Spatial proximity of Cys113, Cys172, and Cys422 in the metalloactivation domain of the ArsA ATPase.

ArsA ATPase activity is allosterically activated by salts of the semimetal arsenic or antimony. Activation is associated with the presence of three cysteine residues in ArsA: Cys113, Cys172, and Cys422. To determine the distance between cysteine residues, wild type ArsA and ArsA proteins with cysteine to serine substitutions were treated with the bifunctional alkylating agent dibromobimane, which reacts with thiol pairs within 3-6 Å of each other to form a fluorescent adduct. ArsA proteins in which single cysteine residues were altered by site-directed mutagenesis still formed fluorescent adducts. Proteins in which two of the three cysteine residues were substituted did not form fluorescent adducts. These results demonstrate that Cys113, Cys172, and Cys422 are in close proximity of each other. We propose a model in which As(III) or Sb(III) interacts with these three cysteines in a trigonal pyramidal geometry, forming a novel soft metal-thiol cage.

Plasmid-mediated resistance to the metalloid salts arsenite and antimonite has been observed in both Gram-negative and Gram-positive bacteria . In the Gram-negative bacterium Escherichia coli, the conjugative R-factor R773 confers resistance by coding for an ATP-coupled arsenite pump that actively extrudes the metalloid oxyanions out of the cell, thereby reducing their intracellular concentration to subtoxic levels . The pump is composed of ArsA, a 63-kDa peripheral membrane protein that is the catalytic subunit of the pump, and ArsB, a 45-kDa integral membrane protein that serves both as the membrane anchor for ArsA and as the anion-conducting pathway. Under physiological conditions, ArsA is a part of a complex with the ArsB protein in the inner membrane of E. coli. When expressed at high levels, ArsA is found predominantly as a soluble protein in the cytosol (Rosen et al., 1988). Soluble ArsA has been purified and shown to be an As(III)-or Sb(III)-stimulated ATPase. The 583-amino acid ArsA has distinct binding sites for its substrate, ATP, and effector, As(III) or Sb(III) (Hsu and Rosen, 1989).
Genetic evidence suggests that allosteric activation occurs via the formation of a three-coordinate complex of As(III) or Sb(III) with the cysteine thiolates of residues 113, 172, and 422 . Although these three residues are located distant from each other in the primary sequence, these results imply that they come close together in the tertiary structure to interact with either As(III) or Sb(III). From the crystal structure of known As-thiol and Sb-thiol compounds (Sowerby, 1994), we would predict that the cysteine thiolates should be 3-4 Å from each other. The homobifunctional crosslinker 4,6-bis(bromomethyl)-3,7-dimethyl-1,5-diazabicyclo [3.3.0]octa-3,6-diene-2,8-dione (dibromobimane (bBBr)) 1 was used to map distance between pair of cysteines in ArsA. bBBr has two equivalent bromomethyl groups that can cross-link a thiol pair located within 3-6 Å of each other (Mornet et al., 1985). bBBr is nonfluorescent in solution but becomes fluorescent when both of its alkylating groups have reacted (Haugland, 1994). This property allows bBBr to be used as a molecular ruler to map distance between cysteine residues.
In this report we demonstrate that ArsA proteins with any one cysteine residue group altered by mutagenesis still formed fluorescent adducts. Proteins lacking any two of the three cysteines at residue 113, 172, or 422 did not form fluorescent adducts. The results demonstrate that Cys 113 , Cys 172 , and Cys 422 are located within a distance of 3-6 Å from each other. A model is proposed where As(III) or Sb(III) interacts with the three cysteine thiolates in a trigonal pyramidal geometry, forming a soft metal-thiol cage at the metalloactivation domain of the enzyme.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Media-The strains and plasmids used in this study are given in Table I. Cultures of E. coli strains JM109 or JM110 (Sambrook et al., 1989) bearing the indicated plasmids were grown at 37°C in LB medium (Sambrook et al., 1989). Ampicillin (125 g/ml) or tetracycline (10 g/ml) was added as required. Sodium arsenite, potassium antimonyl tartrate, or isopropyl ␤-D-thiogalactopyranoside was added at the indicated concentrations. All chemicals were obtained from commercial sources.
