Groundwater pollution by inorganic arsenic (IA) derived from natural origin is a serious issue of public health in the world, especially in developing countries where clean and safe surface water is not available in many places. In these areas, high arsenic concentrations exceeding the WHO guideline (10 μg/L) for drinking [1] have been reported in the groundwater [2–4]. It is well known that IA is one of the carcinogenic chemicals for the human. Chronic arsenic exposure causes skin pigmentation, hyperkeratosis, and cancers in the skin, lung, bladder, liver, and kidney resulting in high mortalities [1,5,6].

In general, humans can metabolize ingested IA to monomethylated arsenic (MMA) and then dimethylated arsenic (DMA), through arsenic (+3 oxidation state) methyltransferase (AS3MT). The pathways are discussed in detail in a later section of this review. Urinary arsenical profile has been used to assess arsenic exposure and its metabolism in individuals. In general, concentration ratios of MMA/IA and DMA/MMA in the urine indicate first and second methylation capacity, respectively. The metabolic action of AS3MT is considered to have an association with accumulation profile of various arsenic compounds in the body and consequently with susceptibility to toxic effects of arsenic. Several previous epidemiological studies showed that people who had high percentage of DMA (%DMA) in the urine could excrete more arsenic from the body because of high arsenic methylation capacity [7,8]. In other epidemiological studies, high concentration or percentage of MMA in the urine of populations with arsenic-related diseases were reported when compared with that of healthy people [9–12].

Vahter [13] showed that relative distributions of IA, MMA, and DMA in the urine of various populations are generally 10–30%, 10–20%, and 60–70%, respectively. On the other hand, there are large variations in the arsenic metabolism at individual and population levels [7]. It is known that biological and environmental factors including age, sex, pregnancy, arsenic exposure level, smoking habits, nutritional status and diet, potentially relate to the inter-individual variation (reviewed by Tseng [14]). With regard to ethnic related variation, it was shown that Andes women in arsenic contaminated area had only a low percentage of MMA[V] (0–11%) in the urine [15], whereas residents ingesting arsenic through drinking water in the northeast Taiwan had an extraordinarily high %MMA[V] (26.9%) in the urine [16]. These results indicate that genetic polymorphisms of certain enzymes that can metabolize arsenic through methylation pathways may be one of factors associated with the variation of arsenic profiles.

Genetic analyses of SNPs in the enzymes that may be involved in arsenic metabolism have been recently developed to evaluate the variation in the metabolism or toxic impacts of arsenic at individual or population level. We have also reported the association of arsenic metabolism-SNPs in AS3MT in Vietnamese [17–21]. However, a limited number of publications regarding the association of SNPs in AS3MT with arsenic methlyation are available. Moreover, the published results are based on fragmentary investigations and thus consistent conclusions have not yet been drawn. Hernandez and Marcos [22] have recently reviewed the role of genetic polymorphisms in enzymes related to arsenic metabolism including AS3MT. Thereafter, the number of publications regarding genetic polymorphisms dependent metabolism and toxicity of arsenic is increasing.

In this short-review, we address the following three topics: (1) enzymatic processes of arsenic methylation by AS3MT; (2) associations of genetic polymorphisms in AS3MT with variation in arsenic methylation; (3) frequency distributions of genetic polymorphisms in AS3MT at population levels. Based on the summarized data, we suggest which SNPs in AS3MT gene play an important role in arsenic methylation. In addition, we discuss future issues to be solved.

Two distinct pathways of IA methylation (oxidative and reductive) have been proposed (Figure 1). A model of oxidative methylation of IA was proposed by Challenger [23] and Cullen and Reimer [24]. By the combination of oxidative methylation and other pathways, arsenate (As[V]) is transformed to dimethylarsinous acid (DMA[III]); arsenate (As[V]) → arsenite (As[III]) → monomethylarsonic acid (MMA[V]) → monomethylarsonous acid (MMA[III]) → dimethylarsinic acid (DMA[V]) → dimethylarsinous acid (DMA[III]) (Figure 1). However, this sequential pathway does not fully explain the reason why the intermediate species, DMA[V] is detected as a major arsenical in the urine of humans. Methylation has been considered as a detoxification process of IA, because toxicities of MMA[V] and DMA[V] are much lower than that of IA. On the other hand, later studies revealed that MMA[III] or DMA[III] are more cytotoxic and genotoxic than IA [25,26], suggesting that oxidative methylation of IA is a bioactivation process.

