of target mRNA thereby hampering protein translation or inducing mRNA destabilization. Aberrant miRNA Selection of candidate genes and SNPs in miRNA expression and/or function are frequently observed in many target binding sites malignancies, including CRC 1(5). The increasing need for newer The list of all mucin genes was retrieved by literature minin3g,6,(7). The diagnostic strategies to target tumors has led to the appearance approach used to select the candidate miRSNPs was similar to the one of miRNAs as potential cancer therapeutics of new generation. previously described (18). Briefly, for each gene, SNPs within predicted Moreover, genetic variations in the′ 3UTR of target genes may miRNA binding sites were identified by using the freely available software: affect miRNA binding, ultimately adding additional variability in MicroSNiper (http://epicenter.ie-freiburg.mpg.de/services/microsniper)/, the differential mRNA and protein expressions. (Mhitrtnps:n/p/csocomrepb(hiot.tupt:h//swcw.edwu.b/migirR.mSNedPi/s)i.nT.nhtenu5.9nod/emtiercstnepdscoSrNeP/)sanwdeProelymthiretn We recently reported associations between single nucleotide filtered for their minor allele frequency (MAF> 5%) in Caucasian po-pu polymorphisms (SNPs) in miRNA target regions of DNA repair lations (dbSNP;http://www.ncbi.nlm.nih.gov/SNP/) to reach an appropripathway genes and CRC risk or clinical outcome1 6(–19). In the ate representation of all genotypes in our set of cases and controls. The present study, we hypothesized that genetic variations in mucin information was primarily derived from 1000genomes project database, genes may also affect cancer susceptibility, patient’s survival and phase 1, CEU population; whenever this was not possible, other reference efficacy of chemotherapy. SNPs in miRNA target regions (miRSNPs) populations were considered (i.e. HAPMAP CEU population). SNPs with the of mucin genes might affect the efficiency of translation of c-or required MAF were further tested for the possibility to belinnkage disresponding proteins, thus affecting individual’s susceptibility to equilibrium (LD) using HaploView (v. 4.2) with the data from HapMap (v. 3, cancer. So far, the impact of miRSNPs in mucin genes on the CRC release R2 in the CEU population). risk has not been tested yet. We have studied this assumption in patients with CRC from the Czech Republic, a country with one of SNP genotyping the highest incidence worldwide for this cance2r0(). Genomic DNA was isolated from peripheral blood lymphocytes using standard procedures. The DNA from cases and controls was randomly placed on plates where an equal number of samples could be run simu-l Material and methods taneously. The selected SNPs were genotyped using the KATSMPgenotyping assay, a competitive allele-specific PCR SNP genotyping system (LGC Study population and data collection Genomics, Hoddesdon, Herts, UK). For quality control purposes, duplicate Blood samples were collected from 1111 patients with histologically -con samples (5% of the total numbers of samples) were repeated for each SNP, firmed CRC attending several oncological departments in the Czech Republic no template controls were included in each plate (NTCs). Statistical analyses Pearson`s chi-square test (1  degree of freedom), with a type-I error threshold set atα  =  0.05, was used to verify whether the genotypes were in Hardy–Weinberg equilibrium in control population. SNPs were excluded from further analyses when: (1) the call rate was <95%; (2) -pol ymorphisms deviated from Hardy–Weinberg equilibrium in controls at P < 0.01; (3) genotypes were discrepant in more than 2% of duplicate samples. The multivariate logistic regression analysis was used to test the association between genotypes and risk of CRC. The covariates analyzed Case–control study in the multivariate model were: sex, age, smoking habit (non-smokers vs. smokers and ex-smokers), BMI, familial history of CRC, education The characteristics of the study participants are presented in level (high, intermediate and low) and living area (country, suburbs, and Table 1. Among the 1111 CRC cases, 397 patients were diagnosed town). The associations between SNPs and CRC risk were calculated by with a tumor in colon, 334 in sigmoideum and 377 with rectal estimating the odds ratios (ORs) and their 95% confidence intervals (CI), cancer (3 cases were missing the information about the site of adjusted for both continuous and discrete covariates. For all the g-eno tumor; however, since they had complete survival data, they types, regression coefficients for additive models were estimated. For remained in the survival analysis). Out of the 1469 controls, 688 all SNPs, the dominant or recessive models were also calculated. The were cancer-free colonoscopy inspected controls (Control group Bonferroni-corrected significance threshold is 0.004 (for 13 SNPs and 1) and 781 were healthy blood donor volunteers (Control group α = 0.05). 2). Compared to subjects of both control groups, CRC cases were The model with the highest likelihood was additionally checked for the significance of possible interaction terms in the multivariate logistic more likely to be older, have a slightly higher BMI while, c-om pared to the Control group 2, they were more likely to have a positive family history of CRC and lower formal education them we could not retrieve any information. The remaining 59 miRSNPs were filtered according to the study criteria previously described (i.e. MAF and LD). In total, 46 SNPs were excluded. The remaining 13 miRSNPs in nine genes (MUC6, MUC7, MUC13, MUC14, MUC15, MUC17, MUC20, MUC21 and MUC24) were included in the study (Supplementary Table I, available at Carcinogenesis Online). regression analysis. Statistical analyses were performed usinghRtt(p:// www.rproject.org). In this study, the outcome variables measured were overall survival (Table 1). (OS, time from diagnosis until death or censorship), and event-free s-ur Results of the associations between the investigated vival (EFS, time of surgery or end of chemotherapy until date of relapse, miRSNPs and CRC susceptibility are reported in Supplementary death or censorship whichever came first). The survival curves for OS and Table II, available atCarcinogenesis Online. A  decreased cancer EFS were derived by the Kaplan–Meier method (R version 2.14-2, Survival risk was observed for rs4071 inMUC14 gene: carriers of the AA package). The relative risk of death was estimated as hazard ratio (HR) genotype were at decreased risk to develop either CRC or rectal using Cox regression (R version 2.14-2, Survival package). Multivariate -sur vival analyses were adjusted for age, gender, smoking and stage. cancer (OR 0.57, 95% CI 0.34–0.95,P = 0.03 and OR 0.42, 95% CI 0.18–1.00, P = 0.05, respectively). This association was confirmed in a recessive model as well (OR 0.55, 95% CI 0.34–0.91,P = 0.02 Results and OR 0.40, 95% CI 0.17–0.94, P  =  0.04, respectively). Another SNP in EMCN/MUC14 gene, rs17552409, was also found assoc-i miRSNP selection ated with decreased risk of CRC (for the AA genotype: OR 0.27, Out of the 20 genes categorized as human mucins, only 12 had 95% CI 0.09–0.80, P = 0.02; for the A-allele: OR 0.27, 95% CI 0.09– miRSNPs in their 3′ UTRs (62 miRSNPs retrieved). For three of 0.79, P  =  0.02) and colon cancer (for the AA genotype OR 0.18, 95% CI 0.04–0.81, P = 0.02; for the variant A allele OR 0.18, 95% CI were associated with decreased patients’ survival and increased 0.04–0.80, P = 0.02). An association with CRC risk was observed risk of recurrence. Moreover, adjuvant chemotherapy was also for a SNP in MUC13. The variant AA genotype of rs1532602 was associated with survivalT(able 2). in fact associated