Learn more: PMC Disclaimer | PMC Copyright Notice
Folate and Cobalamin Modify Associations between S-adenosylmethionine and Methylated Arsenic Metabolites in Arsenic-Exposed Bangladeshi Adults1,2,3
Associated Data
Abstract
Chronic exposure to inorganic arsenic (InAs) through drinking water is a major problem worldwide. InAs undergoes hepatic methylation to form mono- and dimethyl arsenical species (MMA and DMA, respectively), facilitating arsenic elimination. Both reactions are catalyzed by arsenic (+3 oxidation state) methyltransferase (AS3MT) using S-adenosylmethionine (SAM) as the methyl donor, yielding the methylated product and S-adenosylhomocysteine (SAH), a potent product-inhibitor of AS3MT. SAM biosynthesis depends on folate- and cobalamin-dependent one-carbon metabolism. With the use of samples from 353 participants in the Folate and Oxidative Stress Study, our objective was to test the hypotheses that blood SAM and SAH concentrations are associated with arsenic methylation and that these associations differ by folate and cobalamin nutritional status. Blood SAM and SAH were measured by HPLC. Arsenic metabolites in blood and urine were measured by HPLC coupled to dynamic reaction cell inductively coupled plasma MS. In linear regression analyses, SAH was not associated with any of the arsenic metabolites. However, log(SAM) was negatively associated with log(% urinary InAs) (β: −0.11; 95% CI: −0.19, −0.02; P = 0.01), and folate and cobalamin nutritional status significantly modified associations between SAM and percentage of blood MMA (%bMMA) and percentage of blood DMA (%bDMA) (P = 0.02 and P = 0.01, respectively). In folate- and cobalamin-deficient individuals, log(SAM) was positively associated with %bMMA (β: 6.96; 95% CI: 1.86, 12.05; P < 0.01) and negatively associated with %bDMA (β: −6.19; 95% CI: −12.71, 0.32; P = 0.06). These findings suggest that when exposure to InAs is high, and methyl groups are limiting, SAM is used primarily for MMA synthesis rather than for DMA synthesis, contributing additional evidence that nutritional status may explain some of the interindividual differences in arsenic metabolism and, consequently, susceptibility to arsenic toxicity.
Introduction
Worldwide, ∼140 million people are exposed to arsenic at concentrations that exceed the safe drinking water guideline set by the WHO (10 μg/L) (1–3), and >57 million of those exposed live in Bangladesh (4). Exposure to arsenic is associated with cancers of the skin, lung, bladder, liver, and kidney (5–8), in addition to noncancer outcomes including peripheral vascular disease (9), atherosclerosis (10), hypertension (11), peripheral neuropathy (12), and decreased intellectual function in children (13). However, individuals vary in their susceptibility to arsenic-induced health outcomes, and some of this interindividual variation may be explained by differences in arsenic metabolism (14). In contaminated drinking water, arsenic is present as inorganic arsenic (InAs)8. Ingested InAs can be methylated to form mono- and dimethyl arsenical species (MMA and DMA, respectively) (Fig. 1), thereby facilitating arsenic elimination, because DMA has a shorter circulating half-life than does InAs and is rapidly excreted in urine (14). Arsenic-exposed individuals who have a higher proportion of MMA and a lower proportion of DMA in their urine have an increased risk of developing adverse health outcomes (15). Therefore, methylation of arsenic to DMA is considered a detoxification pathway (14).
Both steps of arsenic methylation are catalyzed by arsenic (+3 oxidation state) methyltransferase (AS3MT) and require a methyl group from S-adenosylmethionine (SAM) (16). Synthesis of SAM via one-carbon metabolism depends on folate in the form of 5-methyltetrahydrofolate (5-methylTHF) and cobalamin; the latter acts as a cofactor for methionine synthase, which catalyzes the transfer of a methyl group from 5-methylTHF to homocysteine to generate methionine (Supplemental Fig. 1). Each methylation step requiring SAM yields the methylated product and S-adenosylhomocysteine (SAH) (17). Importantly, SAH is a potent product-inhibitor of most methyltransferases (18), including AS3MT (19). Therefore, the concentrations of SAM and SAH and the ratio of SAM to SAH (SAM:SAH) have been used frequently as indicators of methylation capacity (20–22). Previously, our group reported that folic acid supplementation in folate-deficient adults enhances methylation of InAs to DMA (23). Because SAM synthesis relies on folate- and cobalamin-dependent one-carbon metabolism (Supplemental Fig. 1), folic acid supplementation facilitates arsenic methylation by regenerating this methyl donor. However, the relations between blood SAM and SAH and arsenic methylation have not been examined in a human population. Therefore, the objective of this study was to test the hypothesis that blood SAM, like plasma folate, is associated with increased arsenic methylation, whereas blood SAH is associated with decreased arsenic methylation. We hypothesized that SAM would be negatively associated with the percentage of InAs (%InAs) for all participants. However, we predicted that the relations between SAM and the methylated metabolites (%MMA and %DMA) would differ between individuals who were sufficient or deficient for folate and cobalamin, because the relation between SAM and the methylated arsenic metabolites may depend on whether or not SAM is limiting.
