Alternate energy coupling of ArsB, the membrane subunit of the Ars anion-translocating ATPase.

The arsenical resistance (ars) operon of the conjugative R-factor R773 confers resistance to arsenical and antimonial compounds in Escherichia coli, where resistance results from active extrusion of arsenite catalyzed by the products of the arsA and arsB genes. Previous in vivo studies on the energetics of arsenite extrusion showed that expression of both genes produced an ATP-coupled arsenite extrusion system that was independent of the electrochemical proton gradient. In contrast, in cells expressing only the arsB gene, arsenite extrusion was coupled to electrochemical energy and independent of ATP, suggesting that the Ars transport system exhibits a dual mode of energy coupling depending on the subunit composition. In vitro the ArsA-ArsB complex has been shown to catalyze ATP-coupled uptake of 73AsO2(-1) in everted membrane vesicles. However, transport catalyzed by ArsB alone has not previously been observed in vitro. In this study we demonstrate everted membrane vesicles prepared from cells expressing only arsB exhibit uptake of 73AsO2(-1) coupled to electrochemical energy.

The arsenical resistance (ars) operon of the conjugative R-factor R773 confers resistance to arsenical and antimonial compounds in Escherichia coli, where resistance results from active extrusion of arsenite catalyzed by the products of the arsA and arsB genes. Previous in vivo studies on the energetics of arsenite extrusion showed that expression of both genes produced an ATPcoupled arsenite extrusion system that was independent of the electrochemical proton gradient. In contrast, in cells expressing only the arsB gene, arsenite extrusion was coupled to electrochemical energy and independent of ATP, suggesting that the Ars transport system exhibits a dual mode of energy coupling depending on the subunit composition. In vitro the ArsA-ArsB complex has been shown to catalyze ATP-coupled uptake of 73 AsO 2 ؊1 in everted membrane vesicles. However, transport catalyzed by ArsB alone has not previously been observed in vitro. In this study we demonstrate everted membrane vesicles prepared from cells expressing only arsB exhibit uptake of 73 AsO 2 ؊1 coupled to electrochemical energy.
Resistance to arsenical and antimonial compounds in bacterial cells is mediated by active extrusion of oxyanions of As(III) or Sb(III) from the cells (1)(2)(3). These efflux systems are encoded by ars operons. The ars operon of the conjugative R-factor R773 of Escherichia coli has been shown to encode an anion-translocating ATPase composed of two types of subunits, ArsA and ArsB. ArsA, the catalytic subunit, is a 63-kDa As(III)/Sb(III)stimulated ATPase (4). ArsB is a 45-kDa inner membrane protein that catalyzes oxyanion translocation (5). The ArsA-ArsB complex has been shown both in vivo (6 -8) and in vitro to be an obligatorily ATP-coupled primary pump (9).
However, recent results from several laboratories have shown that an arsB gene in the absence of an arsA gene is sufficient for resistance. While the ars operons of plasmids R773 (10) and R46 (11) have arsA genes, the homologous operons of the staphylococcal plasmids pI258 (12) and pSX267 (13) and the chromosomal ars operon of E. coli (14,15) do not contain an arsA gene. Cells expressing the pI258 and E. coli chromosomal ars operons extrude arsenite (15,16). There are at least two possible interpretations of these results. First, there could be a chromosomal arsA gene or homologue. Although this possibility cannot be ruled out, there are no data supporting it. Second, those ArsB proteins could function independently of ArsA, perhaps as a secondary carrier protein. The chromosomally encoded ArsB is 79% identical and overall 90% similar to the R773 ArsB. It seems intuitively unlikely that such close homologues would catalyze different reactions. When the R773 arsB gene was expressed in the absence of arsA, it conferred arsenical resistance and active extrusion (8). These findings suggest that the ArsB protein alone can catalyze energy-dependent efflux in the absence of a catalytic subunit.
For these reasons the energetics of ArsB-catalyzed efflux was compared with that of the ArsAB pump (8). In cells expressing arsB alone, arsenite transport was coupled only to electrochemical energy, not chemical energy, suggesting that a chromosomally encoded ArsA protein is probably not involved. Interestingly, the transmembrane structure of the R773 ArsB is topologically identical to secondary membrane carriers, with 12 membrane spanning segments (5). From the aggregate of these results, we postulated that ArsB functions as a secondary arsenite transporter in the absence of an ArsA subunit, a novel dual mechanism of energy coupling of a transport system (8).
