Nutrition, One-Carbon Metabolism and Arsenic Methylation

Abstract

Exposure to arsenic (As) is a major public health concern globally. Inorganic As (InAs) undergoes hepatic methylation to form monomethyl (MMAs)- and dimethyl (DMAs)-arsenical species, facilitating urinary As elimination. MMAs^III is considerably more toxic than either InAs^III or DMAs^V, and a higher proportion of MMAs in urine has been associated with risk for a wide range of adverse health outcomes. Efficiency of As methylation differs substantially between species, between individuals, and across populations. One-carbon metabolism (OCM) is a biochemical pathway that provides methyl groups for the methylation of As, and is influenced by folate and other micronutrients, such as vitamin B₁₂, choline, betaine and creatine. A growing body of evidence has demonstrated that OCM-related micronutrients play a critical role in As methylation. This review will summarize observational epidemiological studies, interventions, and relevant experimental evidence examining the role that OCM-related micronutrients have on As methylation, toxicity of As, and risk for associated adverse health-related outcomes. There is fairly robust evidence supporting the impact of folate on As methylation, and some evidence from case-control studies indicating that folate nutritional status influences risk for As-induced skin lesions and bladder cancer. However, the potential for folate to be protective for other As-related health outcomes, and the potential beneficial effects of other OCM-related micronutrients on As methylation and risk for health outcomes are less well studied and warrant additional research.

1. Introduction

Arsenic (As) exposure through drinking water and food is a worldwide public health concern. In a process dependent on one-carbon metabolism (OCM), ingested inorganic As (InAs) is methylated to form monomethyl (MMAs) and dimethyl (DMAs) arsenicals. Complete methylation of InAs to DMAs facilitates urinary As elimination and reduces As-related toxicity. OCM is a biochemical pathway that is influenced by nutrients including folate, vitamin B₁₂, choline, betaine and creatine. This article is an update of our review published in 2018,¹ and provides an overview of As exposure, methylation, toxicity, and risk for As-related health outcomes. We then summarize the evidence that nutritional status and interventions that influence OCM nutrient levels can increase As methylation capacity and potentially reduce the risk of As-related toxicity. We will primarily focus on human studies of folate. These findings have important implications for many countries with the highest levels of As exposure, as these regions also tend to have a high prevalence of folate deficiency and could benefit from major public health interventions such as food folate fortification programs currently mandated in more than 80 other countries.²

1.1. Routes of Arsenic Exposure

1.1.1. Arsenic in Drinking Water

Arsenic is a naturally occurring metalloid found in insoluble forms in soil, which can leach into groundwater once mobilized.³ It is estimated that over 140 million people in at least 70 countries are exposed to > 10 μg/L of As via drinking water, the WHO⁴ and United States (US) Environmental Protection Agency (EPA)⁵ safety standard. Much of the global burden of As-related morbidity and mortality is carried by those countries that are the most severely affected by As exposure, including Bangladesh, Cambodia, China, India, Myanmar, Nepal, Pakistan, Taiwan, Argentina, and Vietnam.⁶ In Bangladesh, large-scale exposure to As through drinking water began in the 1970s. The United Nations Children’s Fund (UNICEF) and the Department of Public Health Engineering installed tube wells to reduce infant mortality related to diarrheal disease connected to contaminated surface water. Arsenic contamination of well water was not identified until the 1990s, with concentrations that ranged from below the limit of detection (LOD) to over 1,000 μg/L.⁷ Elevated As concentrations in groundwater are also found in the Western, Midwestern, and Northeastern US.⁸ In the US, it has been estimated that about 730,000⁹ people are exposed to As-contaminated water > 10 μg/L served by community water systems, and more than 2.1 million¹⁰ people exposed to water As > 10 μg/L from private wells. This is related to challenges in implementing the drinking water standard and the use of private wells, which are not included in EPA drinking water regulations.¹¹

1.1.2. Arsenic in Food and Other Beverages

In regions where drinking water As concentrations are low, ingestion of As through As-contaminated food and beverages can represent a relatively large proportion of As exposure. In human experimental^12,13 and epidemiological studies,^14,15 consumption of rice and rice products has been associated with increased urinary As. Arsenic found in rice is primarily arsenate (InAs^V), arsenite (InAs^III), or dimethylarsinic acid (DMAs^V).¹⁶ Consumption of rice-based products is a particular concern for infants and children who have significantly higher relative As intake as a result of a higher per-body-mass food intake and a less varied diet.¹⁷ However, in prospective cohort studies, rice consumption has not been associated with increased risk for cancers or CVD, although these studies were based on food frequency questionnaires and did not measure As concentrations in rice or As biomarkers.^18–20 Arsenic has also been detected in apple juice²¹ and some red wines,²² with As concentrations exceeding 10 μg/L. Arsenic concentrations in common foods and beverages assessed by the US Food and Drug Administration (FDA) and EPA have been previously summarized.¹

1.1.3. In Utero and Early Life Arsenic Exposure

Arsenic species can readily pass through the placenta, and maternal blood concentrations of total As and As species are highly correlated with those of umbilical cord blood.^23,24 Arsenic does not readily enter the mammary gland; consequently, low concentrations of As, e.g., 1 to 3 μg/L, have been found in breast milk.^24–26

1.1.4. Arsenic Exposure via Inhalation

Arsenic can be absorbed by the respiratory tract. This is a concern for occupations that involve As use or production such as smelting, and individuals living in nearby communities.²⁷ Inhalation exposure may also be due to burning of As-containing biomass, as occurs in regions where cattle are exposed to As through drinking water and consumption of contaminated rice straw.²⁸

1.2. Health Effects of Arsenic Exposure

1.2.1. Health Effects of Arsenic Exposure in Adults

Arsenic is a Group 1 human carcinogen and toxicant, and affects numerous organs.²⁹ Exposure to As is associated with an increased risk of health outcomes such as skin lesions (melanosis, leukomelanosis, and keratosis); cardiovascular disease (CVD); impaired intellectual function; diabetes and diabetes-related outcomes;³⁰ and cancers of the bladder, lung, liver and skin (Figure 1).^31–33 Increased risk for several As-related cancers, e.g. lung and bladder cancer, have been shown to persist long after exposure to As has been reduced.³⁴ There is some evidence suggesting that As exposure is also associated with increased risks of breast³⁵ and prostate³⁶ cancers. In Taiwan, chronic exposure to As is associated with blackfoot disease.³⁷

Figure 1.

Arsenic Target Tissues and Comorbidities. Chronic As exposure has been associated with increased risk of skin lesions (melanosis, leukomelanosis, and keratosis), cardiovascular disease, impaired intellectual function, inflammation, diabetes, and cancers. Ingested As accumulates in multiple tissues including the spleen, liver, lungs, bladder, esophagus, skin, and bone marrow.

1.2.2. Health Effects of Arsenic Exposure in Children and Adolescents

Arsenic exposure in childhood and adolescence has been linked to non-alcoholic fatty liver disease,³⁸ atherogenic lipid profile³⁹ and increased lifetime risk of cancer.⁴⁰ A potential association between arsenic exposure and high blood pressure has also been observed in a cross-sectional study of adolescents.⁴¹ Arsenic exposure has been also associated with adverse neurological effects in adolescents⁴² and cognitive effects in children.^43,44

1.2.3. Health Effects of In Utero Arsenic Exposure

Arsenic exposure, as measured by maternal urinary As, has been associated with decreased birth weight and gestational age, and increased infant mortality,⁴⁵ although results from epidemiological studies have been inconsistent.^46–48 A recent meta-analysis found a negative association between maternal As exposure and birth weight (summary regression coefficient = −25.0 g; 95% CI: −41.0, −9.0).⁴⁹ Epidemiological studies have also found in utero As exposure to be associated with impaired immune function.⁵⁰ In utero and early-life exposure has been associated with increased mortality from chronic renal disease and lung, bladder, liver, and laryngeal cancers in adulthood.^51,52 Additionally, As exposure may play a role in the development of gestational diabetes mellitus (GDM).⁵³

1.3. Arsenic Methylation and Elimination

Arsenic in drinking water is primarily present as InAs, either as arsenate or arsenite. Once ingested, InAs is methylated in a process that facilitates urinary As elimination.⁵⁴ As described by Challenger in 1945, As undergoes sequential reduction and oxidative methylation reactions (Figure 2).⁵⁵ OCM is essential for the synthesis of the methyl donor, S-adenosylmethionine (SAM). Arsenic-3-methyltransferase (AS3MT) catalyzes the transfer of a methyl group derived from SAM to InAs^III, forming monomethylarsonic acid (MMAs^V).^55–57 MMAs^V is then reduced to monomethylarsonous acid (MMAs^III), an intermediate with very high cytotoxicity and genotoxicity.^58–60 AS3MT then catalyzes a second methylation reaction, forming DMAs^V.^55–57 DMAs^V is substantially less toxic than MMAs^III or InAs^{III + V}, and is rapidly excreted in urine.⁶⁰ The role of As methylation in urinary As elimination is particularly evident in mice deficient in AS3MT; these mice accumulate As within tissues at levels 16–20 times greater than wild-type mice resulting in severe systemic toxicity.^61,62

Figure 2.

