Arsenic Metabolism by Human Gut Microbiota upon in Vitro Digestion of Contaminated Soils

Background Speciation analysis is essential when evaluating risks from arsenic (As) exposure. In an oral exposure scenario, the importance of presystemic metabolism by gut microorganisms has been evidenced with in vivo animal models and in vitro experiments with animal microbiota. However, it is unclear whether human microbiota display similar As metabolism, especially when present in a contaminated matrix. Objectives We evaluated the metabolic potency of in vitro cultured human colon microbiota toward inorganic As (iAs) and As-contaminated soils. Methods A colon microbial community was cultured in a dynamic model of the human gut. These colon microbiota were incubated with iAs and with As-contaminated urban soils. We determined As speciation analysis using high-performance liquid chromatography coupled with inductively coupled plasma mass spectrometry. Results We found a high degree of methylation for colon digests both of iAs (10 μg methylarsenical/g biomass/hr) and of As-contaminated soils (up to 28 μg/g biomass/hr). Besides the formation of monomethylarsonic acid (MMAV), we detected the highly toxic monomethylarsonous acid (MMAIII). Moreover, this is the first description of microbial thiolation leading to monomethylmonothioarsonic acid (MMMTAV). MMMTAV, the toxicokinetic properties of which are not well known, was in many cases a major metabolite. Conclusions Presystemic As metabolism is a significant process in the human body. Toxicokinetic studies aiming to completely elucidate the As metabolic pathway would therefore benefit from incorporating the metabolic potency of human gut microbiota. This will result in more accurate risk characterization associated with As exposures.

volume 118 | number 7 | July 2010 • Environmental Health Perspectives Research Arsenic (As), a ubiquitous environmental con taminant, presents significant human health risks: Chronic exposure is associated with the development of cancer in the bladder, liver, kidney, and lungs (Chen et al. 1992). Regions with a high geogenic As background show an increased risk for elevated exposure by con sumption of drinking water and diet. An addi tional exposure scenario in urban areas near smelting and mining activities is the ingestion of contaminated soil and dust by children, who display typical handtomouth behavior. Although inorganic As (iAs) may be the pre dominant form in contaminated soils, As spe ciation changes during gastro intestinal transit are not well charac terized. The gut represents a highly reducing environment and harbors a complex microbial community, which may contribute to the pre systemic biotransfor mation of ingested As (systemic metabolism being defined as all metabolic reactions carried out by human cells). Presystemic As specia tion analysis must therefore be considered an essential part of the risk evaluation process, especially with respect to toxicity, which is speciation dependent. In short, methylated trivalent species-monomethylarsonous acid (MMA III ), dimethylarsinous acid (DMA III ), and arsenous acid (iAs III )-are two orders of magnitude more cytotoxic than is As acid (iAs V ) (Naranmandura et al. 2007a). The methylated penta valent species-monomethyl arsonic acid (MMA V ) and dimethylarsinic acid (DMA V )-present a 10fold lower toxic ity than iAs V , whereas trimethyl arsine oxide (TMAO) is essentially non toxic (Hirano et al. 2004).
In the human body, iAs is sequentially methylated and predominately excreted as DMA V in urine. This methylation process was originally considered a detoxification process, but the formation of reactive intermediates (MMA III and DMA III ) has forced researchers to reconsider methylation as an activation pro cess (Styblo et al. 2002). In addition, a recent study on human urine analysis after iAs expo sure revealed new sulfurcontaining methylated As metabo lites, monomethyl mono thio arsonic acid (MMMTA V ) and dimethyl mono thio arsinic acid (DMMTA V ) (Naranmandura et al. 2007b;Raml et al. 2007), for which the mechanism of formation and toxicological profile are not yet fully charac terized. Given the toxicological importance of As speciation changes, it is clear that a complete risk charac terization after As exposure must include the possibility of presystemic metabolism by the microberich environment of the gastro intestinal tract.
The colon harbors a vast (10 14 bacterial cells) and incredibly diverse (> 1,000 species) microbial community, which has the abil ity to metabolize xeno biotics far more exten sively than any other part of the body (Sousa et al. 2008). Thus far, the presystemic bio t ransforma tion of As was primarily studied with gut micro biota from animal models. Rowland and Davies (1981) reported the reduction of iAs V to iAs III by rat cecal bac teria as well as limited formation of MMA and DMA. In another study with rats orally exposed to DMA V Chen et al. (1996) detected demethylated (iAs V , MMA V ) and methylated (TMAO) urinary metabolites. Finally, the thiolation of methylated As oxides (DMA V , TMAO) in the cecal contents of a mouse (Kubachka et al. 2009a(Kubachka et al. , 2009b and the observed thiolation through in vivo experi ments (Kuroda et al. 2004;Naranmandura et al. 2007b) have been reported.
