Involvement of N-6 Adenine-Specific DNA Methyltransferase 1 (N6AMT1) in Arsenic Biomethylation and Its Role in Arsenic-Induced Toxicity

Background In humans, inorganic arsenic (iAs) is metabolized to methylated arsenical species in a multistep process mainly mediated by arsenic (+3 oxidation state) methyltransferase (AS3MT). Among these metabolites is monomethylarsonous acid (MMAIII), the most toxic arsenic species. A recent study in As3mt-knockout mice suggests that unidentified methyltransferases could be involved in alternative iAs methylation pathways. We found that yeast deletion mutants lacking MTQ2 were highly resistant to iAs exposure. The human ortholog of the yeast MTQ2 is N-6 adenine-specific DNA methyltransferase 1 (N6AMT1), encoding a putative methyltransferase. Objective We investigated the potential role of N6AMT1 in arsenic-induced toxicity. Methods We measured and compared the cytotoxicity induced by arsenicals and their metabolic profiles using inductively coupled plasma–mass spectrometry in UROtsa human urothelial cells with enhanced N6AMT1 expression and UROtsa vector control cells treated with different concentrations of either iAsIII or MMAIII. Results N6AMT1 was able to convert MMAIII to the less toxic dimethylarsonic acid (DMA) when overexpressed in UROtsa cells. The enhanced expression of N6AMT1 in UROtsa cells decreased cytotoxicity of both iAsIII and MMAIII. Moreover, N6AMT1 is expressed in many human tissues at variable levels, although at levels lower than those of AS3MT, supporting a potential participation in arsenic metabolism in vivo. Conclusions Considering that MMAIII is the most toxic arsenical, our data suggest that N6AMT1 has a significant role in determining susceptibility to arsenic toxicity and carcinogenicity because of its specific activity in methylating MMAIII to DMA and other unknown mechanisms.


Research
Inorganic arsenic (iAs) compounds are con sidered known human carcinogens that tar get multiple sites, including the lung, skin, and urinary bladder [International Agency for Research on Cancer (IARC) 2004;Pershagen 1981;Smith et al. 1992;Smith and Steinmaus 2009;Straif et al. 2009]. In addition, chronic exposure to high levels of iAs has been associ ated with the development of multiple dis eases and deleterious health effects in humans (Abernathy et al. 1999;Kapaj et al. 2006).
In humans, as in many animal species, iAs is metabolized to mono methyl arsonous acid (MMA) and dimethyl arsonic acid (DMA). The most cited conceptual model of arsenic methylation involves the reduction of penta valent iAs (iAs V ) to trivalent iAs (iAs III ), with subsequent methylation (Drobna et al. 2009 Among these metabolites, MMA III is the most toxic arsenic species (Drobná et al. 2005;Ferrario et al. 2008;Kligerman et al. 2003;Petrick et al. 2001). It is generally accepted that arsenic (+3 oxidation state) methyltrans ferase (AS3MT) is responsible for catalyz ing methyl group transfer from Sadenosyl methio nine (SAM) to iAs (Thomas et al. 2007). However, a recent study by Drobna et al. (2009) showed that knockout of As3mt in the mouse does not completely abolish the methylation of iAs, suggesting that there are alternative pathways for arsenic methylation in these animals. Although Bentley and Chasteen (2002) and Hall et al. (1997) suggested that arsenic methylation could be due to gastro intestinal tract microbiota, they also specu lated that unidentified methyl transferses may be responsible for the methylated arsenicals found in As3mtknockout mice.
We conducted a genome wide, parallel phenotypic screen of yeast deletion mutants to identify the genes required for the growth of yeast in the presence of MMA III and iAs III (Jo et al. 2009). We found a yeast strain with deletion of MTQ2, which was highly resistant to iAs III . MTQ2 encodes a SAM dependent methyltransferase and has been shown to be involved in the methylation of release factor eRF1 in yeast (Polevoda et al. 2006). The human ortholog of the yeast MTQ2 is N6 adeninespecific DNA methyl transferase 1 (N6AMT1), a putative methyl transferase. Although the bacterial homologs of N6AMT1 have been shown to methylate DNA N6adenine (Stephens et al. 1996), the current data do not indicate its function in the methylation of adenine in the DNA of mammalian cells (Ratel et al. 2006).