Participants and Methods
Study region.
Our study site, which is the site of the Health Effects of Arsenic Longitudinal Study (HEALS) cohort (24), is currently a 35-km2 area within Araihazar, Bangladesh, which is situated ∼30 km east of Dhaka.
Participants.
The Folate and Oxidative Stress (FOX) Study is a cross-sectional study in 378 participants selected from 5 water arsenic (wAs)–exposure categories [<10 μg/L (n = 76), 10–100 μg/L (n = 104), 101–200 μg/L (n = 86), 201–300 μg/L (n = 67), and >300 μg/L (n = 45) (25)] who were recruited between February and July of 2008. This study had 2 major aims: 1) to study the dose-response relation between wAs exposure and oxidative stress (25, 26) and 2) to study the hypotheses outlined herein.
Participants between the ages of 30 and 65 y were eligible. The following individuals were excluded: 1) women who were pregnant, 2) participants taking nutritional supplements, and 3) participants with known diabetes, cardiovascular or renal disease, or other diseases known to be associated with oxidative stress. Bangladeshi field staff physicians obtained informed consent after reading an approved consent form to study participants. This study was approved by both the Bangladesh Medical Research Council and the Institutional Review Board of Columbia University Medical Center.
General characteristics of study participants.
General characteristics of the study participants (Table 1) were obtained by questionnaire. BMI was calculated by using the measured height and weight of each participant. Dietary intakes of folate and cobalamin were determined by FFQ (27).
TABLE 1
Value | Median (range) | |
Age, y | 43 ± 8 | 42 (30–63) |
Education, y | 3.4 ± 3.6 | 3.0 (0.0–16.0) |
BMI, kg/m2 | 20.4 ± 3.5 | 19.7 (13.8–35.3) |
wAs, μg/L | 140 ± 125 | 114 (0–700) |
uAs, μg/L | 205 ± 229 | 124 (3–1990) |
uCr, mg/dL | 54 ± 43 | 41 (4–224) |
bAs, μg/L | 13.5 ± 9.9 | 10.8 (1.2–57.0) |
bInAs,2 μg/L | 4.3 ± 2.2 | 3.8 (1.4–15.8) |
bMMA,2 μg/L | 6.1 ± 3.8 | 5.1 (0.8–25.1) |
bDMA,2 μg/L | 4.6 ± 2.7 | 3.8 (1.1–22.7) |
bInAs,2 % | 29.5 ± 4.1 | 29.4 (19.7–46.7) |
bMMA,2 % | 39.3 ± 5.4 | 39.7 (20.5–51.5) |
bDMA,2 % | 31.2 ± 5.6 | 31.3 (18.0–46.5) |
uInAs, μg/L | 37 ± 44 | 22 (0–327) |
uMMA, μg/L | 31 ± 39 | 16 (0–303) |
uDMA, μg/L | 140 ± 166 | 89 (2–1290) |
uInAs, % | 17.7 ± 5.6 | 17.2 (6.7–51.8) |
uMMA, % | 14.1 ± 5.0 | 13.4 (3.6–30.0) |
uDMA, % | 68.3 ± 7.9 | 69.4 (38.3–88.0) |
Dietary folate,3 μg/d | 267 ± 88 | 241 (86–674) |
Dietary cobalamin,3 μg/d | 1.50 ± 0.77 | 1.32 (0.09–6.25) |
Plasma folate, nmol/L | 12.8 ± 7.3 | 11.1 (2.4–60.6) |
Plasma cobalamin, pmol/L | 203 ± 113 | 176 (44–1180) |
Plasma homocysteine, μmol/L | 11 ± 13 | 9 (3–165) |
Blood SAM, μmol/L | 1.30 ± 0.53 | 1.18 (0.44–3.69) |
Blood SAH, μmol/L | 0.31 ± 0.18 | 0.27 (0.07–1.37) |
Male, % | 50.7 | — |
Ever smoker, % | 38.2 | — |
Ever used betel nut, % | 43.1 | — |
Own television, % | 58.4 | — |
Underweight,4 % | 32.9 | — |
Folate deficient,5 % | 30.3 | — |
Cobalamin deficient,6 % | 34.8 | — |
Sample collection and handling.