Presystemic As metabolism in the human body has been less investigated (Hirner et al. 2004). Nevertheless, Michalke et al. (2008) reported that human gut microbes actively volatilize bismuth and other metal loids, including As, through methyla tion and Background: Speciation analysis is essential when evaluating risks from arsenic (As) exposure. In an oral exposure scenario, the importance of presystemic metabolism by gut microorganisms has been evidenced with in vivo animal models and in vitro experiments with animal microbiota. However, it is unclear whether human microbiota display similar As metabolism, especially when present in a contaminated matrix. oBjectives: We evaluated the metabolic potency of in vitro cultured human colon microbiota toward inorganic As (iAs) and Ascontaminated soils. Methods: A colon microbial community was cultured in a dynamic model of the human gut. These colon microbiota were incubated with iAs and with Ascontaminated urban soils. We determined As speciation analysis using highperformance liquid chromatography coupled with inductively coupled plasma mass spectrometry. results: We found a high degree of methylation for colon digests both of iAs (10 µg methylarsenical/g biomass/hr) and of Ascontaminated soils (up to 28 µg/g biomass/hr). Besides the formation of monomethylarsonic acid (MMA V ), we detected the highly toxic mono methyl arsonous acid (MMA III ). Moreover, this is the first description of microbial thiolation leading to monomethyl monothioarsonic acid (MMMTA V ). MMMTA V , the toxicokinetic properties of which are not well known, was in many cases a major metabolite. conclusions: Presystemic As metabolism is a significant process in the human body. Toxicokinetic studies aiming to completely elucidate the As metabolic pathway would therefore benefit from incorporating the metabolic potency of human gut microbiota. This will result in more accurate risk characterization associated with As exposures. hydrogenation. Moreover, Meyer et al. (2008) postulated that gut methanogens play a crucial role in metalloid volatilization, thereby exert ing toxic effects to the human body-not only by direct interaction with the host but also by disturbing the endogenous gut microbiota composition and metabolism. Finally, a thor ough in vitro exploration with the Simulator of the Human Intestinal Microbial Ecosystem (SHIME), a dynamic human gastrointestinal simulator, revealed a high microbial meta bolic potency toward metal(loid)s (DiazBone and Van de Wiele 2009). This was demon strated by the finding of significant volatiliza tion of As, selenium, bismuth, tellurium (Te), and antimony; the formation of highly toxic AsH 3 (arsine) and (CH 3 ) 2 Te (dimethyl tellu ride); and the discovery of two new As-sulfur metabolites.
These data indicate the need for more studies with human gut microorganisms, which can confirm the presystemic metabo lism as observed with animal gut micro biota. Therefore, in the present study we investigated the metabolic potency of human gut micro organisms toward iAs and As from contami nated urban soils, assessing the importance of presystemic As biotransformation upon an oral exposure scenario and the actual specia tion of As that enters the bloodstream upon gastro intestinal digestion.

Materials and Methods
Chemicals and media. We used degassed, ultrapure 18 mΩ water (DDI; Millipore, Bedford, MA, USA) to prepare the chromato graphic mobile phase and the standard stock solutions. American Chemical Society-grade ammonium nitrate and ammonium dihydro gen phosphate (Fisher Scientific, Pittsburgh, PA, USA) and technicalgrade EDTA, tet rasodium salt dehydrate (Fisher Scientific, Fair Lawn, NJ, USA) were used in the chro matographic mobile phase. We obtained stock solutions of iAs (As III and As V ) from Spex Industries (Metuchen, NJ, USA) and certi fied stock solutions of MMA V and DMA V from Chem Service (West Chester, PA). W.R. Cullen (Department of Chemistry, University of British Columbia, Vancouver, BC, Canada) provided tetramethylcyclotetraarsaoxane [cyclo(CH 3 AsO) 4 ] crystals that were syn thesized and characterized as described else where (Cullen et al. 1989); these crystals were stored at -21°C and were hydrolyzed by degassed, deionized water at the time of analy sis to obtain a stock solution of a MMA III and MMA V mixture (Cullen et al. 1989). We pur chased sodium arsenate (Na 2 HAsO 4 ·7H 2 O), methionine, methyl cobalamine, and gluta thione from SigmaAldrich (St. Louis, MO, USA). Arsenate stock solutions were prepared in deionized water at 4,500 mg As/L and 45 mg As/L. Soils. The U.S. Environmental Protection Agency kindly provided four Ascontaminated soils that originated from urban areas around former smelting sites. We sieved all soils at 250 µm before in vitro gastrointestinal incu bation; this sieving reflects the size of particles that most likely sticks to the hands of exposed humans (Kelly et al. 2002). Soil specifications are reported in Table 1.