Our goal in the present study was to explore the mechanism by which N6AMT1 confers resistance to arsenic toxicity. We enhanced N6AMT1 gene expression in UROtsa cells, given its relatively low expres sion in these cells. The UROtsa cell line, originally isolated from a primary culture of normal human uro epithelium, does not methylate arsenic because of the absence of AS3MT expression (Drobná et al. 2005;Styblo et al. 2000) and has been used as a model for bladder epithelium and arsenic induced bladder cancer (Bredfeldt et al. 2006;Eblin et al. 2008;Sens et al. 2004). Here, we show that N6AMT1 is a human methyl transferase specifically involved in the bio methylation of MMA III to DMA. Given that MMA III is the most toxic arsenical and its implication in arsenic toxicity and carcino genicity, N6AMT1 may have a significant role in modulating arsenicinduced toxicity and carcinogenicity.

Cultures of yeast strains and human UROtsa cells. The wildtype BY4743 yeast strain was
Background: In humans, inorganic arsenic (iAs) is metabolized to methylated arsenical species in a multi step process mainly mediated by arsenic (+3 oxidation state) methyltransferase (AS3MT). Among these metabolites is monomethylarsonous acid (MMA III ), the most toxic arsenic species. A recent study in As3mt-knockout mice suggests that unidentified methyltransferases could be involved in alternative iAs methylation pathways. We found that yeast deletion mutants lacking MTQ2 were highly resistant to iAs exposure. The human ortholog of the yeast MTQ2 is N-6 adenine-specific DNA methyltransferase 1 (N6AMT1), encoding a putative methyltransferase. oBjective: We investigated the potential role of N6AMT1 in arsenic-induced toxicity. Methods: We measured and compared the cytotoxicity induced by arsenicals and their metabolic profiles using inductively coupled plasma-mass spectrometry in UROtsa human urothelial cells with enhanced N6AMT1 expression and UROtsa vector control cells treated with different concentrations of either iAs III or MMA III . results: N6AMT1 was able to convert MMA III to the less toxic dimethyl arsonic acid (DMA) when over expressed in UROtsa cells. The enhanced expression of N6AMT1 in UROtsa cells decreased cyto toxicity of both iAs III and MMA III . Moreover, N6AMT1 is expressed in many human tissues at variable levels, although at levels lower than those of AS3MT, supporting a potential participation in arsenic metabolism in vivo. conclusions: Considering that MMA III is the most toxic arsenical, our data suggest that N6AMT1 has a significant role in determining susceptibility to arsenic toxicity and carcinogenicity because of its specific activity in methylating MMA III to DMA and other unknown mechanisms. and protected from light before use. Yeast cells were treated with either iAs III or MMA III at concentrations ranging from 0 to 300 μM. Once UROtsa cells reached 70-80% conflu ence in culture, they were treated with iAs III at concentrations from 0 to 100 μM or MMA III at 0-5 μM.
Yeast growth assay. Yeast strains were pre grown in YPD media to midlog phase, diluted in fresh media to an optical density at 595 nm (OD 595 ) of 0.0165, and inoculated into a 48well microplate. Stock solutions of arsenicals were added to each culture with at least three replicate wells per dose.
Plates were incubated in a Tecan GENios spectrophotometer (Tecan Systems Inc., San Jose, CA) set to 30°C with intermittent shak ing, and OD 595 measurements were taken at 15min intervals for 24 hr. Raw absorbance data were averaged for all replicates, corrected for background, and plotted as a function of time. The area under the curve (AUC) was cal culated for the cultures in each well using Prism software (version 5.01; GraphPad Software, Inc., La Jolla, CA), and the treatments were averaged and expressed as a percentage of the control.

