A C⋅As lyase for degradation of environmental organoarsenical herbicides and animal husbandry growth promoters

Significance

Organoarsenicals are used as herbicides, pesticides, antimicrobial growth promoters, and chemical warfare agents. Environmental organoarsenicals are microbially degraded, but the molecular mechanisms of breakdown are unknown. We previously identified a two-step pathway of degradation involving sequential reduction and C⋅As bond cleavage. Here we report cloning of the gene and characterization of the gene product for a C⋅As lyase, ArsI, a member of the family of type I extradiol dioxygenases. ArsI is the only enzyme shown to be involved in degradation of the reduced forms of the herbicide monosodium methylarsonic acid and the antimicrobial growth promoter roxarsone. As arsI genes are widely distributed in bacteria, ArsI-catalyzed organoarsenic degradation is proposed to have an impact on the arsenic biogeocycle.

Abstract

Arsenic is the most widespread environmental toxin. Substantial amounts of pentavalent organoarsenicals have been used as herbicides, such as monosodium methylarsonic acid (MSMA), and as growth enhancers for animal husbandry, such as roxarsone (4-hydroxy-3-nitrophenylarsonic acid) [Rox(V)]. These undergo environmental degradation to more toxic inorganic arsenite [As(III)]. We previously demonstrated a two-step pathway of degradation of MSMA to As(III) by microbial communities involving sequential reduction to methylarsonous acid [MAs(III)] by one bacterial species and demethylation from MAs(III) to As(III) by another. In this study, the gene responsible for MAs(III) demethylation was identified from an environmental MAs(III)-demethylating isolate, Bacillus sp. MD1. This gene, termed arsenic inducible gene (arsI), is in an arsenic resistance (ars) operon and encodes a nonheme iron-dependent dioxygenase with C⋅As lyase activity. Heterologous expression of ArsI conferred MAs(III)-demethylating activity and MAs(III) resistance to an arsenic-hypersensitive strain of Escherichia coli, demonstrating that MAs(III) demethylation is a detoxification process. Purified ArsI catalyzes Fe²⁺-dependent MAs(III) demethylation. In addition, ArsI cleaves the C⋅As bond in trivalent roxarsone and other aromatic arsenicals. ArsI homologs are widely distributed in prokaryotes, and we propose that ArsI-catalyzed organoarsenical degradation has a significant impact on the arsenic biogeocycle. To our knowledge, this is the first report of a molecular mechanism for organoarsenic degradation by a C⋅As lyase.

The metalloid arsenic is the most common environmental toxic substance, entering the biosphere primarily from geochemical sources, but also through anthropogenic activities (1). Arsenic is a group 1 human carcinogen that ranks first on the Agency for Toxic Substances and Disease Registry Priority List of Hazardous Substances (www.atsdr.cdc.gov/SPL/index.html). Microbial arsenic transformations create a global arsenic biogeocycle (1). These biotransformations include redox cycles between the relatively innocuous pentavalent arsenate and the considerably more toxic and carcinogenic trivalent arsenite (2, 3). In addition, many microbes, both prokaryotic and eukaryotic, have arsM genes for inorganic arsenite [As(III)] S-adenosylmethionine methyltransferases that methylate inorganic As(III) to mono-, di-, and tri-methylated species (4, 5). The genes encoding arsenic transforming enzymes are widely distributed, and these arsenic biotransformations have been proposed to play significant roles in the arsenic biogeocycle and in remodeling the terrain in volcanic areas such as Yellowstone National Park and regions of the world with high amounts of arsenic in soil and water such as West Bengal and Bangladesh (3, 6).