Arsenic Methylation. According to the Challenger pathway, arsenic (+3 oxidation state) methyltransferase (AS3MT) catalyzes the oxidative methylation of arsenite using s-adenosylmethionine (SAM) as the methyl donor, forming methylarsonic acid (MMA^V), and s-adenosylhomocysteine (SAH). MMA^V is then reduced to methylarsonous acid (MMA^III) before a subsequent oxidative methylation step yielding dimethylarsinic acid (DMA^V) and SAH.

Total urinary As adjusted for urinary dilution is a well-established biomarker of As exposure in humans. Arsenic methylation patterns are analyzed as the relative distribution (%) of InAs, MMAs and DMAs in urine. Analytical advances have also allowed for measurement of total As and As species in blood. Although the distribution As species differs in urine and blood, urine and blood As species concentrations are highly correlated (Spearman ρ range: 0.68 – 0.81).⁶³ However, the half-life of DMAs in blood is relatively short, resulting in considerably lower %DMAs and higher %MMAs in blood than urine.⁶³

Trivalent arsenicals are readily oxidized to pentavalent forms by environmental oxygen, and therefore it is difficult to distinguish between the valence states particularly in studies where biological samples are stored for prolonged periods. Most methods cannot measure detectable levels of MMAs^III or DMAs^III,⁶⁴ thus, this was not traditionally done in epidemiological studies. While InAs^III elutes as a separate peak from InAs^V, most human studies report levels of InAs as InAs^III+V.

1.4. Influence of Arsenic Methylation on Arsenic Toxicity

Arsenic species vary considerably in their toxicity, and trivalent arsenicals are considerably more toxic than pentavalent species. Studies in human cell lines have determined the most to least toxic forms to be: MMAs^III, DMAs^III, InAs^III, InAs^V, MMAs^V and DMAs^V.^58,65 Epidemiological studies and a meta-analysis of cancer risk⁶⁶ have found positive associations between %MMAs in urine and risk of cancers of the bladder,^67–70 breast,^35,71 lung,^69,72 and skin,^73,74 in addition to skin lesions,^75–77 atherosclerosis,⁷⁸ and peripheral arterial disease,^79,80 although for the latter there were some inconsistencies. In Figure 3, we plotted the estimated relative risk for disease from several of these studies illustrating that the effect estimates associated with higher %MMAs in urine for these health outcomes range from approximately 1.6 for skin lesions to as high as 5.5 for skin cancer. Note that this is not a comprehensive representation of all studies linking decreased As methylation to health outcomes. Not all studies reported %MMAs, e.g. some studies reported increased risk of bladder cancer in relation to increased primary methylation,⁸¹ decreased secondary methylation index (DMA/MMAs), and/or decreased %DMAs in relation to e.g. risk for skin lesions⁸² and urothelial carcinoma⁸³. In contrast, associations with hypertension^84–87 were generally null, and more recently, inverse associations have been found between %MMAs in urine and risk of diabetes, and diabetes-related outcomes including metabolic syndrome and insulin resistance, with relative risks ranging from 0.35 to 1.05.^{30,33,88–91} These findings are also consistent with an inverse association between %MMAs and adiposity (not shown in the figure).^88,91–96 However, the relationship between As methylation and metabolic outcomes is not fully understood, and we cannot at present fully rule out reverse causality or confounding. A limitation common to most of the studies in Figure 3 is that they employ prevalent cases^{35,67–69,71–74} and therefore temporality cannot be firmly established; however, experimental data^58,61,62,65 and nested case-control studies in which As species were measured prior to disease onset^75,76 also support the toxicity of higher %MMA and risk for multiple As-related health outcomes.

Figure 3.

Summary plot of estimated relative risks and 95% confidence intervals for adverse health outcomes reported to be associated with MMAs in urine (Citations provided in Section 1.2).

Efficiency of As methylation is variable within and across species, individuals, and populations. In humans, genetic variation is a strong determinant of As methylation capacity^95,97–100 and As-related health outcomes (e.g., skin lesions and cancer).^75,100

1.5. Mechanisms of Action of Arsenic Influenced by One-Carbon Metabolism

Arsenic’s mechanisms of action are likely multifactorial and are not completely understood. Biological mechanisms involved in As-related health outcomes may involve DNA repair, oxidative stress, cell cycle disruption, endocrine disruption, inhibition of thymidylate synthesis, epigenetic dysregulation, microbiome disruption, cytotoxicity, and proliferative regeneration.^101–110 We briefly summarize mechanisms which are likely influenced by OCM, recognizing that the role of nutrition on these processes is often overlooked.

1.5.1. Gut Microbiome Disruption

Disruption of the gut microbiome (GM) has been shown to influence OCM in mouse models. In GM-disrupted mice, there was downregulation of the hepatic expression of OCM-related genes, including folate receptor 2 (FOLR2), betaine homocysteine methyltransferase (BHMT), and methylene tetrahydrofolate reductase (MTHFR). Additionally, GM-disrupted mice exposed to As had a significant decrease in hepatic SAM levels compared to GM-disrupted mice that were not exposed to As. GM disruption may potentially influence the biotransformation of As through impacts on OCM.¹⁰⁸ GM disruption was also found to increase the toxic effects of As as indicated by increased mortality and alteration in expression of cancer-related genes.^108,109

There is also evidence that the biotransformation of As in mammals can be influenced by gut bacteria. GM composition, influenced by a wide range of genetics or pathogenic infections, has been associated with As methylation patterns, indicative of a possible role of GM phenotypes in As methylation.^111,112

1.5.2. Epigenetic Dysregulation.

Early-life exposure to As increases the risk of diseases later in life, with As-related health effects persisting long after exposure has ceased.^113,114 Studies in animals and human cells strongly suggest that the etiology of As-related diseases, such as lung cancer, involves epigenetic dysregulation (e.g.^115,116), which could be influenced by nutritional status. Studies of epigenetic dysregulation associated with chronic As exposure have been previously reviewed.^105,117,118 In summary, As exposure has been shown to induce genome-wide DNA hypomethylation in cell-culture studies, likely contributing to genomic instability.¹¹⁹ In epidemiological studies, As has been associated with alterations in global DNA methylation,^120–123 loci-specific DNA methylation,^{117,124–127} global 5-hydroxymethylcytosine (a DNA demethylation intermediate product),¹²⁸ and with histone modifications.¹²⁹ In several studies of Bangladeshi adults, we found that As was positively associated with peripheral blood mononuclear cell (PBMC) global DNA methylation, a potential compensatory mechanism to prevent hypomethylation, and an effect that was modified by folate status.^120,121,130 Epigenome-wide association studies of As exposure have found associations with DNA methylation levels at individual loci, although a common signature of As exposure has not been identified.^{117,120,121,130–132} Although few epidemiological studies have investigated mediation of the association As and health outcomes by epigenetic dysregulation, a study conducted in Bangladesh found an association between As exposure and methylation of four CpG sites and risk for skin lesions (Bonferroni-adjusted significance threshold: P_Bonferroni <1×10⁻⁷).¹²⁶