Human tissue array and real-time quantitative polymerase chain reaction (PCR) assay.
We used TaqManbased realtime quanti tative polymerase chain reaction (rtqPCR) to quantify N6AMT1 and AS3MT expres sion on a panel of 48 normal human tissues using the Human RapidScan Plate (OriGene Technologies, Inc., Rockville, MD). The human tissues were selected from multiple individuals of different ethnicity and pooled together. We obtained the primers and probes used for amplification of N6AMT1, AS3MT, and ACTB (βactin; control) from Applied Biosystems (Foster City, CA). Gene expression of N6AMT1 and AS3MT was calculated rela tive to ACTB using the ΔΔC T method.
N6AMT1 gene expression vector constructs and stable cell lines. Human N6AMT1 cDNA (GenBank accession no. NM_013240; National Center for Biotechnology Information 2011) was PCR amplified with primers 5´AACGCAGCGAAGGACTAT3´ and 5´CAGTAGTTCTGGGCACAC3´. The PCR product was gel purified (Qiagen, Valencia, CA) and cloned into pcDNA 2.1 vector (Invitrogen) according to the manu facturer's instructions, and the sequence was confirmed. The pcDNA 2.1 vector containing the N6AMT1 gene was excised using NotI/ BamHI restriction enzymes (New England Biolabs, Ipswich, MA) and subjected to gel purification. The nucleo tides of the N6AMT1 gene containing BamHI and NotI overhangs were annealed and ligated to a linearized pRetro XIRESZsGreen vector (Clonetech, Mountain View, CA) digested with BamHI and NotI (New England Biolabs). The pRetro XIRESZsGreen vector is a fluorescent retro viral expression vector that allows both a gene of interest and the ZsGreen gene to be expressed. The resultant constructs were ampli fied, purified, and sequenced. UROtsa cells were transfected with this constructed vector or a control vector using Lipofectamine 2000 reagent (Invitrogen) according to the manufac turer's instructions. After incubation at 37°C for 8 hr, the super natant fraction containing the retro viral vector was removed and replaced with normal growth medium. Cells grown for 48-72 hr were assessed by fluorescence microscopy. The ZsGreen fluorescent marker yields a bright green fluorescence, permitting direct monitoring of the delivery efficiency. Finally, the cell populations were sorted by the DAKOCytomation MoFlo High Speed Sorter (Dako North America, Carpinteria, CA), and the green fluorescent cells were puri fied and collected for continuing culture. The green fluorescent cells were used for additional experimentation.
Semiquantitative reverse-transcription (RT)-PCR. UROtsa cells with either N6AMT1 or plasmid vectors were col lected, and total RNA was isolated from these cells using the Qiagen RNAEasy Mini kit. We performed a reverse transcrip tion reaction using SuperScript II Reverse Transcriptase (Invitrogen) according to the manufacturer's instructions. The PCR con ditions for DNA amplification in the linear range were established on the GeneAmp PCR System 7600 (PerkinElmer, Inc., Wellesley, MA). The primers for DNA amplification were as follows: N6AMT1, 5´AACGCAGCGAAGGACTAT3´ and 5´CAGTAGTTCTGGGCACAC3´; AS3MT, 5´GTGTCTGGGTG GTGCTT TATA CTG3´ and 5´TGGAGGGCAGA ACCCAATT3´; and the housekeeping gene ACTB, 5TCACC CACACTGTGC CCATCTACGA3 and 5CAGCGGAAC CGCTCAT TGCCAATGG3. RTPCR products were analyzed on 1% agarose gels.
Cytotoxicity assay. We performed the 3(4,5dimethyl thiazol2yl)2,5diphenyl2H tetrazolium bromide (MTT) assay to assess cell viability after arsenic treatment. Cells were cul tured in 96well plates in a volume of 100 μL medium/well at a density of 5 × 10 4 cells/mL. Twentyfour hours after incubation with iAs III or MMA III (six replicates/arsenical concentra tion), 10 μL sterile MTT dye (SigmaAldrich; 5 mg/mL) was added to each well and plates were incubated at 37°C for 4 hr. The culture medium was then removed, and 200 μL dime thyl sulfoxide was added and thoroughly mixed for 10 min. Spectrophotometric absorbance at 570 nm was measured in a microplate reader.

