|Home | About | Journals | Submit | Contact Us | Français|
Microbial arsenate reduction affects the fate and transport of arsenic in the environment. Arsenate respiratory (arr) and detoxifying (ars) reduction pathways in Shewanella sp. strain ANA-3 are induced by arsenite and under anaerobic conditions. Here it is shown that an ArsR family protein, called ArsR2, regulates the arsenate respiratory reduction pathway in response to elevated arsenite under anaerobic conditions. Strains lacking arsR2 grew faster in the presence of high levels of arsenite (3 mM). Moreover, expression of arrA and arsC (arsenate reductase-encoding genes) in the ΔarsR2 mutant of ANA-3 were increased in cells grown under anaerobic conditions and in the absence of arsenic. Mutations in putative arsenic binding amino acid residues in ArsR2 (substitutions of Cys-30 and Cys-32 with Ser) resulted in ANA-3 strains that exhibited anaerobic growth deficiencies with high levels of arsenite and arsenate. DNA binding studies with purified ArsR2 showed that ArsR2 binding to the arr promoter region was impaired by trivalent arsenicals such as arsenite and phenylarsine oxide. However, ArsR2 binding occurred in the presence of arsenate. A second known regulator of the arr operon, cyclic AMP (cAMP)-cAMP receptor protein (CRP), could bind simultaneously with ArsR2 within the arr promoter region. It is concluded that ArsR2 is most likely the major arsenite-dependent regulator of arr and ars operons in Shewanella sp. strain ANA-3. However, anaerobic growth on arsenate will require coregulation with global regulators such as cAMP-CRP.
A worldwide public health crisis has emerged due to chronic consumption of arsenic-contaminated drinking water (reviewed in reference 2). Metal-reducing bacteria are partially responsible for this problem because they can catalyze the reductive dissolution of arsenate-bearing minerals, resulting in the liberation of arsenic into the surrounding water (reviewed in reference 23). The interconversion of pentavalent arsenate [As(V)] to trivalent arsenite [As(III)] is an important process that is thought to exacerbate the release of arsenic from mineral surfaces into groundwater (reviewed in reference 22). Compared to arsenate, arsenite is more toxic and mobile in the environment, and it is the most common form of arsenic in anaerobic contaminated aquifers. In order to predict the occurrence of arsenic mobilization in aquifers, it is critical to develop a mechanistic understanding of the environmental and biological factors that control microbial arsenate reduction and arsenite production.
Biological arsenate reduction occurs via two distinct pathways: one that allows the bacterium to use arsenate as a terminal electron acceptor and a second pathway that allows the cell to detoxify intracellular arsenate (reviewed in references 23 and 31). In the model Fe(III)- and arsenate-reducing bacterium Shewanella sp. strain ANA-3, the detoxification pathway is encoded by an ars operon (26), which includes a cytoplasmic arsenate reductase (ArsC), an arsenite efflux pump (ArsB), an ATPase (ArsA), and an arsenite binding protein (ArsD) (8, 15, 16). The arsenate respiratory reduction pathway in Shewanella sp. strain ANA-3 is composed of at least two proteins: ArrA, a molybdenum-containing arsenate reductase, and ArrB, a ~26-kDa Fe-S-containing subunit (1, 14). ArrA and ArrB are soluble in the periplasm (17). A membrane-associated c-type tetraheme cytochrome, CymA, is also required for arsenate respiration in several arrAB-containing Shewanella strains (20).
Regulation of arsenate reduction pathways in metal reducers is not well understood. In a previous report with Shewanella sp. strain ANA-3, several key environmental conditions that strongly affected the expression of the respiratory (arrA) and detoxifying (arsC) arsenate reductase genes were identified (28). Oxygen and nitrate negatively affected arrA expression. However, under anaerobic conditions, arrA was induced by both arsenite and arsenate. In contrast, arsC transcription was activated only with arsenite and under a variety of aerobic and anaerobic growth conditions. Moreover, regulation of arr requires adenylate cyclase activity and the cyclic AMP (cAMP) receptor protein (CRP), which appears to be activated by cAMP production under anaerobic conditions. The fumarate-nitrate regulator (FNR, called EtrA in Shewanella) plays a minor role in arr regulation (19).
The arsenite-dependent regulation of Shewanella arr and ars operons has patterns similar to those for the Escherichia coli ars operons. Regulation of the E. coli ars operon is mediated by an arsenite-dependent repressor, ArsR, which in the absence of arsenite represses ars transcription (37). Derepression of the ars operon occurs in the presence of arsenite, antimonite, and phenylarsine oxide (PAO). ArsD has also been implicated as an arsenite regulator; however, more recent studies have shown that it also acts as an arsenite metallochaperone in E. coli, transporting arsenite to ArsA. Based on these observations, we hypothesized that ArsR mediated the arsenite-dependent regulation of ars and arr. Here we report on the first identification and functional characterization of an arsenic-specific regulator for arsenate reduction pathways in Shewanella sp. strain ANA-3.
