Domenico Gatti, Jian Wang, Xingjue Xu, Tongqing Zhou

     3-Deoxy-D-manno-octulosonic acid (KDO) is an 8-carbon sugar present in the lipopolysaccharide (LPS) of a wide variety of Gram-negative bacteria.  KDO provides a link between lipid A, the membrane embedded moiety of LPS, and the elongated polysaccharide chain that protrudes from the bacterial outer membrane into the surrounding environment.  3-Deoxy-D-manno-octulosonate 8-phosphate synthase (KDO8P synthase) is a key enzyme in the biosynthesis of KDO.  KDO8P synthase catalyzes the condensation of phosphoenolpyruvate (PEP) with arabinose 5-phosphate (A5P) to form KDO8P (precursor to KDO) and inorganic phosphate.  Previous studies have established that the synthesis of KDO8P proceeds through cleavage of the C-O bond of PEP and that the anomeric oxygen of the product derives from bulk solvent.  However, the detailed mechanism of this reaction is unknown.  In particular, there is controversy as to whether the reaction involves a linear or cyclic intermediate.  Furthermore, there is no information available regarding the identity of the amino acids involved in binding the substrates in the active site and in catalyzing the reaction.  Purely from the standpoint of enzymology, KDO8P synthase is an exciting enzyme to study since there is still much to be learned about the transfer of 3-carbon units in metabolism.  Moreover, the essential role played by KDO8P synthase in the synthesis of the LPS of Gram negative bacteria makes this enzyme an attractive target for antimicrobial chemotherapy; to fully realize the therapeutic potential of agents directed against KDO8P synthase requires detailed knowledge of the enzyme catalytic mechanism.  We are employing X-ray crystallographic analysis as the principal means for determining the details of the KDO8P synthase catalytic mechanism.  The specific goals of the project are:

 I. To determine the three-dimensional structure of the Escherichia coli KDO8P synthase by means of X-ray diffraction techniques.  Complete native data sets have been collected at the resolution limit of 3.0 Å from crystals flash-frozen at 140 K.  A 3-fold axis of non-crystallographic symmetry is observed in these crystals in accord with the trimeric architecture of the enzyme that has been established from solution studies.

 II. To determine the mechanism of KDO8P synthase.  The structure of KDO8P synthase will be determined in the presence of the individual natural substrates (A5P and PEP) and products (Pi and KDO8P), and with combinations of substrates and non-reactive substrate/product analogs and inhibitors.  The structure of KDO8P synthase will also be determined at different pH values to provide information on the chemical groups that contribute to the stabilization of the transition states in the active site.  Finally, site-directed mutants of KDO8P synthase will be analyzed by crystallographic means to test the importance of individual residues in the enzyme catalytic mechanism.

BACKGROUND
    Gram negative bacteria possess two membranes that separate the cytoplasmic compartment from the extracellular environment.  The inner membrane is the site for active transport of ions and nutrients, and for the synthesis of complex lipids and of cell wall components.  Its properties are those of a typical phospholipid bilayer.  The outer membrane has, instead, rather distinct characteristics; while the inner monolayer is composed mostly of phospholipids, the outer monolayer contains a unique molecule known as the lipolysaccharide (LPS).  LPS is often referred to as the endotoxin of Gram-negative bacteria.  This term was introduced in the 19th century to identify the bacterial component responsible for the shock syndrome associated with Gram-negative sepsis.  LPS is a potent activator of macrophages and results in the rapid release of tumor necrosis factor, interleukins and other protein and lipid mediators.

