Introduction
Methylenetetrahydrofolate reductase (MTHFR) is a key enzyme in the remethylation reaction, that catalyses the reduction of methylenetetrahydrofolate to methyltetrahydrofolate, which is the methyl donor for the conversion of homocysteine to methionine (1). An increased plasma total homocysteine (tHcy) is a risk marker for cardiovascular disease, neural tube defects and other birth defects (2). There is evidence that increased serum Hcy levels are associated with declining cognitive function and dementia (3). Deficiency of B vitamins, in particular folate, and/or mutations in genes coding for enzymes or proteins involved in metabolism, are major causes of elevated concentrations of tHcy (4–8). There are three universally common polymorphisms in the gene for MTHFR and two of them, 677C>T (rs1801133) and 1298A>C (rs1801131), are generally known to affect Hcy concentration to a varying degree (1,5,9–13). However the third, 1793G>A (rs2274976), is less common and has not been well studied, and thus whether it affects tHcy concentration remains controversial (14–16).
Subjects with the TT genotype have normal tHcy if their folate status is optimal (17). The 1298A>C polymorphism is considered not to cause elevated tHcy concentrations, except when present with the 677T-allele in ‘compound heterozygotes’ (18,19). Our previous study investigated the impact of MTHFR haplotypes on plasma tHcy concentrations in Swedish children and adolescents, and evidence was found for a tHcy-raising effect of the 1298C-allele and a tHcy-lowering effect of the 1793A-allele (20). It was hypothesised that in adults too, the haplotype-based approach in combination with data on serum (S)-folate and S-cobalamin levels would facilitate the elucidation of the impact of the MTHFR 1298A>C and 1793G>A polymorphisms on tHcy levels. The present study reports the findings in a representative epidemiological sample of healthy Spanish adult subjects from the Canary Islands.
Materials and methods
Subjects
The blood samples for DNA analysis were obtained from 723 subjects (395 females and 328 males) belonging to the Canary Islands Nutrition Study (ENCA). Serum samples were obtained from 523 subjects (297 females and 226 males). ENCA is a cross-sectional study from the Canary Islands (Spain) which was conducted to survey the nutritional status and selected metabolic and genetic variables of the population of the Canary Islands. The sampling procedures and participation rates have been described previously (21,22). The present study was approved by the Research Ethics Committee of the Hospital Universitario Insular of Gran Canaria (Las Palmas, Canary Islands, Spain). Written informed consent was obtained from the participants.
Homocysteine and B vitamin analyses
As described previously (21), the blood samples were obtained in the morning after subjects had fasted for 12 h. The S-folate levels were measured at the Haematology Unit of the Hospital Universitario Insular of Gran Canaria (Canary Islands, Spain) through an automated ionic capturing method with Abbott AXSYM equipment (Abbot, Berkshire, England, UK). Cobalamin was analysed via the micro-particle enzyme immune assay method with Abbott AXSYM equipment also at the Haematology Unit. Homocysteine was analysed at the University of Barcelona’s Clinical Hospital (Barcelona, Spain), with polarized fluorescence immunoassay in an AXSYM (Abbot) analyser. The vitamin and tHcy values were available for 523 subjects.
DNA extraction and genotyping
Total blood DNA was extracted and purified from 200 μl whole blood anticoagulated with EDTA, using the QIAamp DNA Blood Mini kit according to the manufacturer’s instructions (Qiagen Inc., Valencia, CA, USA). The purity was assessed by the ratio of A280/A260, which was typically 1.7–1.8. All of the polymerase chain reaction amplifications were performed with HotStar Taq DNA polymerase kit (Qiagen Inc., Valencia, CA, USA) and an Eppendorf Mastercycler. The reaction volume was 50 μl for all polymorphisms. MTHFR 677C>T was amplified according to the Pyrosequencing® Assay Protocol ‘Genotyping of the C677T variant in the human methylenetetrahydrofolate reductase (MTHFR) gene’, version 1 (Biotage AB, Uppsala, Sweden; www.biotage.com). Approximately 30 ng of genomic DNA was used as a template. For the MTHFR 1298A>C and 1793G>A polymorphisms, our own genotyping procedures were used, using the Pyrosequencing platform as described previously (23).
Statistics
For examining of the Hardy-Weinberg equilibrium, a χ2 test was applied. Plasma tHcy, S-folate and S-cobalamin concentrations required transformation in order to achieve normal distribution. Following ln transformation, the residuals demonstrated a satisfactory pattern and ln values were used in all statistical analyses. In all tables and figures, untransformed data are provided.