Production and characterization of colon micro biota for SHIME. The in vitro colon microbial community used in this study was cultured and maintained in a modified SHIME, which consisted of four compart ments simulating the stomach, small intes tine, and both proximal and distal colon. A detailed description of the SHIME, the carbohydratebased medium, and the in vitro colon microbiota has been described previ ously (Van de Wiele et al. 2004). Briefly, fecal microbiota previously obtained from a 29yearold male volunteer (who had no his tory of anti biotic treatment in the 6 months before the study) were inoculated in the dif ferent colon compartments. The SHIME reactor was fed carbohydratebased medium three times per day to provide digested nutri tion for the colon microbiota. After 3 weeks of adaptation, a stable microbial community was obtained in the respective colon com partments. We found microbial fermentation activity of the distal colon (shortchain fatty acid production and ammonium production) and community composition to be consistent with that of previous SHIME runs and the in vivo situation (Molly et al. 1994 Noncontinuous incubation studies. Metabolic potency of fecal microbial inocu lum. The first experiment constituted a screening phase to test whether the fecal microbial community from human ori gin actively metabolized As. The microbial community was isolated from a fecal sample as previously described by Molly et al. (1994). Thirty milliliters of microbial fecal suspension was sampled, placed in 60mL serum bot tles, and incubated with NaH 2 AsO 4 ·7H 2 O (iAs V ; 90 mg iAs V /L), similar to the method of Herbel et al. (2002). Serum bottles were capped with butyl rubber stoppers that are impervious to O 2 and subsequently made anaerobic by flushing with N 2 gas for 30 min. Samples were then incubated at 37°C on a rotary shaker (150 rpm) for 48 hr. We compared the effect of specific methyl group donors toward microbial As methy lation by comparing methionineamended (5 mmol/L) and methyl cobalaminamended (5 mmol/L) samples with control samples (incubation of the sample in the presence of heatsterilized fecal microbiota). The effect of glutathione as a reducing agent was eval uated by comparing glutathioneamended samples (10 mmol/L) with control samples. Duplicate incubations were performed on two different days to evaluate the reproduc ibility. A scheme of the experimental setup is presented in Supplemental Material, Figure 1 (doi:10.1289/ehp.0901794).
Metabolic potency of colon microbiota toward As from contaminated soils. The objec tive of the second experiment was to screen for microbial speciation changes of iAs V at more relevant concentrations (i.e., 50-500 µg/L) by mimicking conditions of oral exposure to environmental samples. In addition, four Ascontaminated soil samples (one slag soil and three from urban sites) were subjected to a gastro intestinal digestion procedure. To better mimic in vivo conditions, all gastrointestinal stages-gastric, small intestine, and colonwere simulated. We combined the in vitro gastro intestinal method (IVG) from Ohio State University with the SHIME to subsequently simulate stomach and small intestine (IVG) and colon (SHIME) conditions, respectively. The IVG method was previously validated against in vivo data for As bioaccessibility (Rodriguez and Basta 1999), whereas the SHIME has been validated against in vivo data for microbial community composition and metabolic activity toward drugs and phyto estrogens (Molly et al. 1994;Possemiers et al. 2006). Soils were incubated in the gastric and intestine solution (30 mL) of the IVG pro tocol at a liquidtosoil (L/S) ratio of 150 (Rodriguez and Basta 1999). These intesti nal digests from the IVG protocol were sub jected to colon conditions by adding 30 mL of the colon suspension sampled from the dis tal colon compartment of the SHIME reac tor, resulting in an L/S ratio of 300 for the soil digests. The vessels containing the colon digests were capped with butyl rubber stop pers and subsequently flushed with N 2 for 30 min to obtain anaerobic conditions and incubated on a shaker at 150 rpm at 37°C for 18 hr. See Supplemental Material, Figure 1 (doi:10.1289/ehp.0901794) for a schematic of the experimental setup.