with a decreased risk of cancer (OR 0.75, 95% After adjusting for sex, age, smoking and CRC stage, the CI 0.56–1.00, P = 0.05). After stratification according to the tumor strongest association with patient´s survival was observed for site, the association was still observed in colon cancer patients rs886403 in MUC21 (Tables 3 and 4). CRC patients carrying the (OR 0.72, 95% CI 0.51–1.00, P  =  0.05), and resulted stronger for CC genotype displayed a shorter survival and higher recurrence the dominant model (OR 0.79, 95% CI 0.63–0.99,P = 0.04). Finally, risk (OS: HR 1.69; 95% CI 1.13–2.46; P = 0.01 and EFS: HR 1.99; 95% a decreased risk of CRC was observed in a recessive model for CI 1.38–2.84; P = 0.0002, respectively) when compared with ca-r rs974034 in MUC24 (OR 0.76, 95% CI 0.58–1.00, P = 0.05). riers of the most frequent genotype. The observed associations None of the above associations remained significant after were more striking, after stratification for tumor site, in patients applying correction for multiple testing (Bonferroni’s correction). with colon cancer (OS: HR 2.63; 95% CI 1.69–4.10P; < 0.0001 and We have also repeated the analyses in the subgroup of EFS: HR 2.65; 95% CI 1.72–4.07; P  =  0.0001 and in the recessive patients with diagnosed mucinous CRC histology (47 patients). model OS: HR 2.70; 95% CI 1.77–4.12; P < 0.0001 and EFS: HR 2.43; However, in this subgroup of patients, we did not find any s-ig 95% CI 1.61–3.64; P < 0.0001). Particularly for EFS, CRC patients nificant association with the CRC risk for any of the analyzed also showed a similar significant trend across genotypes in the SNPs (data not shown). This last result should be cautiously-con Kaplan–Meier curves (log-rank tesPt  =  0.03; Median survival sidered due to the low frequency of mucinous CRC in our study time for TT carriers=353 months; MST for CT =231 months; MST group. for CC carriers=79 monthsF;igure 1A). A similar trend was found also for colon cancer patients (log-rank tePs =t 0.03; MST for CC Survival analysis carriers = 115 months; MST not reached for the other genotypes; The mean (median) OS and EFS for patients were 86.5 (80.5) Figure 1B) but not for rectal cancer patientFsig(ure 1C). and 72.6 (62.4) months, respectively. Age, gender, T, N, M status, Among colon cancer patients, carriers of the TC genotype chemotherapy treatment and CRC stage were associated with of MUC6 rs4077531 showed a decreased survival (OS: HR 1.38; OS and EFS in the preliminary univariate assessment of cova-ri 95% CI 1.00–1.90; P = 0.05). On the other hand, individuals with ates (Table  2). Advanced age, male gender and current smo-k rectal cancer and carrying variant CC genotype of rs4729655 in ing status were related to a shorter OS. Likewise, men were MUC17 displayed a longer survival when compared with the also at higher risk of relapse or metastasis (OS: HR 1.54; 95% CI reference genotype (OS: HR 0.27; 95% CI 0.14–0.54;P = 0.0002). 1.23–1.92; P = 0.0001; EFS: HR 1.35; 95% CI 1.09–1.68; P = 0.006). Overall, rectal cancer patients also showed a similar significant Four established prognostic factors (T, N, M status and stage) trend across genotypes in the Kaplan–Meier curves (log-rank P n a v 7 0 3 4 8 3 5 6 6 1 c E 1 1 1 1 2 1 n 7 4 4 1 3 7 1 8 0 1 e 4 7 6 3 1 6 1 2 3 5 3 3 1 6 2 8 5 2 8 5 5 0 3 5 1 5 6 1 6 6 1 4 1 5 5 1 0 9 0 9 9 0 4 6 1 7 20 1 6 11 3 15 18 3 13 7 8 20 1 2 1 2 5 2 5 7 7 5 8 8 6 4 6 6 4 0 8 8 4 8 8 5 3 8 3 3 6 6 3 2 6 0 2 9 2 1 5 2 6 4 2 6 0 2 4 9 5 4 3 5 a C N 12 92 01 93 02 01 04 81 1 02 95 1 04 91 2 12 95 2 22 92 9 83 25 9 73 12 2 42 95 2 94 21 21 26 3 2 2 6 r s e t c .07 .18 .06 .58 .71 .07 .38 .07 .45 .76 .44 .84 .16 .84 .24 .65 G G A A A A A A G G C C A G G A r s e t c a t c e a 8 5 7 2 4 7 0 1 1 1 R N 1 2 2 3 3 1 0 0 0 0 0 0 0 0 . n o i b)(I95RCH% feR ...)(–138730201 ...)(–103530470 ...)(–172180290 feR ...)