On the contrary, Hayakawa et al. [27] reported a reductive methylation model for the metabolism of arsenic compounds. In this model, trivalent arsenicals are conjugated with glutathione (GSH) and then are methylated; As[III] → arsenotriglutathione (As[III](GS)₃) → monomethylarsenodiglutathione (MMA[III](GS)₂) → dimethylarsenoglutathione (DMA[III](GS)) (Figure 1). MMA[III](GS)₂ and DMA[III](GS) are then oxidized to MMA[V] and DMA[V], respectively. Naranmadura et al. [28] investigated hepatic and renal metabolites of arsenic after an intravenous injection of As[III] (0.5 mg As/kg body weight) in rats. The authors proposed that trivalent arsenic species can bind to thiol groups in proteins. The protein-bound arsenicals are released from the proteins by conjugation with GSH to form As[III](GS)₃ or MMA[III](GS)2 or DMA[III](GS) [28]. Hence, in the reductive methylation, MMA[V] and DMA[V] could be the end products. This hypothesis is consistent with the results that DMA[V] is found as a major arsenical in the urine. This sequential reaction is considered to be a detoxification process of IA, because of the low toxicities of MMA[V] and DMA[V].

Through both oxidative and reductive metabolic pathways, AS3MT plays an important role on arsenic methylation (Figure 1). Lin et al. [29] succeeded in purifying arsenic (III) methyltransferase (previously called as Cyt19) from the liver cytosol of adult male Fischer 344 rats. AS3MT is an S-adenosyl-L-methionine dependent enzyme and could methylate trivalent arsenicals [29,30]. Human AS3MT gene is composed of 32-kb with 11 exons [30]. Variety of genetic polymorphisms including SNPs and variable number of tandem repeats (VNTRs) have been identified in this gene [30,31]. This indicates that these genotypes may influence the methylation capacity of arsenic compounds and subsequently the arsenic toxicity. Recently, Kojima et al. [32] reported that methylation of IA by AS3MT induces oxidative DNA damage and increase their carcinogenicity in vitro. In the following section, we summarize in vitro and epidemiological studies on the AS3MT genetic polymorphism-dependent metabolism of arsenic compounds.

Notation of SNPs in human AS3MT is individually defined among studies. In this review, we use the notation by SNP ID which indicates the polymorphism identification number relative to the location in the consensus sequence (AY817668) with the first base of the consensus number 1 and dbSNP rs# cluster id which is available on Search for SNP on NCBI Reference Assembly [33]. Chromosome positions of genetic polymorphisms in AS3MT are indicated in Figure 2.

As we have mentioned above, several SNPs in AS3MT may influence arsenic metabolism. However, only limited information on frequencies of genetic polymorphisms in AS3MT was available for the number of the SNPs and the association with ethnicity. We have reported distribution of SNPs in AS3MT in some Asian [Japanese (JP), South Koreans (SKR), Chinese (CH), Mongolians (MN), Tibetans (TIB), Nepalese (NP), Vietnamese (VN), Sri Lanka-Tamils (SLT), Sri Lanka-Sinhalese (SLS)] and African [Ovambos (OVA) and Ghanaians (GH)] populations since 2007 [18,20,62–65].