Arsenicals, both inorganic and organic, have been used in agriculture in the United States for more than a century (7). Historically, the use of inorganic arsenical pesticides/herbicides has been largely replaced by methylated arsenicals such as monosodium methylarsonic acid (MSMA), which is still in use as an herbicide for turf maintenance on golf courses, sod farms, and highway rights of way, and for weed control on cotton fields (7). More complex pentavalent aromatic arsenicals such as roxarsone [4-hydroxy-3-nitrophenylarsonic acid, Rox(V)] have been largely used since the middle of the 1940s as antimicrobial growth promoters for poultry and swine to control Coccidioides infections and improve weight gain, feed efficiency, and meat pigmentation (8, 9). These aromatic arsenicals are largely excreted unchanged and introduced into the environment when chicken litter is applied to farmland as fertilizer (8). Pentavalent organoarsenicals are relatively benign and less toxic than inorganic arsenicals; however, aromatic (8–10) and methyl (11, 12) arsenicals are degraded into more toxic inorganic forms in the environment, which may contaminate the foods and water supplies. Although microbial degradation of environmental organoarsenicals has been documented (8, 9, 11, 13), no molecular details of the reaction have been reported. We recently demonstrated that a microbial community in Florida golf course soil carries out a two-step pathway of MSMA reduction and demethylation (14). Here we report the isolation of an environmental methylarsonous acid [MAs(III)]-demethylating bacterium Bacillus sp. MD1 (for “MAs(III) demethylating”) from Florida golf course soil and the cloning of the gene, termed arsenic inducible gene (arsI), responsible for MAs(III) demethylation. The gene product, ArsI, is nonheme iron-dependent dioxygenase with C⋅As lyase activity. ArsI cleaves the C⋅As bond in a wide range of trivalent organoarsenicals, including the trivalent roxarsone [Rox(III)], into As(III), which strongly suggests that the environmental pentavalent phenylarsenicals such as Rox(V) also undergo a two-step pathway of sequential reduction and ArsI-catalyzed dearylation, in analogy with the demethylation of MSMA by a microbial community. Thus, ArsI-catalyzed C⋅As bond cleavage is a newly identified mechanism for degradation of organoarsenical herbicides and antimicrobial growth promoters.

Results

Isolation of a MAs(III)-Demethylating Bacillus from Golf Course Soil.

We previously reported the isolation of methylarsonic acid [MAs(V)]-reducing bacterium Burkholderia sp. MR1 and the MAs(III)-demethylating bacterium Streptomyces sp. MD1 from Florida golf course soil (14). Together, these two activities result in degradation of the herbicide MSMA. In this study, a second bacterial strain capable of MAs(III) demethylation was isolated. Like Streptomyces sp. MD1, this isolate demonstrated no MAs(V) transformation when cultured alone (Fig. S1A, curve 2), but nearly completely transformed MAs(V) into As(III) when cocultured with Burkholderia sp. MR1 (Fig. S1A, curve 1), suggesting that the isolate possesses MAs(III) demethylating activity. In confirmation, the isolate demethylated MAs(III) (Fig. S1B). The isolate was identified as a Gram-positive Bacillus by 16S ribosomal DNA sequence analysis (GenBank accession no. KF899846) (Fig. S2), and was designated Bacillus sp. MD1 [i.e., MAs(III) demethylating].

Cloning the ArsI MAs(III) Demethylase.