2. Nutritional Influences on Arsenic Methylation via One-Carbon Metabolism

2.1. Nutrition and One-Carbon Metabolism

The synthesis of SAM through OCM is essential for the methylation of hundreds of substrates including As, DNA, RNA, histones, lipids, small molecules and proteins¹³³ (Figure 4). Critical micronutrients involved in OCM include folate, vitamin B₁₂ (cobalamin), betaine, choline, vitamin B₆ (pyridoxal phosphate), riboflavin, and niacin. In the folate cycle, a methyl group is transferred from serine to tetrahydrofolate (THF) by serine hydroxymethyltransferase (SHMT) to form 5,10-methylene-THF, which is subsequently reduced to 5-methyl-THF by MTHFR. 5-methyl-THF is the most prevalent form of naturally occurring dietary folate.¹³⁴ Fortification of foods uses the synthetic and chemically stable form of folate, folic acid (FA), which can be reduced by dihydrofolate reductase to THF. The methyl group of 5-methyl-THF is transferred to homocysteine (Hcy) by methionine synthase utilizing B₁₂ as a cofactor, generating methionine and THF. For this reaction, an alternative methyl donor is betaine, which is obtained exogenously from the diet or endogenously synthesized from choline. Methionine is then activated by methionine adenosyltransferase to form SAM.¹³⁵ Along with the methylated products formed from all methylation reactions, S-adenosylhomocysteine (SAH) is also produced, and is hydrolyzed to form Hcy. SAH is a potent product inhibitor of most methyltransferase enzymes; therefore, 5-methyl-THF is critical for pulling the pathway forward. SAM concentrations are tightly regulated by numerous long-range allosteric interactions. Under conditions of folate deficiency where the pool of SAM may become limiting, methylation reactions, such as As methylation, may decrease. However, other critically essential reactions, such as DNA methylation, have very low K_Ms for SAM and can thus operate at full capacity even when SAM concentrations are quite low. Although OCM requires many micronutrients, folate is a primary nutritional factor involved in influencing circulating Hcy concentrations. For example, in a cross-sectional survey of Bangladeshi adults with high prevalence of elevated Hcy, 15% and 5% of the variance in Hcy were explained by plasma folate and B₁₂, respectively.¹³⁶ However, the relative contribution of micronutrients to the variance in Hcy likely varies by additional factors such as age, diet, sex and geographical region.

Figure 4.

One-Carbon Metabolism (OCM). In the folate cycle, a methyl group is transferred from serine to THF by serine hydroxymethyl-transferase, with pyridoxal phosphate (PLP) as a coenzyme, forming 5,10-methylene-THF. This is either used for the synthesis of thymidylate or is reduced to 5-methyl-THF. 5-methyl-THF is the most prevalent form of naturally occurring dietary folate. Fortification of foods uses the synthetic and chemically stable form of folate, folic acid (FA), which can be reduced by dihydrofolate reductase to THF. The methyl group of 5-methyl-THF is transferred to homocysteine in a reaction catalyzed by methionine synthase (MTR) and utilizing B₁₂ as a cofactor, generating methionine and THF. Alternatively, in the liver, betaine can donate a methyl group for the remethylation of homocysteine in a reaction catalyzed by betaine homocysteine methyltransferase (BHMT). Methionine adenosyltransferase enzymes activate methionine to form S-adenosylmethionine (SAM), the methyl donor for numerous acceptors, including arsenicals, guanidinoacetate (GAA, the precursor to creatine), and DNA, in reactions that involve substrate-specific methyltransferase enzymes. These methylation reactions generate the methylated products and S-adenosylhomocysteine (SAH), a potent product inhibitor of most methyltransferases. SAH is hydrolyzed to generate homocysteine which is either remethylated to regenerate methionine or is directed to the transsulfuration pathway and to glutathionine synthesis. (Inset) Betaine can either be obtained from the diet or can be generated through endogenous synthesis from choline. However, because choline is synthesized de novo through a pathway in which PEMT catalyzes three sequential SAM-dependent methylation reactions, endogenous choline synthesis is a significant consumer of SAM. PEMT uses three molecules of SAM to synthesize phosphatidyl choline (PC) from phosphatidylethanolamine (PE), and is transcriptionally regulated by estrogen.

2.1.1. Major Consumers of SAM.

The two major consumers of SAM are guanidinoacetate methyltransferase (GAMT), important to creatine synthesis, and phosphatidylethanolamine methyltransferase (PEMT), important to phosphatidylcholine (PC) synthesis. In omnivores, about half of creatine requirements are obtained through dietary intake, primarily through the consumption of meat.¹³⁷ Therefore, urinary creatinine concentrations reflect both creatine intake through the diet and endogenous synthesis of creatine.¹³⁷ Approximately 50% of all SAM-derived methyl groups are consumed by the methylation of guanidinoacetate (GAA), the final step of creatine synthesis.^138,139 Dietary intake of creatine inhibits the synthesis of GAA, and consequently inhibits creatine biosynthesis; this has been shown to spare methyl groups and lower Hcy in rodents.^140–142 The other major consumer of SAM, PEMT, uses three molecules of SAM to synthesize PC from phosphatidylethanolamine, and is transcriptionally regulated by estrogen.¹⁴³ Other methyltransferase enzymes, including AS3MT, consume only a small fraction of SAM. For example, based on one-carbon kinetics,¹⁴⁴ we estimated that even at high levels of As exposure (500–1,000 μg/L), methylation of 80% of a daily dose of InAs to DMAs would require approximately 50 μmol SAM, or roughly 2–4% of the SAM normally turning over in a well-nourished adult per day.¹⁴⁵

2.2. One-Carbon Metabolism and Arsenic Methylation in Animal Models

Animal models provided the first evidence that As methylation and toxicity were influenced by nutritional regulation of OCM. Vahter & Marafante found that rabbits fed diets deficient in methyl donors (i.e., methionine, choline, or protein) had significantly lower total urinary As and DMAs excretion, and increased uptake of As in tissues.¹⁴⁶ Similar findings were also reported in mice.¹⁴⁷ The effect of folate on As methylation and neural tube defects (NTDs) was subsequently tested by Finnell’s group using mice nullizygous for several folate-binding proteins involved in cellular uptake of folate from the circulation, e.g., Folbp-1 and −2, and/or reduced folate carrier proteins.^148–151 Folbp-1 knockout mice were more susceptible to As-induced NTDs.¹⁴⁸ Mice exposed to As and maintained on folate deficient diets had a reduction in urinary DMAs excretion in all genotypes assessed.^148–151 While the link between As exposure and NTD risk has not yet been firmly established in humans, a study by Mazumdar et al. in Bangladesh found that the efficacy of FA in preventing NTDs was reduced by As exposure.¹⁵² Among mothers with low folate status, As exposure was associated with a greater risk of having an infant with myelomeningocele.¹⁵³

Animal models have limited utility for studying the effects of OCM-related micronutrients on As methylation efficiency for four main reasons: (1) As methylation capacity varies drastically between species; (2) humans are more likely to develop As-related cancers than animals; (3) rodents are unlikely to develop folate deficiency, and (4) it is difficult to replicate chronic (i.e., decades-long), low-dose As exposure in rodent studies. There is new promise for a resolution to the problem of high As methylation efficiency among rodents with the recent generation of a mouse model carrying a humanized BORCS7/AS3MT locus by syntenic replacement, as the As methylation profile of these mice more closely resembles that of humans.¹⁵⁴

2.3. One-Carbon Metabolism and Arsenic Methylation in Humans

This section will summarize the literature on the influence of OCM-related nutrients on As methylation efficiency throughout the life course, with a primary focus on folate. The research conducted in humans on the associations between creatine, vitamin B₁₂, choline, and betaine with As methylation profiles is also summarized below. All results are significant with p-values < 0.05 unless otherwise stated. Although the influence of these OCM-related micronutrients on As methylation are not as well studied as folate in regard to As methylation, they provide potential avenues for future research.