All E. coli and Shewanella strains and plasmids used in the study are described in Table Table1.1. E. coli strain BL21 was a kind gift from Karen Ottemann, University of California, Santa Cruz, Santa Cruz, CA.
E. coli strains were grown in Luria-Bertani (LB) medium or 2× YT medium (29). Shewanella sp. strain ANA-3 strains were grown at 30°C in LB or minimal salts medium (referred to as TME) consisting of 1.5 g liter−1 NH4Cl, 0.6 g liter−1 NaHPO4, 0.1 g liter−1 KCl, 0.5 g liter−1 yeast extract, 10 mM HEPES, 20 mM lactate, and 10 ml liter−1 each of trace mineral and vitamin solution (20). ANA-3 aerobic liquid cultures were shaken at 250 rpm. Preparation of anaerobic media, electron acceptors, medium additions, and anaerobic culturing were done as previously described (20).
Aerobic cultures were grown overnight in TME medium. The optical densities at 600 nm (OD600) of each culture were adjusted to 0.6 and standardized to each other by the addition of medium to ensure that inoculation levels for each strain were equal. Cells were then inoculated at 1/100 dilution into anaerobic Balch tubes containing 10 ml. Aerobic growth curves were done with cultures in 250-ml flasks containing 20 ml medium and shaken at 250 rpm. Growth was monitored using a Spectronic 20D+. Control cultures were also grown and monitored in anaerobic medium without added electron acceptors. Higher concentrations of arsenite (3 mM) were used in order to observed differences in growth phenotypes in various mutants of ANA-3 relative to the wild type.
In-frame, nonpolar deletions of arsR1, arsR2, arsR3, arsR4, arsR5, arsR6, and arsD were generated using previously developed methods (20, 27) with the primers listed in Table S1 in the supplemental material. To generate stains with multiple deletions, the plasmid carrying the new deletion was transformed into the appropriate Shewanella sp. strain ANA-3 null mutants.
Complementation plasmids pArsR1 and pArsR2 were generated by cloning PCR products containing the respective gene with ~200 bp upstream into the SpeI site of pBBR1-MCS2; primers are listed in Table S1 in the supplemental material. Transformation of the complementation vectors into Shewanella sp. strain ANA-3 was performed as previously described (26). All plasmid constructs were verified by PCR, restriction mapping, and sequencing.
Mutations to the cysteine (Cys) residues at amino acid positions 30 and 32 in arsR2 in pArsR2 were generated using the following modified protocol of the QuikChange site-directed mutagenesis method (Stratagene) as previously described (20) and primers listed in Table S1 in the supplemental material. For the double mutation (C30S/C32S), pArsR2-C30S was used as the DNA template in the PCR to generate pARsR2-C30S/C32S. Plasmids were extracted and sequenced to confirm the correct mutation. The resulting plasmids were transformed into strain AN-ARSR2 as previously described.
The methods for culturing, preparing cells, RNA extraction, and quantitative reverse transcriptase PCR (RT-PCR) have been described previously (20, 28). The DNA gyrase gene, gyrB, was used as a reference for normalizing gene expression; it has been used in several past studies involving quantitative gene expression in Shewanella sp. strain ANA-3 (19, 20, 28). Lower concentrations of arsenite were used (1 mM) in order to minimize the toxic effects of arsenite, such as slow and delayed growth, which could introduce artifacts on gene expression quantification.
A bacterial expression vector for arsR2 was generated by fusing the N-terminal domain of ArsR2 to glutathione S-transferase (GST). The crp expression plasmid and purification of CRP were previously described (19). The arsR2 gene was amplified using the primers arsR2-BamHI-F1 (5′-GGG GAT CCA TGA TTA ATC CTA CGC AAT-3′) and arsR2-BamHI-R1 (5′-GGG GAT CCT CAT TGT TCA CTA CAA CAT T-3′) (BamHI sites are underlined). The BamHI-digested PCR product was cloned into pGEX-6P-2 (GE Healthcare) to create pGEX-arsR2. The orientation and correct sequence were verified through DNA sequencing. Induction of ArsR2-GST, purification of the GST-fused protein, removal of GST, and sodium dodecyl sulfate-polyacrylamide gel electrophoresis were carried out as described by Murphy et al. (19). The purity of the ArsR2 preparation was determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Aliquots of the protein were stored in 20% glycerol at −80°C.