The summarized results are shown in Table 1. The genotype frequencies followed the Hardy–Weinberg Principle except for 4602 (rs7085104) in MN, 6144 (rs17878846) in CH and TIB, 7395 (rs12767543) in SKR and VN, 9749 (rs17881367) in MN, 12390 (rs3740393) in JP, 12590 (rs3740392) in JP, 14215 (rs3740390) in JP, SKR, MN, and GH, 26790 (rs11191445) in JP, 35587 (rs11191453) in JP, SKR, MN, TIB, and VN, 37616 (rs4568943) in JP, CH, Mongolians, and OVA, and 37950 (rs17879819) in OVA. Because no genotyping error was observed in duplicate analysis, the reason that some genotype frequencies did not follow the Hardy–Weinberg Principle may probably be due to the small sample size. The allele frequencies of SNPs of Japanese in Tokyo, Japan (JPT (J)), Han Chinese in Beijing, China (CHB (H)), Chinese in Metropolitan Denver, Colorado (CHD (D)), Luhya in Webuye, Kenya (LWK (L)), Maasai in Kinyawa, Kenya (MKK (K)), Yoruban in Ibadan, Nigeria (YRI (Y)), African ancestry in Southwest USA (ASW (A)), Tuscan in Italy (TSI (T)), Utah residents with Northern and Western European ancestries from the CEPH collection (CEU (C)), Gujarati Indians in Houston, Texas (GIH (G)), and Mexican ancestry in Los Angeles, California (MEX (M)) that are available in the International HapMap Project [31], were compared with our studies (Figure 3). While the distribution patterns of AS3MT 4602 (rs7085104), 9749 (rs17881367), 12390 (rs3740393), 12590 (rs3740392), and 27215 (rs11191446) polymorphisms were not drastically different among the populations, notable variations were found for other SNPs. For AS3MT 3963 (rs7098825), C alleles were low in SLT, SLS, OVA, GH, YRI (Y) and CEU (C), particularly in MN. NP has no C allele of AS3MT 4740 (rs12416687). Relatively high frequencies of AS3MT 7395 (rs12767543) A allele and 8979 (rs72920657) A allele were found in CH and VN, respectively. AS3MT 6144 (rs17878846) A allele, AS3MT 37616 (rs4568943) A allele, and AS3MT 37950 (rs17879819) C allele frequencies in OVA and GHA populations were higher than Asians, but comparable with those in MN. Frequencies of AS3MT 10209 (rs3740394) A allele, AS3MT 26790 (rs11191445) C allele, and AS3MT 35991 (rs10748835) A allele in Asians (except for AS3MT 35991 (rs10748835) A allele in OVA) were higher than those in other ethnic groups, while the opposite trend was observed for AS3MT (rs11191439) T allele. For AS3MT 14215 (rs3740390), the allele distribution was different even within Chinese (CH vs CHB(H) and CHD(D)) and African groups (OVA and GH vs LWK(L), MKK(K), YRI(Y), and ASW(A)). TIB also showed a different pattern of AS3MT 14215 (rs3740390) allele from other Asian populations. Relatively lower C allele frequency of AS3MT 25986 (rs7085854) was detected in JP, JPT, SKR, and MN. TIB showed different allele distribution of AS3MT 35587 (rs11191446). To understand these variations, further research is necessary.

In this review, we summarized recent results on association of AS3MT genetic polymorphism with arsenic methylation. Although there are some inconsistencies between the genotypes and metabolism in human case studies, we found consistent results on two SNPs, AS3MT 12390 (rs3740393) in intron and 14458 (rs11191439, Met287Thr) in exon, in all the nations (Figure 4 and Table 2). This indicates that these SNPs may be ethnically independent polymorphisms, but may affect arsenic methylation.

AS3MT 12390 (rs3740393) is an intronic polymorphism, but the genotype with C allele in this SNP may have a higher second methylation capacity as predicted from the consistent results of high DMA[V]/MMA[V] in Mexicans, Argentines, Taiwanese, and Vietnamese (Table 2). Although AS3MT 12390 (rs3740393) belongs to a LD cluster in these populations, other SNPs within the cluster are different among populations (Table 2). Therefore, AS3MT 12390 (rs3740393) may be involved in an independent reaction from other SNPs in the same LD on a worldwide population basis. The frequency of AS3MT 12390 (rs3740393) CC or CG types was much higher than the GG type in Argentina’s population; the allele frequencies of C and G were 72% and 28%, respectively [40]. This allele frequency distribution was drastically different from other populations (Table 1 and Figure 3). This unique distribution may contribute to the results indicating that Argentina’s population has a higher %DMA and a lower %MMA in the urine, when compared with yet another study [15]. It is noteworthy to study whether this specific genotype distribution of AS3MT 12390 (rs3740393) is observed only in this group (Argentinean Andes) and how this unique SNP selection took place.