Arsenic species profile analysis by highperformance liquid chromatography/inductively coupled plasma-mass spectrometry (HPLC-ICP-MS) methods. UROtsa cells with
N6AMT1 and UROtsa cells with vector were grown in RPMI 1640 medium supplemented with 10% fetal bovine serum and anti biotics. Culture medium was collected after exposure to iAs III or MMA III for 24 hr or 3 days and stored at -80°C until analysis. Cells treated with iAs III or MMA III for 3 days were collected, lysed in RIPA buffer, and extracted in methanol by incubating overnight at 4°C in a rotational shaker. After centrifugation at maximum speed for 5 min at 4°C, supernatants were transferred to micro centrifuge tubes and stored at -80°C until analysis. Before analysis, samples were diluted 1:5 with water-methanol to bring the methanol concentration to 2.5%, incubated at 5°C to precipitate poorly soluble material, and filtered (0.45 μm). Analysis was performed by HPLCICPMS (Agilent 1090 HPLC and Agilent 7500CE ICPMS run in normal mode; both from Agilent Technologies, Santa Clara, CA) under conditions that resolved neutral, tri valent, and penta valent iAs species. The ion pairing method (Le et al. 2000) was used with major modifications to improve the resolution of the species. Briefly, calibrants were prepared from neat materials [As 2 O 3 (Aldrich) and As 2 O 5 (Acros), SigmaAldrich; DMA V and MMA V · 6H 2 O, Chem Service, West Chester, PA] in deionized water (≥ 18 MΩ). For spe ciation, we used a Phenomenex GeminiNX column (3 μm, C18, 110Å, 150 × 4.6 mm; Torrance, CA) with a corresponding guard column at 40°C. Concentrations of arsenic species in stock solutions were standardized against NIST traceable commercial ICPMS standards (VWR BDH Aristar Plus; Ultra Scientific, Kingstown, RI). Serial dilutions were made into deionized water. iAs III and iAs V species were quantified by separate cali brant series, and iAs III concentration in the calibrants was corrected for any conversion to iAs V . The HPLC conditions were isocratic (5 mM tetrabutyl ammonium hydroxide, 10 mM ammonium carbonate, 2.5% metha nol, pH 9.2, 1 mL/min) for 5 min; then a step gradient (5 mM tetra butyl ammonium hydroxide, 30 mM ammonium carbonate, 2.5% methanol, pH 8.75, 1.2 mL/min) for 5 min to elute iAs V was followed by step gradi ents (5 mM tetra butyl ammonium hydroxide, 30 mM ammonium carbonate, 2.5% metha nol, pH 9.2, 1.2 mL/min) for a 5min equil ibration to the initial pH and finally to the initial mobile phase for 5 min (1.2 mL/min). The data were analyzed by LC ChemStation A.09.03 and ICPMS ChemStation B.03.03 software (Agilent Technologies).
Data analysis. Statistical analyses were performed using oneway analysis of variance. Data represent mean ± SE of at least three independent experiments.

Deletion of yeast MTQ2 leads to increased resistance to arsenic treatment.
We evaluated the growth pheno type of the MTQ2deletion mutants in the presence of either iAs III or MMA III (Figure 1). Deletion strains and their iso genic wildtype counter part, BY4743, were treated with equi oxic doses equivalent to the concentrations that resulted in 20% growth inhibition (IC 20 ) and 2 × IC 20 , which were 300 and 600 μM for iAs III and 150 and 300 μM for MMA III , respectively. iAs III treat ments had no effect on the growth of MTQ2 deletion mutants but significantly decreased growth of the wildtype strain. In compari son, the growth of both MTQ2deletion mutants and wildtype yeast treated with MMA III decreased to the same degree despite the higher toxicity of MMA III .