(–055980370 feR ...)(–133230360 — ...)(–133230360 feR — feR ..)(–118310310 ...)(–201971018 ...)(–178390011 feR ...)(–290921971 feR ...)(–176090680 ...)(–231461961 ...)(–167200690 feR ...)(–232601781 feR ...)(–176110680 ...)(–118600411 ...)(–137160290 feR ...)(–119690421 feR ...()–157210590 ...()–136430590 ...()–167019590 feR ...()–561045790 iiiffssrtaognnmm o 4 5 9 8 .2 .4 .4 .1 e p y t iten eEvn 107 167 69 236 274 69 345 9 0 9 354 0 217 112 21 133 329 21 185 129 36 165 314 36 137 153 52 205 290 52 172 144 30 174 316 30 illba .e e pa vaa stag ll a N 52 45 21 67 71 21 91 3 A 2 9 5 4 1 5 2 2 0 32 494 0 601 294 41 335 895 41 479 373 80 453 852 80 339 45 12 58 79 12 44 40 7 47 85 7 fo dn 7 5 3 6 5 9 8 1 9 7 1 % a 0 , 0 g 1 n i CC TC +GG +GA +GG +GA +CC +TC CC TC CC TC .d to ko d e U 7 s M r in dd ,eg s a a t t , l o x re ya r o t m f n s d a r e c e t if b s i m ju n g u d i S aNbA .01 .15 .31 .49 .15 .17 .06 .56 .56 .00 .86 .95 .90 .93 .95 .87 w l 2 3 0 3 5 0 1 4 4 8 5 4 2 5 7 2 7 7 7 6 6 2 3 6 0 2 2 5 7 2 2 7 8 0 6 7 3 6 7 8 6 4 0 4 4 lA aN 13 24 61 85 37 61 95 72 2 92 68 2 85 92 3 33 78 3 32 43 13 57 76 13 55 3 3 8 7 1 1 9 .67 .92 .67 .79 .48 .03 .95 .82 .89 .89 .29 .59 .99 .76 .79 .98 .49 .89 P 0 0 0 0 0 0 0 0 0 — 0 0 0 0 0 0 0 0 0 7 7 9 9 .7 .2 .5 .2 0 0 0 0 4 4 0 4 2 6 5 2 9 5 4 4 9 3 8 9 2 9 1 0 1 1 7 3 4 0 6 4 5 0 4 5 1 7 0 1 1 1 1 d s e t t c n e itea pEx 69 43 4 47 112 4 27 58 26 84 85 26 11 p l a t c e a 8 1 1 2 9 1 8 5 6 1 3 6 9 1 R N 1 1 1 2 1 2 2 2 0 2 0 2 2 0 0 2 5 7 2 5 3 0 0 0 8 2 7 9 0 7 8 2 2 4 0 2 5 2 1 3 7 1 1 8 9 1 0 8 1 6 1 2 3 8 2 1 5 3 8 6 3 1 1 2 1 1 1 2 1 1 1 2 2 3 0 0 P .360 .920 .640 .320 .860 .860 .950 .750 .505 — .505 .703 .108 .502 .109 .403 .000 .008 .000 a p l l a 4 4 3 7 8 3 4 4 0 5 9 0 8 3 A N 5 3 3 8 2 4 2 6 6 2 8 1 2 5 7 3 5 5 8 7 5 3 7 7 1 0 31 85 85 40 25 70 40 64 66 76 42 30 76 27 5 1 7 7 1 1 9 0 8 9 5 2 3 8 4 3 4 8 3 4 1 5 7 1 . 4   le e N C 7 4 5 D 4 I 3 P 5 4 1 2 b n S U 1 a e b s T G d M r G G G A C C C T G G G G G A /42CU 461CD s7372 M r s t n e i t a p r e c n a c n o l o a C s t n e i t a p r e c n a c l a t c e a R s t n e i t a p l l a A P b) I C % 5 9 ( R H b) I C % 5 9 R H d e t c e p x E s t n e v E N P d e t c e p x E s t n e v E N P d e t c e p x E s t n e v E N e p y t o n e G Mucins are high molecular weight glycoproteins predominantly expressed at the epithelial surface of tissues that provide p-ro tection for colon surface under normal physiological conditions (22). Several lines of evidence point towards a biological role of mucins in CRC ( 22,23). These include: observations on mucinous CRC in in vitro and in vivo experiments and alterations of mucin structure in polyps and cancer. Although mucinous CRC have a higher Dukes stage at diagnosis and show a worse prognosis, the cause of such negative phenotypes is not comprehensively elucidated. Mucins have an important role in epithelial cell -pro tection and maintenance of homeostasis by covering human colon surface by gel mucous layer. Lower cell adhesion may contribute to incorrect cellular organization and structur-e, pro liferation and survival, and ultimately gene expression alt-era tion. The appropriate cell adhesion is necessary for numerous physiological processes and can be deranged in many diseases, including thrombosis, inflammation, and cancer2(4). Recently, miRNAs have emerged as important regulators responsible for an altered mucin expression during the malignant development (14). To date, no study has investigated the impact of miRSNPs within mucin genes on CRC susceptibility or clinical outcome. In the present study, we have found that some miRSNPs in mucin genes were associated with CRC risk and they had an impact on patient´s survival. Interestingly, a reduced CRC risk was observed for individuals bearing homozygous variant gen-o types of MUC13 rs1532602, EMCN/MUC14 rs4071 and rs17552409 