Sample treatment. To preserve the spe ciation of As in the colon digests, all samples were flash frozen with liquid nitrogen upon incubation and subsequently stored at -80°C. Before analysis with highperformance liquid chromatography (HPLC) coupled with induc tively coupled plasma (ICP) mass spectrom etry (MS), samples were thawed and diluted appropriately with 20 mmol/L (NH 4 ) 2 CO 3 at pH 9.0 to minimize sulfur-oxygen exchange while awaiting analysis (Conklin et al. 2008). Upon complete thawing, the sample was vor texed and centrifuged for 10 min at 10,400 rel ative centrifugal force with an Eppendorf 5810R centrifuge (Brinkman Instruments, Westburg, NY, USA) to separate soluble As species from soilbound As. The supernatant was filtered through a MillexLCR 0.45 µm fil ter (Millipore) with a LuerLok 10mL syringe (BD, Franklin Lakes, NJ, USA). Finally, filtrates were diluted with the mobile phase and injected into the HPLC. The sum of the As species in the filtrate observed chromato graphically was considered the bioaccessible fraction. We measured total As concentration in the digest filtrates using ICP optical emis sion spectroscopy (ICPOES). This allowed us to calculate chromatographic recovery, which quantifies the extent to which the sum of the chromatographic As species comprises the total amount of As in the digest filtrates.
As speciation analysis by HPLC/ICP-MS. Sample supernatants were analyzed with HPLC (Agilent 1100) and ICPMS (Agilent 7500ce; Agilent, Palo Alto, CA, USA) for As specific detection at m/z 75. Separation of As oxides was performed on a PRPX100 HPLC column (250 mm × 4.1 mm, 5 µm). The mobile phase was a solution of NH 4 NO 3 (10 mmol/L), NH 4 H 2 PO 4 (10 mmol/L), and EDTA (500 mg/L) at pH 4.57 in distilled water (sepa ration 1). The flow rate was 1.0 mL/min, and the sample injection volume was 100 µL. The retention times of the separated compounds were 3.6 min for As III , 4.2 min for DMA V , 5.5 min for MMA III , 7.1 min for MMA V , and 8.9 min for As V , similar to those previ ously reported by Yathavakilla et al. (2008). We used this separation for quantification of the As species of interest. Arsenic sulfides were identified by retentiontime matching between samples and fortified samples. Using chromato graphic separation 1, monothioarsonic acid eluted at 15 min, whereas MMMTA V eluted at 18.6 min. See the Supplemental Material (doi:10.1289/ehp.0901794) for details on syn thesis, chromatographic confirmation of these As sulfides, and sample analysis.
A second chromato graphy [separation 2; see Supplemental Material (doi:10.1289/ ehp.0901794)] was used for ICPMS and electrospray ionization (ESI)MS detection, because the mobile phase of separation 1 was not compatible with ESIMS detection.

Results
In the first experiment, we assessed the meta bolic potency of the human fecal microbial inoculum toward high levels of iAs V (90 mg/L). iAs V was efficiently (> 94%) reduced to iAs III after the 48hr incubation with both active and sterilized fecal microbiota (Table 2), prob ably because of the highly reducing conditions (redox potential was -180 mV). Incubation with sterilized fecal microbiota did not lead to thiolated or methylated arsenicals. In contrast, incubation of iAs V with active fecal microbiota resulted in the production of mono thio arsonic acid (mTA; mean ± SD) in non amended (2.2 ± 3.1 mg/L) and methionineamended (0.8 ± 1.1 mg/L) samples. Interestingly, we observed methylation only in the presence of methyl cobalamin. Addition of the methyl cobalamin displayed a significant methylation of iAs (18%), with MMA V (13.0 ± 1.4 mg/L) being more dominant than MMA III (2.6 ± 1.4 mg/L). The addition of both methyl cobalamin and glutathione as a reducing agent increased the methylation to 28%, with MMA III (10.5 ± 5.4 mg/L) becoming equally as important as MMA V (11.3 ± 5.6 mg/L).