Cy3-labeled DNA probes were generated using primers arsRbind-F1* (5′-[Cy3] GTT GAC TTG ATT CTC TTT CT-3′), arsRbind-F1 (5′-GTT GAC TTG ATT CTC TTT CT-3′), and arsRbind-R2 (5′-AGC ACC ACA TAC GCC AGT A-3′). The F1R2 probe was PCR synthesized using arsRbind-F1* and arsRbind-R2; the probe was purified and concentrated using the QIAquick gel extraction kit according to the manufacturer's PCR purification instructions. Nonlabeled competitor DNA was created using the same method, using arsRbind-F1 and arsRbind-R2. Electrophoretic mobility shift assay (EMSA) was performed in a final volume of 20 μl containing 10 mM Tris-Cl (pH 7.6), 0.5 mM Na2EDTA, pH 8.0, 10% glycerol, 100 mM KCl, 50 μg/ml bovine serum albumin, 50 μg/ml poly(dI/dC), 0.1 pmol DNA probe, and the indicated amounts of purified protein. cAMP (1 μM) was added to the reaction mixture where indicated, as well as various concentrations of arsenate, arsenite, PAO, and competitor DNA. The mixtures were incubated on ice for 30 min and loaded on a nondenaturing 5% polyacrylamide gel. Electrophoresis was carried out at 4°C and 95V in a 0.5 Tris-borate-EDTA buffer. Gels were photographed on a Typhoon variable-mode fluorescence imager (Amersham Biosciences).
Previously, we showed that the arsenate respiratory (arrA) and arsenate detoxifying (arsC) reductases were induced in the presence of arsenite under anaerobic conditions (i.e., growth on fumarate in the presence of arsenite) (28). Because the only characterized arsenite-sensing regulators are within the ArsR/SmtB/CadC family of proteins (3, 6), we hypothesized that the arsenite-dependent induction of arrA and arsC in Shewanella sp. strain ANA-3 was mediated by an ArsR. Neither the ars nor the arr operon of Shewanella sp. strain ANA-3 contains a gene encoding an ArsR. Analysis of the Shewanella sp. strain ANA-3 genome sequence revealed six genes encoding putative proteins with ArsR-like features, such as a helix-turn-helix motif and several conserved cysteine residues known to bind arsenite and other metals. For ease of reference, these are designated arsR1, arsR2, arsR3, arsR4, arsR5, and arsR6. Of the six, three (arsR1, arsR2, and arsR3) are located in a 23-kb region (Fig. (Fig.1)1) that is highly conserved among several arsenate-respiring Shewanella strains (CN-32 and W3-18-1) that also contain arrAB and arsDABC operons. Two of these arsR-like genes, arsR1 (Shewana3_2337) and arsR2 (Shewana3_2335), cluster with two other genes, arsC1 and arsC2, in the arrangement arsR1C1R2C2. Analysis using Pfam (http://pfam.sanger.ac.uk/) (9) showed that the putative proteins encoded by arsC1 and arsC2 are part of the low-molecular-weight protein tyrosine phosphatase (PTP) family of proteins, which also include ArsCs. The eukaryotic arsenate reductase of Saccharomyces cerevisiae Acr2p (Arr2p) is also a similar part of the PTP superfamily (18). arsC1 and arsC2 may not encode arsenate reductases, because a strain lacking arsC (from the arsDABC cluster) and arrA does not reduce arsenate (28, 32). The third arsR-like gene (arsR3 [Shewana3_2351]) is also in the “arsenic island” and is the first gene within a cluster of six that includes a gene encoding a putative acriflavin resistance protein. The other three arsR-like genes are distributed throughout the genome and in highly conserved loci of other Shewanella genomes. For example, arsR4 (Shewana3_0029, HTH_5 family) is located next to a gene encoding HemG, a protoporphyrinogen oxidase. arsR6 (Shewana3_3141) is located next to coding sequences for a peptidase and a homolog of MurJ (also called MviN), which is involved in peptidoglycan biosynthesis (24). Lastly, Shewana3_0532 (ArsR5) is the first gene within a cluster that encodes various efflux pumps and permeases possibly contributing to arsenic resistance.
We were interested in determining whether these arsR-like genes played functional roles in growth on arsenate, arsenic detoxification, and regulation of arsenic metabolism. Null mutations were generated for each of these genes. However, none were essential to anaerobic growth on arsenate; this outcome would be expected if these genes encode repressors. A strain lacking all six arsR-like genes could still grow anaerobically on arsenate (data not shown).
Quantitative transcription profiling of the six arsR-like genes was done to determine if expression of one or more of these arsR-like genes was arsenic dependent (Fig. (Fig.2).2). The expression of arsR1 and arsR4 was either low or not responsive to arsenic, respectively. In contrast, arsR3, arsR5, and arsR6 appeared to be increased either under anaerobic conditions and/or with arsenic. One of the arsR-like genes, arsR2, was highly expressed in cells grown with arsenic (arsenite and arsenate). Anaerobic expression with arsenate was nearly 10-fold higher than in cells grown under aerobic conditions. Using the arsenate reduction mutant (ARM1 ) of Shewanella sp. strain ANA-3, we observed that arsR2 was expressed only in the presence of arsenite and not in cultures amended with arsenate and another electron acceptor such as fumarate (data not shown). Because of its high expression in cells grown with arsenic (namely, arsenite), we focused our investigations on determining the functional role of arsR2 in arsenate respiration, arsenite resistance, and regulation of arr and ars operons. We also included arsR1 in some of our analyses because of its close proximity to arsR2.
