Heavy Metals in the Environment - Chapter 15 pps

The first bacterial P-type ATPase identified was the Kdp K⫹-translocating ATPase, a pump for up-take of monovalent ions of the hard metal potassium 6.. Other hard metal bacte-rial P-type A

Microbial Resistance Mechanisms for Heavy Metals and Metalloids

Mallika Ghosh and Barry P Rosen

Wayne State University School of Medicine, Detroit, Michigan

1 INTRODUCTION

In this chapter the mechanisms of resistance to ions of the heavy metals zinc, cadmium, lead, copper, arsenic, and antimony in bacteria will be described In

addition, the pathways of arsenical resistance in the prokaryote Saccharomyces cerevisiae will be discussed Although often grouped as heavy metals, the ions

of these metals are better characterized as soft metal ions, which are those with

high polarizing power (a large ratio of ionic charge to the radius of the ion), in

contrast to the hard metal ions of Groups I and II such as Na⫹and Ca2 ⫹ This distinction between hard and soft metals is important biologically When consid-ering how these metals interact with biological molecules such as proteins, hard metals most frequently bind to proteins weakly through ionic interactions with functional groups such as carboxylates of glutamate or aspartate residues In con-trast, soft metals form much stronger, nearly covalent, bonds with functional groups such as the thiolates of cysteine residues and the imidazolium nitrogens

of histidine residues These strong interactions with proteins account for much

of the biological toxicity of soft metals

All of these ions are toxic in excess, and bacterial metal resistances proba-bly arose early in evolution owing to widespread geochemical sources of metals

Resistance genes to inorganic salts of soft metals, including arsenic, antimony, lead, cadmium, copper, nickel, zinc, bismuth, and mercury, are found both on extrachromosomal plasmids and in chromosomes of bacteria, archaea, and eu-karyotes For metals such as copper and zinc, which are required in low amounts but toxic in high amounts, the efflux systems are components of the homeostatic mechanisms that maintain intracellular concentrations at optimal levels Recent reviews have been written on bacterial metal resistances (1–3), which allows this chapter to focus on a few specific mechanisms

2 P-TYPE ATPases FOR MONOVALENT SOFT METALS:

Cu(I) AND Ag(I) PUMPS

P-type ATPases comprise a superfamily of enzymes that transport cations (4) Some pump cations into cells, some into organelles, others pump cations out of cells, and yet others are cation exchangers: it is not possible to predict the direc-tionality of transport from inspection of the sequence Every member of the super-family has conserved sequences that include an ATP-binding domain, an aspar-tate residue that is the site of phosphoenzyme formation, and a phosphatase domain There are at least five branches to the family (5) One branch includes the hard metal translocating ATPases, and another the soft metal cation-translocating ATPases (Fig 1) Both branches can be further subdivided into branches that comprise monovalent and divalent pumps (1) The first bacterial P-type ATPase identified was the Kdp K⫹-translocating ATPase, a pump for up-take of monovalent ions of the hard metal potassium (6) Other hard metal bacte-rial P-type ATPases include the MgtA and MgtB pumps for uptake of divalent ions of the soft metal magnesium (7)

The soft metal ATPases were identified more recently They have common features not present in the hard-metal ATPases, in particular characteristic cyste-ine- or histidcyste-ine-rich metal-binding motifs at the N-terminus and a Cys-Pro-Cys (or His) sequence in the sixth membrane-spanning segment (Fig 2) The presence

of the CPC(H) sequence has led to the designation CPx-type ATPases (8)

How-ever, given that the substrates of this class of enzymes are all soft metal ions,

soft metal P-type ATPases seems an appropriate designation The most widely

recognized members of the soft metal P-type ATPases are the human copper pumps ATP7A (or MNK) (9) and ATP7B (or WIND) (10) Inheritable mutations

in the genes for these pumps produce Menkes or Wilson disease, respectively Both MNK and WND have six CXXC metal-binding sequences in their N-termini and a CPC sequence in a membrane-spanning segment (11) Peptides correspond-ing to the N-terminal metal bindcorrespond-ing domains have been shown to bind a variety

of metal ions, including Cu(II), Cu(I), Ag(I), and Zn(II) (12,13) The function

of the N-terminal metal binding domains is not known Possibilities include: (1)

F IGURE 1 The family of soft metal ion-translocating P-type ATPases This family is growing rapidly, so only a few representative ones are shown to illustrate the nature of the lineage The dendograms were made using the CLUSTAL4 algorithm (84) with DNASIS software from Hitachi Software Engi-neering Co., Ltd On the left is the hard metal ion ATPase subfamily, repre-sented by Pmal, the fungal H⫹-translocation ATPase (85), NaK, the mamma-lian Na,K- ATPase (86), and SERCA1, the endoplasmic reticulum calcium pump (87) On the right is the soft metal ion ATPase subfamily Two major branches are shown, those with histidine-rich N-termini such as the Cu(I)/