Differential level of N6AMT1 mRNA expression in human tissues. A direct search for sequence homology and conserved func tional domains revealed that the human N6AMT1 gene is orthologous to the yeast MTQ2 gene. We used the webbased online tool Protein Function Prediction (PFP; Kihara Bioinformatics Laboratory 2010), to analyze and predict its potential func tions (Hawkins et al. 2009). The suggested molecular functions of N6AMT1 include protein hetero dimerization and methio nine Smethyltransferase activity, with an almost certain (100%) predicted prob ability of being a methyl transferase. Given N6AMT1's suggested function as a methyl transferase, we were interested in exploring its potential involve ment in arsenic biom ethylation. Considering that the primary methyl transferase responsible for arsenic metabo lism in human cells is AS3MT, we did pairwise alignment analysis of N6AMT1 and AS3MT using EMBOSS Pairwise Alignment Algorithms, an online tool (European Bioinformatics Institute 2010). The two proteins shared about 25% similarity. Of the three sequence motifs found in most AS3MT homologs, only motif ILDLGSGSG is highly conserved in N6AMT1 [LEVGSGSG; see Supplemental Material, Figure 1(doi:10.1289/ ehp.1002733)], whereas (D/N)PPY is pres ent in N6AMT1 but not in AS3MT. These differences suggest that the mechanism by which N6AMT1 methylates arsenic may dif fer from that of AS3MT, if N6AMT1 is, in fact, involved in the methylation of arsenicals.
A search of the Expressed Sequence Tags (EST) Database (National Center for Biotechnology Information 2010) revealed sequences matching the cDNA of N6AMT1 in many human tissues with varied expres sion levels. To experimentally measure the mRNA expression of N6AMT1 across tissues, we performed rtqPCR analysis using cDNA from a panel of 48 human tissues contained in a tissue array (Figure 2). We confirmed the amplification products to be N6AMT1 by DNA sequencing. Using the liver as a refer ence, the expression of N6AMT1 was normal ized to the expression level of ACTB [with cycle threshold (Ct) values ranging from 18 to 20] and found to be relatively highly expressed in tissues such as the parathyroid, pituitary, adrenal gland, and kidney, and weakly expressed in tissues such as the skin, lung, and mammary gland. We also meas ured AS3MT mRNA levels using the same tissue panel in order to compare N6AMT1 Figure 1. Deletion of the MTQ2 gene in yeast results in increased resistance to arsenite (iAs III ), shown as the growth phenotype of MTQ2 mutant yeast cells (B) and the wild-type BY4743 cells (A) treated with 300 or 600 μM iAs III or 150 or 300 μM MMA III . Growth curves show the OD 595 for each treatment as a function of time for 24 hr. Bars represent the mean ± SE AUC for three technical replicates. At the doses tested, iAs III treatment did not alter the growth pattern of MTQ2 mutants but led to a dose-dependent reduction in growth of the wild-type strain; the growth patterns of both yeast strains were similar after MMA III exposure. # p< 0.001, compared with control.

Overexpression of N6AMT1 in UROtsa cells increases resistance to arsenic treatment.