     The LPS of most Gram negative bacteria consists of four distinct regions:

1)  Lipid A, the hydrophobic part of the amphipathic LPS molecule, is a phosphorylated diglucosamine disaccharide containing seven fatty acid chains (Figure 2).  Phosphoryl groups are located at position 4’ of the non-reducing end and position 1 of the reducing glucosamine residue.  The phosphate group in position 4’ is partially substituted by an acylated 4-amino-4-deoxy-L-arabinose.  The phosphate at position 1 is partially substituted by phosphoryl ethanolamine.  The fatty acyl chains constitute approximately half of the outer monolayer of the outer membrane.
2)  The inner core region, attached to lipid A, is composed of two or three molecules of 3-deoxy-D-manno-octulosonate (KDO) and two or three heptose residues.  This region is fairly constant among species.
3)  The outer core region, a pentasaccharide whose covalent structure is more variable than that of the inner core.  The outer core provides the attachment site for the O antigen.
4)  The O antigen, a large polymer of repeating oligo(tri- to hexa-)saccharide units, which determines the major somatic antigen specificity of the organism.  The composition of the O antigen varies between species and also between strains.  Modification of the O antigen may help Gram negative bacteria to evade the immune system.  However, some pathogenic species lack the O antigen altogether.  The LPS of these organisms consists only of lipid A and KDO.

Fig. 2.  Structure of lipid A

    LPS is first assembled in the inner membrane, by the sequential addition of sugars to lipid A, and then transported to the outer membrane by an unknown mechanism.  Synthesis of lipid A proceeds in two steps.  Acylation of the glucosamine molecules to the full complement of seven fatty acids stops at an intermediate known as lipid IVA that has only four acyl chains.  Further acylation requires the addition of two molecules of KDO via the activated sugar nucleotide CMP-KDO.  The final structure containing the fully acylated lipid A and two KDO unit is the minimal LPS required for bacterial growth and is commonly known as the Re-endotoxin (see Fig. 1).
    The LPS biosynthetic pathway is an attractive target for multi-drug therapy, since mutants that produce incomplete LPS are both less pathogenic and more susceptible to known antibiotics.  Compounds that inhibit the synthesis of the lipid A have been reported recently.  However, although these drugs were shown to be bactericidal against Escherichia coli, they were ineffective in reducing lipid A content when used against living cells of Pseudomonas and Serratia.  As an alternative, the biosynthetic pathway for KDO is an excellent candidate for the development of new therapeutic agents that may be more globally effective as antibiotics.  Of prime significance for such ventures is that KDO is a sugar found exclusively in Gram negative bacteria and not in mammalian cells.  Notably, since all the enzymes of KDO biosynthesis can be purified easily for inhibition studies, work on the metabolic pathway for this sugar is greatly facilitated.

    An early step in the KDO biosynthetic pathway involves the synthesis of the phosphorylated precursor, 3-deoxy-D-manno-octulosonate 8-phosphate (KDO8P), and is catalyzed by a specific synthase.  Dephosphorylation of the precursor, yielding KDO, and synthesis of CMP-KDO (from CTP and KDO) occur prior to insertion of the sugar into LPS.  E. coli have been isolated with mutations in KDO8P synthase that confer temperature sensitive growth.  Such strains fail to synthesize KDO at the non permissive temperature, which leads to the inhibition of LPS, RNA and protein synthesis with the consequent arrest of cell growth.  These studies demonstrate that KDO8P synthase is a key enzyme involved in the maintenance of bacterial homeostasis.  The importance of this enzyme is further reflected by the high degree of conservation across different species: the only two amino acid sequences available for KDO8P synthase, from E. coli and Haemophilus influentiae, are 82% identical.
    KDO8P synthase from E. coli is composed of three identical subunits encoded by the kdsA gene.  The translated monomeric product contains 284 amino acids with a calculated Mr = 30,830 Da.  KDO8P synthase (EC 4.1.2.16) catalyzes the condensation of D-arabinose 5-phosphate (A5P) and phosphoenolpyruvate (PEP) to form a new phosphorylated eight-carbon saccharide (KDO8P) and inorganic phosphate (Pi) (Figure 3).

