Thymidylate Synthase Promoter Polymorphism, Interaction with Folate Intake, and Risk of Colorectal Adenomas Free

TS² is a key enzyme in folate metabolism and catalyzes the conversion of dUMP to dTMP. This conversion is essential for the provision of thymidine, a nucleotide needed for DNA synthesis and repair. TS is also a target for major chemotherapeutic drugs, including 5-fluorouracil. The substrate for TS, 5,10-methylene-THF, is a central metabolite in folate metabolism and is diverted into different pathways, including thymidine synthesis, purine synthesis, and toward the provision of methyl groups for DNA methylation. Colon cells are subject to rapid turnover and, thus, are the site of high rates of DNA synthesis. Furthermore, DNA repair is essential in the colonic environment, and DNA repair mechanisms can be affected by nucleotide availability (1, 2, 3). Folate deficiency has been shown to increase several measures of DNA damage (4, 5), and a low folate status is consistently associated with an increased risk of colorectal cancer (6, 7). The human TS promoter region includes several mechanisms for gene regulation, including a cis-acting enhancer element (8). This TSER is polymorphic, containing two or three 28-bp tandem repeats among Caucasian and Asian populations (9, 10). The less common double repeat is associated with 2.6-fold lower TS expression in HeLaS3 cells than the triple repeat (9), and in vivo TS mRNA levels in tumor tissue were 3.6-fold lower among 2rpt/2rpt individuals compared with 3rpt/3rpt individuals (11). We also reported recently on a second TS polymorphism, a 6-bp deletion at bp 1494 in the 3′UTR that may be associated with decreased mRNA levels in colorectal tumors (12, 13). We hypothesized that a change in TS expression may affect colon carcinogenesis and investigated the risk of colorectal adenomas associated with the two TS polymorphisms in a large case control study of colonoscopically screened men and women.

Subject recruitment for the Minnesota Cancer Prevention Research Unit case control study has been described previously (14). Briefly, cases and controls were recruited through a multiclinic private gastroenterology practice in metropolitan Minneapolis, Digestive Healthcare. Patients (age 30–74 years) who were scheduled for colonoscopy between April 1991 and April 1994 were screened for eligibility and recruited prior to colonoscopy. Recruitment at all 10 sites was initiated at the time of scheduling with the intention of recruiting subjects with both patient and recruiter blind to the final diagnosis. The study protocol was approved by the internal review boards of the University of Minnesota, and each DH endoscopy site and written informed consent was given by each study participant.

Eligibility criteria for both cases and controls were: (a) resident of Twin cities metropolitan area; (b) age 30–74 years; (c) English speaking; (d) no known genetic syndrome associated with colonic neoplasia; and (e) no history of cancer (except nonmelanoma skin cancer), adenomatous polyps, or inflammatory bowel disease. Indications for colonoscopy have been published previously (15), and 68% of all colonoscoped patients participated.

At the colonoscopy visit, questionnaires were collected, blood was drawn, and the colonoscopy findings were recorded. Only participants with a complete colonoscopy reaching the cecum were eligible. All polyps were removed and evaluated histologically. Patients with polyps showing invasive carcinoma were not included. Adenomatous polyp cases are those found to have ≥1 adenomatous polyp; controls were polyp free at colonoscopy.

Data collection included information on dietary intake, physical activity, smoking habits, anthropometric measurements, medical information, demographic information, and reproductive history. When data were incomplete, study staff followed up by phone. The dietary questionnaire was an adaptation of the Willett food-frequency questionnaire, which has been studied previously for validity and repeatability. Furthermore, folate values derived from this questionnaire correlated with RBC folate levels (an indicator of long-term folate status r = 0.56).