and MUC24 rs974034. After stratification for tumor localization, some of these results were confirmed in colon cancer patients (rs1532602 in MUC13, in a dominant model) or rectal cancer patients (rs4071 in EMCN/MUC14 in recessive model). After applying correction for multiple testing, these associations were lost. On the other hand, such conservative correction may not be required considering: (1) the exploratory nature of our study, (2) the fact that all the SNPs were selected for their high prior -prob ability of functional significance, and (3) based on differential binding of miRNAs to their predicted polymorphic target sites. BothEMCN/MUC14 rs4071 and rs17552409 polymorphisms were also tested for their potential SNP-SNP interaction on CRC-sus ceptibility. No effect on CRC risk was observed when these two SNP were analyzed together. Similarly, we have also repeated the analyses in the subgroup of patients with diagnosed mu-ci nous CRC histology. However, due to the low representation of mucinous CRC in this subgroup of patients (47 patients), we did not find any significant association with the CRC risk for any of the analyzed SNPs. Mucinous CRC tends to occur in younger patients, are often seen in the proximal colon, are more dia-g nosed at an advanced stage and are more frequently asso-ci ated with hereditary non-polyposis colorectal cancer (HNPCC) and young-age sporadic colorectal cancer. The low proportion of mucinous CRC patients in the present study could be inflicted by the fact that CRC cases comprise rather older patients (age miR-125 and let-7) are predicted to bind in the region surrou-nd range 18–47 comprises 86 CRC patients while age range 65–91 ing rs886403. However, none of them have been validated in comprises 427 CRC patients). relation toMUC21, so far. The stronger and novel finding of this study is represented In this study, rectal cancer patients carrying the CC g-eno by the associations of some miRSNPs with clinical outcome. In type of MUC17 rs4729655 displayed a better OS when compared particular, CRC patients carrying the CC genotype for rs886403 with the carriers of the other genotypes. MUC17, an intestinal in MUC21 displayed a shorter survival and higher recurrence membrane-bound mucin, has been shown to enhance mucosal risk. The observed association was strikingly pronounced in restitution by stimulation of cell migration and inhibition of colon cancer patientsM.UC21 has been identified quite recently apoptosis (27). MUC17 is highly expressed on the surface ep-i (25). The gene encodes for a transmembrane mucin related to thelium of normal colonic mucosa but its expression becomes the biosynthesis of N-glycan precursor (dolichol lipid-linked altered in colorectal neoplasia. Interestingly, an increased oligosaccharide, LLO). According to STRING Interaction Network expression of MUC17 was associated with a longer OS in (www.genecards.org), there is a close cooperation amongMUC21 patients with stage III and IV colorectal adenocarcinoma2s8)(. and MUC17, MUC20, MUC16, GALNT5 and ST3GAL3. Scarce infor- These results point to its possible role in cancer progression mation is available for this gene, thus the postulation of hypo-th and prognosis. Although the physiological function oMfUC17 esis that this gene might be associated with patient´s survival is still unclear, it may serve as a physical barrier against mi-cro is quite difficult. Cells transfected by MUC21 were significantly organisms and as cell-surface sensor. MUC17 may also conduct less adherent to each other and to extracellular matrix -com signals in response to external stimuli that lead to cellular ponents than control cells, suggesting that MUC21 prevents responses, including proliferation, differentiation, apoptosis or integrin-mediated cell adhesion to extracellular matrix c-om secretion of cellular products such as other membrane-bound ponents (26). Cell adhesion and cell-cell interactions also play mucins ( 29). Kitamoto et  al. (29) proposed several miRNAs as vital roles in many later steps in cancer progression, facil-itat potential regulators oMfUC17 expression, but none of them has ing the entry and survival of cancer cells into the bloodstream, been validatedin vivo. and their arrest and establishment at distant orga2n4s).