Oligonucleotide-directed Mutagenesis-Mutations in the sequence of the arsA gene were introduced by site-directed mutagenesis using the Altered Sites TM in vitro Mutagenesis System (Promega), as described earlier . Plasmid pALTER-AB, containing the arsA and arsB genes in the multiple cloning site of vector pALTER-1 (Promega) was used as the template . The mutagenic oligonucleotides used and the respective changes (underlined) introduced were as follows: C26S, CGTGGCGGAGGAAATAGA; C113S, CTCTGTTGTGGATGCACCTGA; C172S, TGGGCCGAGAGAG-GACGCGCC; C422S, TTCCTCAGTGGAAGGTGAGCG. Oligonucleotides were synthesized in the Macromolecular Core Facility of Wayne State University School of Medicine. The identity of the mutations was confirmed by DNA sequencing of the entire arsA gene of each mutant.
DNA Manipulation and Sequence Analysis-Plasmid DNA was prepared using a Wizard DNA purification kit (Promega). Plasmid isolation, DNA restriction endonuclease analysis, ligation, and transformation were performed as described (Sambrook et al., 1989;Chung et al., 1989). For DNA digestion with ClaI, appropriate plasmids were introduced into E. coli strain JM110. Restriction enzymes and nucleic acid-* This work was supported by U.S. Public Health Service Grant AI19793. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence and reprint requests should be addressed: Dept. of Biochemistry and Molecular Biology, Wayne State University, School of Medicine, 540 E. Canfield Ave., Detroit, MI 48201. Tel.: 313-577-1512; Fax: 313-577-2765; E-mail: brosen@med.wayne.edu. modifying enzymes were obtained from Life Technologies, Inc. except for Sse8387I, which was purchased from Takara Biomedicals. The Sequenase kit (version 2.0, U.S. Biochemical Corp.) was used for double-stranded DNA manual sequencing, as described previously . For automated DNA sequencing, doublestranded DNA was isolated with a QIAGEN plasmid kit and sequenced using an ALFexpress system and Cy5 labeled sequence kit (Pharmacia Biotech Inc.).
Purification and Assay of ArsA-Altered ArsA proteins were purified from cultures of E. coli strain JM109 bearing pALTER-1 derivatives containing the mutated arsA genes. Cultures were grown at 37°C overnight with aeration in 0.25 liters of LB medium containing 0.125 mg/ml of ampicillin. The cultures were diluted into 2.5 liters of prewarmed LB medium containing 0.125 mg/ml of ampicillin. At an A 600 of 0.6 -0.8, production of the mutant protein was induced by the addition of 0.1 mM isopropyl-␤-D-thiogalactopyranoside. ArsA was purified essentially as described (Hsu and Rosen, 1989) and stored at Ϫ70°C until use. Each ArsA was judged to be Ͼ95% homogeneous by Coomassie Blue staining of samples separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) (Laemmli, 1970). The concentration of ArsA in purified preparations was determined using a modification of the method of Lowry et al. (1951) or from the absorption at 280 nm using a molar extinction coefficient of 33,480 (Rosen et al., 1988). ATPase activity was assayed using the coupled assay of Vogel and Steinhart (1976), as described previously (Hsu and Rosen, 1989).
Polyacrylamide Gel Electrophoresis and Immunological Blotting-Cultures of cells containing the appropriate plasmids were grown in LB medium at 37°C to an A 600 ϭ 0.6, at which point expression of the ars genes was induced with 0.1 mM isopropyl-␤-D-thiogalactopyranoside. After growth at 37°C for 3 h, 0.5 ml of cells were pelleted and suspended in 0.5 ml of SDS sample buffer. Samples were prepared by boiling in SDS sample buffer for 5 min and separated by SDS-PAGE. Immunoblotting was performed using an enhanced chemiluminescence assay (DuPont NEN) and exposed on x-ray film at room temperature as described previously (Dey et al., 1994) Reaction of ArsA with bBBr-bBBr was purchased from Molecular Probes, Inc. bBBr was dissolved in dimethyl formamide and used as a 10 mM stock solution. Before reaction of ArsA with bBBr, dithiothreitol was removed using a spin column (Penefsky, 1977). ArsA (10 M) was incubated at room temperature with a 2-fold molar excess of bBBr in 50 mM MOPS-KOH buffer, pH 7.5, containing 0.25 mM disodium ethylenediaminetetraacetate (buffer A). For activity measurements, portions of the reaction mixture were removed at intervals, the reaction was quenched by the addition of 20 mM dithiothreitol, and residual activity was determined.