During each participant’s visit to our field clinic, a physician collected a venous blood sample. Spot urine samples were collected in 50-mL acid-washed polypropylene tubes and frozen at –20°C. After blood samples underwent initial processing in the field clinic, aliquots of blood and plasma were immediately frozen at −80°C. Samples were then transported on dry ice to Dhaka by car where they were again stored in −80°C (blood and plasma) or −20°C (urine) freezers. In Dhaka, samples were packed on dry ice and flown to Columbia University.
wAs.
Field sample collection and laboratory analysis procedures are described elsewhere in detail (28, 29). Water samples were analyzed by high-resolution inductively coupled plasma MS after 1:10 dilution and addition of Ge to correct fluctuations in instrument sensitivity. The detection limit of the method is typically <0.2 μg/L. Arsenic standards of known concentration were run multiple times in each batch. The intra- and interassay CVs were 6.0% and 3.8%, respectively.
Total blood arsenic.
Total blood arsenic (bAs) concentrations were measured by using a Perkin-Elmer Elan DRC II inductively coupled plasma mass spectrometer equipped with an AS 93+ autosampler, as described previously (30). The intra- and interassay CVs were 3.2% and 5.7%, respectively.
Total urinary arsenic.
Total urinary arsenic (uAs) concentrations were measured by graphite furnace atomic absorption spectrometry (31) using the AAnalyst 600 graphite furnace system (PerkinElmer), as previously described (25). The intra- and interassay CVs were 3.8% and 5.1%, respectively. A method based on the Jaffe reaction was used to measure urinary creatinine (uCr) concentrations (32).
Blood and urine arsenic metabolites.
Four arsenic metabolites [arsenite (AIII), arsenate (AsV), monomethylarsonous acid plus monomethylarsonic acid (MMAIII+V), and dimethylarsinous acid plus dimethylarsinic acid (DMAIII+V)] were measured in blood and urine by coupling HPLC to dynamic reaction cell inductively coupled plasma MS, as described previously (33). The reduced and oxidized forms of MMA and DMA cannot be separated by HPLC, so each metabolite is measured as 1 variable (i.e., as MMAIII+V or DMA III+V, respectively). All 4 bAs metabolites could only be measured for individuals with total bAs concentrations ≥5 μg/L. Each metabolite (with AsIII + AsV combined as InAs) was calculated as a percentage of the total measured urinary or blood arsenic. The intraassay CVs for urinary AsIII, AsV, MMA, and DMA were 3.6%, 4.5%, 1.5%, and 0.6%, respectively; those for blood were 0.9%, 11.5%, 3.6%, and 2.6%, respectively. The interassay CVs for urinary metabolites were 9.7%, 10.6%, 3.5%, and 2.8%, respectively, whereas those for blood were 3.7%, 23.2%, 2.9%, and 3.5%, respectively.
Plasma folate and cobalamin.
A radio protein-binding assay (SimulTRAC-S; MP Biomedicals) was used to measure folate and cobalamin, as previously described (25, 33). The within- and between-day CVs for folate were 9% and 14%, respectively, and for cobalamin these were 5% and 9%, respectively.
Plasma total homocysteine.
Homocysteine concentrations were measured in plasma by HPLC with fluorescence detection (34). The within- and between-day CVs were 2% and 9%, respectively.
SAM and SAH.
SAM and SAH were measured as described by Wise et al. (35) in whole blood. Briefly, samples were thawed and mixed on a vortex, and 400 μL of blood was added to 200 μL of 0.1 mol/L sodium acetate, pH 6.0, and 160 μL 40% trichloroacetic acid. After 30 min of incubation on ice, tubes were centrifuged for 10 min at 20,817 × g. An aliquot of 200 μL of the supernatant was filtered by using a 0.45-μm Ultra free MC filter (Millipore) and centrifuged for 3 min at 2655 × g for measurement of SAM. For SAH, the remaining supernatant was extracted twice with 100 μL of diethyl ether followed by filtration with another 0.45-μm filter. SAM and SAH were separated by reversed-phase HPLC on a 25 × 0.46 cm (5-μm particle size) column (Beckman Instruments) by using a mobile phase consisting of 50 mmol/L NaH2PO4 and 10 mmol/L heptane sulfonic acid in 18% methanol, adjusted to pH 4.38 with phosphoric acid at a flow rate of 0.9 mL/min. The running column was preceded by a precolumn filter (ChromTech). By using a 996 Photodiode Array UV absorbance detector (Waters), SAM and SAH were detected at 254 nm and were quantitated by comparing the integrated areas under HPLC peaks with standard curves generated by using purified SAM and SAH (Sigma). The interassay CVs for SAM and SAH were 9.6% and 16.1%, respectively.