We first examined the effects of deleting arsR2 (and arsR1) on anaerobic growth either with arsenate or in the presence of arsenite and other electron acceptors (i.e., oxygen and fumarate). The arsR2 null mutant exhibited a growth phenotype similar to that of the wild-type strain with arsenate as the terminal electron acceptor (Fig. (Fig.3A).3A). In comparison to wild-type and ΔarsR1 Shewanella sp. strain ANA-3, the lag phase of growth in the presence of high arsenite levels (3 mM) for the arsR2 null mutant (called AN-ARSR2) was diminished by ~2 to 3 h (Fig. (Fig.3B).3B). This larger amount of arsenite was required to observe a growth phenotype in the mutant versus wild-type strains. The ΔarsR1 strain had a slightly enhanced growth phenotype relative to the wild type (Fig. (Fig.3B).3B). The growth curves for each strain were indistinguishable when each strain was grown in the absence of arsenic but with fumarate as the electron acceptor (Fig. (Fig.3C).3C). Complementation of arsR2 in AN-ARSR2 and of arsR1 in AN-ARSR1 restored growth to wild-type levels (Fig. (Fig.3D).3D). Based on the growth observations, arsR2 and arsR1 most likely function as a repressor, and their absence should lead to elevated expression of the arsenate detoxification and respiration genes.
To test our hypothesis that arsR1 and arsR2 function as a repressor for the arr and ars operons, we investigated arrA and arsC gene expression in AN-ARSR1, AN-ARSR2, and wild-type ANA-3 grown under different conditions. Figure Figure44 shows the expression pattern for arrA and arsC in AN-ARSR1, AN-ARSR2, and wild-type ANA-3 when grown anaerobically and with vigorous aeration in atmospheric air. The expression patterns of arrA and arsC under aerobic (i.e., with oxygen as the sole electron acceptor), fumarate, and arsenate growth conditions were similar for AN-ARSR1 and wild-type ANA-3. In these two strains, arsenic exposure and anaerobic conditions resulted in the greatest expression of arrA and arsC. However, in fumarate-grown cultures, arrA and arsC expression was nearly 10-fold higher in AN-ARSR2 than in the wild type and AN-ARSR1. It is unclear what role arsR1 plays in regulating arrA and arsC, because its effects on transcription were similar to those in wild-type Shewanella sp. strain ANA-3. Based on the transcription patterns in the arsR2 mutant strain, we conclude that arsR2 should encode an arsenic-dependent repressor that coregulates the arr and ars operons.
Previous reports by Rosen and colleagues (8, 36) showed that ArsD, like ArsR, functioned as an arsenic-dependent regulator of the E. coli ars operon. In recent years, they also showed that ArsD functioned as a metallochaperone that is involved in arsenite resistance (16). Because of its close proximity to the ars and arr operons in ANA-3, we were interested in determining the physiological role of arsD with respect to anaerobic growth with arsenic. An ANA-3 mutant lacking arsD (AN-ARSD1) exhibited an anaerobic growth response in the absence of arsenic (i.e., growth on fumarate) that was nearly identical to that of the wild-type ANA-3 strain (Fig. (Fig.5A).5A). For anaerobic growth conditions with arsenate as the terminal electron acceptor, we observed very similar growth responses for the ΔarsD strain and the wild-type and ΔarsR2 strains (Fig. (Fig.5B).5B). When grown anaerobically with fumarate amended with high arsenite levels (3 mM), growth of the ΔarsD strain was significantly delayed compared to that of the wild-type and ΔarsR2 strains (Fig. (Fig.5C).5C). The ΔarsD strain had an extended lag in growth, about 7 h longer than the wild-type strain, which took about 9 h to achieve the mid-log phase of growth (an OD600 of ~0.1). In the double arsD arsR2 null mutant, growth on arsenite was enhanced relative to that of the arsD single mutant but was deficient relative to that of the wild-type and ΔarsR strains (Fig. (Fig.5C).5C). From these growth curve analyses, we concluded that arsD is required for wild-type levels of arsenite resistance. This outcome is consistent with the effects of arsD in E. coli. However, further work will be needed to test whether the Shewanella ArsD functions as an arsenite metallochaperone, a repressor, or both.