However, all of the data suggesting a role of ArsB in secondary anion translocation were from physiological studies. In this study, we provide the first direct in vitro evidence that ArsB catalyzes electrophoretic anion transport. We have constructed an arsB expression plasmid that has enabled us to measure arsenite transport activity in everted membrane vesicles. NADH respiration provided electrochemical energy for ArsBmediated transport in vesicles. Transport was sensitive to the addition of uncouplers and depolarizing permeant anions, indicating that in these everted vesicles anion transport is coupled to an electrochemical proton gradient, positive interior.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes and nucleic acid modifying enzymes were purchased from Life Technologies, Inc. and Promega. Oligonucleotides were synthesized in the Macromolecular Core Facility of Wayne State University School of Medicine. Carrier-free 73 AsO 4 3Ϫ was obtained from Los Alamos National Laboratories. All other chemicals were obtained from commercial sources.
DNA Manipulations-Plasmid DNA propagation, restriction enzyme treatment, ligation, and transformation were performed by minor mod-ifications of published procedures (17). Plasmid DNA was isolated with QIAGEN Plasmid kit (QIAGEN) for sequence analysis by the method of Sanger et al. (18). Oligonucleotide-directed mutagenesis was performed using the Altered Sites TM II in vitro Mutagenesis System (Promega) according to the manufacturer's directions. All mutations introduced were confirmed by sequencing using a Cy5-labeled sequencing kit and ALFexpress system from Pharmacia Biotech Inc.
Construction of an arsB Expression Plasmid-Plasmid pBC101 was digested with EcoRI and KpnI, and a 2.9-kilobase fragment containing arsB was cloned into vector plasmid pALTER-1 (Promega), yielding plasmid pAM-EK. The BamHI site in the arsA gene in pAM-EK was inactivated by Klenow fragment treatment. The resulting plasmid pAMB1 was used for site-directed mutagenesis. Single-stranded DNA prepared from E. coli strain JM109 containing pAMB1 was used as a template. Multiround mutagenesis was done to introduce EcoRI and HindIII sites immediately after the arsA and arsB genes, respectively. The two mutagenic oligonucleotides used and the respective changes (underlined) introduced were as follows. To introduce an EcoRI site: 5Ј-GCT GGG TAA ATT AAT GAA TTC ACG TAG GGC AGC-3Ј; to introduce a HindIII site: 5Ј-CTG TCA CAT TGT AAT AAG CTT CTG ATA TGA GCA AC-3Ј.
The resulting plasmid pAMB11 was digested with EcoRI and HindIII and religated to construct plasmid pAMBT1 that contains only arsB and its Shine-Dalgarno sequence. The EcoRI-HindIII fragment containing arsB was cloned into EcoRI-HindIII-digested vector plasmid pKK223-3 (Pharmacia), yielding plasmid pKMB1.
Transport Assays-Everted membrane vesicles were prepared essentially as described previously (19). Membrane vesicles were prepared fresh for transport assays; storage at Ϫ80°C resulted in loss of activity. Transport assays were performed at room temperature. Unless otherwise noted, the reaction mixture contained 5 mM NADH, 0.1 mM sodium 73 AsO 2 Ϫ , and 0.9 mg of membrane protein in a final volume of 0.6 ml of a buffer consisting of 75 mM HEPES-KOH, pH 7.5, containing 0.15 M K 2 SO 4 , 0.25 M sucrose, and 2.5 mM MgSO 4 . Unless otherwise noted, the reaction was initiated by addition of NADH. At intervals, 0.1-ml samples were withdrawn, filtered through nitrocellulose filters (0.22 m pore size, Corning Costar), and washed with 5 ml of the same buffer. The filters were dried, and the radioactivity was quantified in a liquid scintillation counter. A blank value, obtained by filtering 0.1 ml of assay mixture without membrane vesicles, was subtracted from all points. To determine the initial rates, 30-s time points were used.