Statistical analyses.
Of the 378 participants recruited for the FOX study, 353 had complete information for the predictor variables (SAM and SAH), urine outcome variables (%uAs metabolites), and potential confounders. Descriptive statistics were calculated for general characteristics of this study sample, including for arsenic metabolites and nutrition variables; these values are reported as medians (ranges) for continuous variables and as frequencies (%) for categorical variables. The Wilcoxon rank-sum test was used to detect differences in quantitative variables, including SAM and SAH, by dichotomous characteristics, including folate and cobalamin nutritional status (i.e., deficient vs. sufficient). Folate and cobalamin deficiencies were defined by using cutpoints from Christenson et al. (36), as follows: plasma folate <9 nmol/L and plasma cobalamin <151 pmol/L. Spearman correlation coefficients, reported as rho (ρ), were used to assess bivariate relations between quantitative variables including SAM and SAH concentrations, arsenic metabolites, and other continuous measures. To examine the bivariate relations between blood InAs (bInAs) or blood MMA (bMMA) and the ratio of blood DMA (bDMA) to bMMA (bDMA:bMMA), scatterplots and corresponding LOESS curves were plotted in R (R Foundation) using the default smoothing parameter 0.7. Spearman correlations were used to evaluate the statistical significance of the bivariate relations.
Linear regression models were used to evaluate the relation between each of the predictors (SAM, SAH, SAM:SAH) and the outcome variables [% urinary InAs (uInAs), % urinary MMA (uMMA), % urinary DMA (uDMA), %bInAs, %bMMA, %bDMA). The estimated regression coefficient for each predictor of interest is reported as β (95% CI) and P value. Because bAs metabolites could only be measured for individuals with total bAs concentrations ≥5 μg/L, the sample size for bAs outcomes (%bAs metabolites) was smaller (n = 276) than the sample size for uAs outcomes (%uAs metabolites) (n = 353). To meet model assumptions, an ln transformation was applied to arsenic metabolites with skewed distributions (%uInAs, %uMMA, %bInAs) for approximate normality. To satisfy the linearity assumption for regression models, SAM, SAH, SAM:SAH, age, BMI, folate, and cobalamin were ln-transformed, and a square root transformation was applied to wAs. Potential control variables included sex, age, BMI, folate and cobalamin (measured continuously), television ownership (an indicator of socioeconomic status in this population), years of education, cigarette smoking status (ever or never smoker), uCr, estimated glomerular filtration rate, and amount of time (days) that blood samples were stored at −80°C before SAM and SAH analysis by HPLC. All linear regression models were adjusted for age, smoking status, sex, wAs exposure, and blood sample storage time at −80°C. Results were not altered appreciably after further adjusting for television ownership, years of education, BMI, folate, cobalamin, or estimated glomerular filtration rate, so these variables were not included in the final models. The associations between SAM, SAH, SAM:SAH, and arsenic metabolites were also assessed stratified by folate and cobalamin status. The Wald test was used to detect differences between strata in the covariate-adjusted regression coefficient for SAM with respect to methylated bAs metabolites. Because a subset of individuals were deficient for both folate and cobalamin, stratified analyses were also performed in the following 4 groups: deficient for both nutrients, deficient for folate only, deficient for cobalamin only, sufficient for both nutrients. Due to the small sample size of the group deficient for both folate and cobalamin (n = 32), the final models for the 4 groups were adjusted for a restricted set of variables: sex, wAs, and blood sample storage time at −80°C, because the exclusion of the covariates age and cigarette smoking status from regression models did not appreciably alter the coefficient for SAM. SAS and R were used to conduct all statistical analyses, and a significance level of 0.05 was used.
Results
General characteristics of study participants.