    It is of note that a reaction similar to KDO8P synthesis constitutes the first step in the shikimate pathway for the synthesis of aromatic amino acids: this reaction is the formation of 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAH7P) from erythrose-4-phosphate and PEP, catalyzed by the metalloenzyme DAH7P synthase (EC 4.1.2.15).  However, despite the similarity in the types of metabolic intermediates that are involved in these two reactions, the KDO8P and DAH7P synthases utilize different catalytic strategies.  This point is evident from the fact that only DAH7P synthase employs a metal ion.  Moreover, there is no amino acid sequence homology between the two enzymes.

    Previous studies on the mechanism of KDO8P synthase have established the following points:
1.  The Michaelis constants for the substrates, PEP (Km = 6 mM) and A5P (Km = 30 mM) have been determined.  Additional kinetic data have led to the suggestion that there is a sequential mechanism in which the binding of PEP precedes the binding of A5P and that the release of inorganic phosphate precedes the release of KDO8P.
2.  The reaction of KDO8P synthesis is irreversible.  This is shown by the lack of PEP/Pi exchange during the incubation of KDO8P synthase with [32Pi] in the presence of PEP and A5P.  Also, no scrambling of 18O from the bridging to the non-bridging positions of the phosphate of [18O]-PEP occurs in the presence of A5P.
3.  The condensation step is stereospecific, involving the addition of the si face of PEP C-3 to the re face of the A5P carbonyl.
4.  Phosphate release occurs by cleavage of the C-O bond of PEP.  When [18O]-PEP, specifically labeled in the enolic oxygen, is employed as the substrate for KDO8P synthase all the 18O is recovered in the released Pi that is released.
5.  The anomeric oxygen of the product KDO8P originates from bulk solvent.  When the reaction is carried out in the presence of H218O, KDO8P is recovered with [18O]-C2.

 Hedstrom and Abeles have proposed that a water molecule attacks at C-2 of PEP, followed by the addition of PEP C-3 to the electrophilic aldehyde of A5P, to yield an open chain intermediate.  Phosphate release is then accomplished by nucleophilic displacement by the 6-OH group of the intermediate (Mechanism I, Figure 4).

Fig. 4. Synthesis of KDO8P.  Mechanism I: Pi release occurs by displacement by the 6-OH forming the pyranose configuration

Fig. 5.  Structures of KDO8P and of some analogs and inhibitors.  The squiggle between carbon and oxygen in the structure shown in a indicates there are two possible configurations of the tetrahedral carbon.

Since the reaction catalyzed by KDO8P synthase is essentially irreversible, it has not been possible to use the natural substrates to analyze the putative intermediate under equilibrium conditions.  Information that challenges the existence of an open-chain intermediate has come from studies using inhibitors of KDO8P synthase (Figure 5).  Ray and coworkers reported that the dihydro analogs of KDO8P (Fig. 5, a), which mimic the putative linear intermediate in Mechanism I, are not inhibitors of the enzyme.  In contrast, the product KDO8P and the cyclic 2-deoxy analogs of KDO8P (Fig. 3, b,c) are weak inhibitors of the enzyme with Ki values between 0.3 and 0.6 mM.  These observation have led others to postulate an alternative mechanism based on the formation of a cyclic intermediate (Mechanism II, Figure 6).  According to this mechanism, an initial nucleophilic attack by the 3-hydroxyl of A5P on the C-2 of PEP is followed by condensation of C-3 of PEP on the carbonyl carbon of A5P, forming a cyclic pyranose intermediate (Fig. 6).  Phosphate release can then occur by either an SN2 displacement by water from "underneath" the ring (as shown in Fig. 6) or by oxonium formation on the ring oxygen followed by the addition of water to the anomeric carbon from either side of the ring (not shown).  In considering the probability of Mechanism II for KDO8P synthase it is noteworthy that a nucleophilic attack by a secondary hydroxyl upon C-2 of PEP was shown to occur in 5-enolpyruvylshikimate 3-phosphate synthase and in UDP-N-acetylglucosamine enolpyruvate transferase.  Furthermore, the fact that the isosteric C-2 phosphonate analog of the putative cyclic 2,8 diphosphate intermediate (Fig. 5, d) is the most potent inhibitor of KDO8P synthase (Ki = 5 mM) is in accord with Mechanism II.