MAs(III) is more toxic than inorganic As(III) in Chang human hepatocytes (15) and hamsters (16). We showed that it is more toxic in Escherichia coli cells as well (Fig. S3) (17), suggesting that demethylation is a detoxification process. By using the difference in toxicity between MAs(III) and As(III), we cloned the gene responsible for MAs(III) demethylation. A genomic DNA library from Bacillus sp. MD1 was constructed and transformed into E. coli, and the transformants were selected for MAs(III) resistance. Two MAs(III)-resistant clones were obtained, and their ability to demethylate MAs(III) confirmed. Because the pattern of HindIII-digested inserts of the two clones was identical, both probably have the same insert, and one was chosen for further analysis. The restriction fragment of ∼12 kbp was subcloned, and the region responsible for MAs(III) demethylation was narrowed to approximately 4 kbp. This DNA was sequenced and found to contain five putative genes (Fig. S4). The first would encode a 99-residue ArsR As(III)-responsive transcriptional repressor except for an apparent frame shift that truncates the putative gene product after 72 residues. This is followed a putative gene similar to an ars gene of unknown function termed arsI (i.e., arsenic inducible gene) from Cupriavidus metallidurans CH34 (18) (Fig. S5). The 161-residue gene product has been annotated as a bleomycin resistance protein, glyoxalase or type I extradiol ring-cleaving dioxygenase. Three other putative reading frames are encoded on the opposite strand and annotated as encoding a FAD-dependent monooxygenase and a GCN5-related N-acetyltransferase, more distantly, a partial sequence for a hemin ABC transporter ATP-binding subunit. Genes highly homologous to those three were rarely found in ars operons, indicating that these genes are not involved in arsenic detoxification. In contrast, type I extradiol ring-cleaving dioxygenases catalyze cleavage of the C⋅C bond of aromatic ring (19), and MAs(III) demethylation can be predicted to involve cleavage of the analogous C⋅As bond, so it is reasonable to propose that arsI is the responsible gene.

Type I extradiol dioxygenases belong to the vicinal oxygen chelate superfamily and use a divalent metal ion to catalyze a reaction involving direct metal ion chelation by vicinal oxygens of the substrate (19). Type I extradiol dioxygenases include two-domain and one-domain enzymes, and ArsI resembles a one-domain enzyme. Most are two-domain proteins composed of N-terminal and C-terminal domains exemplified by methylcatechol 2,3-dioxygenase encoded by akbC from Rhodococcus sp. strain DK17 (20) and 2,3-dihydroxybiphenyl 1,2-dioxygenase encoded by bphC from Pseudomonas sp. strain KKS102 (21) (Fig. S5). Although the N- and C-terminal domains are structurally similar to each other, only the C-terminal domain binds metal and functions in catalysis (19). The divalent metal binding site of these dioxygenase contains a triad of three charged amino acid residues. A homology search using Protein Basic Local Alignment Search Tool identified His5-His62-Glu115 as a putative metal binding site in the Bacillus ArsI, which corresponds to those of AkbC and BphC (Fig. S5). His5 is replaced by a glutamine residue in the putative ArsI orthologs from Thermomonospora curvata DSM 43183 and Streptomyces coelicolor A3 (2) (Fig. S5).

A Basic Local Alignment Search Tool Link to Protein Alignments and Structures search identified nearly 650 putative ArsI orthologs in 487 bacterial species, with no representatives in other kingdoms. In the Bacillus ArsI sequence there are four vicinal cysteine pairs. One of the four (Cys96-Cys97 in the Bacillus ArsI) is conserved in all putative ArsI orthologs, and the other three pairs in Bacillus ArsI, located near the C terminus, are not conserved in all ArsI orthologs (Fig. S5). Conserved cysteine residues in arsenic resistance proteins often function as binding sites for trivalent arsenicals (22–24). The ArsI conserved Cys96-Cys97 cysteine pair is located at the equivalent position as the aspartate residue used by AkbC and BphC to form their substrate binding sites (Fig. S5). It is reasonable, therefore, to propose that Cys96 and Cys97 form the MAs(III) binding site in ArsI.