2.3.1. Folate

Isolated case reports provided the earliest human data on the role of folate in As-related toxicity. For example, a 1965 report in Lancet described an 18-year-old female treated for psoriasis with Fowler’s solution who developed clinical signs of As-related toxicity. She was found to be folate deficient and her symptoms responded to intramuscular FA treatment.¹⁵⁵ Another case study presented a 16-year-old girl with severe MTHFR deficiency and homocystinuria exposed to an As-containing pesticide that developed severe symptoms of As-related toxicity.¹⁵⁶ Epidemiologic studies of As exposure and folate status (described below) emerged later.

2.3.1.1. Folate and Arsenic Methylation during Adulthood

The relationship between folate and As methylation capacity among adults has been studied in observational and intervention studies, as has previously been reviewed,¹ and is here updated in Table 1. In 2002, a cross-sectional study by Smith’s group investigated As-contaminated drinking water of 11 families in Chile (N=44). Arsenic methylation was weakly correlated between fathers and mothers; however, adjustment for plasma folate and Hcy increased the strength of these the correlations, suggesting these factors account for some of the variation in As methylation capacity.¹⁵⁷ In an early case-control study in West Bengal, India, Smith’s group found that higher serum folate was associated with significantly lower %MMAs.¹⁵⁸ In 2005, another study conducted by Smith’s group in the western US assessed diet using questionnaires, and found that lower dietary protein, iron, zinc, and niacin were associated with higher urinary %MMAs and lower %DMAs.¹⁵⁹ Unlike the previous studies conducted by Smith’s group, no significant associations with dietary folate intake were found; however, these study participants likely had fairly high folate intake due to the 1998 US mandate for the FA fortification of foods.

Table 1.

Summary of epidemiological studies on folate and arsenic methylation

Study	Design	Population	Measure of folate status	Summary of results related to folate*	Summary of results related to other OCM micronutrients*
Mitra et al. 2004;196 Basu et al. 2011158	Case-control study for As-induced skin lesions	Individuals in West Bengal, India with drinking water As <500 μg/L (N=405)	24-hr dietary recall; serum folate	• Lowest quintile of folate intake was not significantly associated with skin lesions risk • Highest to lowest quintile of folate intake had a significant positive trend in OR for skin lesions • Compared to the highest tertile, the lowest tertile of serum folate was associated with lower %InAs and higher %MMAs in urine	• Lowest quintile of animal protein intake was associated with risk of skin lesions • Highest to lowest quintile of animal protein intake had a significant positive trend in OR for skin lesions • Compared to the highest tertile, the lowest tertile of urine creatinine was associated with higher %InAs, and lower %MMAs and %DMAs in urine • Lowest tertile of plasma Hcy was associated with lower urine %MMAs compared to the highest tertile • Compared to the highest tertile, the lowest tertile of riboflavin intake was associated with lower %InAs and higher %DMAs in urine • Compared to the highest tertile, the lowest tertile of animal protein intake was associated with lower urine %MMAs
Steinmaus et al. 2005159	Crosssectional study	Individuals in Western U.S. (N=87)	Dietary questionnaire	• Folate intake was not significantly associated with urine As species proportions	• Compared to highest quartile, the lowest quartile of protein intake was associated with higher %MMAs and lower %DMAs in urine
Gamble et al. 2005160	Crosssectional study	Individuals in Bangladesh (N=300)	Plasma folate	• Plasma folate was associated with lower %InAs and %MMAs, and higher %DMAs in urine	• Total homocysteine was associated with higher %MMAs and lower %DMAs in urine • Cysteine was associated with lower %InAs and higher %MMAs in urine • Urinary creatinine was associated with lower %InAs and higher %DMAs in urine
Gamble et al. 2006;161 Gamble et al. 200763	Randomized controlled trial	Folate deficient individuals in Bangladesh (N=300)	Supplementatio n of 400 μg FA/day for 12 weeks	• Compared to placebo, FA treatment was associated with a greater decrease in %InAs and %MMAs, and increase in %DMAs in urine • Compared to placebo, FA treatment was associated with a greater decrease in total blood As and blood MMAs concentrations	• Urinary creatinine was associated with lower %InAs and higher %DMAs in urine
Li et al. 2008176	Cohort study	Pregnant women in Bangladesh (N=864)	Plasma folate at gestational week 14	• Compared to the highest tertile, the lowest tertile of plasma folate was associated with higher urine %InAs among women with urine As >209 μg/L • Compared to the highest tertile of plasma folate, B12, and zinc, the lowest tertile of plasma folate, B12, and zinc were associated with higher urine %InAs and primary methylation index	• Urinary creatinine was associated with lower %iAs, and higher %DMAs, primary methylation index, and secondary methylation index
Pilsner et al. 2009130	Nested case-control study for As-induced skin lesions	Individuals in Bangladesh (N=548)	Plasma folate	• Compared to high folate, low folate (< 9 nmol/L) was associated with increased risk of skin lesions • Compared to the referent group of low Hcys + high folate, low homocysteine + low folate, hyperhomocysteinemia + high folate, and hyperhomocysteinemia + low folate were all associated with increased risk of skin lesions	• Compared to low homocysteine, hyperhomocysteinemia was associated with increased risk of skin lesions • A fold increase in urinary creatinine was associated with decreased risk of skin lesions • Compared to the lowest quartile of urinary creatinine, the second and third quartile of urinary creatinine were associated with decreased risk of skin lesions
Gardner et al. 2011178	Cohort study	Pregnant women in Bangladesh (N=324)	Plasma folate	• In multivariate mixed effects linear models, plasma folate was not associated with change in urine As proportions during pregnancy	• In multivariate mixed effects linear models, plasma B12 was not associated with change in urine As proportions during pregnancy
Peters et al. 2015;162 Bozack et al. 2017163	Randomized controlled trial	Individuals in Bangladesh (N=610)	Supplementatio n of 400 μg FA/day, 800 μg FA/day, or 400 μg FA + 3g creatine/day for 12 weeks	• Compared to placebo, 800 μg FA/day treatment was associated with a greater decrease in total blood As • Compared to placebo, FA treatment was associated with a greater decrease in %InAs and %MMAs, and increase in %DMAs in urine	• Compared to placebo, creatine treatment was associated with a greater decrease in urine %MMAs
López-Carrillo et al. 2016165	Crosssectional study	Women in Northern Mexico (N=1,027)	Dietary questionnaire	• Compared to low folate intake, High folate intake ≥ 400 μg/day was associated with lower urine %InAs and higher total methylation	• B12 intake was associated with lower urine %InAs and higher urine %DMAs, second methylation, and total methylation • Choline intake was associated with lower urine %InAs and higher urine %DMAs, second methylation, and total methylation • Methionine intake was associated with lower urine %InAs and higher urine %DMAs, second methylation, and total methylation
Kurzius-Spencer et al. 2017166	Crosssectional study	Adults and children in the U.S. (N=2420)	Red blood cell and serum folate; 24-hour dietary recall	• Folate intake was associated with lower urine %InAs and higher urine %DMAs among adults • Red blood cell and serum folate were not associated with urine As species proportions • In multivariate models, red blood cell folate was associated with lower urine %InAs • In multivariate models, folate intake and serum folate were not associated with urine As species proportions or secondary methylation index	• Dietary B6 was associated with lower urine %InAs among adults • Plasma Hcy was associated with higher urine %MMAs and lower secondary methylation index among children • In multivariate models, urinary creatinine was associated with lower urine %InAs and %MMAs, and higher urine %DMAs and secondary methylation index • In multivariate models, B12 intake was associated with higher urine %InAs and lower urine %DMAs • In multivariate models, B6 intake was associated with lower urine %InAs • In multivariate models, plasma Hcy was associated with higher urine %MMAs, and lower urine %DMAs and secondary methylation index
Spratlen et al. 2017167	Cohort study	American Indians (N=405)	Dietary questionnaire	• In fully adjusted models, folate intake was not associated with urine As species proportions • Compared to low folate intake and low B6 intake, high folate intake and high B6 intake were associated with lower %InAs and higher %DMAs in urine	• Compared to the lowest tertile, the highest tertile of B2 intake was associated with lower %MMAs and higher %DMAs in urine • Compared to the lowest tertile, the highest tertile of B6 intake was associated with lower %InAs and %MMAs, and higher %DMAs in urine • First principal component of OCM nutrients (representing intake of all OCM nutrients) was associated with lower %InAs and %MMAs, and higher %DMAs in urine
Laine et al. 2018177	Cohort study	Pregnant women and infant pairs in US (N=188)	Cord serum folate	• Compared to infants born to folate replete mothers, infants born to mothers in the lowest tertile of serum folate had significantly higher mean levels of %MMA in cord serum	• Higher maternal Hcys was positively associated with maternal total urinary and cord serum As, and with %MMAs in cord serum
Bozack et al. 2019;164 Bozack et al. 2019181	Randomized Control Trial	Adults in Bangladesh (N=622)	Plasma folate	• Relative to the placebo group, the 400- and 800-μg FA treatment groups had increases in As methylation capacity between baseline and week 12, as measured by decreases in urinary %InAs and %MMAs, and increases in urinary %DMAs	• No differences were found in treatment effects between the 400 μg FA and creatine + FA groups.
Saxena et al. In Press.173	Crosssectional study	Adolescents in Bangladesh (N=679)	RBC and plasma folate	• Plasma folate was negatively associated with %InAs and positively associated with %DMAs in male participants	• In male participants, Hcys was positively associated with %InAs • In female participants, plasma Hcys was inversely associated with %MMA