Titration of ArsA with 5,5Ј-Dithiobis(2-nitrobenzoic acid) (DTNB)-The reactivity of the sulfhydryl groups in the native and denatured enzyme was determined by reaction with DTNB. ArsA (1-2 ϫ 10 Ϫ5 M in buffer A) was reacted with DTNB, and the reaction was monitored by the increase in absorbance at 412 nm. The total number of cysteine residues was determined by denaturing the enzyme with a solution of 6 M guanidine HCl in buffer A prior to reaction with DTNB. The concen-tration of sulfhydryl groups was calculated using ⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 (Riddles et al., 1979).
Measurements of Fluorescence-Fluorescence measurements were performed using an SLM 8000 spectrofluorimeter. ArsA proteins were treated with a 2-fold molar excess of bBBr. Portions were removed over a period of 15 min and diluted 10-fold with buffer A, and the fluorescence emission was measured at 470 nm with an excitation wavelength of 390 nm. The slit width for excitation and emission was 4 mm.
Proteolysis with Trypsin-Trypsin digestion was performed at room temperature in buffer A containing 0.8 mg/ml of ArsA. The protein: protease ratio was 100:1 (w/w). ArsA was preincubated for 20 min with 0.5 mM potassium antimonyl tartrate, with 5 mM ATP alone, or with a combination of both. Digestion was initiated by the addition of N-tosyl-L-phenylalanine chloromethyl ketone-treated trypsin (Sigma) and terminated by the addition of a 2-fold excess of soybean trypsin inhibitor. Samples were analyzed by SDS-PAGE using a 12% polyacrylamide gel and Coomassie Blue staining.

Inactivation of ArsA by Dibromobimane-
The effect of bBBr on the Sb(III)-stimulated ATPase activity of wild type ArsA was examined. ArsA was rapidly and completely inactivated by a 2-fold molar excess of bBBr. A plot of 1/[V t /V 0 ] against time FIG. 1. Inactivation of ArsA by bBBr. Treatment with bBBr and assays of ArsA activity were performed as described under "Experimental Procedures." Following removal of dithiothreitol with a spin column, purified ArsA protein (10 M) was incubated with ATP and/or potassium antimonyl tartrate for 20 min at room temperature. bBBr (20 M) was added, and the incubation continued. ArsA activity was measured at the indicated times. Additions were as follows: none (å), 0.5 mM potassium antimonyl tartrate (q), 5 mM ATP (ࡗ), 0.5 mM potassium antimonyl tartrate and 5 mM ATP (f).

FIG. 2. Correlation of bBBr modification of cysteine residues and inactivation of ATPase activity.
Purified C26S ArsA (10 M) was incubated with a 2-fold molar excess of bBBr. At the indicated times portions were withdrawn. One portion was mixed with dithiothreitol to quench the residual bBBr, and ATPase activity was assayed. A second portion was mixed with an equal volume of chloroform, the upper aqueous phase was treated with guanidine hydrochloride, and the number of free thiols remaining was determined by DTNB titration. å, ATPase activity; q, number of cysteines modified. was linear, consistent with second-order reaction kinetics ( Fig.  1), where V t is the rate at various times of inactivation by bBBr, and V 0 is the rate at zero time. Partial protection was observed when the protein was preincubated with either ATP or Sb(III). The t1 ⁄2 of inactivation by bBBr alone was 2.1 min, whereas the half-time of inactivation reaction increased to 2.6 min and 3.5 min when preincubated with either antimonite or ATP, respectively. When both ATP and antimonite were added together, synergistic protection was observed, with t1 ⁄2 increasing to 16.2 min. These results indicate that together antimonite and ATP synergistically induce a conformational change that decreases the accessibility of the cysteine thiolates to bBBr. The protection experiments were done in the absence of MgCl 2 ; thus, the observed effects occurred in the absence of ATP hydrolysis.