Nonconserved cysteine pairs in other arsenic resistance proteins such as ArsR, ArsD, and ArsM are not critical for the activity, and removal of these nonconserved cysteine residues improves protein production and crystallization (25–27). A truncated arsI was constructed lacking the sequence for residues from Glu125 to the C terminus, which included the three nonconserved cysteine pairs. This construct, designated ArsI₁₂₄, was found to confer MAs(III) resistance and demethylation (Fig. 1). E. coli strain AW3110, an arsenic-hypersensitive strain lacking the chromosomal arsRBC operon (28), was transformed with plasmid pMAL-arsI₁₂₄ expressing ArsI₁₂₄ as a maltose-binding protein (MBP) fusion. This C-terminally truncated MBP fusion is henceforth termed ArsI for simplicity. Cells expressing ArsI nearly completely transformed MAs(III) into inorganic As(III) (Fig. 1A, curve 1), whereas no MAs(III) transformation occurred in control cells (Fig. 1A, curve 2), confirming that the single arsI gene encodes a C⋅As lyase that catalyzes demethylation of MAs(III). AW3110 cells bearing plasmid pMAL-arsI₁₂₄ were more resistant to MAs(III) than cells bearing vector plasmid pMAL-c2x (Fig. 1B). These results support our hypothesis that transformation of MAs(III) to As(III) is a detoxification process that confers herbicide resistance. It also demonstrates that the nonconserved C-terminal vicinal cysteine pairs are not essential for enzymatic activity, and the truncated ArsI was used subsequently for this study.

Fig. 1.

ArsI confers MAs(III) demethylating activity and resistance in E. coli. (A) MAs(III) biotransformation by E. coli strain AW3110 (Δars) bearing plasmid pMAL-arsI₁₂₄ (curve 1), vector plasmid pMAL-c2x (curve 2), or no cells (curve 3) was analyzed by HPLC-ICP-MS (Materials and Methods). The x axis represents column retention time and the y axis represents relative amounts of arsenic expressed as counts per second (cps). (B) Growth of E. coli AW3110 bearing pMAL-arsI₁₂₄ (●) or pMAL-c2x (▽) with the indicated concentrations of MAs(III). Error bars represent the SD of three assays.

Characterization of C⋅As Bond Cleavage.

The requirements for demethylation activity of purified ArsI were examined (Fig. 2). Purified ArsI requires a ferrous ion and a reductant such as glutathione or cysteine for the C⋅As lyase activity. The activity with Fe²⁺ and cysteine was enhanced approximately eightfold by addition of the strong reducing reagent tris(2-carboxyethyl)phosphine (TCEP), and further assays were conducted in the presence of TCEP and cysteine in addition to Fe²⁺. Some dioxygenases have enzymatic activity with various metals (19), but ArsI uses Fe²⁺ exclusively (Fig. S6). Reductant is likely required to maintain the enzyme (cysteine thiolates and Fe²⁺) and the substrate MAs(III) in their reduced forms. Even in the presence of reductant, some MAs(III) oxidized to MAs(V), probably nonenzymatically by air, but the remainder of the MAs(III) was quantitatively transformed into As(III) over a period of 60 min (Fig. 3).

Fig. 2.

Effect of Fe²⁺ and reductants on the MAs(III)-demethylating activity of purified ArsI. The amount of As(III) produced upon demethylation of 10 μM MAs(III) by 1 µM purified ArsI in each indicated condition was quantified by HPLC-ICP-MS, as described in Materials and Methods. Activities are expressed as the percentage of As(III) produced [4.0 µmol As(III)] with Fe²⁺, cysteine, and TCEP. Error bars represent the SD of three assays.

Fig. 3.

Time course of MAs(III) demethylation by purified ArsI. The time course of MAs(III) demethylation was conducted as described in Materials and Methods. The reaction was initiated by addition of ArsI (arrow). △, total arsenic; ○, MAs(III); ▼, MAs(V); and ●, As(III). Error bars represent the SD of three assays.

Members of the family of type I extradiol ring-cleaving dioxygenases catalyze cleavage of C⋅C bond of the aromatic ring by incorporating an atom of molecular oxygen into each carbon of the target bond (19). O₂ consumption by purified ArsI clearly increased when the substrate MAs(III) was added (Fig. S7). The initial immediate decrease in O₂ was caused by the reducing reagent used to prepare MAs(III) and As(III). This supports the proposition that ArsI is a dioxygenase that uses oxygen for cleavage of the C⋅As bond, in which one oxygen atom from dioxygen is added to the arsenic and the other is added to the carbon (Fig. S8). Although the organic product following MAs(III) cleavage has not been identified, it is predicted to be formaldehyde (Fig. S8, Upper). The K_m for MAs(III) was determined to be 2.9 ± 0.8 μM, with a k_cat of 0.25 ± 0.02 min⁻¹. The low turnover number for the enzymatic process of C⋅As bond cleavage catalyzed by purified ArsI may reflect the fact that the enzyme, cofactor, and substrate are all oxygen-sensitive. In vivo, these may be protected by the low redox potential of the cytosol.