To verify that the protein encoded by arsR2 functions as a regulator, we investigated the DNA binding activities of the purified protein using a fluorescence-based EMSA. After overexpressing ArsR2 as a recombinant protein from E. coli, EMSA was performed with a fluorescence-labeled probe generated from the DNA region connecting arrA and arsD. Figure Figure66 shows that ArsR2 binds to the 202-bp DNA probe, F1R1, that includes 178 bp upstream of the arrA start codon. In the absence of arsenite, there was one predominant upward-shifted DNA band (Fig. (Fig.6A,6A, lane 7) relative to when no ArsR2 was included in the EMSA (Fig. (Fig.6A,6A, lane 1). Decreasing arsenite (5 to 0.1 mM) (Fig. (Fig.6A,6A, lanes 2 to 6) resulted in increased intensity of the shifted DNA probe with a concomitant decrease in intensity of the free DNA probe. A faint intermediately shifted DNA band was observed in most of lanes that contained ArsR2, which may indicate multiple ArsR2 binding sites on the DNA probe.
Other arsenicals were also tested in the ArsR2 EMSA, such as arsenate [As(V)] and PAO (Fig. (Fig.6B).6B). There was no observable effect on the downward shift of the DNA probe in EMSAs containing several arsenate concentrations (Fig. (Fig.6B,6B, lanes 3 to 5). In contrast, the gel shift was inhibited when the trivalent arsenical PAO was included in the EMSA (Fig. (Fig.6B,6B, lanes 8 to 10). These results indicated that ArsR2 is specific for trivalent arsenic oxyanions, which is consistent with the transcription patterns being specific to arsenite and not arsenate and interacting partners that donate at least two oxygen ligands such as PAO.
Several cysteine residues have been shown to be part of the arsenic binding motif in the E. coli ArsR, Cys-32 and Cys-34 (30). Certain mutations in these residues resulted in the protein retaining DNA binding ability that is not affected by the presence of arsenite. These mutations prevented the ars operon from being efficiently transcribed (30). Alignments among other ArsR proteins, including the E. coli ArsR, showed that Cys-30 and Cys-32 for ANA-3 ArsR2 are likely trivalent arsenic binding sites (Fig. (Fig.7).7). However, there are several other Cys residues in ArsR2 (Cys-10, Cys-19, Cys-55, Cys-102, Cys-114, and Cys-115) that are not present in other ArsR proteins (Fig. (Fig.7A).7A). We made amino acid substitutions C30S and C32S and a double substitution (C30S/C32S) in ArsR2 using the complementing plasmid pAN-arsR2 as a template for site-directed mutagenesis. Growth curve analyses of the Shewanella sp. strain ANA-3 ΔarsR2 mutant harboring the various mutant alleles of arsR2 on plasmids are shown in Fig. 7B and C. Growth on arsenate for arsR2-C30S was similar to that of the wild type and AN-ARSR2 containing a plasmid copy of wild-type arsR2. However, the AN-ARSR2 strain with the arsR2-C32S allele had a twofold-longer generation time (2.2 h) than the same strain with the wild-type arsR2 allele (0.95 h). In contrast to the arsR2 strain with the single C32S mutation in ArsR2, the double C30S/C32S arsR2 strain grew to an OD600 similar to that of the ΔarsR2 strain carrying a wild-type copy of arsR2. Strains with Cys-Ser mutations grown with fumarate amended with arsenite showed more pronounced physiological responses. The arsR2-C32S strain exhibited the weakest growth; the OD600 never increased above 0.05 during the 14-h incubation. The generation time of this strain (2.8 h) was almost 3 times that of the wild-type strain (1.1 h). Unlike cultures grown with arsenate, the arsR2-C30S/C32S strain had a significantly extended lag phase (6 to 8 h) in cultures containing high levels of arsenite. The strain eventually grew with a generation time similar to that of the arsR2-C30S strain (1.2 h). We conclude that C32S is required for wild-type levels of growth on arsenate and resistance to arsenite under anaerobic conditions. However, the C30S mutation can alleviate the arsenic growth defect of an arsR2-C32S strain.
The FNR-like anaerobic regulator encoded by the etrA gene, although not essential in ANA-3, was previously shown to cause moderate anaerobic growth deficiencies on arsenate (20). This is likely due to regulation by other factors, such as ArcA, cAMP-CRP, and adenylate cyclases, which all have been shown to play various roles in anaerobic respiration in Shewanella (7, 10, 11, 19, 25). We first wanted to determine if arsR2 exerted a genetic interaction with etrA mainly in its impact on anaerobic growth in the presence of arsenic through arsenate respiration or with high levels of arsenite. For arsenate growth phenotypes, the double ΔarsR2 ΔetrA mutant grew as well as the single ΔetrA mutant but not as well as the single ΔarsR2 mutant (which grew as well as the wild-type strain) (Fig. (Fig.8A).8A). Similar trends were also observed for growth with nitrate (Fig. (Fig.8B)8B) and fumarate (data not shown). When arsenite was included in cultures grown on nitrate, the ΔarsR2 ΔetrA strain grew at a level intermediate between those of the wild-type and ΔarsR2 strains (Fig. (Fig.8C).8C). The ΔarsR2 ΔetrA strain had a much shorter lag phase (~4 h) than the wild type (~8 h) but not as short as that of the arsR2 single mutant strain (~2 to 4 h) (Fig. (Fig.8C).8C). Similar trends were also observed with fumarate-grown cultures amended with arsenite (data not shown). These results are consistent with enhanced arsenite resistance in strains carrying arsR2 mutations.
