ATPase assays

The ATPase activity was estimated using a coupled assay (Vogel and Steinhart, 1976), as described (Hsu and Rosen, 1989). MBP-ArsD was buffer exchanged into 50 mM MOPS-KOH, pH 7.5, 0.25 mM EDTA using a Micro Bio-Spin Chromatography Column (Bio-Rad) and then added at a final concentration of 3 μM into an assay mixture containing the same buffer plus 5 mM ATP, 1.25 mM phosphoenolpyruvate, 0.25 mM NADH, 10 units of pyruvate kinase and lactate dehydrogenase with or without various concentrations of potassium antimonyl tartrate or sodium arsenite. ArsA was added to a final concentration of 0.3 μM. The mixture was prewarmed to 37°C, and the reaction was initiated by the addition of 2.5 mM MgCl2, which was measured at 340 nm. The linear steady state rate of ATP hydrolysis was used to calculate specific activity. The reaction volume was 1 ml for assays in 2 ml cuvettes or 0.2 ml for microplate reader assays.

CHAPTER 3 RESULTS

3.1 ArsD confers elevated resistance to arsenic upon cells expressing the arsenical pump.

To examine whether ArsD and ArsA have linked functions in arsenic detoxification, the arsD gene was co-expressed with the arsAB genes from compatible plasmids under control of heterologous promoters. The plasmids were expressed in E. coli strain AW3110, in which the chromosomal arsRBC operon had been deleted (Carlin et al., 1995). By itself, arsB confers low-level resistance, while arsAB expression confers resistance at considerably higher levels (Figure 2A and B) (Dey and Rosen, 1995). Cells co-expressing arsD with arsB were no more resistant to arsenite than cells express only arsB, while cells co-expressing arsDAB were significantly more resistant to higher concentrations of arsenite compared to cells expressing only arsAB. Since an immunoblot established that arsD did not affect the levels of ArsA produced (Figure 5), the data are consistent with interaction of ArsD with ArsA to increase the efficiency of the ArsAB pump.

3.2 ArsD confers an competitive advantage to cells growing in subtoxic concentrations of arsenite.

Arsenic is a ubiquitous toxic metal contaminant and health hazard in drinking water worldwide (Smedley and Kinniburgh, 2002). When arsenite

resistance was compared between cells expressing the arsAB genes or the arsDAB genes, the latter showed a modest increase in resistance, with the greatest differences observed at millimolar concentrations of arsenite, amounts of arsenite that are toxic under laboratory settings (Figure 2A and B). Does the presence of the arsD gene confer an evolutionary advantage on the host organism for growth in concentrations of arsenite frequently found in the environment? A molecular competition experiment was devised to examine this question. Two sets of cells of E. coli strain AW3110 were allowed to compete with each other in a mixed culture for growth in the presence of a sub-toxic concentration (10 µM) of arsenite, which is in the range of what is found in the environment (Smedley and Kinniburgh, 2002). One set of cells had a plasmid with arsAB under control of the tac promoter, while the other set had arsDAB in the same vector. Each day the culture was diluted 1000-fold, and the relative amounts of the arsDAB and arsAB plasmids were analyzed by restriction digestion (Figure 3A). After nine days of growth, cells with arsDAB had largely replaced those with only ArsAB (Figure 3B), indicating that the arsD gene provides a competitive advantage for growth in the low concentrations of arsenite that are ubiquitous in soil or surface waters.

3.3 ArsD enhances the ability of the pump to lower the intracellular concentration of arsenite

To demonstrate that ArsD enhances the ability of the pump to lower the intracellular concentration of arsenite, the effect of the ars genes on arsenite