Fig. 6. Synthesis of KDO8P.  Mechanism II:  Pi release may occur from attack by water from "underneath" the ring.
Evidence against the reaction course shown as Mechanism II for KDO8P synthase is provided by other observations:

1.  Although pyruvate would be expected to compete with PEP for binding to the active site, pyruvate is not a competitive inhibitor of the reaction.  This point suggests that the phosphate moiety is important for the binding of non-cyclic compounds.  Therefore, it is possible that the 2-dihydro analogs of KDO8P, which mimic the putative linear intermediate (Fig. 5, a), are not inhibitors of the reactions because they lack of the 2-phosphate.

2.  The a (Fig 5 f) and b (Fig 5 e) analogs of KDO with phosphate in the C-2 position are both inhibitors of KDO8P synthase.  Their overall binding is not particularly good (Ki values of 0.16 mM and 1.3 mM for the b and a diastereomer respectively), probably due to the lack of the 8-phosphate.  Despite the fact that b-KDO-2-P (Fig. 5, e) is remarkably similar in structure to the putative cyclic intermediate proposed by Baasov and others (Fig. 6), its phosphate moiety is not hydrolyzed by the enzyme.

     From the above discussion it may be concluded that additional work must be done to establish the catalytic mechanism for KDO8P synthase.  Of particular importance is discerning between the non-cyclic and cyclic reaction intermediates that have been postulated.  This point is addressed specifically by studies proposed in the current grant, which will use crystallographic analysis to visualize the intermediate in the active site of the enzyme.  Mutant forms of KDO8P synthase, whose catalytic defects may reflect an increased life-time of the reaction intermediate, are already available for this work.

     Interest for pursuing the structure determination of KDO8P synthase derives, in part, from the fact that inhibition of the enzymes involved in the synthesis and incorporation of KDO into LPS is detrimental to the survival and growth of Gram-negative bacteria.  The inhibition of KDO synthesis is of particular note since it may also produce a decrease in pathogenicity due to the accompanying loss of the O antigen, which is the LPS component that helps bacteria evade the host immune system.  Furthermore, underacylated precursors of lipid A accumulate in the absence of KDO, and these have lower endotoxin activity relative to mature LPS or the fully acylated and KDO-conjugated Re-endotoxin.  A specific inhibitor of 3-deoxy-D-manno-octulosonate cytidylytransferase (CMP-KDO synthetase Fig. 4) was designed on the basis of mechanistic studies of the purified enzyme and represents the first antibacterial agent that specifically inhibits LPS synthesis.  The structure determination of KDO8P synthase will provide the physico-chemical foundations for the rational design of a new class of antibiotics that attack Gram-negative bacteria by interfering specifically with the function of KDO8P synthase.

     Of added significance is the fact that the E. coli KDO8P synthase bears sequence similarity to two other enzymes that also recognize a pyruvyl moiety.  These are the bifunctional 3-deoxy-D-arabino-heptulosonate 7-phosphate synthase/chorismate mutase from Bacillus subtilis and the chorismate mutase from Staphylococcus xylosus.  Chorismate mutases are a very heterogeneous class of enzymes, which differ widely both in primary sequence and tertiary structure (see below).  In particular, the aforementioned chorismate mutases from B. subtilis and S. xylosus, which are similar to KDO8P synthase, are not homologous to any of the other chorismate mutases for which a structure is available.  Therefore, the studies of KDO8P synthase will also provide a structural and mechanistic model for a new class of enzymes that catalyze the transfer, or rearrangements, of three-carbon units.
 