To examine the correlation between loss in activity and the number of cysteines modified, the C26S ArsA was treated with bBBr. We had shown earlier that Cys 26 is not involved in either activation or catalysis and that the C26S enzyme is fully active . For that reason C26S was used as the control. C26S was treated with a 2-fold molar excess of bBBr, and the rate of inactivation was determined. The number of cysteines modified by bBBr was determined by removing portions at intervals and extracting bBBr with chloroform. The protein was then denatured with 6 M guanidine hydrochloride, and the concentration of free thiol was determined with DTNB. C26S was completely inactivated following modification of two of the three cysteines (Fig. 2). The bBBr modification reaction of C26S exhibited biphasic kinetics, where one of the cysteines reacted rapidly, with more than 50% of the activity lost in the first min. The other cysteine reacted more slowly, following which activity was completely lost. The kinetics of inactivation suggest rapid displacement of the bromine of bBBr by a thiol, followed by slow displacement of the second bromine with another thiol. C26S modified with bBBr did not form significant amounts of dimers or higher order oligomers as determined by SDS-PAGE (data not shown), suggesting the absence of intermolecular cross-linking. The monomer was highly fluorescent on SDS gels (data not shown), indicating that any two of the three available cysteines can be cross-linked within a monomer.
Isolation of Altered ArsA Proteins-To determine which cysteines in a monomer react with bBBr to produce a fluorescent adduct, combinations of altered proteins were engineered. The four cysteinyl residues in ArsA were replaced by all possible combinations of two, three, or four seryl residues. There were six double substitutions (C26S/C113S, C26S/C172S, C26S/ C422S, C113S/C172S, C113S/C422S, and C172S/C422S), four triple substitutions (C26S/C113S/C172S, C26S/C113S/C422S, C26S/C172S/C422S, and C113S/C172S/C422S), and one cysteine-free C26S/C113S/C172S/C422S ArsA. Cells bearing the mutated arsA genes (Table I) in combination with a wild type arsB gene were characterized phenotypically for arsenite resistance (Fig. 3). Cells expressing the wild type arsA and arsB genes could grow in media containing concentrations of sodium arsenite in excess of 8 mM; cells without an ars operon could not grow in media with more than 1 mM sodium arsenite. Cells bearing the mutant arsA genes were each resistant to 3 mM sodium arsenite but sensitive to 5 mM sodium arsenite, a phenotype intermediate between sensitive and fully resistant cells. The resistance patterns of these new mutants are essentially the same as that of single mutants, as described earlier .
Analysis of the Altered ArsA Proteins-The steady state level of production of the altered ArsA proteins was approximately the same as the wild type ArsA, as shown by Western blot analysis using an antiserum against wild type ArsA (data not shown). There was no difference between the wild type and

2505-bp AvaI fragment from pC26S/C113S cloned into AvaIdeleted pC172S/C422S
This study altered proteins in mobility on SDS-PAGE, nor was abnormal degradation of the altered proteins observed. Each mutant protein was purified to Ͼ95% homogeneity as described under "Experimental Procedures." Limited Trypsin Digestion of the Altered ArsA Proteins-Accessibility to trypsin has been used as a measure of the structure of ArsA and of the conformational changes induced by binding of nucleotide and metalloid effector (Hsu and Rosen, 1989). When wild type ArsA was exposed to trypsin in the absence of ligands, the protein was digested to a 30-kDa species (Fig. 4A). In the absence of ligands the accessibility of C26S/ C113S to trypsin was essentially identical to the wild type (Fig.  4B). ATP afforded the same protection for C26S/C113S as the wild type (Fig. 4, C and D), indicating that binding of ATP produced a similar conformational change in both. While the addition of antimonite had no effect on the rate of proteolysis (Fig. 4A), the addition of both ATP and antimonite produced synergistic protection from trypsin (Fig. 4C), as had been shown previously (Hsu and Rosen, 1989). In contrast, the combination of ATP and antimonite did not provide additional protection over that afforded by ATP alone with the C26S/ C113S ArsA (Fig. 4D). The other combinations of double, triple, and quadruple serine-substituted ArsA proteins showed the same pattern of trypsin accessibility as C26S/C113S (data not shown). The loss of this cooperative effect of ligands in the proteins lacking any of the three cysteine residues at position 113, 172, or 422 indicates decreased binding of the metalloid effector in these serine-substituted proteins.