(The As there is an established interplay betweeMnUC21 and miRNA-mediated regulation ofMUC21 expression has not been MUC17 genes according to STRING, the polymorphisms signif-i investigated yet. Several miRNAs (such as miR-4647, miR-588, cantly associated emerging from the follow up study were also explored for their potential SNP–SNP interaction on patients been already observed by us in a previous study on rectal cancer survival. However, no effect was observed for these two part-icu (34) but it emerged also from the results of the present study. lar genes in combination. There are in fact different survival rates associated to the d-iffer A strong association with either shorter OS or EFS was ent kind of cancer site and specific miRSNPs. Our data contr-ib observed forMUC20 gene (rs6782006). CRC patients, particularly utes to improved understanding of the role of specific miRSNPs those with colon cancer, carrying the variant GG genotype of in rectal and colon cancer pathogenesis. Treating the 2 sites as this miRSNP had worse survival. MUC20 is a novel mucin protein independent entities may improve discovery of biomarkers used highly expressed in kidney and colon tissues. Based on whole- for early detection and prognosis. genome expression profiling of CRC, MUC20 was significantly We are aware of certain limitations of the present invest-iga upregulated in CRC patients with poor prognosis10(). A  rela- tion. In the case-control study, healthy subjects differed from tionship betweenMUC20 overexpression and poor survival was cases in age and gender distribution, as well as other par-am found in many human tumors, including ovarian cancer 3(0), eters such as BMI. However, we attempted to control tentative non-small cell lung cancer31(), and gastric cancer3(2). MUC20 age effect by matching cases and controls by age quartiles overexpression predicts poor prognosis in endometrial cancer through bootstrap sampling (830 cases and 830 controls). In and enhances EGF-triggered invasive behavior through act-iva particular, we obtained similar results to those presented in tion of EGFR–STAT3 pathway (33). Increased expression levels of this work in 8 out of 10 resamplings. Moreover, patients were MUC20 promoted metastasis of CRC cells, whereas knockdown collected from the same centers (with follow up data collected of this gene attenuated migration and invasion abilities of CRC by the same physicians) and were highly homogeneous for cells 1(0). Like for the other mucin genes, the number of studies their ancestry, thus with the exclusion of possible popu-la regarding miRNA-mediated control of the expression oMfUC20 tion stratifications. In addition, the inclusion of ‘colonoscopy is still scarce. negative’ individuals ensured disease-free control indivi-du From a clinical point of view, malignancies in the colon and als because a negative colonoscopy result is the best available the rectum represent two distinct entities that require differentproof of the CRC absence (35). Nevertheless, there is concern treatment strategies. The distinction between colon and rectum that the colonoscopy negative control group is based on ex-ist is largely anatomical but it impacts both surgical and radiot-her ing medical conditions of patients, which required examin-a apeutic management with often different prognoses. This has tion and thus may carry unknown CRC risk factors. Since this group of individuals may not necessarily represent