 

     The arsenical resistance (ars) operon of the clinically isolated Escherichia coli plasmid R773 encodes a system for the active extrusion from cells of the toxic oxyanions arsenite (As(III)O21-) and antimonite (Sb(III)O21-) via an ATP-driven pump.  The ars operon has five genes.  The arsA and the arsB gene products are, respectively, the catalytic subunit (ATPase) and the membrane subunit of the pump.  The arsC gene product extends the range of resistance to arsenate (As(V)O43-) and tellurite (Te(V)O43-) by catalyzing the reduction of these oxyanions to substrates for the ArsA-ArsB pump.  The products of the arsD and arsR genes are gene repressors that control the expression of the other three genes.
     The goal of our research is to elucidate the molecular mechanisms of the ArsA, ArsB, ArsC detoxifying system by analyzing the crystal structure of the individual components.  This study includes the following components:

I. Structural Studies with the ArsA ATPase:  The main goal is to define the regions of the protein involved in function, with specific attention to:  a) the activator binding domain;  b) the nucleotide binding sites;  c) the site of interaction with ArsB.  Crystals of ArsA diffract to at least 1.9 Å.  Native data sets have been collected with crystals frozen at 140 K.

II. Structural Studies with the ArsC Reductase:  ArsC appears to be the first identified member of a new class of reductases that use a single thiol as a donor of reducing equivalents.  Crystals of this protein diffract to at least 2.0 Å.  Native data sets have been collected both at room temperature and with crystals frozen at 100 K.  Of particular interest will be the identification of the oxyanion binding site (As(V)O43-) and of the residues involved in electron transfer.  ArsC requires glutathione and glutaredoxin to reduce arsenate to arsenite and appears to channel arsenite directly into the ArsA-ArsB pump.  Efforts will be made to identify putative regions in ArsC that make contact with glutaredoxin, and with the components of the ArsA-ArsB pump.

III. Structural Studies with ArsB:  This membrane protein is the only component of the ars detoxifying system for which crystals are not yet available.  Methods for the overproduction and purification of ArsB will be developed in the early phase of the project as a prelude to the structural determination of this protein.

BACKGROUND:
     Arsenic-based compounds were the first substances to be used effectively in antimicrobial chemotherapy.  The arsenical Salvarsan was Paul Ehrlich's "Silver Bullet" against syphilis and sleeping sickness.  Although Salvarsan is no longer used, leishmaniasis is treated with the antimonial Pentostam, and trypanosomal diseases are treated with the arsenical melarsen oxide.  Therefore, the emergence of resistance to arsenical and antimonial compounds is a phenomenon of great concern in the treatment of these tropical diseases.  In his Nobel Lecture on December 11, 1908 Ehrlich pointed out that "if a substance is able to kill, this can happen only because it accumulates in cells" and suggested that arsenic resistance might arise from the inability of resistant cells to concentrate arsenicals.  This prediction was borne out almost a century later by the explosion of studies on drug resistance, and it is now widely recognized that one of the most frequently employed strategies for drug resistance by both prokaryotes and eukaryotes is the transport of the toxic compound out of the cell.  For example, a common mechanism based on active extrusion from cells underlies resistance to arsenicals and antimonials in bacteria, protozoa, and mammals.  Drug resistance in Leishmania involves first reduction of As(V) or Sb(V), then conjugation of the trivalent semimetal to a thiol such as GSH or trypanothione, and finally ATP-dependent extrusion by a metal-thiol pump.  In gram-positive and gram-negative bacteria plasmid-mediated resistance to arsenic and antimony is wide-spread and appears to be mediated by the extrusion of the unconjugated anion.
    The detoxifying system encoded by the ars operon of plasmid R773 in E. coli confers resistance against oxyanions of arsenic (arsenite and arsenate) and antimony (antimonite).  The ars operon codes for two regulatory (ArsR and ArsD) and three structural (ArsA, ArsB and ArsC) proteins.  The arsA and arsB gene products form a membrane bound anion pump.  ArsC reduces arsenate to arsenite, which is then extruded by the Ars pump.

Fig. 1.  The ars  operon of E. coli  plasmid R773 and its products.