Descriptive statistics for general characteristics of the FOX participants are shown in Table 1. By design, the participants were between the ages of 30 and 63 y. Approximately 32.9% of the participants were underweight (BMI <18.5 kg/m2). Median (range) dietary intakes of folate and cobalamin were 241 (86–674) μg/d and 1.32 (0.09–6.25) μg/d, respectively (Table 1); thus, in FOX participants the average dietary intakes of these B vitamins were below the RDA for adults (folate: 400 μg/d; cobalamin: 2.4 μg/d) (37, 38). Prevalences (%) of folate and cobalamin deficiencies were 30.3% and 34.8%, respectively. Due to the study design, the mean wAs exposure was 140 μg/L, which is 14-fold higher than the WHO standard of 10 μg/L. In blood, on average, the %MMA exceeded the %InAs or %DMA. Conversely, in urine, the mean %DMA was higher than the mean %InAs or %MMA.
SAM and SAH concentrations and relations with other indices of one-carbon metabolism.
The relations between SAM, SAH, and other bivariate variables are reported in Supplemental Table 1. SAH concentrations were significantly higher in folate-deficient individuals compared with folate-sufficient individuals. SAM concentrations did not differ by folate nutritional status. However, SAM concentrations were significantly higher in cobalamin-sufficient individuals.
Correlations between nutrition variables are reported in Supplemental Table 2. SAM and folate were not significantly correlated. However, SAM was positively correlated with cobalamin (ρ = 0.17, P = 0.01). SAH was negatively correlated with folate (ρ = −0.15, P < 0.01) and positively correlated with homocysteine (ρ = 0.18, P < 0.001).
Correlations between nutrition variables and arsenic metabolites.
Correlations between nutrition variables and arsenic metabolites are reported in Table 2. Folate was negatively correlated with %InAs and %MMA and positively correlated with %DMA in blood and urine. In contrast, homocysteine was positively correlated with %MMA and negatively correlated with %DMA in blood and urine. Cobalamin was positively correlated with %MMA in urine but not in blood. SAM was positively correlated with %MMA in blood and urine and negatively correlated with %uInAs. SAH was not correlated with any of the arsenic metabolites.
TABLE 2
Blood metabolites (n = 276) | Urinary metabolites (n = 353) | |||||
%InAs | %MMA | %DMA | %InAs | %MMA | %DMA | |
Folate2 | −0.14* | −0.22#x2020 | 0.29#x2021 | −0.15** | −0.21#x2021 | 0.24#x2021 |
Cobalamin3 | −0.06 | −0.04 | 0.07 | −0.10 | 0.12* | −0.03 |
Homocysteine | 0.05 | 0.20** | −0.22#x2020 | −0.03 | 0.22#x2021 | −0.13* |
SAM | −0.03 | 0.13* | −0.08 | −0.11* | 0.12* | −0.01 |
SAM4 | −0.03 | 0.13* | −0.08 | −0.11 | 0.12* | −0.01 |
SAH | −0.04 | 0.01 | 0.02 | −0.06 | 0.02 | 0.04 |
SAH4 | 0.05 | −0.07 | 0.03 | −0.04 | 0.02 | 0.01 |
BMI | −0.09 | −0.06 | 0.12 | −0.06 | −0.18#x2020 | 0.15** |
Relation between blood arsenic metabolites.
bDMA:bMMA decreased with increasing concentrations of bInAs or bMMA (Fig. 2). The Spearman correlations between bInAs or bMMA and bDMA:bMMA were negative and significant (P < 0.05).
SAM, SAH, and SAM:SAH as predictors of arsenic methylation.
In regression analyses, after adjusting for age, sex, wAs, cigarette smoking status, and blood sample storage time at −80°C, there was a significant negative association between log(SAM) and log(%uInAs) (β: −0.11; 95% CI: −0.19, −0.02; P = 0.01) (Table 3). No other significant associations were observed between SAM and the arsenic metabolites (Table 3). SAH and SAM:SAH were not significantly associated with any of the arsenic metabolites (Table 3). The results from the urinary metabolite analyses were very similar after restricting the sample size to participants with measured blood arsenic metabolites. The results were not appreciably altered when SAM and SAH were included simultaneously in the regression models, nor were the results altered after adjusting for folate or cobalamin as continuous predictors in the model.