the general 12. Kang, H. et  al. (2011) Loss of E-cadherin and MUC2 expressions co-r population, we included also healthy cancer-free indivi-du related with poor survival in patients with stages II and III colorectal als recruited among volunteers from blood centers. However, carcinoma. Ann. Surg. Oncol., 18, 711–719. there were no statistically significant differences in genotypes 13. Matsuda, K. et al. (2000) Clinical significance of MUC1 and MUC2 mucin frequencies of mucin genes between the two control groups, and p53 protein expression in colorectal carcinoma. Jpn. J. Clin. Oncol., 30, 89–94. which allowed pooling of the controls to increase the statistical14. Macha, M.A. et  al. (2015) Emerging potential of natural products for power of the study. targeting mucins for therapy against inflammation and cancer. Cancer Mucin synthesis and secretion by CRC affects the phenotype Treat. Rev., 41, 277–288. of the disease, as demonstratedin vitro and in vivo. Expanding 15. Slattery, M.L. et al. (2011) MicroRNAs and colon and rectal cancer-: dif our knowledge on mucin involvement in CRC may help us to ferential expression by tumor location and subtype. Genes. Chrom-o better understand the etiopathogenesis of this disease and somes Cancer, 50, 196–206. thereby contribute to the development of new treatment str-ate 16. Naccarati, A. et  al. (2012) Polymorphisms in miRNA-binding sites of gies. It is well known that mucinous CRC have a higher Dukes nucleotide excision repair genes and colorectal cancer risk. Carc-ino stage at diagnosis and a worse prognosis. The present results genesis, 33, 1346–1351. 17. Vymetalkova, V. et  al. (2014) Variations in mismatch repair genes identified plausible candidate SNPs potentially affecting miRNA and colorectal cancer risk and clinical outcome. Mutagenesis, 29, binding in mucin genes that were related either to CRC susc-ep 259–265. tibility or to patient’s survival. Moreover, our study supports the 18. Pardini, B. et  al. (2013) Variation within 3’-UTRs of base excision emerging idea of a ‘miRNA network’ that may contribute to CRC. repair genes and response to therapy in colorectal cancer patients: Further studies are needed to replicate these SNPs as predictive A potential modulation of microRNAs binding. Clin. Cancer Res., 19, biomarkers in independent populations, to functionally char-ac 6044–6056. terize the significant genetic variants and to find the biologic 19. Naccarati, A. et  al. (2016) Double-strand break repair and colorectal mechanisms underlying the associations. cancer: gene variants within 3’ UTRs and microRNAs binding as mod-u lators of cancer risk and clinical outcome. Oncotarget, 7, 23156–23169. 20. Pardini, B. et al. (2009) NBN 657del5 heterozygous mutations and co-lo Supplementary material rectal cancer risk in the Czech Republic. Mutat. Res., 666, 64–67. Supplementary Tables 1 and 2 can be found athttp://carcin. 21. Pardini, B. et al. (2015) Polymorphisms in microRNA genes as predictors oxfordjournals.org/ of clinical outcomes in colorectal cancer patients. Carcinogenesis, 36, 82–86. 22. Gupta, B.K. et al. (2012) Increased expression and aberrant localization Funding of mucin 13 in metastatic colon cancer. J. Histochem. Cytochem., 60, 822–831. Internal Grant Agency of the Ministry of Health of the Czech 23. Niv, Y. (2000) Mucin and colorectal cancer. Isr. Med. Assoc. J., 2, 775–777. Republic (IGA NT 13424 and AZV MZ 15-26535A) and the Grant 24. Labelle, M. et al. (2012) The initial hours of metastasis: the importance Agency of the Charles University GAUK 112515. of cooperative host-tumor cell interactions during hematogenous-dis Conflict of Interest Statement: None declared. semination. Cancer Discov., 2, 1091–1099. 25. Itoh, Y. et al. (2008) Identification and expression of human epiglyc-a References nin/MUC21: a novel transmembrane mucin. Glycobiology, 18, 74–83. 