ArsI Cleaves the C⋅As Bond of Aromatic Organoarsenicals.

The ability of ArsI to catalyze C⋅As bond breakage in more complex trivalent aromatic organoarsenicals was examined. ArsI could degrade the Rox(III) to inorganic As(III) in vivo (Fig. 4A) and in vitro (Fig. 4B). With Rox(III) as substrate, assuming that ArsI is a ring-cleavage dioxygenase, the organic product can be predicted to be 4-hydroxy-5-nitro-hexa-2,4-dienal (Fig. S8, Lower). Rox(III) is more toxic than As(III) (Fig. S3) (17), and AW3110 cells bearing pMAL-arsI₁₂₄ were resistant to Rox(III) (Fig. 4C), demonstrating that cleavage of the C⋅As bond in Rox(III) is a detoxification process. The values of K_m and K_cat for Rox(III) degradation by ArsI were 6.4 ± 1.7 μM and 0.22 ± 0.04 min⁻¹, respectively, which are compatible with the kinetic values for MAs(III) demethylation. ArsI-expressing AW3110 cells cleaved the C⋅As bond in the reduced forms of other aromatic arsenical, including nitarsone (4-nitrophenylarsonic acid; Fig. S9A) and p-arsanilic acid (4-aminophenylarsonic acid; Fig. S9B). These results demonstrate that ArsI is a generalized C⋅As lyase.

Fig. 4.

Degradation of and resistance to Rox(III). (A) Rox(III) biotransformation by E. coli AW3110 (Δars) bearing pMAL-arsI₁₂₄ (curve 1), pMAL-c2x (curve 2), or no cells (curve 3). (B) The time course of Rox(III) degradation by purified ArsI was conducted as described in Materials and Methods. The reaction was initiated by addition of ArsI (arrow). △, total arsenic; ○, Rox(III); ▼, Rox(V); and ●, As(III). (C) Resistance to Rox(III) was assayed by growth of E. coli AW3110 (Δars) bearing pMAL-arsI₁₂₄ (●) or pMAL-c2x (▽). Error bars represent the SD of three assays.

Discussion

The arsM gene for arsenic methyltransferases is widespread and is found in members of every kingdom (1). This implies that arsenic methylation is an ancient process occurring since the first organisms arose nearly 3.5 billion years ago. Why, then, is the majority of environmental arsenic not methylated? One reason is that inorganic arsenic is renewed through geothermal sources. Another contributing factor is that organoarsenicals are continuously broken down by microbial transformations. In this report, we identify what is, to our knowledge, the first known enzyme, ArsI, for cleavage of the C⋅As bond. The gene for ArsI is found in many aerobic bacterial species that exist in communities in which methylating bacteria produce MAs(V) that reducing bacteria transform into the more toxic MAs(III). ArsI-expressing bacteria detoxify the trivalent methylated species by transforming it back to inorganic arsenic. The continual interplay creates a biogeocycle that maintains a balance between organic and inorganic species.

During the past half century, mankind has upset that balance by the introduction of massive amounts of organoarsenicals. Methylated arsenicals such as MAs(V) and DMAs(V) (i.e., dimethylarsinic acid) have long been used as herbicides and insect pesticides. Many of these are for weed control, especially for cotton, ornamental plants, lawns, and golf-course turf. Approximately 3,000,000 lbs. of MAs(V), mainly as MSMA, and 100,000 lbs. of DMAs(V) have been applied annually in the United States (www.epa.gov/oppsrrd1/REDs/organic_arsenicals_red.pdf). Aromatic compounds such as roxarsone, nitarsone, and p-arsanilic acid have been used in animal husbandry for disease prevention, growth promotion, enhanced feed utilization, and improved meat pigmentation. The majority of these compounds is not retained by the animals but excreted into the environment unchanged. Based on broiler production and Rox(V) feed dosage, it is estimated that ∼2,000,000 pounds of Rox(V) were released into environment in the United States annually by the poultry industry alone (9).