ATPase Activity of Altered ArsAs-The purified proteins were analyzed for their ability to catalyze metalloid-activated ATP hydrolysis. Of the six double serine-substituted ArsAs, only C26S/C113S, C26S/C172S, and C26S/C422S showed Sb(III)-stimulated ATPase activity (Table II). Each altered protein had a K m for ATP that was within the same order of magnitude as that of the wild type enzyme, while the concentration of antimonite required for half-maximal activation differed considerably. Both C26S/C113S and C26S/C172S exhibited approximately a 20-fold increase in the concentration of antimonite required for half-maximal activation, and the C26S/ C422S enzyme exhibited more than a 200-fold increase. The results with arsenite as an activator were consistent with the results with antimonite; C26S/C113S and C26S/C172S both exhibited nearly a 10-fold increase in the concentration of arsenite required for half-maximal activation, whereas C26S/ C422S required about 20-fold more than the wild type enzyme (data not shown). The intrinsic rates of hydrolysis of the altered enzymes were slightly less than the wild type, but the major effect of replacement of the cysteine residues was on the Sb(III)-activated rates ( Table II). The Sb(III)-stimulated V max of the C26S/C172S enzyme was near that of the wild type. Enzymes with any two of the three essential cysteines altered simultaneously to serines had no measurable Sb(III)-stimulated ATPase activity, including the double-substituted C113S/ C172S, C113S/C422S, and C172S/C422S; triple-substituted C26S/C113S/C172S, C26S/C113S/C422S, C26S/C172S/C422S, and C113S/C172S/C422S; and cysteine-free C26S/C113S/ C172S/C422S enzymes.
Dibromobimane Modification of ArsA-The increase in fluorescence of bBBr on cross-linking of spatially proximate thiols was used for distance mapping of cysteines in ArsA. If the cysteine triad of Cys 113 , Cys 172 , and Cys 422 are indeed each As(III) or Sb(III) ligands, they should be close enough to each other to form fluorescent adducts with bBBr. C26S, which still has the cysteine triad, reacted with bBBr to form a fluorescent species with an absorption maximum at 393 nm and emission maximum at 470 nm (Fig. 5, curve 1). Mutations into the codons of each of the three cysteines in the triad were introduced into the arsA C26S gene, and the gene products with only two of the three cysteines remaining became highly fluorescent following bBBr modification, as illustrated with C26S/C172S (Fig. 5, curve 2). C26S/C113S and C26S/C422S gave superimposable results (data not shown). In contrast, removal of two of the three cysteines of the triad resulted in ArsA proteins with substantially reduced levels of fluorescence (Fig. 5, curves 3-5), and the cysteine-free ArsA had almost no increase in fluorescence over base line following reaction with bBBr (Fig. 5, curve 6).
Steady state experiments with bBBr have a relatively high background level of fluorescence. This is due in part to decomposition of bBBr in the presence of light and in part to a slow, nonspecific displacement of the second bromine by water following reaction of the first with cysteine. bBBr can also react with other nucleophilic residues to produce fluorescence. Since cysteine thiolates react much more rapidly with bBBr than do other nucleophilic residues, the rate of bBBr reaction with ArsA was examined (Table III). Single serine-substituted ArsAs C26S, C113S, C172S, and C422S and the doubly substituted ArsAs C26S/C113S, C26S/C172S, and C26S/C422S reacted with bBBr at rates only slightly slower than wild type. C113S/C172S was 3-fold less reactive, whereas C113S/C422S and C172S/C422S were 7-fold less reactive compared with wild type. Triple serine substitutions C26S/C113S/C172S, C26S/ C113S/C422S, and C26S/C172S/C422S were on the average 9-fold less reactive. Both C113S/C172S/C422S and the cysteine-free ArsA had more than a 60-fold decrease in reactivity. The near complete absence of steady state fluorescence in the cysteine-free protein (Fig. 5) and the low rate of modification of the triple serine-substituted ArsAs by bBBr clearly indicate that bBBr is specific for thiols in ArsA and that other functional groups are not modified. Moreover, the kinetics of the bBBr reaction show that at least two of the three cysteines at resi- dues 113, 172, and 422 are required for rapid cross-linking by the bifunctional reagent bBBr.