Detection of ArsB in Everted Membrane Vesicles-Because production of ArsB-specific antibodies has not been successful, an alternate approach to detect production of ArsB was used in which the arsC gene was fused in frame to the 3Ј-OH end of arsB. To construct the arsBC fusion, the intergenic sequence between arsB and arsC was deleted using the unique-site elimination method of Deng and Nickoloff (20) using single-stranded DNA from plasmid pJHW101 (arsRDABC) (5) as template. The termination codon of arsB and the intergenic region between arsB and arsC were deleted using a primer with the sequence: 5Ј-ATA AAT AGT GAT GTT GCT CAT CAA TGT GAC AGA GAG ACG TAG CGC GAG CGC GGC CAG 3Ј.
This primer also eliminated the BglI site near the 3Ј-OH end of arsB. A second primer, designed to delete the unique EcoRI site in plasmid pJHW101, had the sequence: 5Ј AAC TTG GAG TTC CCC TGT AAT 3Ј.
In both primers, the alteration eliminating the restriction site is underlined. Cells of E. coli strain BMH71-18mutS were transformed with the mutagenesis reaction mixture, and the cell suspension was inoculated into LB medium containing kanamycin and grown overnight. Plasmid DNA was prepared and digested by EcoRI, and the product was transformed into JM109. Plasmid DNA was prepared and screened for loss of the EcoRI and BglI sites. The arsBC fusion was confirmed by DNA sequencing. The resulting plasmid, pAO-BC, was used to clone the arsBC gene fusion into plasmid pKMB1. A Csp45I-HindIII fragment from plasmid pAO-BC was inserted into Csp45I-HindIII-digested pKMB1, yielding plasmid pKMO-BC.
The ArsBC chimeras were detected by immunoblotting of membrane protein separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). Membrane protein was suspended in SDS sample buffer and heated at 80°C for 10 min, following which the proteins were separated by SDS-PAGE on 15% polyacrylamide gels (21). Proteins were electrophoretically transferred to a nitrocellulose membrane (0.2 m pore size, Schleicher & Schuell) overnight at 25 V and 4°C. Antisera raised against purified ArsC was used to detect the ArsBC chimera. Immunoblotting was performed utilizing an enhanced chemiluminescence assay (DuPont NEN) and exposed to x-ray film at room temperature, as described previously (22).
Other Methods-Protein content was determined by a micromodification of the procedure of Lowry et al. (23) using bovine serum albumin as a standard. 73 AsO 2 Ϫ was prepared by reduction of 73 AsO 4 3Ϫ (24).

Construction of an arsB Expression
Plasmid-ATP-driven 73 AsO 2 Ϫ transport has been measured in membrane vesicles from cells expressing the arsA and arsB genes from the native Dual Energetics of ArsB ars promoter. When the arsA gene was deleted from the R773 ars operon, the cells retained a low level resistance to arsenite (8), similar to that conferred by the chromosomal ars operon (15), which also lacks an arsA gene (14). However, no arsenite transport was observed in membrane vesicles from cells expressing arsB from the ars promoter regardless of the source of energy (data not shown). Previous attempts to express the R773 arsB gene at increased levels have been unsuccessful.
The reasons for this are obscure, but possible explanations include instability of the polycistronic mRNA (25) and lethality of ArsB itself (26). In an attempt to increase expression, arsB was cloned behind the tac promoter of plasmid pKK223-3, producing plasmid pKMB1. This plasmid was transformed into three strains of E. coli in which the chromosomal ars operon had been disrupted: strains AW10, AW3110, and LE392⌬uncIC⌬ars. However, only strain AW10 could be stably transformed, perhaps because, among the three strains, only AW10 contains a lacI q gene to control basal level of expression of arsB. Even in the absence of exogenously added IPTG, expression of arsB from pKMB1 produced low level arsenite resistance similar to that conferred by the chromosomal ars operon, and higher level resistance was observed when arsA was expressed in trans (Fig. 1). Although it is difficult to quantify the amount of ArsB due to the lack of a specific ArsB antiserum, relative amounts of ArsB produced from the ars promoter could be compared with those from the tac promoter in pKMB1. In both plasmids, the arsB gene was fused in frame to the downstream arsC gene, and the chimeras were detected with antiserum directed against ArsC (Fig. 2). The arsBC gene fusion produced arsenite resistance comparable to the wild type arsB gene (data not shown). In neither case could the ArsBC chimera be visualized by Coomassie staining of the gels. Even using very sensitive chemiluminescent methods, the ArsBC chimera produced from the ars promoter could not be detected (Fig. 2, lane 2). However, under control of the tac promoter, the ArsBC chimera could be detected immunologically (Fig. 2, lanes 3 and 4). ArsB was produced even in the absence of inducer, with a severalfold increase following IPTG induction. As shown above, phenotypic expresssion of arsenite resistance from the tac promoter likewise did not require induction (Fig. 1). 