ArsI also catalyzes cleavage of the C⋅As bond in trivalent form of these manmade aromatic arsenicals: roxarsone (Fig. 4), nitarsone (Fig. S9A), and p-arsanilic acid (Fig. S9B). Degradation of such aromatic arsenicals has been demonstrated under aerobic (10) and anaerobic conditions (8–10). In analogy with the demethylation of MAs(V), we propose that these pentavalent aromatic arsenicals undergo a two-step pathway of sequential reduction and ArsI-catalyzed dearylation. We have not observed reduction of Rox(V) by the MAs(V)-reducing bacterial isolates from Florida golf course soils, and so we propose that there are yet-unidentified bacteria capable of reducing Rox(V) to Rox(III). To date, all arsI-carrying bacteria, including Bacillus sp. MD1, are aerobes. It is possible that alternate pathways exist in anaerobes.

Although the use of MSMA in the United States is now restricted to cotton fields, golf courses, sod farms, and highway rights of way, and roxarsone has been largely replaced by the related nitarsone, these organoarsenicals are still produced and used in other countries such as India and China. Moreover, organoarsenicals have been used for chemical warfare for more than a century. During the Vietnam War, the United States released more than 1.2 M gallons of DMAs(V), which was called Agent Blue, one of the “rainbow herbicides” to kill the rice, bamboo, and banana crops (29). Agent Blue was used until recently in the United States for spraying of cotton fields and golf courses, especially in Florida. Diphenylated arsenic compounds such as Clark I (diphenylchloroarsine) and Clark II (diphenylcyanoarsine) were produced as chemical warfare agents during World Wars I and II. These were disposed of in land and seas after those wars. However, diphenylarsinic acid, which results from chemical transformation and accumulates in the environment, has been demonstrated to degrade into inorganic forms by bacterial isolates from contaminated sites (30). We speculate that bacteria with arsI genes are involved in the environmental degradation of these dimethyl and diphenyl arsenicals.

The molecular mechanisms of MSMA reduction, the initial reaction of the two-step MAs(V) breakdown pathway, remains unknown (14). Reduction of MSMA has been also observed in plants such as rice (31). As is the case with MSMA, the toxicity of Rox(V) is also increased by reduction to Rox(III) (Fig. S3), and ArsI-catalyzed degradation of Rox(III) to As(III) confers arsenic resistance (Fig. 4C). We propose that the marked increase in toxicity in the trivalent forms of these relatively nontoxic pentavalent organoarsenicals is linked to bioactivation by reducing organisms. To function as herbicides, pesticides, and antimicrobial growth promoters, the pentavalent compounds must be reduced to the active species (17).

We analyzed the distribution of more than 100 arsI genes and found that all are in ars operons. There seems to be no specific ars gene associated with arsI, in contrast to arsD and arsA, which always go together because their functions are interrelated (32). MerB is a C⋅Hg lyase that confers resistance to organomercurials (33). Although superficially similar to ArsI, MerB is an unrelated enzyme with a different reaction mechanism (34). Similarly, merB is usually associated with merA, and mer operons that include merB are mostly broad-spectrum resistances to inorganic and organomercurials (35, 36). In contrast, narrow-spectrum operons that lack merB confer resistance to only inorganic mercury.