Analysis of covariance (ANCOVA) was used to examine differences in tHcy between the age groups, gender, ln S-folate, ln S-cobalamin and the MTHFR genotypes and haplotypes. Gender had a significant effect on tHcy concentrations and therefore the subjects were stratified by gender. Age group, ln S-folate and ln S-cobalamin all had a significant effect on the tHcy levels and therefore all ANCOVA calculations adjusted for age in four groups (18–25, 25–45, 45–55 and 65–75 years), ln S-folate and ln S-cobalamin. It was previously found in this cohort that alcohol intake, smoking and BMI did not predict homocysteine concentrations (21), therefore these variables were not included in the present study.
When analysing one genotype’s effect on tHcy, adjustments were made for the other two MTHFR genotypes, either by including them in the ANCOVA’s or by stratification, as indicated. Since S-folate and S-cobalamin levels did not differ between males and females, cut-offs for quintiles were generated using all of the subjects. Logistic regression was performed for analysis of the effect of MTHFR 677TT genotype on S-cobalamin levels. S-cobalamin was stratified into two groups below and above 150 pmol/l, which has been previously suggested as a cut-off level for deficiency (24). All of the mean values are estimated marginal means. To examine for a linear trend in ln tHcy concentrations between quintiles of S-folate or S-cobalamin in the subgroups MTHFR 677 CC+CT or MTHFR 677TT, one way ANOVA was performed. Statistical significance was interpreted as values of P<0.05 and confidence intervals at 95%. Statistical analyses were performed using SPSS 15.0 for Windows (SPSS, Inc., Chicago, IL, USA).
Results
MTHFR genotypes and haplotypes
Basic clinical characteristics of the studied population have been published previously (21). Briefly, noted characteristics of the population were as follows: there was a significant difference in median tHcy between males and females (13.1 and 10.9 μmol/l, respectively) and also in median folate intake between males and females (161.6 and 141.9 μg/day, respectively); there were no significant differences for cobalamin intake, S-folate, erythrocyte folate or S-cobalamin. The genotype prevalences and allele frequencies for all three studied MTHFR polymorphisms in the 723 ENCA subjects are revealed in Table I. All loci were in Hardy-Weinberg equilibrium when analysing the entire population.
The frequency of the MTHFR 677T minor allele was q=0.357 and 0.360, respectively, in subjects below and above 40 years of age (χ2=0.01). In females aged below and above 40 years of age, the frequency of the MTHFR 677T-allele was q=0.377 and 0.356, respectively (χ2=0.76), and in males below and above 40 years of age the frequency was q=0.333 and 0.364, respectively (χ2=1.13).
tHcy and MTHFR genotypes or haplotypes
A one-way ANCOVA was performed with MTHFR 677C>T as the fixed factor and gender, age in the four groups, ln S-folate, ln S-cobalamin, and genotypes of MTHFR 1298A>C and 1793G>A as covariates. Adjusted R2 for the model was 0.333. The MTHFR 677C>T polymorphism demonstrated a statistically significant interaction with gender (P=0.02) so in the following studies the results were therefore stratified according to gender. Adjusted for covariates, the MTHFR 677C>T polymorphism had a significant effect on tHcy in males (P=0.005) and adjusted R2 for the model was 0.342. Adjusted for covariates, the MTHFR 677C>T had no significant effect in females.
To isolate the single effect of the MTHFR 1298A>C polymorphism on tHcy, a one-way ANCOVA was performed separately in the genotype groups 677CC and 677CT, with 1298A>C as a fixed factor and age in the four groups, ln S-folate, and ln S-cobalamin as covariates (Table II). The 1298A>C polymorphism had a significant effect on tHcy in males with the 677CT genotype, with a mean tHcy of 1.7 μmol/l higher than the 1298AA wildtype subjects. In males with the 677CC genotype, 1298A>C had no significant effect on tHcy concentrations. In females, there was no significant effect of MTHFR 1298A>C on tHcy concentrations.
To investigate the effect of the MTHFR 1793G>A polymorphism on tHcy concentrations, one-way ANCOVA’s were performed with the diplotypes CCG/CCG and CCG/CCA as fixed factors and age in the four groups, ln S-folate and ln S-cobalamin as covariates. There was no significant effect on tHcy of the MTHFR 1793G>A genotype in neither females nor males, but the sample number of subjects was small. When the study population was stratified in younger vs. older subjects (below and above 52 years of age according to the mean age of menopause) and MTHFR 677C>T genotype (CC+CT vs. TT) ANCOVA (with MTHFR 1298A>C, ln S-folate and ln S-cobalamin as cofactors) demonstrated a significant Hcy-lowering effect of the MTHFR 1793 GA genotype in males <52 years, with a mean of 2.4 μmol/l lower than the males with the MTHFR 1793 GG genotype (Table III).