DISCUSSION
From the results of mutagenesis of cysteine codons in arsA we demonstrated that Cys 113 , Cys 172 , and Cys 422 are required for allosteric activation of the ArsA ATPase . We proposed that activation by arsenite or antimonite occurs through soft metal-thiol interaction of As(III) or Sb(III) with the thiolate of those three cysteine residues. Although the three cysteine residue are located far apart in the primary sequence, they must come close enough in the tertiary structure to interact with the soft metals. From x-ray crystal data of As(III) or Sb(III) complexed to dithiols, the lengths of an As-S bond and Sb-S bond are 2.23 and 2.45 Å, respectively, while S-As-S and S-Sb-S angles are 92.7 and 84.8°, respectively (Sowerby, 1994). Moreover, using a combination of sitedirected mutagenesis and arsenic x-ray absorption spectroscopy, we have recently shown that the ArsR repressor binds As(III) to the thiolates of the cysteine triad of Cys 32 , Cys 34 , and Cys 37 with As-S bonds of 2.25 Å (Shi et al., 1996). With the knowledge of these examples we propose a model for interaction of ArsA with As(III) or Sb(III) in which the thiolates of Cys 113 , Cys 172 , and Cys 422 are coordinated to the metalloid in a trigonal pyramidal geometry (Fig. 6). This model predicts that the thiols should be within 3.1-3.5 Å of each other. To test this hypothesis, dibromobimane was used as a molecular ruler. bBBr rapidly inactivated the ATPase activity of ArsA, and FIG. 4. Substrate protection from trypsin cleavage. Purified wild type ArsA (A and C) or C26S/C113S (B and D) were treated with trypsin for varying times without substrate or effector (A and B, lanes 1-5), with 0.5 mM potassium antimonyl tartrate (A and B, lanes 6 -9), with 5 mM ATP (C and D, lanes 1-5), or with both ATP and potassium antimonyl tartrate (C and D, lanes 6 -9) and analyzed by SDS-PAGE on 12% polyacrylamide gels stained with Coomassie Blue. Times of incubation were 0, 0.2, 15, 30, and 60 min in lanes 1-5, and 0.2, 15, 30, and 60 min in lanes 6 -9. The position of migration of standards is indicated.   synergistic protection from bBBr inactivation was observed when both the nucleotide substrate and metalloid effector were present (Fig. 1). Since bBBr reacts preferentially with cysteines, the protection experiment suggested that it is reacting with the cysteines located in the metalloactivation domain. Inactivation required modification of only two of the three essential cysteines. One cysteine reacted rapidly, whereas the other reacted more slowly (Fig. 2). This might reflect modification of a cysteine by one bromomethyl group of bBBr, followed by reaction of the remaining bromomethyl group with an adjacent cysteine. Dibromobimane is nonfluorescent until it cross-links two adjacent nucleophiles located within 3-6 Å of each other. The thiols of the cysteine triad would therefore be expected to react with bBBr to form fluorescent adducts. ArsA reacted with bBBr became highly fluorescent, and nearly all of the fluorescence was found in the 63-kDa band on SDS-PAGE, indicating intramolecular cross-linking of cysteines. To determine which cysteines reacted with bBBr, mutations were introduced to create proteins in which two, three, or four cysteines were altered to serines, greatly decreasing resistance to arsenite and antimonite (Fig. 3). The gene products were not grossly altered in structure. The rate of production of tryptic peptides was the same in the wild type and altered ArsAs, showing that surface accessibility of arginine and/or lysine residues was not modified. Binding of ATP produces a conformational change that decreases the rate of trypsin digestion (Hsu and Rosen, 1989). None of the serine substitutions affected ATP protection, suggesting that the mutations did not reduce substrate binding (Fig. 4). The synergistic protection afforded by ATP and Sb(III) reflects a conformational change produced by effector binding. This cooperative protection was lost in each of the proteins in which any of the three cysteines of the triad were replaced by serine residues (C113S/C172S, C113S/C422S, and C172S/ C422S). Only the C26S/C113S, C26S/C172S, and C26S/C422S enzymes retained any catalytic activity. The K m value for ATP for each was within an order of magnitude of the wild type enzyme. In contrast, the concentration of semimetal oxyanion required for activation was increased by 1-2 orders of magnitude, suggesting that As(III) or Sb(III) cannot interact with ArsA when any two cysteine thiolates of the triad are missing.