These preliminary data convinced us that the selected microbial community had the potency to actively metabolize iAs V . We therefore inoculated the SHIME reactor with this fecal microbial inoculum; after 3 weeks of adaptation, a stable microbial community was obtained in the proximal and distal colon compartments. We regularly sampled the distal colon compartment to perform colon incubations on iAs V and Ascontaminated soil samples that had already gone through a gas tric and intestinal digestion. Characterization of the colon digests consisted of determin ing As bio accessibility and As speciation. The bioaccessibility determination was based on the sum of all chromatographically detected (HPLC/ICPMS) As species in the filtrates (0.45 µm) of the colon digests; therefore, the chromatographic recovery was calculated first. The sum of the concentrations of chromato graphically detected As species in the colon  filtrates was divided by the total As concen tration in the colon filtrates, as measured by ICPOES. The chromatographic recoveries for all colon digests, except for that of soil 4, were satisfactorily high: 93 ± 7% (mean ± SD) on average [the recovery of the soil 4 digest excluded; see Supplemental Material, Table 2 (doi:10.1289/ehp.0901794)]. Hence, most As species present in these digest super natants could be detected with the HPLC/ ICPMS protocol. Bioaccessibility calcula tions for these digests displayed the highest value (75.5%) for the iAs V incubated colon digest, whereas colon incubation of the con taminated soils resulted in As bioaccessibility values of 24% (soil 1), 44% (soil 2), and 36% (soil 3) (Table 3). In sharp contrast, As bio accessibility in the colon digest of soil 4 was only 0.3%. Even when taking into account the low chromatographic recovery of 15%, we obtained a low As bioaccessibility of 2.4%, which is still an order of magnitude lower than the bioaccessibility values for the other soil digests. Overall, colon bioaccessibility val ues (Table 3) for the four soils were consis tently lower than the corresponding intestinal bioaccessibility values (Table 1) obtained with the IVG method.
The most important part of this study consisted of the As speciation analysis of the colon digests after the gastrointestinal incu bation of iAs V and the four contaminated soils. The original analytical protocol was optimized to detect the presence of iAs III , iAs V , MMA III , and MMA V . We detected an additional As species, MMMTA V , in many of the colon digests. We initially identified MMMTA V using a combination of reten tiontime matching and by fortifying the sample with the suspected standard using separation 1 with ICPMS detection, but we used a second chromato graphy (separation 2) for ICPMS and ESIMS detection. Figure 1 shows HPLC/ICPMS mass chromatograms of m/z 75 ( 75 As) and HPLC/ESIMS mass chromatograms of m/z 155 ([MH] -) for an MMMTA V standard and a SHIME extract using separation 2. The retention times of the MMMTA V in the standard and MMMTA V in the sample were slightly offset because the matrix of the soil extract caused the decreased retention of MMMTA V on the C 18 column. Tandem MS (MS/MS) of m/z 155 yielded a product ion of m/z 137 (loss of H 2 O) and, to a lesser extent, a product ion of m/z 121 (due to CH 2 AsO 2 -) and m/z 140 (loss of CH 3 ). The molecular mass of 155 and cor responding fragments were consistent with other reports for MMMTA V (Yathavakilla et al. 2008).
We detected significant As methylation upon colon incubation of 225 µg iAs V /L (Figure 2). The sum of the concentrations of MMA V (31.0 µg/L), MMA III (4.5 µg/L), and MMMTA V (43.7 µg/L) exceeded that of iAs V (39.0 µg/L) and iAs III (34.8 µg/L). In contrast, iAs species were predominantly present in colon digests of soils 1, 2, and 3, whereas they were the only As species in the colon digest of slag soil 4 ( Figure 2). The colon digest of soil 1 displayed a methylation percentage of 4.7% with MMA V (17 µg/L) and MMMTA V   (23 µg/L) as detected methylarsenicals. The methylation percentage for colon digests of soil 2 (22.8%) and soil 3 (21.2%) was higher, with soil 2 displaying MMA V (111 µg/L), MMA III (9 µg/L), and MMMTA V (158 µg/L) and soil 3 displaying only MMA V (28 µg/L) and MMMTA V (68 µg/L). Finally, no meth ylated As species were detected in the colon digests of slag soil 4. Summarizing the in vitro As speciation changes by human gut microorganisms, we calculated the specific production rates of methylated arsenicals by taking into account the initial microbial biomass and As con centrations. We obtained a methylation rate of 10 µg methylarsenicals/g biomass/hr for the colon digest of iAs V (Table 3). Although no methylarsenicals were detected in colon digests of the slag soil 4, the presence of the other soil matrices did not necessarily lower the abovementioned methylation rate. We obtained methylation rates of 4, 29, and 10 µg/g/hr for colon digests of soils 1, 2, and 3, respectively (Table 3).