In contrast to etrA, crp was previously shown to be essential for growth on arsenate and fumarate in Shewanella sp. strain ANA-3 (19). When the crp gene was deleted in a ΔarsR2 strain and the strain was grown with arsenate, the double mutant was unable to grow, which is similar to the phenotype of a Δcrp-only strain (Fig. (Fig.8A).8A). Similar trends were observed with fumarate (data not shown). This indicates that CRP is dominant over ArsR2 for the arsenate respiration pathway. The use of nitrate as a terminal electron acceptor was investigated because the Δcrp strain could still grow, although not as well as the wild-type strain (Fig. (Fig.8B).8B). In contrast, arsR2 was not required for growth on nitrate, because the growth curve for the ΔarsR2 strain was indistinguishable from that of the wild type. When the ΔarsR2 Δcrp double mutant was grown in nitrate medium amended with high levels of arsenite (3 mM), no obvious growth advantage was observed in comparison to the ΔarsR2 strain (Fig. (Fig.8C).8C). After 12 h of growth with arsenite, cultures of the Δcrp and ΔarsR2/Δcrp strains had OD600 values 12% and 38% of that of the ΔarsR2 single-deletion strain (Fig. (Fig.8C).8C). From these observations it appears that the absence of the ArsR2 repressor provides a significant growth advantage at extremely high arsenite concentrations (3 mM). However, in a crp mutant background this enhanced growth conferred by ΔarsR2 was diminished in cultures grown on certain electron acceptors (e.g., arsenate and fumarate).
Because cAMP-CRP and ArsR2 appear to be major regulators of arsenic metabolism in ANA-3, we were interested in determining the DNA binding dynamics of the two regulators for the arr promoter region. In vitro, both regulators should be able to bind simultaneously to the same arr promoter region under certain conditions: with elevated cAMP and without arsenite. To investigate this prediction, EMSA was done with CRP, ArsR2, and the F1R1 DNA probe under various incubation conditions that are known to affect how the regulators individually interact within the arr promoter region. Although ArsR2 and cAMP-CRP can bind independently of each other (Fig. (Fig.9,9, lanes 4 to 10) neither cAMP nor arsenite, respectively, affected binding to the F1R1 DNA probe (Fig. (Fig.9,9, lanes 5 and 6 for ArsR and lanes 9 and 10 for CRP). A slightly supershifted band was seen in an EMSA containing CRP, ArsR2, and cAMP, indicating that both proteins can bind simultaneously to the arr promoter region (Fig. (Fig.9,9, lanes 11). Excluding cAMP from the EMSA with ArsR2 and CRP resulted in a shifted DNA band that resembled the ArsR2-bound probe (Fig. (Fig.9,9, lane 12). Similarly, when arsenite was included in an EMSA with ArsR2 and cAMP-CRP, a smaller shifted band was observed (Fig. (Fig.8,8, lane 13), which most likely corresponded to a shift with only cAMP-CRP bound to the probe. From these observations, we propose that in vivo ArsR2 should repress arr transcription under non-arsenate-reducing conditions, even when CRP is activated by cAMP and bound to the operator in vivo.
The objective of this work was to determine how the arsenate respiration and detoxification pathways are regulated in response to arsenite in the metal-reducing bacterium Shewanella sp. strain ANA-3. Here we show for the first time that an ArsR regulates the arsenic-specific expression of the arr operon. In a previous study, we showed that the arr system was expressed mainly under anaerobic conditions and further activated by arsenite (28). Moreover, it was observed that the ars system was expressed both aerobically and anaerobically, requiring the presence of arsenite. The simplest explanation for the arsenite-dependent expression of arr and ars systems is that an ArsR-type repressor would regulate arr and ars operons. ArsRs are known to bind and release DNA in the absence or presence of arsenite. We identified at least one arsenite-dependent regulator, ArsR2, and found that the gene was not required for anaerobic growth with arsenate and also growth in the presence of elevated arsenite. In fact, the arsR2 null strain grew better in the presence of elevated arsenite levels (Fig. 3B and D). This outcome was predictable because transcription of arrA and arsC was elevated in the arsR2 null mutant (Fig. (Fig.4).4). From these results we concluded that ArsR2 is a repressor. EMSA further showed that ArsR2 could bind the arr promoter region and that binding was arsenite dependent.