We also measured and compared the level of N6AMT1 mRNA in several cell lines, includ ing 293 (human embryonic kidney cells), HeLa, UROtsa, and HL60 (human pro myeloc ytic leukemia cells). N6AMT1 expres sion in UROtsa cells is relatively low, with a Ct value of about 33 (data not shown). This cell line also has almost no detectable level of AS3MT, making it an excellent model to study the role of N6AMT1 in arsenic toxicity and metabolism in mammals. We enhanced N6AMT1 expression in UROtsa cells using a retrovirusbased vector ( Figure 3A) and found the level of N6AMT1 mRNA in UROtsa cells to be significantly increased by approximately 5fold in clone 2, as measured by semiquan titative RTPCR ( Figure 3B). We also meas ured AS3MT gene expression in these two cell lines and found no detectable mRNA level in either cell line ( Figure 3B). We further con firmed these PCR results by realtime PCR analysis (data not shown). Unfortunately, we could not detect N6AMT1 protein levels in these cells using two commercially available anti bodies. Transfected UROtsa cells did not have an altered doubling time or morphology in culture. The UROtsa cells with N6AMT1 (N6AMT1enhanced cells) and the UROtsa cells with vector cells (vector control) were treated with either iAs III or MMA III at a series of concentrations for 24 hr. Arsenical treat ments induced a dosedependent decrease in viability of both cell lines. However, increased expression of N6AMT1 in UROtsa cells resulted in higher viability after iAs III and MMA III treatment at almost all concentra tions tested, compared with the UROtsa vec tor control cells ( Figure 3C,D). This increased arsenic resistance was more apparent in cul tures treated with MMA III than with iAs III , approximately 2 and 1.3fold, respectively.
Enhanced N6AMT1 in UROtsa cells methylates MMA III to DMA. We collected medium and cell extracts from cultures of UROtsa cells with N6AMT1 and UROtsa cells with vector treated with different concentra tions of either iAs III or MMA III for up to 3 days and then meas ured and analyzed arsenic meta bolic profiles using ICPMS [for representative chromatograms, see Supplemental Material, Figure 3 (doi:10.1289/ehp.1002733)]. Methylated metabolites were undetectable either in the media (Table 1) or in the cell extract ( Table 2) from cultures of UROtsa cells with N6AMT1 or of UROtsa cells with vector, after treatment of with iAs III , suggest ing that N6AMT1 does not methylate iAs III . When UROtsa cells with vector control were treated with MMA III , levels of MMA V but not dimethylarsinic acid (DMA V ) increased in the media (Table 3) and cells (Table 2). Treatment with MMA III of UROtsa cells with N6AMT1 resulted in similarly increased levels of MMA V andled to increased levels of DMA V in the media after 24 and 72 hr (Table 3) and, to a lesser degree, in the cell extracts after 72 hr (Table 2). At 24 hr, the level of DMA V was similar at each dose level, but after 3 days treatment levels in both media and cell extract increased in relation to the initial con centration of MMA III , in a dosedependent manner. Moreover, the amount of DMA V in the culture medium increased 5fold after 3 days of treatment (1 μM MMA III ) compared with 1 day of treatment.

MMA III is the most toxic arsenic metabolite in vivo and in vitro.
In humans, iAs is metabo lized to methylated arsenical species in a multi step process. Methylated arsenicals, especially MMA III , may be more toxic than iAs both in vivo and in vitro (Drobná et al. 2005;Ferrario et al. 2008;Kligerman et al. 2003;Petrick et al. 2001). In cultured human cells, MMA III is the most toxic arsenical (Drobná et al. 2005;Ferrario et al. 2008;Petrick et al. 2001) and inhibits several key cellular proteins, such as glutathione reductase ) and thio redoxin reductase (Lin et al. 1999(Lin et al. , 2001. Several studies have shown that MMA III is capable of inducing genetic damage and changes in signal trans duction by either direct or indirect mecha nisms (Ahmad et al. 2002;Kligerman et al. 2003;Nesnow et al. 2002). In addition, exposure to MMA III for 52 weeks induced malignant transformation of UROtsa cells (Bredfeldt et al. 2006). Epidemiological studies have suggested that individuals who  BLOQ, below the limit of quantitation (the m/z 75 signal at the retention time of the species is not distinguishable from the baseline signal). "Present" indicates that the peak was seen at MMA III retention time; no concentration is given because the samples were not analyzed by the MMA III assay.  (Steinmaus et al. 2006). MMA III has been proposed as the ultimate geno toxic form of arsenic (Kligerman et al. 2003), and the existing evidence indicates that biomethylation of iAs to MMA III is likely to alter the adverse effects of environmental arsenic exposure on human health. AS3MT is primarily responsible for methylating iAs to MMA III and DMA in humans. The arsenic methyltransferase AS3MT is recog nized as the primary enzyme responsible for conversion of iAs to its methy lated metabolites MMA III and DMA (Lin et al. 2002;Wood et al. 2006). Studies have shown that singlenucleotide poly morphisms in AS3MT lead to different urinary arseni cal profiles (Agusa et al. 2009;Schläwicke Engström et al. 2009;Wood et al. 2006), some of which are associated with increased risk of pre malignant skin lesions (Valenzuela et al. 2009). Although these data suggest that AS3MT plays a critical role in arsenic methyla tion and toxicity, a recent study showed that As3mtknockout mice retain some ability to methylate arsenicals, suggest ing the existence of other methyl transferases that could be involved in alternative arsenic metabolism pathways (Drobna et al. 2009).