^*Except where otherwise indicated, all results were significant at P < 0.05.

To better understand the role of OCM nutrients such as folate in As methylation and toxicity, our group has conducted multiple studies in a Bangladeshi population with chronic As exposure through drinking water. In 2006, we reported the results of a cross-sectional study in Bangladesh, in which we observed that plasma folate was negatively associated with %MMAs and positively associated with %DMAs in urine.¹⁶⁰ We went on to conduct a randomized, double-blind, placebo-controlled trial of folate-deficient adults (plasma folate <9 nmol/L) to study the effect of 400-μg FA per day (the US recommended dietary allowance) on As methylation. The treatment group had significantly larger decreases in %InAs and %MMAs and larger increases in %DMAs in urine following 12 weeks of supplementation compared to placebo, with treatment effects found as early as one week of intervention.¹⁶¹ Furthermore, compared to the placebo group, total blood As and blood MMAs concentrations were lowered following FA supplementation by 14% and 22%, respectively.⁶³ In the Folic Acid and Creatine Trial (FACT), a randomized controlled trial (RCT) in Bangladesh, participants were selected independent of folate status, and received 400 μg FA, 800 μg FA, 3 g creatine, 3 g creatine + 400 μg FA, or placebo daily for 24 weeks, including 12-week wash-out period where half of the FA-treated groups were switched to placebo to evaluate the effects of supplementation cessation on As methylation. Each participant received an As-removal water filter. After 12 weeks, the treatment group receiving 800 μg FA per day had a larger decrease in total blood As than the placebo group.¹⁶² Additionally, between baseline and week 12, all FA treated groups had increases in As methylation capacity, i.e., decreases in urinary %InAs and %MMAs, and increases in urinary %DMAs, relative to the placebo group.^163,164 Further emphasizing the importance of continuous adequate folate nutritional status, FACT found that following the 12-week wash-out period, proportions of As species in urine reverted to pre-intervention levels. The results of these studies have important policy implications, as food FA-fortification programs may be necessary to achieve a sustained effect on folate status and improved As methylation capacity in As-exposed regions.

Our findings of the influence of folate on As methylation capacity have been confirmed in epidemiological studies including those with populations exposed to lower levels of As. A cross-sectional study in Mexico used a food frequency questionnaire to estimate the micronutrient intake and found that higher folate intake was significantly associated with lower %InAs and a higher ratio of DMAs/InAs.¹⁶⁵ An analysis of the 2003–2004 US National Health and Nutrition Examination Survey (NHANES) found that in unadjusted models, dietary folate intake was negatively associated with %InAs and positively associated with %DMAs in urine, and in adjusted models, red blood cell (RBC) folate was negatively associated with urinary %InAs.¹⁶⁶ In the Strong Heart Study (SHS), a population-based prospective cohort study of American Indian adults residing within 13 tribal communities with low-to-moderate As exposure, a food frequency questionnaire was used to estimate nutrient intake in American Indians, and high combined intake of folate and vitamin B₆ was associated with lower urinary %InAs and higher %DMAs.¹⁶⁷ In contrast, no significant associations were found between dietary folate intake and As methylation efficiency in a study in Bangladesh. However, accurate measurement of dietary folate intake is difficult in Bangladesh due to local traditions of prolonged cooking that degrades naturally occurring food folate.¹⁶⁸ This underscores the benefits of assessing circulating folate concentrations over that of dietary intake in order to more accurately assess folate nutritional status.

2.3.1.2. Folate and Arsenic Methylation during Childhood and Adolescence

The literature regarding OCM and As methylation during childhood and adolescence is limited. Children are reportedly more efficient at methylating As than adults.^169–172 In a cross-sectional study of 6-year-old children in Bangladesh (N=165), plasma folate was inversely correlated with %InAs and positively correlated with %DMA in urine. Although the latter did not achieve statistical significance, the magnitude of the association (r = 0.12, p = 0.14) was similar to that observed in cross-sectional studies of adults (r = 0.14, p < 0.05), and the study sample size of children was relatively small. In a study of 9-year-old children by Vahter’s group, plasma folate was inversely associated with %InAs in urine and positively associated with %DMA. While less data is available for adolescents, we recently conducted a cross-sectional study of Bangladeshis between the ages of 14–16. In males, plasma folate was negatively associated with %InAs and positively associated with %DMAs in urine, consistent with our previous studies in adults. While no significant associations between folate and urinary As species were observed among females, RBC folate was inversely associated with total blood As.¹⁷³ While speculative, it is possible that these findings may be related to short term fluctuations in estrogen, as estrogen is known to influence both OCM and choline synthesis,^174,175 and choline likely influences As methylation (see sections 2.1.1. and 2.3.4.). However, additional research in adolescents is needed to better understand these findings.

2.3.1.3. Folate and Arsenic Methylation in Prenatal Development and Early Life

Maternal OCM is altered during pregnancy due to increased requirements needed to support fetal development. Maternal plasma folate levels change considerably throughout pregnancy as folate is preferentially transferred to the fetus.¹⁷⁵ Several studies have investigated the role of OCM-related micronutrients in the efficiency of As methylation in pregnant women. A cross-sectional study conducted by Vahter’s group that evaluated women at 14 weeks gestation in Bangladesh (N=753) found an inverse association between plasma folate and %InAs in urine. Compared to women that were sufficient in folate, vitamin B₁₂, and zinc, those that were deficient in all three nutrients had significantly higher urinary %InAs and lower primary methylation indices.¹⁷⁶ A study conducted by Laine et al. in a pregnancy cohort in Mexico, found that the distribution of As species in cord serum may be influenced by maternal OCM status: compared to infants born to folate sufficient mothers, those born to folate deficient mothers had significantly higher mean levels of %MMA in cord serum. Furthermore, higher maternal Hcys was positively associated with maternal total urinary and cord serum As, along with %MMAs in cord serum.¹⁷⁷ A subsequent study conducted by Vahter’s group assessed Bangladeshi women at gestational weeks 8, 14, and 30. Although neither plasma folate nor B₁₂ were associated with As species proportions, gestational week was significantly negatively associated with %InAs and %MMAs in urine, and positively associated with %DMAs.¹⁷⁸ Due to dramatic changes in OCM during the course of pregnancy, it is difficult to interpret these observed associations between gestational week and As methylation. Further complicating the evaluation of OCM nutrient status during pregnancy, all women in this study received a daily supplement of 400-μg FA starting from week 14, potentially influencing As methylation efficiency.

2.3.2. Creatine and Creatinine (see also Section 2.1.1.)

Urinary creatinine is a degradation product of creatine that reflects both dietary creatine consumption and creatine synthesized endogenously. It is secreted into urine at a relatively constant rate over the course of a day and urinary creatinine concentrations are commonly used to adjust for urine dilution. Urinary creatinine has consistently been found to be negatively associated with urinary %InAs and positively associated with urinary %DMAs, as observed by members of our group^160,161,170 and others.^158,179 For reasons that remain somewhat unclear, urinary creatinine is a remarkably robust predictor of As methylation efficiency.