Genomic DNA from lymphocytes was extracted using the PureGene kit (Gentra Systems, Inc., Minneapolis, MN). For the analysis of the TSER 28-bp repeat polymorphism, a fragment containing the repeats was amplified using primers TS forward primer: 5′-GTGGCTCCTGCGTTTCCCCC-3′ and TS reverse primer: 5′-GGCTCCGAGCCGGCCACAGGCATGGCGCGG-3′ (9). The PCR reactions contained 1 × GeneAmp buffer (Applied Biosystems, Foster City, CA), 1.5 mm MgCl₂, 200 μm deoxyribonucleotide triphosphates, 100 nm each primer, 10% DMSO, 1 unit of Amplitaq DNA polymerase (Applied Biosystems), and 100 ng of genomic DNA. Cycling conditions were one cycle of 94°C for 2 min, 35 cycles of 94°C for 30 s, 63°C for 30 s, and 72°C for 30 s, and a final extension at 72°C for 5 min. The amplified fragments were separated on a 3% NuSieve agarose gel (Biowhittacker, Rockland, ME). The fragment containing three and two repeats were 243 and 215 bp, respectively.

The 3′UTR polymorphism was analyzed by RFLP as described (12). Briefly, a fragment containing the 6-bp deletion was amplified using primers 5′-CAAATCTGAGGGAGCTGAGT-3′ and 5′-CAGATAAGTGGCAGTACAGA-3′ in a reaction containing 1 × GeneAmp buffer, 2.5 mm MgCl2, 150 mm deoxyribonucleotide triphosphates, 300 nm each primer, 1 unit of AmpliTaq DNA polymerase, and 100 ng of genomic DNA. The cycling conditions were 1 cycle of 94°C for 5 min, 30 cycles of 94°C for 30 s, 58°C for 45 s, and 72°C for 45 s, and a final extension at 72°C for 5 min. The amplified fragments were digested with DraI, and the products were separated on a 3% NuSieve agarose gel. The expected fragment sizes were 70 and 88 bp for the wild-type allele and 152 bp for the variant allele.

Unconditional logistic regression models were used to obtain maximum likelihood estimates, ORs, and 95% confidence intervals. Multivariate adjustment included age, sex, body mass index (kg/m²), percentage of calories from fat, hormone replacement therapy (ever/never), and pack-years of smoking. These variables altered some risk estimates by >10%. In general, confounding effects were small and often only apparent in stratified analyses; nonetheless, a consistent multivariate adjustment was maintained throughout all analyses presented. Other factors evaluated but not found to be confounders were waist:hip ratio, regular use of aspirin or nonsteroidal anti-inflammatory drugs (at least once/week versus never), hours of physical activity, and the dietary intake variables kcal, dietary fiber, vitamin B₆, vitamin B₁₂, and alcohol. The MTHFR C677T polymorphism was also not found to confound the associations.

Effect modification of the relation between nutrient intakes and risk of adenomas by genotype was evaluated by testing for different slopes with increasing nutrient intake across genotypes. Because the allele frequency of the TSER polymorphism is known to vary by race (10), we restricted the study population to Caucasians (who comprised 97% of the original population) for a sample size of 510 cases and 604 controls. All tests of statistical significance were two sided. SAS, version 8.2 (SAS Institute Inc., Cary, NC), was used for the analyses.

The study population and their dietary intakes have been described previously (15), and selected characteristics are presented in Table 1. The allele frequency of the TSER polymorphism among the controls was 0.48 with 52% heterozygotes (2rpt/3rpt) and 21% homozygous variant (2rpt/2rpt). The allele frequency of the TS 1494 6-bp deletion was 0.30 with 40% heterozygotes (6 bp/0 bp) and 10% homozygous variant (0 bp/0 bp). The two polymorphisms were in linkage disequilibrium: among those with a TSER 3rpt/3rpt genotype, 29% were found to have the TS 6 bp/6 bp genotype and 26% the 0 bp/0 bp genotype; among individuals with a TSER 2rpt/2rpt genotype, these proportions were 73 and 3%, respectively (P < 0.001). However, all genotype combinations were observed. Both variants were in Hardy-Weinberg equilibrium.