73 AsO 2 Ϫ Transport in Everted Membrane Vesicles-The results of the in vivo study of energetics of transport suggested that ArsB alone could catalyze arsenite extrusion coupled to electrochemical energy (8). To test this hypothesis, 73 AsO 2 Ϫ uptake was measured in everted membrane vesicles prepared from cells of E. coli strain AW10 bearing plasmid pKMB1. In these experiments, the cells were induced with IPTG to maximize expression of ArsB. The membrane vesicles exhibited time-and NADH-dependent 73 AsO 2 Ϫ accumulation (Fig. 3). Membrane vesicles prepared from cells harboring vector plasmid pKK223-3 showed no transport. Transport required NADH oxidation, as shown by the complete inhibition by KCN. The uncoupler carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) completely reversed NADH-dependent 73 AsO 2 Ϫ uptake (Fig. 3), as did the combination of valinomycin plus nigericin (Table II). These results clearly show that arsenite transport catalyzed by ArsB alone is coupled to the electrochemical proton gradient established by NADH respiration. In contrast, little transport was observed with ATP (Table II).

Effect of Permeant Anions and Weak Base on 73 AsO 2
Ϫ Transport-The effect of permeant anions and a permeant weak base was examined (Table II). In these experiments, the vesicles were prepared in a sulfate-containing buffer. Under such conditions, the electrochemical gradient has been shown to be primarily in the form of a membrane potential, positive interior, with little or no pH gradient (27). NH 4 ϩ , which dissipates the remaining ⌬pH, had a small effect on the initial rate of 73 AsO 2 Ϫ accumulation. NH 4 Cl was more inhibitory, attributable to an effect of Cl Ϫ as a permeant anion. The effect of SCN Ϫ and ClO 4 Ϫ , anions even more permeant than Cl Ϫ , considerably reduced transport activity.
Concentration Dependence for Arsenite-The concentration dependence for arsenite exhibited saturation kinetics, with an apparent K m of 0.14 mM (Fig. 4). This is essentially identical with the apparent K m of 0.1 mM for the ArsA-ArsB pump (9), suggesting that the mechanism of transport by ArsB is independent of the source of energy.
Effect of pH and Oxyanions on 73 AsO 2 Ϫ Transport-Transport activity was maximal at pH 7 and decreased between 7 and 9 (Fig. 5) In contrast, potassium antimonyl tartrate was found to stimulate 73 AsO 2 Ϫ transport 4-to 5-fold. The effect was specific for Sb(III) salts; sodium potassium tartrate had no effect on 73 AsO 2 Ϫ transport, and the same stimulation was observed with antimony trichloride, which would be expected to hydrate to an antimonite oxyanion. This stimulatory effect of antimonite was further investigated. ArsB was required for Sb(III)-stimulated 73 AsO 2 Ϫ accumulation (Fig. 6). Transport required NADH oxidation in the presence and absence of antimonite, and in both cases FCCP inhibited. Thus, this effect appears to be a property of the ArsB-mediated transport system and not a nonspecific effect of Sb(III). The degree of stimulation required stoichiometric amounts of antimonite and arsenite (Fig. 7). Importantly, at each concentration of arsenite examined, the maximal stimulation occurred at an equimolar concentration of Sb(III) (Fig. 7).
The solution chemistry of arsenicals and antimonials is not well characterized. In solution it might be expected that the arsenite oxyanion (AsO 2 Ϫ ) would be hydrated to As(OH) 2 O Ϫ , which is in equilibrium with the protonated form As(OH) 3 . However, other reasonable forms could be suggested, such as substrates for the transport system than arsenicals, such mixed salts could explain the substantial stimulation of 73 As(III) uptake by Sb(III). In both transcriptional regulation by ArsR and allosteric regulation by ArsA, Sb(III) is more effective than As(III). Thus, it would not be surprising if antimonials were a better substrate for ArsB than arsenicals. Al-  ternatively, an allosteric effect of Sb(III) on the carrier itself cannot be ruled out. Sb(III) allosterically activates the ArsA ATPase, but it does so by binding to a triad of cysteine thiolates (4), while there are no essential thiols in ArsB (28). Further arguing against an allosteric effect is the fact that maximal stimulation requires stoichiometric amounts of As(III) and Sb(III) at all concentrations of As(III). These results suggest that a mixed salt containing both As(III) and Sb(III) can be transported by the carrier.