In summary, to our knowledge, ArsI is the first and, at this point, only identified gene product involved in organoarsenic degradation. ArsI also expands the range of arsenic resistance from narrow spectrum (inorganic) to broad spectrum (both inorganic and organoarsenicals). MerB-catalyzed organomercurial degradation predominates in mercury-polluted sites (36). Because arsI was cloned from a golf course treated with MSMA, we speculate that ArsI-catalyzed organoarsenical degradation predominates in sites contaminated with organoarsenicals, so ArsI degradation of organoarsenicals contributes to the arsenic biogeochemical cycle, completing the alternation of arsenic methylation and demethylation.

Materials and Methods

Reagents.

All chemicals were obtained from Sigma-Aldrich unless otherwise mentioned. Rox(V) was purchased from Acros Organics. The trivalent forms of methylarsonic acid, roxarsone, nitarsone, and p-arsanilic acid were prepared by a slight modification of the previous procedure (37). Briefly, 0.2 mM arsenical was mixed with 27 mM Na₂S₂O₃, 66 mM Na₂S₂O₅, and 82 mM H₂SO₄, following which the pH was adjusted to 6 with NaOH.

Cloning the Gene for MAs(III) Resistance.

A MAs(III)-demethylating bacterium Bacillus sp. MD1 was isolated as described in SI Materials and Methods. The total DNA extracted from Bacillus sp. MD1 (14) was partially digested with HindIII (New England BioLabs) and ethanol-precipitated. The pelleted DNA was suspended in water and layered on a discontinuous sucrose gradient (10%, 20%, 30%, and 40%, wt/vol) containing 50 mM Tris, pH 8.0, 0.1 M EDTA, and 0.1 M NaCl, and centrifuged at 210,000 × g for 5 h at 17 °C. After centrifugation, the solution was carefully removed from the top of the layer and separated into fractions. Fractions containing DNA fragments of more than 2 kbp were combined and concentrated by ethanol precipitation. The resulting DNA was ligated to vector plasmid pUC118 Hind III/BAP (Takara Bio), and the ligation mixture was transformed into E. coli strain TOP10 (Invitrogen) by using a MicroPulser (Bio-Rad).

The transformants were spread on three agar plates of ST 10⁻¹ medium (14) containing 50 µg/mL ampicillin, 0.2% d-glucose, and 0, 1, or 2 µM MAs(III). These plates were incubated at 30 °C until colonies formed. The plate with the fewest number of colonies was replicated onto new plates containing 1, 2, or 3 µM MAs(III) and incubated at 30 °C until colonies formed. Replicate plating was repeated twice. The clones obtained from these selections were cultured in Luria–Bertani (LB) medium supplemented with 100 µg/mL ampicillin at 37 °C overnight; then, the cultured cells were transferred to ST 10⁻¹ medium supplied with 1 µM MAs(III) and incubated at 25 °C for 8 h. MAs(III) demethylation in the culture medium was analyzed by HPLC inductively coupled plasma (ICP) MS as described in SI Materials and Methods. Plasmid DNA was extracted from two clones with MAs(III)-demethylating activity, and the inserts were subcloned into vector plasmid pUC19 (New England BioLabs) using appropriate restriction enzymes (New England BioLabs) and transformed into E. coli TOP10. The MAs(III)-demethylating activity of the obtained subclones was analyzed by HPLC-ICP-MS as described in SI Materials and Methods. By subcloning, the activity was narrowed to a 4-kbp fragment, which was sequenced by primer-walking technique (GenBank accession no. KF899847). The arsI gene was cloned for further investigation. CLUSTAL W (38) was used to build up multiple sequence alignment of ArsI homologs.

ArsI Expression, Purification, and Characterization.