AS3MT 14458 (rs11191439), locates in the exon region, is synonymous substitution (Met to Thr substitution) at the amino acid base 287, and its functional difference has been investigated in some studies. By reviewing the in vitro [30,34] and human case studies [17–21,35–40,43–49], we found that C allele containing genotype in this SNP consistently showed a higher first methylation capacity than the T allele carrier. Interestingly, the effect of this SNP has been observed even in subjects exposed to low levels of arsenic [46]. Hence, it seems that the association of AS3MT 14458 polymorphism with arsenic methylation may occur irrespective of arsenic exposure status. This allele frequency was relatively low in Asian populations (Table 1 and Figure 3), indicating that among different global populations, Asians may be genetically less sensitive to arsenic methylation.

Increased %MMA in the urine linked to cancer risk has been suggested in previous epidemiological studies [9–12]. Thus, AS3MT 14458 (rs11191439) C allele carriers, who have a higher methylation capacity from IA to MMA, may be at higher risk of arsenic related cancers. The association of this SNP with human health effect was investigated in Mexicans [37] and Indians [48]. According to the study by De Chaudhuri et al. [48], there was no significant interaction between this SNP and skin lesions. On the other hand, Valenzuela et al. [37] found higher frequency of C allele type of AS3MT 14458 (rs11191439) TC + CC in people with cancers, although the level of significance was not high (p = 0.055). Hence, at present, no rigid evidence has been provided to link the AS3MT 14458 (rs11191439) dependent variation in arsenic methylation to arsenic-induced carcinogenesis. Apart from an evidence of cancer, Sampayo-Reyes et al. [39] found that AS3MT 14458 (rs11191439) is linked with DNA damages among children living in arsenic contaminated sites in Mexico (p = 0.034). Although these authors did not analyze arsenic compounds in urine and thus did not take into consideration the methylation capacity, they suggested that increased production of MMA[III] during methylation processes in AS3MT 14458 (rs11191439, Met287Thr) C allele carrier might induce genotoxicity.

Up to now, it still remains unknown how AS3MT 12390 (rs3740393) and 14458 (rs11191439) can influence the structure and function of this enzyme. Also, several results of AS3MT genotype-dependent arsenic methylation are interesting, because some non-exonic (intron, UTR, or 5′ and 3′ gene) SNPs may play important roles during methylation. The study on splicing variants of AS3MT gene and their function is one such issue to be pursued.

There are several tasks for future study regarding the role of AS3MT polymorphisms on arsenic methylation. As shown in this review, most studies encompassed only a small number of subjects and SNPs in AS3MT and thus, they merely provide fragmental and weak significant results at particular level(s). As shown in Table 2, results on the genotype association of AS3MT 4602 (rs7085104), 5194 (rs3740400), 7395 (rs12767543), 12590 (rs3740393), 14215 (rs3740390), 35587 (rs11191453), and 35991 (rs10748835) with arsenic metabolism were not consistent among different country groups. Larger scale studies would be required to gain enough statistical power and strengthen currently proposed assumptions on the linkages of AS3MT SNPs with arsenic methylation and cancer risk.

SNPs in AS3MT can provide useful information on the medical treatment of some diseases or risk assessment. For example, arsenic trioxide has been used for the treatment of acute promyelocytic leukemia, although the systematic mechanism is not fully understood [66]. The information on gene-arsenic interactions presented here can be applied to a personalized medicine for leukemia or other cancers. To better understand the variations in arsenic metabolism and to assess the potential susceptibility to its toxic and carcinogenic effects at individual and population levels, more comprehensive studies would be necessary.