DISCUSSION
From the results of previous in vivo and in vitro studies, we concluded that the arsA and arsB gene products of the R773 ars operon form a membrane-bound complex that functions as an obligatorily ATP-coupled arsenite pump. First, arsenical extrusion from cells of E. coli exhibited dependence on chemical energy; electrochemical energy was neither necessary nor sufficient (6, 7). Although the form of chemical energy could not be unambiguously identified from those physiological studies, there was a correlation between ATP levels and extrusion. Second, the transport system was shown to be a complex of two subunits, ArsA and ArsB (22), where ArsA is an As(III)/Sb(III)stimulated ATPase. Third, everted membrane vesicles containing the ArsA-ArsB complex exhibited energy-dependent accumulation of 73 AsO 2 Ϫ (9). In vitro transport had an absolute requirement for ATP. Again, electrochemical energy was neither necessary nor sufficient. In these cells, the H ϩ -translocating F 0 F 1 ATPase was deleted, so coupling of the ArsA-ArsB complex to ATP hydrolysis was direct.
However, as described above, several findings posed a question with respect to the energy coupling of the Ars system. Most intriguing was the observation that three of five ars operons lack an arsA gene. Only the operons from E. coli plasmids R773 (10) and R46 (11) have arsA genes. The two staphylococcal (12,13) and the E. coli chromosomal (14) operons do not. Obviously it would be difficult to have an ATP-coupled pump without an ATPase subunit. One possibility is that the two types of extrusion systems have different biochemical mechanisms. The close similarity of the R773, R46, and E. coli chromosomal ArsB proteins (each exhibits over 90% similarity to the other two) would suggest that the proteins should have a common mechanism. Even the ArsBs from the staphylococcal plasmids, which are less than 60% similar to the proteins from the E. coli proteins, have essentially superimposable hydropathic profiles, suggesting similar membrane topology (29). Indeed, chimeras of the ArsB proteins from the Gram-positive and Gram-negative bacteria constructed by gene fusions of the arsB genes are functional (29). A reasonable deduction is that all ArsBs have the same biochemical mechanism.
Do ArsBs function as components of primary pumps, as the in vivo energetics studies indicate (6, 7)? Are they secondary carriers, as suggested by their transmembrane topology (5, 28)?
FIG. 8. Dual energy coupling of the Ars transporter. ArsB functions physiologically in either of two modes: as a potential driven secondary carrier or as a subunit of an obligatory ATP-coupled pump. In cells lacking an arsA gene, ArsB translocates arsenical and antimonial oxyanions, with energy derived from the proton-pumping respiratory chain. In cells with both genes, the ArsA-ArsB complex is an aniontranslocating ATPase unable to utilize the membrane potential. Although the substrate is shown as the arsenite anion in this model, the structure of inorganic As(III) and Sb(III) oxysalts in dilute aqueous solution is unknown. These possibilities are not mutually exclusive. Indeed, results of in vivo transport studies suggested that the ArsB protein mediates the electrochemical energy-dependent arsenite efflux in the absence of the ArsA protein while the ArsA-ArsB complex catalyzes chemical energy-dependent transport (8). The pI258-encoded ArsB similarly had been proposed to utilize electrochemical energy (30). The results in this study clearly demonstrate that ArsB functions as a secondary arsenite transporter under conditions where there is only ⌬ (27), consistent with the carrier catalyzing electrophoretic anion movement coupled to the membrane potential. The fact that some ars operons have both arsA and arsB genes and thus encode pumps, while others have only the arsB gene and encode secondary systems suggests that the acquisition of a gene for a catalytic subunit might be a recent evolutionary event. Independence from electrochemical gradients would make the cell less susceptible to depolarizing environmental poisons. Therefore, the Ars system is a novel transport system that physiologically has two possible modes of energy coupling depending on its subunit composition (Fig. 8).