     ArsA is a 63 kDa peripheral membrane protein that catalyzes As(III)-stimulated ATP hydrolysis.  An interesting feature of ArsA is that the amino and carboxyl terminal halves of the protein show 20% identity.  There are two nucleotide-binding consensus sequences (P-loops) in ArsA, one in each homologous half of the protein (see A1 and A2 in Fig. 2), but only a single adenine site has been identified in the center of the polypeptide chain by means of UV cross-linking studies.  Notably, both the A1 and A2 sites bind nucleotides independently, and mutations in either site can eliminate the ATPase activity of the protein.  Genetic and biochemical studies have shown that both the A1 and A2 nucleotide binding sites (NBS) are required for actions associated with arsenic resistance, namely arsenite transport and ATPase activity.  Genetic complementation and biochemical reconstitution experiments also suggest that the formation of a catalytic site requires interaction of the A1 and A2 sites.  The ATPase activity of ArsA is activated allosterically by As(III) and Sb(III) and there is evidence that binding of these metalloids favors the formation of protein dimers.  On the basis of experiments of chemical modification with sulfhydryl reagents, and of site-directed mutagenesis of the cysteine residues, it has been proposed that the allosteric activation of ArsA is mediated by the formation of a three-coordinate complex between Sb(III) or As(III) and the thiolates of cysteines 113, 172 and 422, with the three thiol groups forming the base of a pyramid (Fig. 2).  A similar coordination geometry is observed in the complex of dithiothreitol and As(III).  Analysis of model compounds also suggests that the bond angles and distances of the coordination site must be different with Sb(III) (Sb-S = 2.45 Å) and with As(III) (As-S = 2.25 Å).  The expected differences in the site geometry and in the structural perturbations emanating from this site might provide the physical basis for the fact that Sb(III) is a much better activator of ArsA than is As(III).

     ArsB is a 45 kDa integral membrane protein to which ArsA binds to form the ArsA-ArsB anion pump.  The function of the ArsA-ArsB complex has been studied both in vivo and in vitro.  Studies performed in vivo using cells that lack a functional F0F1 have shown that the ArsA-ArsB pump extrudes AsO2- in a manner that is independent of electrochemical energy from respiration.  Interestingly, in mutants that are defective for ArsA (the ATPase moiety), arsenite extrusion appears to be DmH+-dependent.  In vitro studies with everted membrane vesicles containing both ArsA and ArsB show that the ArsA-ArsB complex catalyzes ATP-dependent, uncoupler-insensitive, accumulation of arsenite.  These observations suggest that an electrochemical proton gradient is not necessary for arsenite transport by the ArsA-ArsB complex.  In contrast, arsenite extrusion by ArsB alone is coupled to electrochemical energy, most likely in the form of a positive potential at the periplasmic face of the membrane.  Thus, ArsB has the ability to function either as a subunit of the ArsA-ArsB ATP-dependent pump or independently as a passive translocator (Fig. 2).  These conclusions are supported by the observation that the chromosomal ars operon of E. coli, and the ars operons of the staphylococcal plasmids pI258 and pSX267, code for only three genes (arsRBC) and are notably lacking the arsA gene.  These operons still provide resistance to arsenic, however the level of resistance is lower in comparison with that observed when the plasmid-encoded ArsA-ArsB pump is present.

Fig. 2.  Dual mode of action of ArsB.

 The membrane topology of ArsB has been investigated through the use of gene fusions and it has been suggested that the ArsB polypeptide spans the bacterial inner membrane 12 times (Fig. 3).  Very little is known about the function of specific amino acid residues in ArsB.  Five arsB genes have been sequenced, three from E. coli plasmid or chromosomal DNA and two from staphylococcal plasmids.  Interestingly there are very few conserved charged residues in the putative membrane spanning regions.  The charged residues may form ion pairs that stabilize the helices or may be part of a water filled channel.  Several helices are amphipathic, such that the channel might be composed of hydrophilic residues contributed from two or more helices.  In contrast to ArsA, in which activation of ATPase activity results from the interaction of the soft metal As(III) with cysteine thiolates, ArsB contains only a single cysteine residue that can be altered without effect on resistance or transport.  The fact that ArsB does not use soft metal-thiol chemistry has led to the suggestion that the arsenic bound to the allosteric site in ArsA is not the arsenic that is transported by ArsB.