accumulation was examined in intact cells. Higher rates of extrusion result in lower accumulation of arsenite (Dey and Rosen, 1995). Cells of the arsenite-hypersensitive strain AW3110 with no ars genes (vector plasmids pSE380 and pACBAD) accumulated approximately 150 pmol As(III)/10⁹ cells/10 min (Figure 4). Cells expressing only arsB accumulated arsenite to approximately 22 pmol As(III)/10⁹ cells/10 min, reflecting the ability of ArsB to catalyze arsenite/proton exchange (Meng et al., 2004). Expression of arsAB resulted in decreased accumulation to approximately 7 pmol As(III)/10⁹ cells/10 min, reflecting more efficient arsenite extrusion by the ArsAB pump than by ArsB alone (Dey and Rosen, 1995). Cells co-expressing arsD and arsAB exhibited substantially less accumulation of arsenite (approximately 1 pmol As(III)/10⁹ cells/10 min) than those with arsAB. Cells expressing arsD with only arsB accumulated arsenite to approximately the same level as cells expressing only arsB, indicating that ArsD does not affect activity of ArsB. The results of immunoblotting showed that arsD does not affect the levels of ArsA produced (Figure 5). These results clearly show that ArsD increases the efficiency of the ArsAB pump.

3.4 Interaction of ArsD with ArsA in vivo

Yeast two-hybrid analysis was applied to demonstrate that ArsD and ArsA physically interact (Figure 6A). ArsA interacted with ArsD but not with the ArsR repressor or the ArsC arsenate reductase. ArsD interacted with ArsA and with itself, which would be expected since ArsD is a homodimer (Chen and Rosen, 1997), but not with ArsR or ArsC. ArsR, which is a homodimer, also interacts with

itself, but not with ArsD or ArsA. These results indicate specific interaction of ArsD and ArsA. When 0.1 mM potassium antimonyl tartrate was added to the medium, the cells grew more slowly, but there was no effect on the ability of BD-ArsD to interact with AD-ArsA (Figure 6B). Thus, BD-ArsD and ArsA interact in the absence of added metalloid. However, the presence of some metal or metalloid in the yeast cytosol that promotes interaction cannot be ruled out.

3.5 Interaction of ArsD with ArsA in vitro

Direct physical interaction between ArsD and ArsA was observed by chemical crosslinking with two different crosslinkers. Since both proteins have metalloid binding sites composed of cysteine residues (Bhattacharjee and Rosen, 1996; Li et al., 2001), it was reasonable to consider that they might interact at those sites. Crosslinking was performed using (4,6-bis(bromomethyl)-3,7-di-methyl-1,5-diazabicyclo[3.3.0]octa-3,6-diene-2,8-dione (dibromobimane or bBBr) (Invitrogen Corporation, Carlsbad, California), a fluorogenic, homobifunctional thiol-specific crosslinking reagent that becomes highly fluorescent when its two alkylating groups react with cysteine residues that are within 3 to 6 Å of each other (Kosower et al., 1980). ArsA forms intramolecular crosslinks with bBBr at its metalloid binding site (Bhattacharjee and Rosen, 1996). When ArsD was treated with bBBr, subjected to sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE), and immunoblotted with anti-ArsD, it formed a number of species that correspond to dimers and higher order species (Figure 7). All ArsD bands, including the monomer, became fluorescent

following reaction, showing that both intra- and intermolecular crosslinking had occurred (Figure 7, top panel). Since ArsD is a functional dimer, intermolecular crosslinking is not unexpected. When an equimolar mixture of ArsD and ArsA was reacted with bBBr, a crosslinked species was detected that reacted with both anti-ArsA and anti-ArsD antibodies (Figure 7, lane 2, second and third panels). This species migrated as a band with an apparent mass of approximately 90 kDa, the predicted mass of an ArsD dimer crosslinked to a monomer of ArsA. It should be pointed out that the intensity of the bands from one blot to the next cannot be directly compared because different polyclonal antibodies react with their antigens differently and because the stripping process removes variable amounts of the antigenic species. As a control, ArsA was reacted with CadC, a Cd(II)-responsive regulatory protein of similar size to ArsD and with a metal binding site formed of cysteines that react intramolecularly with bBBr (Wong et al., 2002). No ArsA-CadC adducts were observed using anti-ArsA (Figure 7, lanes 6-8, second panel) or anti-CadC (lanes 6-8, bottom panel) antibodies. The amount of the ArsD-ArsA crosslinked product was increased by addition of MgATP (Figure 7, lane 4, second and third panels). As expected, addition of Sb(III) did not increase crosslinking since the thiol groups that coordinate the metalloid would have reacted with bBBr (Figure 7, lane 3, second and third panels). In these and other in vitro experiments Sb(III) was used rather than As(III) since both ArsD and ArsA have higher affinity for trivalent antimony, a softer and more thiophilic metal than arsenic.