The arsenite-dependent expression of the Shewanella arr operon shows regulatory characteristics similar to those of arsenic detoxification operons (e.g., ars genes) in other non-arsenate-respiring prokaryotes such as E. coli (35), Staphylococcus (16), Pseudomonas (5), Corynebacterium (21), and Acidithiobacillus (4) and the archeaon Halobacterium sp. strain NRC-1 (33). In these microbes, an ArsR mediates the arsenite-dependent regulation of arsenic detoxification. Several unifying features of ArsR family proteins include their predicted size (12 to 16 kDa), helix-turn-helix domains, and several conserved cysteine residues (3). In general, ArsR represses transcription in the absence of arsenite by binding near the ars promoter region. In vivo, repression is usually alleviated in the presence of arsenite and antimonite but not arsenate. Moreover, ars operons of most characterized prokaryotes contain a single arsR gene. In Shewanella sp. strain ANA-3, neither the ars nor the arr operon contained an arsR. Formerly identified as a repressor in E. coli (8, 36), a gene encoding ArsD is present in the Shewanella sp. strain ANA-3 ars operon. When arsD was mutated, the strain grew slower in the presence of arsenite than the wild-type. This phenotype is consistent with enhanced arsenite resistance in E. coli complemented with a plasmid carrying arsDAB relative to arsAB (15). It was also shown that ArsD functions to chaperone arsenite to a membrane-bound ATP-dependent arsenite efflux pump, ArsAB (16). Further work will be needed to identify the functional role of ArsD in arsenic metabolism in Shewanella sp. strain ANA-3.
It appears that Shewanella genomes have several genes annotated as arsR. In Shewanella sp. strain ANA-3, the genome contains six open reading frames annotated as arsR-like. Two of these arsR-like genes (referred to as arsR1 and arsR2) are located within the “arsenic island” (Fig. (Fig.1).1). The phylogenetic analysis of ArsR-like proteins identified in the genomes of 14 other Shewanella strains showed that ArsR1, ArsR2, and ArsR3 were present mainly in the arrAB-containing Shewanella strains: ANA-3, CN-32, W3-18-1, and 200. The genomic locations of the other arsR-like genes (arsR4, arsR5, and arsR6) are highly conserved among many sequenced Shewanella genomes. Moreover, the predicted proteins encoded by the genes flanking these arsR-like genes have little similarity to ars or arr genes. The functional roles of these ArsR-like proteins are not known, but they might be involved in sensing and responding to metals such as Zn, Cd, and/or Co (3). Compared to the other arsR-like genes in Shewanella sp. strain ANA-3, the transcriptional profile for arsR2 is highly expressed in the presence of arsenite and in cells grown on arsenate, which is unlike the case for the neighboring ArsR1. Questions remain as to the mechanism for this high level of expression. It is possible that ArsR2 may regulate itself.
Several cysteine residues in other ArsRs are known to mediate the interactions with trivalent arsenic and antimony oxyanions. In the E. coli ArsR, mutations in these residues (Cys-32 and Cys-34) render the protein insensitive to inducers while preserving the protein's ability to bind DNA (30). These mutated forms of ArsR caused repression of ars transcription in E. coli in the presence of arsenite. These past studies have focused mainly on ArsR DNA binding properties and not physiological outcomes of the mutations. We extended these studies by examining how mutations of two conserved Cys residues in the Shewanella ArsR2 would affect anaerobic growth with arsenate as a terminal electron acceptor and also anaerobic growth in the presence of high arsenite levels. It appeared that Cys-30 was less critical for growth on arsenate, unlike Cys-32, which in arsR2-C32S led to growth inhibition on arsenate. The arsR2-C32S mutant was even more deficient in anaerobic growth in the presence of high arsenite levels. An unexpected result was that when the C30S was introduced into arsR2-C32S (to make arsR2-C30S/C32S), growth was nearly restored to wild-type levels, albeit with an 8-hour increase in the lag phase. Although the protein levels could be different for each ArsR2 variant, we predict that these delayed growth effects are due to the loss of arsenite sensing in mutant forms of ArsR2, which may have led to persistent repression of arr and ars transcription. Work done by Shi et al. (30) provided evidence that the E. coli ArsR with C32G, C34G, and C32G/C34G alterations could still bind the E. coli ars promoter element and that binding was insensitive to inducers. In the Corynebacterium ArsR, the CXXC (X = any amino acid) motif is not conserved. Instead, arsenite binding occurs through inter-ArsR thiol sharing, with one arsenite molecule holding two ArsR molecules together to form a homodimer. These critical Cys residues are located near the N terminus of the protein. The Shewanella ArsR2 also has several Cys residues near the N terminus (Cys-10 and Cys-19) and C terminus (Cys-102, Cys-114, and Cys-115) that might be involved in the shared binding of one molecule of arsenite per two ArsR molecules. Further work will be necessary to determine how these ArsR2 variants interact with arsenite and DNA.