The role of OCM on creatine synthesis at the “expense” of large amounts of SAM-derived methyl groups is described in Section 2.1.1. As dietary creatine, through meat or through dietary supplements, downregulates creatine synthesis, we hypothesized that creatine supplementation may facilitate As methylation by sparing methyl groups in our FACT study. We found that participants receiving 3 g of creatine per day (equivalent to about 1.5 times the normal daily creatine turnover for a 70-kg male) for 12 weeks lowered GAA as predicted, indicating that creatine supplementation downregulated creatine synthesis.¹⁸⁰ Additionally, the creatine treatment group had a mean decrease of %MMAs in urine that exceeded that of the placebo group at weeks 6 and 12. However, changes in %InAs or %DMAs in the creatine treatment group were not significantly different than placebo.¹⁸¹

There are known metabolic interactions between folate and choline/betaine: e.g., when folate is low, choline/betaine are preferentially used to remethylate Hcy. In FACT analyses stratified by baseline choline status, creatine supplementation resulted in a decrease in the primary methylation index in the low choline strata but not in the high choline strata.¹⁸² This may be related to the fact that when choline status is low, PC synthesis by PEMT is upregulated, increasing consumption of SAM (see sections 2.1.1. and 2.3.4).^183,184 It is also possible that creatine treatment effects were tempered by long-range allosteric regulation of OCM. Further research and consideration of alternative mechanisms may be required to fully understand the strong cross-sectional associations between As methylation and urinary creatinine. For example, urinary creatinine reflects meat and methionine intake, which could influence As methylation. The role of creatine/creatinine on As methylation may be multifactorial and warrants further study.

2.3.3. Vitamin B₁₂ (cobalamin)

Methylcobalamin plays an essential role as a co-factor for methionine synthase, a critical step in the process of SAM synthesis, thus, one would predict that vitamin B₁₂ status, like folate, would be positively associated with As methylation capacity. However, vitamin B₁₂ has been inconsistently linked to As methylation. A rodent study found that exposing dams to As via drinking water induced fasting hyperglycemia and insulin resistance among male offspring; this phenotype was rescued with a folate/B₁₂ supplemented diet during gestation. These findings suggest that prenatal exposure to As may have a sex-specific adverse effect on glucose metabolism. The supplemented diet only had marginal effects on As methylation in the dams when compared to dams receiving a folate/B₁₂ adequate diet, but standard chow diets are high in folate and no folate-deficient group of dams was available for comparison.¹⁸⁵

Associations between B₁₂ and As methylation have also been investigated in epidemiological studies. In a cross-sectional study in Bangladesh that oversampled B₁₂-deficient individuals (N=778), we found that plasma B₁₂ concentrations were negatively associated with urinary %InAs, positively associated with %MMAs, and not associated with %DMAs.¹⁸⁶ Surprisingly, an analysis of the 2003–2004 NHANES found that dietary B₁₂ intake was positively associated with urinary %InAs and negatively associated with %DMAs, but found no association between biomarkers of B₁₂ nutritional status, including plasma B₁₂ and methylmalonic acid, and As methylation.¹⁶⁶ Other studies have found dietary B₁₂ to be either unrelated to As methylation profiles¹⁶⁷ or to be associated with increased As methylation.¹⁶⁵

We further assessed associations between B₁₂ and As species in urine and blood in the FACT and the Folate and Oxidative Stress (FOX) studies. In baseline samples from FACT, we found that B₁₂ had nonsignificant correlations with As species in urine and negative associations with the concentrations of As species in blood. Although B₁₂ was positively associated with urinary %MMAs (Spearman correlation coefficient ρ=0.12, p=0.02) in the FOX study, the correlation with blood MMAs concentrations was null (M.V. Gamble, unpublished data). In addition, our group’s recent study in adolescents found B₁₂ to have no significant associations with urinary As species.¹⁷³

Inconsistencies in studies of B₁₂ and As methylation may be related in part to the relatively low nutritional requirements for B₁₂ (e.g., the US RDA for B₁₂ is 2.4 μg/d and the US RDA for folate is 400 μg/d), and it takes considerably longer to become B₁₂ deficient. In addition, dietary sources of B₁₂, almost entirely meat, are very different from folate which is derived largely from vegetables and fortified grains. Consequently, deficiencies of these two micronutrients do not necessarily overlap and there are lifestyle factors that distinguish them from one another that could influence As methylation capacity. Currently, there are no published studies that have investigated the treatment effects of B₁₂ supplementation on As methylation efficiency in exposed populations. Additional research is required to fully understand the role of B₁₂ in As methylation.

2.3.4. Choline and betaine.

In the liver, the remethylation of Hcy to methionine can utilize a methyl group from 5-methyl-THF or from betaine (Figure 4) via betaine-homocysteine S-methyltransferase (BHMT). As ~30% of choline is synthesized de novo through a pathway in which PEMT catalyzes three sequential SAM-dependent methylation reactions, endogenous choline synthesis is a significant consumer of SAM.¹⁸⁷ PEMT has eight estrogen response elements and is strongly upregulated by estrogen, contributing to important sex differences in OCM.¹⁷⁵

There are relatively few epidemiological studies of choline and As methylation. A cross-sectional analysis (N=1,016) nested within the Health Effects of Arsenic Longitudinal Study (HEALS) cohort in Bangladesh, found a positive association between dietary choline intake and the secondary methylation index.¹⁶⁸ In a study of Mexican women, López-Carrillo et al. found an inverse association between dietary choline and urinary %MMAs and positive associations with %DMAs and the secondary methylation index.¹⁶⁵ Findings from these observational studies suggest that dietary betaine has a weaker association than dietary choline with As methylation. For example, in both the HEALS cohort and López-Carrillo et al. study, dietary betaine intake was not associated As methylation efficiency.^165,168

We conducted an 8-week pilot intervention of choline (700 mg/day), betaine (1,000 mg/day) or choline + betaine supplementation in Bangladesh (N=60). Compared to placebo, groups receiving supplementation had significantly different within-person changes in urinary %MMAs and %DMAs (M.N. Hall & M.V. Gamble, unpublished data). While the sample size was small, results suggest that choline + betaine supplementation led to the largest decrease in urinary %MMAs and increase in %DMAs, supporting the hypothesis that As methylation is influenced by both dietary choline and betaine. However, choline supplementation also increased plasma tri-methylamine oxide (TMAO), which has been linked to CVD risk. While it is not clear if the association between TMAO and CVD is causal,¹⁸⁸ this finding tempered our interest in larger intervention studies with choline supplementation.¹⁸⁹

2.3.5. Other One-Carbon Metabolism Micronutrients

Riboflavin (vitamin B₂) is a cofactor for methionine synthesis, while niacin (vitamin B₃) is a precursor for nicotinamide adenine dinucleotide (NAD), which is involved in OCM, other biosynthetic pathways, energy metabolism, and safeguards against ROS.¹⁹⁰ In a random sample from the HEALS cohort (N=1041), increase in dietary riboflavin intake was inversely associated with urine %InAs, positively associated with %MMAs, and was not associated with %DMAs. Additionally, increase in dietary intake of niacin was associated with higher ratios of DMAs/MMAs.¹⁶⁸ While these results suggest that niacin and/or riboflavin intake may influence As methylation capacity, no intervention studies have tested these hypotheses.