Because the interaction of the two proteins appears to involve cysteine

residues, crosslinking was examined with dimethyl adipimidate (DMA) (Pierce Biotechnology, Inc. Rockford, IL), a homobifunctional imidoester that crosslinks free amines within 8.6 Å of each other, including N-termini and ε-amino groups of lysine residues, and does not modify cysteine thiolates (Figure 8). ArsA has 75 amino groups (74 lysines and the amino terminus), and ArsD has 15, so there are a large number of potential sites of crosslinking with DMA. Not surprisingly, a number of crosslinked species were observed that reacted with ArsA or anti-ArsD sera, and several that appeared to react with both. The most prominent was a band observed after staining with Coomassie Blue or immunoblotting with either anti-ArsA or anti-ArsD that migrated with an apparent MW of approximately 130 kDa (indicated by the asterisk in Figure 8, top, second and third panels).

The position of this band is higher than predicted for the ArsA-ArsD complex.

However, bifunctional crossslinking reagents such as DMA are known to retard electrophoretic mobility as a result of intramolecular crosslinks that prevent unfolding by SDS (Sieber et al., 2002). To demonstrate specificity of crosslinking between ArsA and ArsD, no crosslinking of ArsA and ArsC was observed (Figure 8, bottom panel). Again, crosslinking of ArsD and ArsA was enhanced by the presence of nucleotide. There was also an additional enhancement by either As(III) or Sb(III), but this was difficult to quantify by DMA crosslinking. In agreement with the yeast two-hybrid results, these in vitro data suggest direct interaction of ArsD and ArsA through the As(III) binding sites of the two proteins.

The requirement for nucleotide suggests that ArsD interacts preferentially with a nucleotide-bound conformation of ArsA.

3.6 Transfer of metalloids from ArsD to ArsA

To explore whether ArsD-ArsA interactions give rise to transfer of metalloid, the ability of ArsA to abstract Sb(III) from ArsD was determined. For these assays, cytosol from cells expressing a maltose binding protein (MBP)-ArsD fusion were incubated with Sb(III), following which the MBP-(MBP)-ArsD-Sb(III) complex was bound to an amylose column, which was then washed with 10 column volumes of buffer to remove other proteins. When the column was eluted with BSA and MgATP, little Sb(III) came off with in the BSA-containing fractions (Figure 9A). Subsequent application of buffer with maltose then eluted nearly homogeneous MBP-ArsD protein in fraction 11 with Sb(III). In contrast, when the column was eluted with ArsA and MgATP, more Sb(III) came off with ArsA in fraction 2 and less with ArsD in fraction 11, consistent with transfer of metalloid from ArsD to ArsA (Figure 9B). The elution fractions were analyzed with SDS-PAGE (Figure 10). The effect of nucleotides on Sb(III) transfer was examined using similar assays (Figure 11). Mg²⁺ enhanced transfer with ATP but was not effective alone. Little Sb(III) eluted with ArsA and Mg²⁺ without ATP in fraction 2, and most of the metalloid eluted with ArsD in fraction 11 (Figure 11D). Among the various conditions, MgATP was the most effective, followed by MgATPγS, MgADP and ATP alone (Figure 11A, B, C and 12), indicating that the nucleotide enhances transfer but this process is not dependent upon its hydrolysis.

Furthermore, when ArsA was incubated with MgATPγS using a similar metalloid transfer assay, ArsA bound more As(III) in the presence of excess ArsD than in

its absence (Figure 13). In contrast, under the same conditions, ArsD bound less As(III) with excess ArsA than in its absence, consistent with transfer of metalloid from ArsD to ArsA. ArsD, with a Kd of 1.7 μM (Li et al., 2002), has higher affinity for Sb(III) than does ArsA, with a Kd of 540 μM (Walmsley et al., 2001).

However, the affinity of ArsA for Sb(III) is substantially increased by binding of nucleotides (Kd = 8 μM) (Ruan et al., 2006). Considering that ArsA has a greater affinity for metalloids in the presence of nucleotides, it is not surprising that metalloid transfer from ArsD to ArsA is similarly enhanced by nucleotides.