Fig. 3.  Membrane topology of ArsB.

    To function as an active efflux pump, ArsB must interact with the catalytic ATPase subunit, ArsA.  Although ArsA can be isolated as a soluble protein when expressed in excess over ArsB, physiologically it is found as a complex with ArsB.  The ArsA-ArsB intermolecular associations are strong and are broken only in the presence of denaturing agents such as urea.  ArsA binds to the cytoplasmic face of ArsB, which is modeled to include the N- and C-termini of the protein and five inter-helical loops (Fig. 3). Insertional mutagenesis was employed to identify the ArsA binding site(s).  Insertions at the N-terminus do not alter the pump function, while small insertions in the C2 and C5 loops result in a decreased level of resistance that might reflect a function for these loops in ArsA binding.

     ArsC is a small (14 kDa) reductase that converts arsenate to arsenite, the substrate of the ArsA-ArsB pump.  This function of ArsC is particularly beneficial to cells, which become resistant also to the pentavalent state of arsenic.  ArsC type proteins are present both in Gram negative and Gram positive bacteria, although their amino acid sequence is only marginally related (< 20% similarity).  The ArsC protein encoded by plasmid pI258 in arsenate-resistant strains of Staphylococcus aureus catalyzes thioredoxin-linked reduction of arsenate.  In contrast, in Gram negative cells that show resistance associated with the R773 plasmid, the source of the reductant is glutathione (GSH) rather than thioredoxin.  This difference in the source of reducing potential may reflect the availability of reductants in the host strains:  for example S. aureus contains little or no glutathione, while glutathione is the major intracellular reductant in E. coli.
     Glutathione-dependent arsenate reduction by ArsC requires glutaredoxin (GRX, Fig. 1), a small protein with redox active sulfhydryls that participate in the transfer of electrons from GSH.  The KM of ArsC for glutaredoxin is low (0.02 µM) and the kCAT for the reaction is less than 1 sec-1.  The slow turnover rate of ArsC in vitro may reflect either a mechanistic difficulty akin to the reduction of the arsenate species, or the need for direct association with the ArsA-ArsB pump (see below). The KM of ArsC for arsenate is 8 mM.  In E. coli, arsenate influx occurs via phosphate transport systems and the free pool of phosphate in aerobic E. coli cells is approximately 3 mM.   Thus, the ArsC KM for arsenate probably reflects the fact that arsenate concentration in cells may reach values in the millimolar range.  The product of the reaction (arsenite) is a competitive inhibitor of ArsC, with a KI of 0.1 mM.  The higher affinity for product is consistent with the slow turnover rate and the necessity to keep low the intracellular concentration of arsenite.  Phosphate and sulfate are both low affinity competitive inhibitors, and neither is reduced by the ArsC protein.  This result suggests that, while the anion binding site is somewhat nonspecific, the mechanism of catalysis is quite stringent in terms of ligand requirement.
     Spectroscopic analysis has established that ArsC contains no detectable flavin, heme or nonheme iron, and studies with mutants lacking the molybdenum cofactor have ruled out the participation of molybdenum in the ArsC reaction.  N-ethylmaleimide (NEM) was found to strongly inhibit the ArsC reductase activity, supporting a role for cysteine residues in the reaction.  ArsC has two cysteines, C12 and C106, and a data base search indicates that only C12 is conserved in the three closest homologs.  Mutagenesis studies support the view that cysteine 12 is directly involved in the transfer of electrons to arsenate.  Notably, the product of the reaction, arsenite, reacts almost irreversibly with thiol pairs, but shows a much lower affinity for monothiols.  It follows that if a cysteine pair was involved, it would be difficult to release the product.  Thus, ArsC is proposed to act as a single thiol reductase.
     High level resistance to arsenate requires the expression of all three ars genes (arsA, arsB, arsC).  Interestingly, some strains of E. coli that lack the ars operon show endogenous rates of arsenate reduction that equal those in cells expressing the arsC gene (Rosen, unpublished).  However, the arsenate sensitivity of these cells is not cured by the expression of the arsA and arsB genes.  Therefore, it would appear that arsenate reduction in cells that have the ArsA-ArsB pump is not sufficient to establish resistance unless the reduction is catalyzed by ArsC.  One possible explanation for this observation is compartmentalization of the arsenite produced by the ArsC protein; it is proposed that the ArsC protein reduces arsenate to arsenite in association with the ArsA-ArsB pump, channeling arsenite directly into the active site of the pump.