2.3.6. Impact of One-Carbon Metabolism-Related Polymorphisms

Several studies have been conducted to investigate the effect of OCM-related variants on As methylation. A case-control study of skin lesions in Bangladeshi adults (N=594 cases and 1,041 controls) led by H. Ahsan investigated the effect of two SNPs in MTHFR (rs1801133 and rs1801131); neither risk for skin lesions nor As methylation profiles differed significantly between haplotypes or diplotypes.⁷⁵ In the Gene Environment Nutrient Interactions (GENI) study, we genotyped 26 SNPs in 13 OCM-related genes in a nested case-control study of incident skin lesions in Bangladesh (876 matched pairs).⁸² Although SNPs in MTHFR, methionine synthase (MTR), and serine hydroxymethyltransferase 1 (SHMT1) were nominally significant with urine As species proportions, no SNPs were statistically significant after FDR correction. In models predicting skin lesion risk, nominally significant interactions were found for three SNPs [rs1540087 and rs7109250 in folate receptor 1 (FOLR1) and rs1801394 in methionine synthase reductase (MTRR)] with water As, but were no longer significant following FDR correction. The MTR rs1805087 G allele and SHMT1 rs1979277 A allele were associated with higher %InAs, while the MTHFR rs1801133 T allele was associated with higher %MMAs and lower %DMAs in models predicting skin lesion risk. Nominally significant interactions were found for three SNPs [rs1540087 and rs7109250 in FOLR1 and rs1801394 in MTRR] with water As, but were no longer significant following FDR correction. Pierce et al. genotyped 19,992 variants in 4,939 Bangladeshi individuals. One SNP, rs61735836, which is located in the FTCD gene encoding formimidoyltransferase cyclodeaminase, a bifunctional enzyme that channels one-carbon units from formiminoglutamate to the folate pool, was associated with increased urinary %InAs, increased %MMAs, and decreased %DMAs, and the A allele was associated with an increased risk of skin lesions.¹⁹¹

Other groups have also investigated the association between OCM-related variants and As methylation efficiency. Consistent with our findings, Lindberg et al. found that the MTHFR rs1801133 T allele was associated with higher urinary %MMAs and lower %DMAs in Central Europeans (N=415).⁹⁵ Although, SNPs in MTHFR were not associated with As methylation in a study of Argentinian women (N=104), among individuals with the CC (AlaAla) genotype, rs9001 located in CHDH (encoding choline dehydrogenase) was negatively significantly associated with urinary %MMAs and positively with %DMAs.¹⁹² In a case-control study of lung cancer in Argentina (N=142), Smith’s group also evaluated the association between SNPs in six OCM-related genes and As methylation, finding a positive significant association rs4920037 in cystathionine-β-synthase (CBS) and urinary %MMAs.¹⁹³ Additionally, in the SHS a polymorphism in MTR was associated with higher urinary %MMA. Furthermore, a cross-sectional study conducted in Mexican women (N=1,027) found a significant interaction between folate hydrolase 1 (FOLH1) polymorphisms and vitamin B₁₂ intake; compared to carriers of wild-type TT, CT and CC carriers had significantly lower %InAs, and higher DMA/InAs with higher vitamin B₁₂ intake.¹⁹⁴ In addition, a study conducted in Colombia (N=151) found significant gene-gene (AS3MT-GSTM1), gene-demographic factors (AS3MT-age) and gene-habit interactions (AS3MT-alcohol consumption) in models predicting %MMA.¹⁹⁵ The results of these studies suggest that As methylation may be influenced by gene x nutrient x environment interactions.

2.4. One-Carbon Metabolism and Arsenic-related Health Outcomes

Observational studies have identified associations between OCM nutrients and As-related health outcomes, which may be mediated by effects on As methylation. Here we summarize research related to OCM nutrients, skin lesions, cancer, and diabetes.

2.4.1. Skin Lesions

In 2004, Smith’s group reported that, in a nested case-control study of As-related skin lesions (keratosis and melanosis) in West Bengal, India (192 cases and 192 controls), dietary folate intake was inversely associated with risk for skin lesions.¹⁹⁶ Similarly, in a nested case-control study of skin lesions (N=274 cases and 274 controls) of Bangladeshi adults, our group found that low folate, high Hcy (hyperhomocysteinemia), and low urinary creatinine were associated with an increased risk of skin lesions.¹³⁰ We later conducted a larger nested case-control study (N=876 cases and 876 controls) of As-related skin lesions, and also found that participants with hyperhomocysteinemia and lower urinary %DMAs had an increased risk of developing skin lesions within the following 2–7 years.⁸²

OCM may have a potential role in As-related toxicity that is independent of As methylation. In our GENI study of As-related skin lesions, individuals exposed to water As exposure >50 μg/L and having a polymorphism in thymidylate synthetase (TYMS) (rs1001761) had an increased risk of developing skin lesions.⁸² TYMS is critical for DNA synthesis and repair and requires 5,10-methylene-THF to methylate 2-deoxy-uridine-5-monophosphate (dUMP) to form 2-deoxy-thymidine-5-monophosphate (dTMP) (Figure 4).¹⁹⁷ Thus, we hypothesize that at higher As concentrations, TYMS-related DNA damage may be a potential mechanism of As toxicity. A study from the Stover group found that As trioxide at 0.5 μM (equivalent to 75 μg/L As in water) targets thymidylate biosynthesis.¹⁰² In this study, cell cultures exposed to As had reduced folate-dependent dTMP biosynthesis, with resulting uracil misincorporation into DNA and genomic instability. The effect of As on uracil misincorporation and genomic instability was further exacerbated by folate deficiency, revealing another potential mechanism involving folate deficiency and increased risk of developing adverse health outcomes.¹⁰²

2.4.2. Cancer

Folate status was associated with urothelial carcinoma risk in a case–control study (N=177 cases and 488 controls) conducted in Taiwan of a population exposed to low water As concentrations. Higher urinary %DMAs and plasma folate concentrations were associated with a decreased risk for urothelial carcinoma. Additionally, an interaction between urinary As methylation patterns and plasma folate was observed in relation to urothelial carcinoma risk.⁸³ In a population-based case-control study of US adults (N=1213 cases and 1418 controls), individuals with high cumulative lifetime As exposure (>22.42 mg) and insufficient folate intake (≤400 μg/day) had a higher bladder cancer risk compared to participants with both low cumulative As (≤3.52 mg) and sufficient folate intake (>400 μg/day). However, among participants with sufficient folate intake there was no association with high cumulative As exposure and bladder cancer risk.¹⁹⁸

Arsenic and OCM-related nutrients have been shown to be associated with posttranslational histone modifications (PTHMs). An analysis of three PTHMs that are dysregulated in cancers (H3K36me2, H3K36me3, H3K79me2) was conducted within FACT (N=324) using histones isolated from PBMCs. The study found As exposure to be associated with higher levels of %H3K36me2 among men and levels of this mark decreased over time following use of As-removal filters. We also found sex-related associations between OCM-related nutrients and PTHMs. Among male participants, plasma choline was positively associated with %H3K36me2 (FDR < 0.05), whereas among female participants, plasma vitamin B₁₂ was positively associated with %H3K79me2 (FDR < 0.01). The study also found that FA supplementation did not change the PTHMs assessed. This study suggests that OCM and PTHMs may play a role in the sex-specific risks for As-related cancer outcomes.¹⁹⁹

2.4.3. Diabetes and Metabolic Syndromes

The SHS has identified inverse associations between urinary %MMAs and the incidence of diabetes and diabetes-related outcomes.²⁰⁰ In an expansion of the SHS (N=1,838) that includes family members of SHS participants known as the Strong Heart Family Study (SHFS), As exposure was associated with a higher incidence of diabetes, although exposure levels were different across regions and findings stratified by regions were not reported. Additionally, in a pilot study within a subset (N=59) of the SHFS), waist circumference was inversely associated with %MMA and positively associated with %DMA in urine. However, adjustment for OCM-related nutrients, including glutamate and choline-related metabolites, attenuated and reversed the direction of these associations.²⁰¹ BMI has also been associated with As species in female but not in male participants in FACT (N=527) and were similar but not significant in FOX (N=342). The adjusted mean differences (95% CI) in female participants’ urinary MMA% and DMA% for a 5 kg/m² difference in BMI were −1.21 (−1.96, −0.45) and 2.47 (1.13, 3.81), respectively in FACT, and −0.66 (−1.56, 0.25) and 1.43 (−0.23, 3.09) in FOX. However, these associations were null after adjustment for choline suggesting that choline is an important confounder of the association between BMI and As methylation.⁹⁶ The observed association between lower urinary %MMA and higher %DMA with diabetes- and obesity-related outcomes may be influenced by OCM status and further research is required to assess whether these associations are causally related.²⁰¹