To investigate the effect of the MTHFR haplotypes on tHcy, ANCOVA was performed with ln tHcy as a dependent factor, haplotype was entered as fixed factor, and age in the four groups, ln S-folate and ln S-cobalamin were entered as covariates. With the two-locus haplotypes 677T-1298A and 677C-1298C as fixed factors, an R2 of 0.352 was obtained in males (P=0.013) for the TA haplotype whereas in females, no haplotype was significantly correlated with tHcy. In a similar ANCOVA utilizing the three-locus haplotypes, 677T-1298A-1793G and 677C-1298C-1793G as fixed factors, no additional power over that of the MTHFR 677C>T genotype was obtained. The haplotype containing the mutated 1793A-allele was excluded from the analysis due to its low prevalence. Fig. 1 summarizes the mean tHcy concentrations in all the different diplotype subgroups.
tHcy and nutrigenetic interactions
S-cobalamin was divided into groups of above and below 150 pmol/l and logistic regression was performed to elucidate the effect of MTHFR 677TT on S-cobalamin. It was revealed that the MTHFR 677TT genotype was not statistically significantly associated with S-cobalamin.
Fig. 2 reveals that the mean tHcy levels increased with lower quintiles of S-folate, statistically significantly in the MTHFR 677 CC+CT subgroups (P<0.001 for both genders). The mean difference between Q1 and Q5 in males was +4.0 μmol/l in the CC+CT subgroup and +18.8 μmol/l in the TT subgroup, and in females it was +3.8 μmol/l in the CC+CT subgroup and -3.6 μmol/l for the TT group.
Fig. 3 demonstrates that the mean tHcy levels also increased with lower quintiles of S-cobalamin. In the MTHFR 677 CC+CT subgroup, P<0.001 for males and P=0.005 for females; in the MTHFR 677TT subgroup, P=0.005 for males and P=0.015 for females. The mean tHcy difference in males between the cobalamin quintiles Q1 and Q5 was +4.6 μmol/l in the CC+CT subgroup and +21 μmol/l in the TT subgroup, and in females it was +1.8 μmol/l in the CC+CT subgroup and +4.6 μmol/l in the TT subgroup.
tHcy in pre- and postmenopausal females
The female subjects were divided into two groups, below and above 52 years of age and ANCOVA was performed (Table IV). There was a significant difference in tHcy concentrations between the two age groups (P<0.001). None of the polymorphisms MTHFR 677C>T, 1298A>C and 1793G>A were significantly correlated with tHcy in the two groups (data not shown).
Discussion
The present study aimed to clarify the nutrigenetic impact on tHcy concentrations of the MTHFR genotypes and haplotypes adjusted for the known covariates age, sex and serum concentration of folate and cobalamin in the Spanish adult population. Within this population, tHcy levels are known to be dependent on both these vitamins (21).
Our allele figures for the MTHFR 677C>T polymorphism are consistent with another study, which investigated the prevalence of MTHFR 677C>T in Spain (25). A study from Majorca (26) reported different frequencies of the q allele between younger and elderly subjects, a genetically implausible finding not supported by the present data from the Canary Islands, or a study comparing the MTHFR 677T-allele frequencies between newborn and >80-year-old Swedish subjects, which were found to be q=0.291 and 0.270, respectively (27). The present figures on q among younger (below 40 years) vs. elderly (above 40 years) differed only marginally and in a way that evidently invites to the assumption of ‘regression towards the mean’.