TABLE 3
log(%InAs) | %MMA2 | %DMA | ||||
β (95% CI) | P | β (95% CI) | P | β (95% CI) | P | |
Blood (n = 280) | ||||||
log(SAM) | −0.01 (−0.05, 0.04) | 0.75 | 0.77 (−0.90, 2.45) | 0.36 | −0.35 (−2.19, 1.49) | 0.71 |
log(SAH) | 0.02 (−0.02,0.05) | 0.37 | −0.70 (−2.03, 0.62) | 0.30 | 0.17 (−1.28, 1.63) | 0.82 |
log(SAM:SAH) | −0.02 (−0.05, 0.02) | 0.33 | 0.86 (−0.26, 1.99) | 0.13 | −0.28 (−1.52, 0.96) | 0.65 |
Urine (n = 359) | ||||||
log(SAM) | −0.11 (−0.19, −0.02) | 0.01 | 0.06 (−0.04, 0.15) | 0.23 | 0.87 (−1.28, 3.01) | 0.43 |
log(SAH) | −0.03 (−0.10, 0.04) | 0.35 | 0.01 (−0.06, 0.09) | 0.77 | 0.59 (−1.12, 2.29) | 0.50 |
log(SAM:SAH) | −0.03 (−0.09, 0.03) | 0.35 | 0.02 (−0.05, 0.08) | 0.56 | −0.03 (−1.53, 1.47) | 0.97 |
In analyses stratified by folate status, log(SAM) was positively and significantly associated with %bMMA in those who were folate deficient (β: 3.49; 95% CI: 0.40, 6.59; P = 0.03) but not in those who were folate sufficient (β: −0.43; 95% CI: −2.47, 1.61; P = 0.68) (Table 4). Stratifying by cobalamin status revealed similar findings; although not statistically significant, log(SAM) was positively associated with %bMMA in those who were deficient for cobalamin (β: 2.58; 95% CI: −0.44, 5.60; P = 0.09) but not in those who were sufficient for cobalamin (β: 0.15; 95% CI: −1.93, 2.24; P = 0.90). The Wald test for the difference in the association between SAM and %bMMA across folate strata was significant (P = 0.04). SAM was also negatively associated with %bDMA in the cobalamin-deficient group (P = 0.01); this was not observed in the folate-deficient group. The association between SAM and %bDMA differed significantly between cobalamin strata (P < 0.01). For uAs metabolites, the positive associations between log(SAM) and log(%MMA) in folate-deficient participants (β: 0.17; 95% CI: −0.01, 0.35; P = 0.07) and cobalamin-deficient participants (β: 0.15; 95% CI: −0.03, 0.32; P = 0.10) did not achieve significance (Supplemental Tables 3 and 4).
TABLE 4
log(SAM) and %bMMA | log(SAM) and %bDMA | |||
β (95% CI) | P | β (95% CI) | P | |
Folate status2 | 0.043 | 0.443 | ||
Folate sufficient (n = 194) | −0.43 (−2.47, 1.61) | 0.68 | 0.14 (−1.98, 2.26) | 0.89 |
Folate deficient (n = 86) | 3.49 (0.40, 6.59) | 0.03 | −1.51 (−5.22, 2.19) | 0.42 |
Cobalamin status4 | 0.193 | <0.013 | ||
Cobalamin sufficient (n = 179) | 0.15 (−1.93, 2.24) | 0.88 | 0.99 (−1.28, 3.25) | 0.39 |
Cobalamin deficient (n = 101) | 2.58 (−0.44, 5.60) | 0.09 | −4.33 (−7.54, −1.12) | 0.01 |
When individuals were stratified into 4 groups on the basis of joint folate and cobalamin status, there was a strong, positive association between log(SAM) and %bMMA in individuals who were deficient for both folate and cobalamin (β: 6.96; 95% CI: 1.86, 12.05; P < 0.01) (Table 5). The Wald test for overall differences in the association between SAM and %bMMA across the 4 nutrition groups was significant (P = 0.02), mainly due to the difference between those who were deficient for both folate and cobalamin and those who were sufficient for the 2 nutrients (P < 0.01). In individuals deficient for cobalamin and in individuals deficient for both folate and cobalamin, there was a negative association between SAM and %bDMA, although these associations were not significant (0.05 < P < 0.10). However, the Wald test for overall differences in the association between SAM and %bDMA across the 4 nutrition groups was significant (P = 0.01) and was driven by the differences between those who were sufficient for both nutrients and those who were deficient for both nutrients (P < 0.01) or deficient for cobalamin only (P = 0.02).