The fluorescence of the ArsAs was monitored after bBBr modification. Substitution of serine for Cys 26 had little effect on the steady state level of fluorescence (data not shown). C26S/ C113S, in which only Cys 172 and Cys 422 remain, and C26S/ C172S, in which Cys 113 and Cys 422 remain, produced fluorescent products similar to C26S. Thus, only two cysteines of the triad are required for bBBr cross-linking. C172S/C422S showed a significant lowering in the steady state level of fluorescence and a 7-fold decrease in the rate of modification, demonstrating that Cys 172 and Cys 422 are located within 3-6 Å of each other in the tertiary structure of ArsA. C113S/C422S reacted similarly with bBBr; thus, Cys 113 and Cys 422 must likewise be spatially proximate. C26S/C422S, in which Cys 113 and Cys 172 remain, was fluorescent after modification with bBBr, indicating that Cys 113 and Cys 172 are close enough to each other to cross-link. These data clearly demonstrate that Cys 113 , Cys 172 , and Cys 422 are located within a distance of 3-6 Å from each other in the tertiary structure of ArsA, supporting the hypothesis that their thiolates are liganded to As(III) or Sb(III), forming a novel soft metal-thiol structure (Fig. 6). Interestingly, C113S/C172S, which retains Cys 26 and Cys 422 , produced a fluorescent product with bBBr with a rate of reaction only 3-fold less than wild type ArsA (Table III). This is compared with a 7-fold decrease in reactivity in proteins with Cys 26 and either Cys 113 or Cys 172 . This may indicate that Cys 26 , which is not involved in metalloactivation, may be within 6 Å of Cys 422 but further from Cys 113 and Cys 172 .
What is the relationship of metalloid binding and allosteric activation? The catalytically competent form of purified ArsA is a dimer, and the concentration of dimer in solution increases in the presence of Sb(III) or As(III) (Hsu et al., 1991). This suggests that the metalloid binds to the conformation of the protein that preferentially self-associates. Thus, by mass action, effector binding increases the concentration of the active species. Why is Sb(III) nearly 10-fold more effective than As(III) as an activator? Although the overall geometry of the Sb-thiol structure should be similar to that of the As-thiol, the bond angles and distances would be different (Fig. 6). The As-S distance is predicted to be 2.23 Å, with S-S distances of 3.2 Å, while the Sb-S distance would be 2.45 Å, with S-S distances of 3.5 Å. The activator binding site may be slightly too large to accommodate As(III) without strain. The possibility that the metal binding specificity of this site could be altered mutagenically raises interesting possibilities for protein engineering. FIG. 6. Proposed geometry of the allosteric site in ArsA with either As(III) or Sb(III). The allosteric site for arsenite or antimonite in ArsA is proposed to be a trigonal pyramidal structure that is threecoordinate with the sulfur thiolates of Cys 113 , Cys 172 , and Cys 422 as ligands to the metalloid. The structure of the soft metal-thiol complex is postulated from the bond angles and distances found by crystallographic analysis of small molecules containing As-S or Sb-S bonds and from the As-S bond distances in the ArsR repressor.