Multivariate adjusted ORs (95% confidence interval) for 2rpt/3rpt and 2rpt/2rpt compared with 3rpt/3rpt were 0.8 (0.6–1.2) and 0.9 (0.6–1.3), respectively. Risk estimates were similar among men and women and among individuals either greater than age 60 or younger. As shown in Table 2 and Fig. 1, we observed a statistically significant gene-nutrient interaction with dietary folate intake: among individuals with the 3rpt/3rpt genotype, folate intake > 440 μg/day (highest tertile) versus ≤440 μg/day was associated with a 2-fold decreased risk [ORs 1 (ref; low intake) versus 0.5 (0.3–0.9; high intake)]. However, among individuals with the 2rpt/2rpt variant, a 1.5-fold increased risk associated with high folate intake was observed [ORs 0.6 (0.4–0.9) versus 0.9 (0.5–1.5); P for interaction = 0.03]. A similar trend was observed for vitamin B₁₂ intake: among individuals with the 3rpt/3rpt and 2rpt/3rpt genotypes, risk was reduced 2-fold with intakes in the highest tertile (≥9.78 μg/day, compared with <9.78 μg/day), whereas a slight increase in risk (1.3-fold) with intakes ≥ 9.78 μg/day was observed among individuals with the 2rpt/2rpt genotype (P for interaction = 0.08). Folate intake did not confound this association, nor did vitamin B₁₂ confound the folate/TS association. The associations with the TS polymorphism did not differ by intakes of vitamin B₆, methionine, or alcohol (data not shown).

Because we have shown previously that the MTHFR C677T polymorphism affects risk of adenomas in this population (15), we stratified the analysis further by MTHFR genotype (Table 3). Although the cell sizes are small, and the results should be considered exploratory, some trends are suggested: (a) the decrease in risk associated with high folate intakes was again confined to individuals with greater TS expression (3rpt/3rpt genotype), irrespective of MTHFR genotype; and (b) the previously reported increase in adenoma risk associated with low folate intake among individuals with the MTHFR TT genotype (15, 16) was only seen among those with the 3rpt/3rpt genotype (>3-fold). Lower TS expression (2rpt/3rpt or 2rpt/2rpt genotypes) appeared to be inversely associated with risk under conditions of low-folate intakes, most prominently so among individuals with the MTHFR TT genotype. When folate intakes were not considered, the combination of a TSER 2rpt/2rpt genotype and MTHFR TT genotype was the lowest risk group compared with all other genotype combinations (data not shown).

The TS 6-bp deletion polymorphism at bp 1494 in the 3′UTR was not associated significantly with risk of colorectal adenomas [compared with 6bp/6bp wild-type OR = 1 (reference): 6bp/0bp OR = 1.22 (0.93–1.61); 0 bp/0 bp OR = 1.13 (0.73–1.74)]. When stratified by dietary intakes of folate or vitamin B₁₂, or by MTHFR genotype, no consistent patterns were apparent (data not shown). Although one might expect results similar to those for the TSER polymorphism, because of linkage between the two polymorphisms, these findings illustrate that “imperfect” disequilibrium may result in different associations with disease phenotypes. Adjustment for the TSER polymorphism did not modify the associations. We investigated the risk of adenomas associated with all possible TSER/TS 1494del6 genotype combinations, and none of the risks was statistically different from 1 (with a TSER 3rpt/3rpt TS 6bp/6bp genotype as referent; data not shown).

To our knowledge, this is the first study investigating the relevance of TS polymorphisms in colorectal carcinogenesis. As with other polymorphisms in folate metabolism (15, 17), upon stratification by a relevant exposure (folate), an informative pattern emerged. We observed a statistically significant gene environment interaction, with a decreased adenoma risk associated with high folate intakes among individuals with the TSER wild-type (3rpt/3rpt) genotype and the opposite association (increased risk with high folate intake) among individuals with lower TS expression (2rpt/2rpt). Similar trends were observed for vitamin B-12 intake, which is an essential cofactor for another key enzyme in folate metabolism, methionine synthase.