TABLE 5
log(SAM) and %bMMA | log(SAM) and %bDMA | |||
Group2 | β (95% CI) | P | β (95% CI) | P |
Sufficient for both (n = 126) | −1.15 (−3.55, 1.25) | 0.34 | 1.70 (−0.91, 4.31) | 0.20 |
Deficient for folate (n = 52) | 2.26 (−1.38, 5.90) | 0.22 | 0.45 (−3.57, 4.46) | 0.82 |
Deficient for cobalamin (n = 66) | 0.61 (−3.29, 4.52) | 0.76 | −3.48 (−7.43, 0.46) | 0.08 |
Deficient for both (n = 32) | 6.96 (1.86, 12.05) | <0.01 | −6.19 (−12.71, 0.32) | 0.06 |
P for all groups | 0.023 | 0.013 |
Discussion
In vitro (16, 39) and animal (40) studies have established that SAM is necessary for the methylation of InAs to MMA and for the methylation of MMA to DMA and that SAH inhibits both of these methylation steps (19). However, the relations between SAM and the percentage of arsenic metabolites in human populations may be particularly complex for 2 reasons: 1) there is competition between InAs and MMA for methylation, because both methylation steps are catalyzed by AS3MT and require a methyl group from SAM, and 2) these relations may depend on nutritional status, because SAM is synthesized via folate- and cobalamin-dependent one-carbon metabolism. In a cross-sectional study in arsenic-exposed adults in Bangladesh, our group previously observed that folate was negatively correlated with %InAs and %MMA and positively correlated with %DMA in urine, suggesting that folate facilitates the methylation of InAs to DMA (41); we also observed this in the current study. However, in our previous study, we did not analyze blood SAM and SAH concentrations.
With the use of purified recombinant human AS3MT, Song et al. (42) demonstrated that when SAM concentrations are <0.5 mmol/L, the rate of MMA synthesis exceeds the rate of DMA synthesis. Although there are few estimates of human liver SAM concentrations in healthy individuals, reported SAM concentrations in rat liver vary from 60 to 160 μmol/L (43–45). Therefore, according to the findings of Song et al. (42), the rate of MMA production should exceed the rate of DMA production at physiologically relevant concentrations of SAM. Consistent with this, we observed that the mean %MMA exceeded the mean %DMA in blood. Song et al. (42) also observed that with increasing InAs concentrations, the %MMA increases, the %DMA decreases, and the ratio of DMA to MMA decreases. Similarly, Styblo et al. (46) demonstrated that human hepatocytes exposed to increasing concentrations of InAs produce more MMA and less DMA. These studies indicate that the second step of arsenic methylation is inhibited by InAs. Evidence from mathematical modeling also suggests that MMA can inhibit its own methylation, likely due to substrate inhibition (Michael Reed and Fred Nijhout, Duke University, personal communication). Thus, methylation of InAs to MMA may predominate over the methylation of MMA to DMA in individuals who are continuously exposed to high concentrations of InAs, such as the current study participants, because 1) high amounts of InAs compete with MMA for methylation and 2) the second methylation step is inhibited by MMA. Our finding of a decrease in bDMA:bMMA with increasing concentrations of bInAs and bMMA is consistent with this. Furthermore, our observation that SAM is positively associated with %bMMA but only in folate- and cobalamin-deficient participants suggests that limiting SAM concentrations further reduces the ability to methylate MMA to DMA.
Although we had anticipated a strong positive association between SAM and %bDMA in folate- and cobalamin-sufficient participants, many of the FOX participants were drinking from wells with very high water arsenic concentrations; thus, the null association between SAM and %bDMA in sufficient participants may reflect the strong inhibition of DMA synthesis by InAs and MMA. Alternatively, it may reflect saturation of the AS3MT at higher concentrations of SAM. Although Song et al. (42) observed that within the range of physiologically relevant concentrations of SAM, DMA production increases with increasing SAM concentrations, glutathione was the only reductant used in their assays; other groups have demonstrated that arsenic methylation is more efficient in the presence of other reductants, such as thioredoxin (47). Thus, saturation of AS3MT may occur at lower physiologically relevant concentrations of SAM.
Because folate and cobalamin are involved in SAM synthesis, a simplistic prediction was that both folate and cobalamin would be positively correlated with blood SAM concentrations. However, SAM was not correlated with plasma folate in the FOX participants. Increasing SAM concentrations leads to inhibition of methylenetetrahydrofolate reductase through long-range allosteric interactions, such that production of 5-methylTHF decreases when SAM concentrations increase. This negative feedback loop may explain why there is no observable correlation between plasma 5-methylTHF and SAM in our study. Loehrer et al. (48) similarly observed no correlation between SAM and 5-methylTHF in their case-control study examining the relation between folate and coronary artery disease. As expected, in the FOX participants, plasma folate was negatively correlated with both SAH and plasma homocysteine concentrations and was positively correlated with %DMA in blood and urine. Additionally, cobalamin was positively correlated with SAM, and cobalamin-sufficient individuals had significantly higher SAM concentrations than did cobalamin-deficient individuals. This finding is reasonable given that, unlike folate, cobalamin concentrations are not regulated by SAM.