Discussion
The present study demonstrates that human colon micro organisms have the potency to actively metabolize As into methylated arsenicals and thioarsenicals, which indicates that presystemic As metabolism may not be neglected when assessing risks from oral As exposure. We observed this upon colon incubation of both iAs and Ascontaminated soils. These findings parallel those from studies with animal gut microbiota (Hall et al. 1997;Rowland and Davies 1981) and suggest the existence of a presystemic As metabolism in the human body. The most important result was the detection of signif icant levels of MMMTA V in colon digests of both iAs V (25% of bioaccessible As) and of Ascontaminated soils (up to 20% of bio accessible As). To our knowledge, this is the first time that MMMTA V production from iAs V by human colon microbiota has been described. MMMTA V production from this source resembles the methylation and thiola tion of DMA V into trimethylarsine sulfide by mouse cecal microbiota (Kubachka et al. 2009a) and the production of methylated thio arsenicals from DMA V by rat intestinal microbiota (Kuroda et al. 2004;Yoshida et al. 2001). Yet, mammalian cells also have the ability to form methylated thioarsenicals. Kuroda et al. (2004) described rapid detection (5 min) of DMMTA V and dimethyldithioars inic acid (DMDTA V ) after injection of of rats with DMA III , and Naranmandura and Suzuki (2008) reported that DMA III was converted to DMDTA V by human red blood cells.
The finding of presystemic MMMTA V formation by human gut micro organisms raises questions about its toxicological importance. Although the absorption kinetics of MMMTA V and other thiolated arsenicals across the epithelium are unknown, there is evidence that some methylated thio arsenicals elicit a higher toxicity than iAs V because of their more efficient absorption by mam malian cells (Naranmandura et al. 2007a). Preliminary cytotoxicity (Naranmandura et al. 2007a;Yoshida et al. 2003) and genotoxicity (Kuroda et al. 2004) data for DMMTA V show levels of toxicity similar to those of trivalent As species. Our observa tions in the present study emphasize the need to investigate the behavior of MMMTA V in the gut lumen and the absorption rate across the intestinal epithelium. In addition, the mechanism behind the microbial production pathway needs to be elucidated. Interestingly, MMMTA V levels in the colon digests cor related with those of MMA V (R 2 = 0.76), whereas the correlation with levels of MMA III was much lower (R 2 = 0.42). This seems to indicate that MMMTA V in the colon digests arises from the thiolation of MMA V , which would correspond with earlier observations describing the interconversion between oxide and sulfide forms of MMA V , DMA V , and TMAO (Conklin et al. 2008). The sulfide source may originate from microbial sulfate reduction to hydrogen sulfide, which is a common process in the colon environment (Deplancke et al. 2000), and can trigger the formation of thioarseno sugars upon the incubation of arsenosugars with mouse cecal contents (Conklin et al. 2006). The role of sulfatereducing micro organisms in the pre systemic production pathway of MMMTA V must therefore be studied further.
The significant formation of MMA V and MMA III after incubation of iAs V with colon micro organisms was not unexpected. Arsenic methylation by rodent gut microbes (Hall et al. 1997;Rowland and Davies 1981) and human gut microbes (DiazBone and Van de Wiele 2009;Meyer et al. 2008) has been described previously. Taking into account the initial biomass concentration, we observed specific methylation rates of 10 µg methyl arsenicals/hr/g biomass. This roughly corre sponds to 130 pmol/hr/mg biomass, which is > 16 pmol/hr/mg obtained with rat cecal microbiota (Hall et al. 1997). Interestingly, the presence of a soil matrix did not neces sarily result in lower As methylation rates, yet soildependent parameters may have affected the methylation rate. First, comparison of the mineralogy from soils 1, 2, and 3 with that of slag (soil 4) showed an important difference in reactive iron oxide content (Table 1), which is highly efficient in sorbing As (Beak et al. 2006). The reactive iron oxide content in slag soil 4 was particularly high (18,759 µg/kg; Table 1), presumably leading to much lower As availability to colon microorganisms (0.3% bio accessibility) and thus also limiting methyla tion. This observation may confirm earlier observations of slag soil mineralogy sig nificantly decreasing As bioavailability (Davis et al. 1996). A second element in the soil dependent As methylation may be the differ ence in toxic elements. Compared with the first three soils, slag (soil 4) contained high amounts of cadmium, chromium, copper, molybdenum, lead, and zinc (Table 1), which may be toxic to intestinal micro organisms. Our finding of a 70% lower fermentation activity in soil 4 colon digests versus colon digests of the other soils (data not shown) may support the assumption of slagsoil-induced toxicity. The actual relationship between gut microbial As metabolism and soil characteris tics therefore needs further study.