26. Yi, Y. et al. (2010) Mucin 21/epiglycanin modulates cell adhesion. J. Biol. 1. Mehlen, P. et al. (2006) Metastasis: a question of life or death. Nat. Rev. Chem., 285, 21233–21240. Cancer, 6, 449–58. 27. Ho, S.B. et al. (2010) Activity of recombinant cysteine-rich domain p-ro 2. Roessler, S. et  al. (2010) A unique metastasis gene signature enables teins derived from the membrane-bound MUC17/Muc3 family mucins. prediction of tumor relapse in early-stage hepatocellular carcinoma Biochim. Biophys. Acta, 1800, 629–638. patients. Cancer Res., 70, 10202–10212. 28. Li, K. et al. (2011) Aberrant Expression of MUC17 Correlates With -Sur 3. Byrd, J.C. et al. (2004) Mucins and mucin binding proteins in colorectal vival in Advanced Stage Colorectal Adenocarcinomas. Gastroent-erol cancer. Cancer Metastasis Rev., 23, 77–99. ogy, 140, S-145. 4. Debunne, H. et  al. (2013) Mucinous differentiation in colorectal -can 29. Kitamoto, S. et al. (2011) DNA methylation and histone H3-K9 modific-a cer: molecular, histological and clinical aspects. Acta Chir. Belg., 113, tions contribute to MUC17 expression. Glycobiology, 21, 247–256. 385–390. 30. Vlad, A.M. et  al. (2006) MUC1 in endometriosis and ovarian cancer. 5. Hollingsworth, M.A. et al. (2004) Mucins in cancer: protection and -con Immunol. Res., 36, 229–236. trol of the cell surface. Nat. Rev. Cancer, 4, 45–60. 31. Woenckhaus, M. et  al. (2008) Prognostic value of FHIT, CTNNB1, and 6. Kufe, D.W. (2009) Mucins in cancer: function, prognosis and therapy. MUC1 expression in non-small cell lung cancer. Hum. Pathol., 39, 126– Nat. Rev. Cancer, 9, 874–885. 136. 7. Yonezawa, S. et al. (2011) Mucins in human neoplasms: clinical patho-l 32. Marín, F. et al. (2012) Genetic variation in MUC1, MUC2 and MUC6 genes ogy, gene expression and diagnostic application. Pathol. Int., 61, 697– and evolution of gastric cancer precursor lesions in a long-term follow716. up in a high-risk area in Spain. Carcinogenesis, 33, 1072–1080. 8. Yonezawa, S. et  al. (2008) Precursor lesions of pancreatic cancer. Gut 33. Chen, C.H. et al. (2013) MUC20 overexpression predicts poor prognosis Liver, 2, 137–154. and enhances EGF-induced malignant phenotypes via activation of 9. Yonezawa, S. et al. (2010) Significance of mucin expression in pancre-a the EGFR-STAT3 pathway in endometrial cancer. Gynecol. Oncol., 128, tobiliary neoplasms. J. Hepatobiliary. Pancreat. Sci., 17, 108–124. 560–567. 10. Xiao, X. et  al. (2013) Role of MUC20 overexpression as a predictor of 34. Vymetalkova,V. et al. (2016) Epigenome-wide analysis of DNA methyl-a recurrence and poor outcome in colorectal cancer. J.Transl. Med., 11, 151. tion reveals a rectal cancer-specific epigenomic signature. Epigeno-m 11. Yonezawa, S. et  al. (2008) Expression profiles of MUC1, MUC2, and ics, 8, 1193–1207. MUC4 mucins in human neoplasms and their relationship with bio- 35. Slattery, M.L. et  al. (2009) A comparison of colon and rectal somatic logical behavior. Proteomics, 8, 3329–3341. DNA alterations. Dis. Colon Rectum, 52, 1304–1311. P . 809 .008 .503 .003 . 108 - . 108 - . 802 .002 .402 .002 .201 .001 .700 .000 e ptecEx .829 .8071 .497 . 2052 . 6362 . 974 .1413 .219 0 .219 .4503 0 .7292 .8081 .314 .2212 .6337 .314 . 1811 . 1411 . 728 .8619 .2232 .728 .0216 .713 .441 .164 .7319 .441 .4616 .6515 .429 . 1814 . 1231 .429 stceecb j b u s 9 0 0 C 6

This is a preview of a remote PDF: https://academic.oup.com/carcin/article-pdf/38/1/28/9361091/bgw114.pdf

Vymetalkova, Veronika, Pardini, Barbara, Rosa, Fabio, Jiraskova, Katerina, Di Gaetano, Cornelia, Bendova, Petra, Levy, Miloslav, Veskrnova, Veronika, Buchler, Tomas, Vodickova, Ludmila, Naccarati, Alessio, Vodicka, Pavel. Polymorphisms in microRNA binding sites of mucin genes as predictors of clinical outcome in colorectal cancer patients, Carcinogenesis, 2017, 28-39, DOI: 10.1093/carcin/bgw114