SIGNIFICANCE
     Arsenical resistance is a useful model for the study of multiple drug resistance in cancer.  Mammalian tumor cells develop resistance to chemotherapeutic drugs by amplification of the mdr gene that encodes the P-glycoprotein, an active extrusion pump for drugs.  The plasmid encoded Ars pump, composed of the ArsA protein (the ATPase subunit) and the ArsB protein (the membrane subunit), exhibits functional similarity to the P-glycoprotein.  The manner in which the Ars system recognizes and detoxifies diverse substrates is becoming progressively more clear and may shed some light on the mechanism of multidrug recognition by the P-glycoprotein.  It should be pointed out that the closest homolog (of unknown function) to the ArsA protein has been found in C. elegans, and one of the few homologs of the ArsB protein was recently found to be a membrane protein whose mutation results in human type II oculocutaneous albinism.  Finally, an ArsC homolog (of unknown function) was found in an operon that encodes a Streptococcus pneumoniae oligopeptide transport protein.  Arsenical resistance has been observed also in mammalian cell lines Leishmania and trypanosomes.  Strains of Leishmania resistant to the antiprotozoal agent Pentostam are cross-resistant to arsenite and antimonite, and it appears that the resistance is associated with an efflux mechanism that is similar to the Ars system.
The arsenical pump provides also an excellent model system for the study of ion-translocating ATPases.  As a biochemical complex the arsenite-translocating ATPase combines the best features of FOF1 and P-type ATPases.  It has only two types of required subunits, compared with 8 or more for the F0F1 H+-translocating ATPases.  The arsenite pump has a catalytic subunit (the ArsA protein) that can be removed from the membrane and studied as a soluble enzyme, an advantage over P-type cation-translocating ATPases.  F1 can also be removed from the membrane, but it is a complex of different subunits, whereas the arsenite pump catalytic portion is composed of only the ArsA protein.  The ability to overproduce and purify easily the catalytic subunit is another advantage.  Furthermore, since the bacterial ars genes are organized in an operon, the system is amenable to a level of genetic analysis that has not been available for the eukaryotic pumps.  The arsenical pump also has advantages over other bacterial plasma membrane pumps.  Genetic analysis of cation pumps in gram positive organismsis more complicated than in E. coli.   The Kdp K+-ATPase of E. coli and the Mg2+-ATPase of S. typhimurium are excellent systems to study genetically but less amenable to biochemical studies.
     Finally, the evidence suggesting that the arsenic bound to the allosteric site in ArsA is not the arsenic transported by ArsB is noteworthy.  Under this point of view it would appear that the Ars pump uses two distinct arsenic (antimony) chemistries:  1) soft metal binding to allosterically activate the energy-transducing (ArsA) subunit and 2) oxyanion transport through the membrane-spanning ArsB subunit.  This separation of activation and transport domains could be present in other transport ATPases.  For example, the P-type ATPases for Cd(II) (46) and Cu(II) (47) contain cytosolic N-terminal domains with single or multiple cysteine pairs that may function as activation domains.
 

   Structure and Function of p-Hydroxybenzoate Hydroxylase

	In our laboratory we run several flavors of UNIX (Irix, Linux, etc.).  We thank the GIG Animation Gallery and
	for providing us with most of the GIF animations featured in our site.