A final aspect of our study concerns the metabolic potency of fecal microbes toward high levels of iAs V (90 mg iAs V /L) and the influence of cofactors. Non amended colon digests of iAs V resulted in the efficient reduc tion to iAs III and the production of mTA. Similar to the finding of MMMTA V , the for mation of mTA may result from an oxygen forsulfur exchange in iAs V because of the availability of sulfide, originating from the abovementioned microbial sulfate reduction. The absence of mTA in glutathioneamended samples may be explained by the complete reduction of iAs V into iAs III by glutathione as reducing agent. We also evaluated the effect of methyl group donors. In contrast to methi onine, methyl cobalamin may be an effective methyl group donor, resulting in the efficient methylation (19%) of iAs V into MMA V and MMA III ( Table 2). The methylation efficiency increased to 25% upon cosupplementation of methylcobalamin and glutathione. We attributed this to the increased reduction of MMA V into MMA III by glutathione as reduc ing agent. In contrast to the colon digests with low levels of iAs V (225 µg/L), no MMMTA V was detected in the fecal digests. A probable explanation is the difference in experimental setup, the difference in microbial community composition and activity, or a difference in sulfide availability. These observations confirm a previous report that addition of cofactors may increase As methylation by enteric micro organisms, yet it is not a prerequisite for the methylation of low levels of As (micrograms per liter range) (Hall et al. 1997).
The present study provides evidence for the existence of significant presystemic As metabolism by human gut microorgan isms, but the relevance for the total risk of oral As exposure is not yet clear. So far, the in vitro approach for assessing the risks from oral contaminant exposure mainly involved the use of models that focus on gastric and intestinal processes. Methylation of As by intestinal micro organisms was thought to contribute little to the overall methyla tion in vivo (Vahter and Gustafsson 1980) because iAs V and iAs III are rapidly absorbed in the small intestine (Vahter 1983), espe cially when As is ingested in a soluble matrix (e.g., drinking water). However, soilbound and/or dietarybound As may follow a dif ferent digestion scenario in the gut, and a large fraction may end up in the colon lumen, where it is subjected to the resident micro bial community. The finding of MMMTA V and the highly toxic MMA III as metabolites from human colon micro organisms indicates that presystemic methylation will not lead to detoxification. In addition, in vitro stud ies with Caco2 human epithelial colo rectal adenoc arcinoma cells suggest that the absorp tion of methylated arsenicals (DMA V , 10.0%; TMAO, 10.9%) is more efficient than that of iAs III (5.8%) and iAs V (1.6%) (Laparra et al. 2005(Laparra et al. , 2007. Intestinal absorption of methy lated thioarseni cals should be examined in future research. Regarding the variability between indi viduals regarding presystemic As metabolism, we investigated the gut microbiota from only one human. Inter individual variability in gut microbial composition is very high; thus, we expect gut microbiota from different individ uals to display distinct As metabolic profiles. Such inter individual variation in metabolism by human gut microbiota was previously reported for ingested phyto estrogens (Bolca et al. 2007) and, interestingly, also for the metalloid bismuth . Therefore, variability in gut microbial As metabolism should be given the same atten tion as the genetic variations that may gov ern interindividual differences in As response (Hernandez and Marcos 2008).

Conclusion
The present study shows that presystemic metabolism of soilderived As may be rel evant in the human body when significant amounts of As become available to colon micro organisms. The absorption kinetics of methylated arsenicals and thioarsenicals across the gut epithelium and their toxicity need further elucidation. We propose that the metabolic activity of human colon micro organisms be incorporated in development of new toxico kinetic models that assess risks from oral As exposure. Mikov (1994) nicely summarized the importance of gut micro biota, stating that gut microbial metabolism must be considered an integral part of drug/ xenobiotic metabolism and toxicity studies. In this context, knowledge about gut micro bial metabolism must also be translated to metal(loid) biotransformation.