In the present study, an important finding was the significant gender difference in the effect of the MTHFR 677C>T polymorphism on tHcy concentrations, as is consistent with several other studies (28–30). In Spanish females with the MTHFR 677TT genotype, the reduced levels of 5′-methyl-tetrahydrofolate did not affect tHcy levels if adjusted for major covariates. One possible explanation may be estrogen. Lower levels of tHcy are observed in pregnant females, and in premenopausal and postmenopausal females undergoing hormone replacement therapy (31) (Table IV). Estrogen induces the expression of the gene for phosphatidylethanolamine N-methyltransferase (PEMT) which catalyzes the biosynthesis of phosphatidylcholine, a precursor for betaine. Betaine is an alternative source of methyl groups in the remethylation process of Hcy, and it has been proposed that premenopausal females may supply choline from endogenous biosynthesis (32). By contrast, it has been noted that deletion of the PEMT gene in mice reduces Hcy levels by 50% (33), which would argue against estrogen induction of PEMT as an explanation of the present findings. When the effect of the 677T-allele on tHcy in Swedish children and adolescents was studied, a significant tHcy-raising effect of the T-allele in girls was identified, but it was of smaller magnitude in μmol/l than in the boys of the same ages (20). Yang et al demonstrated that the adverse impact of the MTHFR 677TT genotype on homocysteine concentrations was attenuated by dietary folate intake (34), but in this cohort females actually have lower folate intake than males (21) which precludes this as an explanation. Therefore, the causes of the described male/female differences remain elusive and it may well be that alternative food sources (for instance, choline) contributing to the methyl group balance, unaccounted for in this as well as in the majority of other epidemiological studies, may explain, in part, such gender differences (33). Other studies have also demonstrated that the effect of the MTHFR 677C>T polymorphism may vary between populations, for example, people in Mexico City have a high prevalence of the MTHFR 677C>T polymorphism but they also have a low influence of the polymorphism on Hcy concentrations (35).
Notably, another finding was that in the subjects with the MTHFR 677TT genotype, there was an interaction of tHcy with not only low S-folate, but also with low S-cobalamin (Fig. 3) as has also been observed in other studies (24,36–38). The association of low S-cobalamin with increased tHcy was more robust than that of low S-folate (Figs. 2 and 3). The subjects with the MTHFR 677TT genotype raised their tHcy levels quantitatively more with lower cobalamin quintiles than did subjects with the 677CC or CT genotypes, and the effect was of a greater magnitude (in μmol/l) in males than in females.
The findings suggest that in the Canary Islands both folate and cobalamin are major tHcy-determinants in both males and females, and both vitamins should be included in nutrigenetic studies on MTHFR 677C>T, often regarded as biochemically responsive only to S-folate levels.
The MTHFR 1793A-allele was in complete linkage disequilibrium with the 1298C-allele. Consistent with our previous Swedish study (20), it was identified that the 1793 A-allele has a lowering effect on tHcy levels, however in this study it was only statistically significant in males <52 years of age. The 1298A>C polymorphism had a minor elevating effect on tHcy in males, who were 677CT/1298AC compound heterozygotes.
Nutrigenetic interactions and their effect on biomarkers are commonly overlooked. Nevertheless, even small effects may be uncovered in well-characterized populations and, as it appears from our studies, in younger populations and particularly in males. Several body functions decrease with age, e.g. glomerular filtration rate and tubular function which raise tHcy concentrations. In addition, the older population may be taking vitamin supplementation or drugs affecting Hcy metabolism, for instance, Spaniards >65 years of age take a mean of three or more different medications per day (39,40).
Haplotypes from the MTHFR polymorphisms 677C>T, 1298A>C and 1793G>A (CAG, TAG, CCG, CCA) were constructed and their impact on tHcy was analysed. A small additional explanatory power for tHcy concentrations was obtained using the 677–1298 two-locus haplotype system above that which was provided by the MTHFR 677C>T genotype alone. In our previous study in children, the best explanatory power was obtained by the three locus haplotype (20). It is therefore suggested that, when investigating the association of the MTHFR polymorphisms to tHcy in a population sample, the MTHFR haplotypes should be examined in the calculations and not just the genotypes.
Based on the above results and the findings of our previous study (20), the following set of nutrigenetic statements for the Hcy metabolism are proposed: (i) age, sex and factors linked to the ethnicity of the studied subjects, appear to be able to override the nutrigenetic impact of tHcy-raising MTHFR genotypes or haplotypes in particular settings, exemplified here by Spanish adult females; (ii) gene-nutrient interactions on plasma tHcy levels thus may or may not exist in a certain population; (iii) the transferability of nutrigenetic findings between different communities may therefore be limited, and may possibly need to be re-evaluated for each particular community according to age, sex and ethnicity.
In conclusion, the major genetic impact on tHcy concentrations in Spanish subjects was attributable to the MTHFR 677C>T polymorphism but in the full cohort the effect was limited to males only. A haplotype based analysis was marginally superior to genotype based analyses in accounting for the MTHFR impact on tHcy. A nutrigenetic interaction with low S-cobalamin was also demonstrated: in subjects with the lowest S-cobalamin levels, the tHcy increase was higher among 677 TT homozygotes than in subjects with the 677 CC or CT genotypes and the magnitude of this effect was more pronounced in males.