We did not find SAH to be significantly associated with any of the arsenic metabolites, which was surprising given that SAH is a potent inhibitor of most SAM-dependent methylation reactions in vitro (18). However, it is important to note that we measured SAM and SAH in blood, yet arsenic methylation primarily occurs in the liver. Although blood SAM and SAH concentrations are considered indicators of methylation capacity and have been used as biomarkers in several other studies (49–51), we are unaware of any studies that have directly compared liver and whole-blood SAM and SAH concentrations. Although homocysteine is readily exported from cells (52), the transport of intact SAH across the plasma membrane is not well characterized (53). James et al. (53) proposed that homocysteine may, in fact, serve as an exportable form of SAH, because it is more readily transported out of the cell (54) and may therefore be a better indicator of liver SAH concentrations than is blood SAH. As in previous studies, we found plasma homocysteine to be positively correlated with %MMA in blood and urine and negatively correlated with %bDMA. If plasma homocysteine is indeed a better indicator of hepatic SAH concentrations than is blood SAH itself, these findings are consistent with inhibition of the second methylation step of MMA to DMA by SAH in the liver.
The findings of this study have 3 major implications. First, the null associations between SAH and the arsenic metabolites highlight the need for additional research examining the utility of blood SAM and SAH (or alternatively plasma homocysteine) as biomarkers of liver SAM and SAH. Second, the inverse relation between either InAs or MMA and the ratio of DMA to MMA indicates that arsenic metabolism may not be as efficient in populations that are continuously exposed to high concentrations of InAs. Third, the observed positive association between SAM and %bMMA in folate- and cobalamin-deficient individuals suggests that these individuals may be particularly susceptible to arsenic-induced toxicity, because a higher %MMA in urine has been associated with multiple adverse health outcomes (15). Previously, we observed that folate deficiency and hyperhomocysteinemia are risk factors for arsenic-induced skin lesions (55). The findings of this study contribute additional evidence that folate and cobalamin deficiencies, and hyperhomocysteinemia, may help to explain a portion of the interindividual variation in arsenic methylation capacity and in susceptibility to arsenic toxicity. Although eliminating arsenic exposure should remain the primary target for reducing its toxicity, this work and previous studies collectively indicate a significant need for public health interventions directed toward alleviating these micronutrient deficiencies, particularly in arsenic-exposed populations.
Acknowledgments
The authors thank Michael Reed and Fred Nijhout from Duke University for providing valuable insight based on their findings from mathematical models of arsenic metabolism. M.V.G. designed the study; C.G.H., M.M.N., M.N.H., and X.L. designed the statistical analysis plan; V.I., V.S., S.A., and A.B.S. conducted the research; J.H.G.’s laboratory performed the arsenic analyses; C.G.H. and X.L. performed the statistical analyses; C.G.H., M.V.G., and M.N.H. wrote the manuscript; and C.G.H. and M.V.G. had primary responsibility for the final manuscript. All authors read and approved the final manuscript
Footnotes
8Abbreviations used: AsIII, arsenite; AsV, arsenate; AS3MT, arsenic (+3 oxidation state) methyltransferase; bAs, blood arsenic; bDMA, dimethyl arsenical species in blood; bInAs, inorganic arsenical species in blood; bMMA, monomethyl arsenical species in blood; DMA, dimethyl arsenical species; DMAIII, dimethylarsinous acid; DMAV, dimethylarsinic acid; FOX, Folate and Oxidative Stress; InAs, inorganic arsenical species; MMA, monomethyl arsenical species; MMAIII, monomethylarsonous acid; MMAV, monomethylarsonic acid; SAH, S-adenosylhomocysteine; SAM, S-adenosylmethionine; uAs, urinary arsenic; uCr, urinary creatinine; uDMA, dimethyl arsenical species in urine; uInAs, inorganic arsenical species in urine; uMMA, monomethyl arsenical species in urine; wAs, water arsenic; 5-methylTHF, 5-methyltetrahydrofolate.