3.7 ArsD enhances the catalytic activity of ArsA

The effect of ArsD on the catalytic activity of ArsA was investigated. The ATPase activity of ArsA is stimulated by As(III) (Hsu and Rosen, 1989). When ArsD was added to the ATPase assay, ArsD increased the apparent affinity for arsenite 60-fold, from approximately 1.2 mM to 20 µM (Figure 15A). A similar increase in affinity for Sb(III) was observed (Figure 15B). This was not the result of increased thiol buffering of arsenite, since dithiolthreitol could not replace ArsD (Figure 14A and B). ArsD did not greatly affect the Km of ArsA for ATP at a concentration of arsenite (0.5 mM) or antimonite (10 μM) which is below saturation in the absence of ArsD but is sufficient to saturate the enzyme in the presence of ArsD (Figure 16A and B). Significantly, at a sub-saturating concentration of arsenite, ArsD increased the Vmax with ATP by approximately 3-fold. Thus, the functional consequence of the ArsD-ArsA interaction appears to be an increase in the efficiency of the catalytic subunit of the ArsAB pump at low

concentrations of the substrate of the pump, arsenite.

3.8 Cysteine residues in ArsD contribute to metalloid binding sites (MBSs) Alignment of the primary sequence of homologues of the R773 ArsD indicates that they possess an absolutely conserved vicinal cysteine pair, Cys12-Cys13 and a single conserved cysteine, Cys18 (Fig. 2). Two other vicinal cysteine pairs, Cys112-Cys113 and Cys119-Cys120 are found in some homologues but not others. We previously showed that none of the three vicinal cysteine pairs are required for repression (Chen and Rosen, 1997; Li et al., 2001).

Although all the vicinal pairs are capable of binding As(III), Cys12-Cys13 and Cys112-Cys113, but not Cys119-Cys120, are required for derepression by As(III).

Binding of Sb(III) to purified MBP-ArsD derivatives were measured by rapid gel filtration. The binding was measured as a function of Sb(III) concentration. A background binding was observed with only the maltose binding protein (MBP).

The MBP fused wild type ArsD binds metalloid with a stoichiometry of six Sb(III) per dimer and an apparent Kd of 10^-6 M (Figure 17A), suggesting that there are three metalloid binding sites (MBSs) on an ArsD monomer. The actual Kd could be lower since this assay is not very sensitive. An ArsD derivative ArsD1-118, in which Cys119 and Cys120 were replaced with a six-histidine tag, was able to bind four Sb(III) per dimer, suggesting the third vicinal cysteine pair Cys119-Cys120 (MBS3) on ArsD has a metalloid binding capacity of two per dimer. The truncation ArsD1-109, in which a stop codon was added after the codon for residue 109 and only the first cysteine pair Cys12-Cys13 (MBS1) was present, was able

to bind two Sb(III) per dimer. Similarly the mutant ArsD1-118,C12/13A, in which the first cysteine pair had been changed to an alanine pair, the third pair had been replaced with six-histidine tag and only the second pair Cys-112-Cys113 (MBS2) was present, was able to bind two Sb(III) per dimer. These results are consistent with previous observations that each cysteine pair is competent to bind metalloid (Li et al., 2001).

Because the cysteine pair Cys12-Cys13 and the Cys18 are absolutely conserved, further Sb(III) binding assays were performed based on these cysteine mutants. The mutants ArsD1-118,C12A, ArsD1-118,C13A and ArsD1-118,C18A, in which the pair Cys119-Cys120 was replaced with a six-histidine tag and one of the three N-terminal cysteines was changed to alanines, were able to bind only two Sb(III), indicating that either MBS1 or MBS2 was eliminated (Figure 17B).

Similar mutants based on ArsD1-109, in which the Cys112-Cys113 and Cys119-Cys120 were deleted and only two of the three N-terminal cysteines were present, were unable to bind any Sb(III), suggesting that Cys18, in addition to Cys12 and Cys13, is required to form the MBS1 (Figure 18). However, the possibility that Cys18-to-Ala mutant does not fold properly could not be eliminated.