N6AMT1 is capable of methylating MMA III to DMA. In this study, we found that N6AMT1 has the capacity to methylate MMA III to DMA. The expression of N6AMT1 is generally low compared with the expression level of AS3MT in most human tissues. In addition, the low sequence homology shared between these two proteins, about 25%, sup ports differences in substrate specificity and, possibly, in mechanisms of arsenic methyla tion. In contrast to AS3MT, which methy lates iAs to the more toxic MMA III , N6AMT1 methylates MMA III to the less toxic DMA. This is consistent with the increased resistance to MMA III of UROtsa cells over expressing N6AMT1 compared with vector control cells, but other mechanisms likely also con tribute to this increased resistance. Thus, our results suggest that N6AMT1 may play a role in modulating arsenicalinduced toxicity and that decreased N6AMT1 expression or activity could have a significant impact on arsenic induced toxicity and perhaps carcinogenicity under certain conditions or in certain tissues.

N6AMT1 in human cells responded to arsenicals differently from MTQ2 in yeast.
We noted differences in response to arseni cal treatments between yeast and human cells. Specifically, the MTQ2deletion yeast strain is resistant only to iAs III . We did not find evi dence of iAs III methylation in yeast wildtype; that is, DMA and MMA levels in cells and cul ture media were below the limit of quantitation (data not shown). Therefore, the function of MTQ2 in arsenic toxicity may not be related to iAs III methylation. In contrast, over expression of N6AMT1 in UROtsa cells leads to resis tance to both iAs III and MMA III , which pro vides evidence of its involvement in protection from these arsenicals. These results indicate that the orthologous genes-MTQ2 in yeast and N6AMT1 in humans-have different roles in the cellular response to arsenic toxicity.
Conversion of MMA III to DMA by N6AMT1 needs to be further confirmed biochemically. Our analyses showed that enhanced expression of N6AMT1 in UROtsa cells converts MMA to DMA. However, puri fied recombinant N6AMT1, in the presence of SAM and other cofactors, was unable to methylate iAs III or MMA III (data not shown). N6AMT1 dimerizes with tRNA methyl transferase 112 homolog (TRMT112), which appears to be necessary for proper N6AMT1 function (Figaro et al. 2008) and is consistent with hetero dimerization activity as predicted with PFP. Thus, the lack of N6AMT1 dependent MMA III methylation in this test tube experiment may be due to the absence of TRMT112 or another unknown protein. In addition, N6AMT1 over expression increased resistance to iAs III , although UROtsa cells were not able to methylate iAs III to MMA III or to other methylated species, suggesting that mechanisms other than participation in arsenic methylation might be involved. It is not clear whether inter action between N6AMT1 and TRMT112 has a role in arsenic toxicity, but it is certainly worthy of further investigation.

Conclusions
Our data suggest an important potential role of N6AMT1 in modulating arsenicalinduced toxicity by methylating MMA III to the less toxic DMA. However, further investigation is warranted to determine whether N6AMT1 can methylate MMA III in vivo and also to identify the genetic and environmental factors that can alter N6AMT1 expression and/or activity. Our ongoing experiments are focused on the biochemical characterization of N6AMT1, specifically its capacity to meth yl ate MMA III , as well as its ability to modulate arsenic toxicity and carcinogenicity.