Under low or medium folate intakes and in the presence of a MTHFR TT genotype, which can be considered to further impair folate status, the TSER variant genotypes were associated with a decreased risk of colorectal adenoma. Thus, we hypothesize that the lower TS expression (2rpt/2rpt or 3rpt/2rpt genotypes) may have a “folate-sparing” effect that reduces risk of adenoma.

The finding of a decreased risk associated with lower TS expression in the presence of low-folate intake (Fig. 1) was unexpected. Previously, several groups have hypothesized that the inverse association between the C677T MTHFR polymorphism and colorectal carcinogenesis may be attributable to a greater diversion of the relevant substrate (5,10-methylene-THF toward nucleotide synthesis; TS and several enzymes involved in purine synthesis; Refs. 15, 17, and 18). TS converts dUMP to dTMP, and it has been shown that folate deficiency can lead to uracil misincorporation associated with an increased risk of double-strand breaks (4, 5). Our findings, however, suggest that the purine synthesis pathway may be a relevant mechanism linking folate metabolism to colorectal carcinogenesis.

Two potential explanations could reconcile our findings with the existing knowledge about this pathway: (a) it is likely that, within the folate pathway, a balance exists between the provision of 5,10-methylene-THF for thymidine synthesis, purine synthesis, and the provision of methyl groups for methylation reactions. On the basis of biochemical models in pig liver, the cellular concentration of 5,10-methylene-THF regulates the flux of this metabolite into the pathways leading to nucleotide synthesis and methionine regeneration (19). It is unclear whether variations in enzyme function could affect these regulatory mechanisms; and (b) it may be that an adequate provision of 5,10-methylene-THF for purine synthesis is key. Depurination is the most common form of spontaneous DNA damage, with an estimated 10,000 depurinations/cell/day (20, 21). Although an abundance of apurinic endonucleases efficiently repairs this damage, quantitative measurements show that abasic sites are present in cellular DNA, with a steady-state level of 5–10,000 lesions/cell/day (21). Thus, depurination may be the major challenge the body faces on a continuous basis, and the provision of an adequate supply of purines may be an important prerequisite for repair. Accordingly, in the presence of low folate, it may be that purine synthesis is the most protected pathway.

Perturbations of nucleotide pools attributable to folate deficiency have been associated with inhibited DNA repair and increased mutation rates (1, 2). However, the role of purine depletion in mutagenesis is unclear. There is no evidence that folate deficiency results in a greater prevalence of apurinic sites, but, to our knowledge, this has not yet been investigated. An effect of folate deficiency on purine pools has been documented in one study (22). The most promising evidence in support of a link between folate deficiency, purine synthesis, and DNA damage stems from investigations in an in vitro model of folate-deficient erythropiesis, indicating that decreased de novo synthesis of purines plays an equally important role as decreased de novo synthesis of thymidylate in the pathogenesis of megaloblastic anemia (23).

Our study had several limitations. The study population was comprised of individuals undergoing screening by colonoscopy and was thus not necessarily representative of the general population. The major advantage of this clinic-based approach, nonetheless, is the unambiguous distinction between patients with adenomas and polyp-free controls. The indications for colonoscopy were not related to the TS genotype or to intakes of the nutrients investigated. Thus, a bias attributable to differences in indication for colonoscopy is unlikely. A strength of the study is the relatively large study size, permitting us to investigate gene-nutrient interactions. However, for analyses investigating genotype combinations (MTHFR and TS) in conjunction with nutritional status, our statistical power was limited.

In summary, previous research on MTHFR polymorphisms has illustrated the importance of genetic variability in the folate pathway in colorectal carcinogenesis. We here extend these findings to a common polymorphism in TS, which modifies risk depending on folate status (and possibly vitamin B-12 status). Our findings raise questions regarding the molecular pathways linking folate metabolism to colorectal carcinogenesis, especially including whether folate is beneficial in the presence of all metabolic genotypes. It may also have implications for the pharmacogenetics of antifolate drugs, such as 5-fluorouracil.

We thank Angela Bush for technical assistance with the DNA genotyping, Rachel Sparks for general research assistance, and Mari Nakayoshi for assistance with graphics and word processing.