Additionally, in the SHS, a polymorphism in MTR was associated with higher urinary %MMA and lower insulin resistance as measured by Homeostatic Model Assessment Index (HOMA2-IR). While adjustment for OCM variables (SNPs in: PEMT, MTR, MTRR, MTHFR, MTHFD1, SHMT1 and CBS; and nutrient intake: vitamin B₆, vitamin B₂, and folate), did not affect the associations between As methylation and diabetes-related outcomes, including HOMA2-IR, risk for diabetes, and metabolic syndrome, the association between the MTR variant and HOMA2-IR was no longer significant after adjustment for urinary %MMAs.²⁰²

A study conducted in the New Hampshire Birth Cohort (N=1,151) found a significant positive association between maternal toenail As concentrations and GDM.⁵³ Mothers with GDM have been shown to have a 7-fold increased risk of developing Type II diabetes 10 years after pregnancy, and infants born to mothers with GDM have an increased risk of developing glucose intolerance and cardiometabolic disease in the future.^203–206 A review conducted by members of our group evaluated the influence of maternal nutritional status, including folate, on the association between early-life As exposure and risk of diabetes and diabetes-related outcomes in experimental and epidemiological studies. The review found that in four out of five experimental studies, rodent offspring exposed to As in utero had elevated insulin resistance, with evidence suggesting maternal supplementation with folate + B₁₂ was protective against As-induced hyperglycemia in the offspring.¹⁸⁵ Additionally, increased in utero As exposure was related to changes in cord blood gene expression, microRNA and DNA methylation profiles involved in diabetes-related pathways among birth cohorts of As-exposed mothers residing in New Hampshire, Mexico, and Taiwan. The review highlighted a gap in the research regarding the evaluation of prevention strategies such as nutritional interventions that can modify As-related disease risk.²⁰⁷

2.5. Mathematical Models of One-Carbon Metabolism and Arsenic Methylation

To aid in interpretation of the data from our clinical trials in Bangladesh on the effects of FA and/or creatine supplementation on As methylation and excretion, our collaborators, Michael C. Reed and H. Frederik Nijhout at Duke University, developed a whole-body mathematical model of As methylation that includes absorption, storage, methylation, and excretion. Parameters for hepatic As methylation were based on biochemical literature, and compartments for the binding of arsenicals to intracellular proteins were included for model fit. Parameters for transport of As between tissues were adjusted to accurately predict the As urine excretion rates in single-dose experiments of human subjects. The model was able to reproduce decreases in blood As observed in our first clinical trial of folate-deficient individuals, and predicted that As concentrations would decrease by 19% in the liver and 26% other body stores.²⁰⁸

These mathematical models were further used to better understand the modest changes in As methylation observed with creatine supplementation in FACT.²⁰⁹ Our research question was to determine if synthesis of creatine by guanidinoacetate-N-methyltransferase decreased, how much would hepatic SAM increase, and consequently, how much would that increase the activity of AS3MT. When one methyltransferase enzyme is downregulated, there are several long-range allosteric interactions that buffer the flux through other methyltransferases. Therefore, in order to increase As methylation with creatine supplementation, at least one or more of these regulatory mechanisms would need to be disrupted. Because glycine-N-methyltransferase (GNMT) regulates hepatic SAM concentrations and is inhibited by binding to 5-methyl-THF, increasing hepatic 5-methyl-THF concentrations should allow creatine supplementation to increase SAM by inhibiting GNMT. Even in folate sufficient individuals, FA supplementation substantially increases plasma and RBC folate concentrations, but it is unknown to what extent it increases liver folate and liver SAM in replete individuals. Thus, the mathematical models of OCM have helped us to better understand why the effects of creatine supplementation on As methylation, while statistically significant, were modest.¹⁶³ Furthermore, these models have the potential to inform future nutritional interventions.

3. Conclusions and Future Directions

Chronic exposure to As is a global public health concern associated with many adverse health outcomes, including skin lesions, CVD, diabetes and diabetes-related outcomes, and various cancers. Urinary As excretion is facilitated by As methylation in a process that is dependent on OCM, which involves many micronutrients including folate, vitamin B₁₂, choline, betaine, vitamin B₆, riboflavin, niacin and creatine. Particularly among populations living in regions without FA-fortification programs, epidemiological studies and RCTs have provided strong evidence that folate status and FA supplementation increase As methylation capacity, thereby facilitating As elimination and lowering blood As concentrations. However, there is a lack of information regarding the role of other OCM-related micronutrients in As methylation, elimination, and toxicity. A major gap in knowledge is whether the association between OCM, As methylation, and metabolic-related health outcomes such as diabetes are a result of confounding, mediation, or reverse causality. Additionally, there is a lack of definitive evidence that nutritional interventions such as FA supplementation can reduce the risk of As-related health outcomes. There is some evidence from nested case-control studies that better folate status is associated with reduced risk for As-related skin lesions^130,196 and bladder cancer.¹⁹⁸ However, the protective effects of FA for other As-related adverse health outcomes have not been established and will require further research. Innovative study designs may be needed because, while RCTs may be considered the gold standard for providing definitive answers to these critical remaining research questions, new proposals to conduct large scale, long term FA RCTs may raise ethical concerns, as there are clear benefits of FA supplementation on As methylation efficiency.

Mandatory food FA-fortification programs currently exist in more than 80 countries worldwide. Folate fortification has been shown to nearly eradicate folate deficiency and the associated adverse health outcomes in Western countries such as the United States and Canada. However, mandatory FA-fortification programs do not exist in most countries having widespread exposure to As-contaminated drinking water. While the highest priority should be to provide access to safe drinking water, FA-fortification programs in As-endemic countries may help to eliminate folate deficiency with further benefits of aiding in a reduction of the public health consequences of As exposure.

Abbreviations

As

Arsenic

AS3MT

arsenic-3-methyltransferase

BHMT

betaine homocysteine methyltransferase

BHMT

betaine-homocysteine S-methyltransferase

CBS

cystathionine-β-synthase

CHDH

choline dehydrogenase

CVD

cardiovascular disease

DMAs

dimethyl arsenical species

DMAs^V

dimethylarsinic acid

DNMT3A

DNA methyltransferase 3 alpha

dTMP

2-deoxy-thymidine-5-monophosphate

dUMP

2-deoxy-uridine-5-monophosphate

EPA

Environmental Protection Agency

FA

folic acid

FACT

Folic Acid and Creatine Trial

FDA

Food and Drug Administration

FDR

false discovery rate

Folbp

folate-binding protein

FOLH1

folate hydrolase 1

FOLR1

folate receptor 1

FOLR2

folate receptor 2

FOX

Folate and Oxidative Stress

FTCD

formimidoyltransferase cyclodeaminase

GAA

guanidinoacetate

GAMT

guanidinoacetate methyltransferase

GDM

gestational diabetes mellitus

GENI

Gene Environment Nutrient Interactions

GM

gut microbiome

GNMT

glycine-N-methyltransferase

GSH

glutathione

Hcy

homocysteine

HEALS

Health Effects of Arsenic Longitudinal Study

HOMA2-IR

Homeostatic Model Assessment Index

InAs

inorganic arsenic

InAs^III

arsenite

InAs^V

arsenate

LOD

limit of detection

MMAs

Monomethyl arsenic species

MMAs^III

monomethylarsonous acid

MMAs^V

monomethylarsonic acid

MTHFR

methylene tetrahydrofolate reductase

MTR

methionine synthase

MTRR

methionine synthase reductase

NAD

nicotinamide adenine dinucleotide

NHANES

National Health and Nutrition Examination Survey

NTDs

neural tube defects

OCM

one-carbon metabolism

PBMC

peripheral blood mononuclear cell

PC

phosphatidyl choline

PEMT

phosphatidylethanolamine methyltransferase

PTHMs

posttranslational histone modifications

RBC folate

red blood cell folate

RCT

randomized control trial

ROS

reactive oxygen species

SAH

S-adenosylhomocysteine

SAM

S-adenosylmethionine

SHMT

serine hydroxymethyltransferase

SHMT1

serine hydroxymethyltransferase 1

SHS

Strong Heart Study

SHFS

Strong Heart Family Study

THF

tetrahydrofolate

TMAO

tri-methylamine oxide

TYMS

thymidylate synthetase

US

United States

UNICEF

United Nations Children’s Fund