In our current study, it appears that ArsR2 is a specific regulator for arsenite. ArsR2 likely is not an aerobic-anaerobic sensing regulator but would work in concert with other regulators to control expression of arr and ars operons. Under aerobic conditions and in the absence of arsenite, arrA and arsC expression appears to be repressed. However, in the arsR2 null strain, only arsC expression increased under aerobic conditions without arsenite. Under anaerobic conditions in the absence of arsenic, arrA and arsC gene expression was dramatically increased in the ΔarsR2 strain. These observations provide evidence that other regulators are required for anaerobic arr and ars activation, such as the anaerobic sensor FNR homolog EtrA (19) and cAMP-CRP (19, 25). Our studies with arsR2 were further extended by growth curve analyses with arsR2 etrA and arsR2 crp double mutants cultured with arsenate, nitrate, or nitrate amended with a high level of arsenite. Our results showed possible genetic interactions of arsR2 with etrA and crp, but only at high arsenite concentrations (3 mM). At this high arsenite concentration, the arsR2 etrA and arsR2 crp double mutants did not grow as well as the arsR2 single mutant but grew better than etrA or crp single mutants. The differences in growth are likely due to (i) arsenic-independent regulatory effects of EtrA and CRP, (ii) increased expression of the ars operon (and arr) in strains without arsR2, and (iii) the different modes of detoxifying arsenate and arsenite. For the last point, arsenate could be excluded from the cell by decreasing inorganic phosphate transporters (Pit system), which can also transport arsenate (34). This would not induce the ars operon, which requires the presence of arsenite in the cytoplasm formed from arsenate reduction or supplied exogenously in the medium.
In addition to etrA, growth on arsenate requires crp and two adenylate cyclase genes (cya) (19). Deleting arsR2 could not rescue growth on arsenate or with high arsenite levels for the Δcrp strain. We conclude that cAMP-CRP is a dominant regulator for the arr operon. It was shown that cAMP-dependent binding of CRP occurred within a 92- to 178-bp region upstream of the arrA start codon (19). Using this same DNA probe, we showed that both ArsR2 and cAMP-CRP can bind simultaneously (Fig. (Fig.9),9), indicating that the transcription factor binding sites do not overlap. Because the binding reactions were performed under aerobic conditions, ArsR and CRP are not likely direct sensors of redox conditions. From these EMSA results, we predict that full activation of the arr operon would require at least the release of ArsR2 and the binding of cAMP-CRP to the arr/ars intergenic region. In vivo, this would require the presence of arsenite and cAMP. Arsenite would be generated by either ArrA or ArsC and cAMP by the activities of one or more adenylate cyclases. However, additional transcription factors, (e.g., EtrA), might also be necessary to fully express the arr operon during anaerobic growth with arsenate.
Although it is not surprising that in Shewanella an ArsR would mediate the arsenite-dependent regulation of the arsenate respiration pathway, we are intrigued by how this regulatory pathway evolved. The gene synteny for arrAB and arsDABC operons within the “arsenic island” is nearly identical in strains ANA-3, CN-32, W3-18-1, and 200. The gene cluster containing arsR2 is also present nearby the arrAB genes and is well conserved among ANA-3, CN-32, W3-18-1, and 200. Several phage-like genes also exist within this genomic island. Other non-Shewanella arsenate-respiring and arrAB-containing bacteria also contain ars-like genes in close proximity to their arr operons. For example, Geobacter lovleyi and Wolinella succinogenes genome sequences have an arsR-like gene near the core arrAB of the arr operon (reviewed in reference 23). In addition to arsR, some arr operons also have genes predicted to encode two-component sensory-histidine kinases and response regulators. It is clear from genome sequencing projects that there will be several mechanisms for regulating arsenate respiratory reduction in bacteria.
One intriguing question has emerged: why would arr gene expression be regulated in response to arsenite when arsenate is the substrate? The answer may be related to the dilemma of distinguishing phosphate from arsenate; the latter is poisonous to many biological pathways involving phosphate (12). Sensing arsenate may be complicated compared to sensing arsenite because of the chemical similarities of arsenate to phosphate. It is also known that the ars operon provides a growth advantage to Shewanella when respiring arsenate (26). There is likely to be a selective pressure to coexpress the ars arsenite-detoxifying genes during growth with arsenate because of the accumulation of toxic levels of arsenite during the later phases of growth. This would require coregulation of arr and ars transcription; the likely coregulator would be an arsenite-sensing regulator like ArsR.
In conclusion, ArsR and cAMP-CRP (and possibly EtrA) coregulate arsenate metabolism in Shewanella. Similarly to nitrate and iron metabolism, for example, regulation of arr operons in facultative anaerobes is predicted to be complex, because multiple signals, such as phosphate, oxygen, nitrate, cAMP, arsenite, and arsenate, are known to affect arr transcription. To date, we have identified likely candidate regulators for these signals in the genome of Shewanella. However, the arsenate sensor remains elusive.
This work was supported by National Science Foundation grant EAR-0535392 and the Cancer Research Coordinating Committee of the University of California.
Published ahead of print on 28 August 2009.
†Supplemental material for this article may be found at http://jb.asm.org/.