3.9 MBS1 in ArsD is the active site for metalloid transfer and ArsA activation.

When ArsA was incubated with As(III) in the presence of excess ArsD 1-118,C18A in the As(III) transfer assay, ArsA bound much less As(III) than that in the

presence of wild-type ArsD or that in the absence of ArsD (Figure 19). This suggests that this ArsD mutant, which binds As(III) in MBS2 but not in MBS1, does not transfer As(III) to ArsA, but instead competes with ArsA for binding As(III). And of that the cysteines in MBS1 are responsible for delivering metalloids to ArsA.

The ability of ArsD derivatives to activate ArsA ATPase activity was also examined (Figure 20). In the presence of ArsD1-118, a truncated ArsD lacking Cys119-Cys120, the half maximal concentration of As(III) was 10 µM (Figure 20A), similar to that of wild-type ArsD. In the presence of ArsD1-109, a truncated ArsD lacking the Cys112-Cys113 and Cys119-Cys120 pairs, the half maximal concentration of As(III) was 25 μM, also similar to that of wild-type ArsD. These data suggest that MBS2 and MBS3 are not required for ArsD to activate ArsA. In contrast, in the presence of a C12A/C13A derivative (ArsD1-118,C12/13A), the concentration of As(III) required for half-maximal stimulation of ATPase activity was 1.5 mM, similar to that of ArsA in the absence of ArsD. This mutant ArsD, which lacks MBS1, was not able to activate ArsA. These data are consistent with participation of MBS1, but not MBS2 or MBS3, in increasing the affinity of ArsA for metalloid.

Figure 20B shows the effects of single residue substitutions in MBS1 of ArsD on activating ArsA. Mutants ArsD1-118,C12A, ArsD1-118,C13A and ArsD1-118,C18A, in which the MBS1 was eliminated and only MBS2 was active, were unable to activate ArsA. Cysteine-to-serine substitutions were also investigated in the same assay and found to have similar results as those with alanine substitutions

(Figure 20B). Since metalloids alone augment the ATPase activity of ArsA, it seems plausible that Cys12, Cys13 and Cys18 are directly responsible for delivering arsenite to ArsA to enhance its affinity. In other words, the affinity for arsenite in the ArsD-ArsA complex would be largely determined by ArsD rather than by ArsA.

3.10 Effects of elimination of MBSs in ArsD on protein-protein recognition To verify if the ArsD mutants are stable and able to dimerize in yeast, the ability of the cysteine-to-alanine mutants to interact with wild-type ArsD were examined. All of the mutants are able to interact with wild-type ArsD, indicating that they are produced and stable in yeast (Figure 21A). Although the interaction with the mutant C12A/C13A ArsD seemed to be weaker than that with others, yeast two-hybrid results are not quantitative. To determine if the cysteine residues are involved in ArsD-ArsA interactions, the ability of six cysteine-to-alanine and three cysteine-to-serine ArsD mutants to interact with ArsA was examined. The effect of 100 μM Sb(III) or 50 μM As(III) in the medium were also tested. Two mutants were still able to interact with ArsA. These were ArsD 1-118,C112A andArsD1-118,C112A/C113A, in which Cys119 and Cys120 were replaced with a six-histidine tag, and Cys112 or both Cys112 and Cys113 were replaced with alanines (Figure 21B). Both mutants retain Cys12, Cys13 and Cys18. In contrast, the mutants lacking Cys12, Cys13 or Cys18 were unable to interact.

These include ArsD1-118,C13A, ArsD1-118,C18A, ArsD1-118,C12A/C13A, ArsD1-118,C12S, ArsD1-118,C13S, ArsD1-118,C18S (Figure 21B and C). This is consistent with a role for

the MBS1 in interaction with ArsA. However the mutant ArsD1-118,C12A, in which the Cys12 was changed to alanine, seemed to be able to interact with ArsA when incubated longer (Figure 21C) or in the presence of Sb(III) (Figure 22A).

However, the same mutant was unable to interact in the presence of As(III) (Figure 22B). This result is unexplained. Also the similar mutant ArsD1-118,C12S, in which the Cys12 was changed to serine, was unable to interact under all conditions. The requirement of Cys12 for ArsD to interact with ArsA in yeast two-hybrid is not clear at this point.

3.11 Protein-protein interaction domain on ArsD

3.11 Protein-protein interaction domain on ArsD