The arsI gene product is predicted to have 161 aa residues, but, from multiple sequence alignment of ArsI homologs, it appears that there is little sequence conservation in the C-terminal 37 residues (Fig. S5). An arsI derivative encoding only the N-terminal 124 residues, designated arsI₁₂₄, was amplified from Bacillus sp. MD1 genomic DNA by using forward primer GGAGGGAGAATTCATGAAATATGCGC (EcoRI site underlined) and reverse primer TGTGTAAGCTTTTATTCAACAGTTGTC (HindIII site underlined). After digestion with EcoRI (New England BioLabs) and HindIII, the arsI₁₂₄ gene was cloned into vector plasmid pMAL-c2x (New England BioLabs), generating plasmid pMAL-arsI₁₂₄ encoding a hybrid ArsI₁₂₄ with an N-terminal MBP. The plasmids were transformed into the arsenic hypersensitive E. coli strain AW3110(DE3) Δars. The transformants were cultured in LB medium supplemented with 100 µg/mL ampicillin at 37 °C to an absorbance at 600 nm of 0.6, at which point 0.1 mM isopropyl β-d-1-thiogalactopyranoside was added as inducer. After incubation at 37 °C for an additional 4 h, the cells were transferred to ST 10⁻¹ medium and incubated with 1 μM MAs(III), Rox(III), trivalent nitarsone, or trivalent p-arsanilic acid at 25 °C for 1 h. C⋅As lyase activity was analyzed by HPLC-ICP-MS as described in SI Materials and Methods. To assay for resistance to arsenicals, induced cells bearing vector pMAL-c2x or pMAL-arsI₁₂₄ were washed once with the same volume of ST medium (10-fold concentrated ST 10⁻¹ medium), diluted 50-fold into fresh ST medium supplemented with 50 µg/mL ampicillin and 0.2% d-glucose, and incubated with the indicated concentrations of indicated arsenicals at 30 °C for 4 h, following which growth was estimated from the absorbance at 600 nm.

C-terminal truncated ArsI₁₂₄ with an N-terminal MBP was purified as described in SI Materials and Methods. C⋅As lyase activity with purified ArsI was assayed in a buffer consisting of 0.1 M morpholinopropane-1-sulfonic acid (MOPS) and 0.15 M KCl, pH 7.0. To optimize the conditions, enzyme (1 μM) was incubated with or without 0.1 mM Fe²⁺, 1 mM reduced glutathione, 1 mM cysteine, and/or 3 mM TCEP, with the reaction initiated by addition of 10 μM MAs(III). After incubating at 37 °C with shaking at 200 rpm for 30 min, the reactions were terminated with EDTA, and the arsenic species were analyzed by HPLC-ICP-MS. The amount of As(III) in each sample was quantified from the corresponding peak area using Chromera software (Perkin-Elmer). In other assays 0.1 mM Mg²⁺, Ca²⁺, Mn²⁺, Co²⁺ Ni²⁺, Cu²⁺, or Zn²⁺ were used in place of Fe²⁺. The best assay components were 1 mM cysteine, 3 mM TCEP, and 0.1 mM Fe²⁺. For time-based assays, the reaction was initiated by addition of 1 μM ArsI to the reaction solution containing 5 μM MAs(III) or Rox(III) and incubated at 37 °C with shaking at 200 rpm for 60 min. Reactions were terminated with EDTA at the indicated times and the arsenic species were analyzed by HPLC-ICP-MS as described in SI Materials and Methods. Each amount of the indicated arsenic species was quantified from the corresponding peak area.

Kinetic analyses were performed with 1 μM ArsI and varying concentrations of MAs(III) or Rox(III) up to 40 μM, and the amount of As(III) generated at 1 min was used to approximate initial rates. Kinetic constants were calculated by Hanes–Woolf analysis (39).

For analysis of O₂ consumption, 7 μM ArsI was incubated in 0.1 M MOPS, pH 7.0, with 0.15 M KCl, 10 μM Fe²⁺, 0.5 mM cysteine until the base line level of O₂ consumption was constant. MAs(III) and As(III) were prepared by using reducing reagent as mentioned earlier. Then, 7 μM of the prepared MAs(III) or As(III), or reducing reagent only, was injected to initiate catalysis, and the change in O₂ levels was monitored by using a 782 Oxygen Meter (Strathkelvin Instruments).