Ag(I)-translocaing CopB ATPase of E hirae (17) and the histine-rich proteins

Hra1 and Hra2 of unknown origin (20) The other branch has two groups of proteins, the monovalent copper pumps, including the human enzymes MNK (9) and WND (10) associated with Menkes and Wilson diseases, and the

E coli Cu(I)-translocating CopA ATPase (25) The second grouping are the

divalent soft metal ion pumps for the metals Zn(II), Pb(II), and Cd(II), including

the S aureaus plasmid pI258 CadA ATPase (26), the ZntA pumps of E coli (30,31), and Proteus mirabilis (88).

the metal binding domains serve as sensors to activate the pump; (2) they serve

as the initial binding site, transferring the metal ion to the translocation domain, which probably includes the CPC motif; (3) in eukaryotes they may be involved

in trafficking of the pump to the appropriate membrane In support of the last possibility, the metal-binding domains of either the Menkes or Wilson proteins

do not appear to be essential for copper transport, but their removal appears to

F IGURE 2 Generic model of a soft metal ion-translocating P-type ATPase This model of a soft metal ion-translocating P-type ATPase illustrates the common features of these pumps (adapted from ref 11) Most pump either monova-lent or divamonova-lent soft metals from the cytosol, either into an intracellular

com-partment or out of the cell Some, such as E hirae CopA, pump ions into the

cell They have an N-terminal region with one or more cytosolic metal-bind-ing domains, most of which are cysteine-rich motifs but some are histidine-rich sequences They have eight transmembrane segments (TMS) The cyto-solic loop that connects TMS4 and TMS5 is the conserved phosphatase domain TMS6 has the consensus CPC or CPH sequence, which could be in-volved in metal ion translocation In the cytosolic loop that connects TMS6 and TMS7 regions involved in catalysis, the phosphorylation and ATP-bind-ing domains are found In addition, there is a conserved His-Pro sequence that is found only in soft metal ion-translocating P-type ATPases

affect trafficking (14,15) On the other hand, the N-terminal metal-binding do-mains may have a different function in bacteria, most of which lack intracellular membranes that would require trafficking signals

The first bacterial proteins identified as members of the monovalent cation-translocating ATPase subfamily were the CopA and CopB copper pumps of

Enterococcus hirae (16,17) The genes for these two P-type ATPases are in an

operon that also contains the gene for a transcriptional regulator, CopY, and a copper ion chaperone, CopZ While CopA and CopB are cotranscribed, they have considerable differences both structurally and physiologically A disruption in

the copA gene renders cells copper-requiring, suggesting that CopA is required for copper import pump In contrast, a copB disruption renders cells copper

sensi-tive, suggesting that the 745-residue CopB protein is responsible for copper efflux from cells That two such similar pumps transport their substrates in opposite directions illustrates the point that the directionality of transport cannot easily be deduced from inspection of the primary sequence of the proteins Together the two pumps provide for copper homeostasis

CopB has a histidine-rich N-terminus Of the 25 histidine residues in CopB,

16 are located in the first 100 residues, before the first transmembrane segment

It seems reasonable that this region binds copper In the sixth putative transmem-brane region is a CPH sequence that corresponds to the CPC of other soft metal P-type ATPases CopB has been shown to be an efflux pump involved in

resis-tance to copper in E hirae (18) To demonstrate the direction of transport, Solioz

and co-workers prepared everted membrane vesicles and showed that the vesicles accumulated64Cu(I) and110Ag(I) The affinity for metal ion was in the micromo-lar range, and for ATP the Kmwas approximately 10µM Vanadate, the classical inhibitor of P-type ATPases, inhibited CopB activity CopB has been purified and reconstituted into proteoliposomes (19) CopB formed an acylphosphate reaction intermediate with theγ-phosphate of ATP, and formation of the phosphorylated intermediate was sensitive to vanadate The purified protein exhibited a low level

of ATPase activity that was also inhibited by vanadate However, ATP hydrolysis was not stimulated by copper ion, although a P-type ATPase would be expected

to require its metal ion substrate for activity

The sequences of two homologues of CopB, Hral and Hra2, have been

reported (20) These were originally described as Escherichia coli proteins How-ever, they are not found in the E coli genome (21), and their derivation is

un-known Another soft metal P-type ATPase with a histidine-rich N-terminus is

the Ag(I) resistance pump SilP from Salmonella typhimurium (22) Of the 25

histidines in SilP, 15 are located in the N-terminal region, including a highly charged stretch of residues with the sequence EHHHHHDHHE However, the N-terminus of SilP exhibits little sequence similarity with the N-terminus of other soft metal P-type ATPases, including CopB, and SilP confers resistance to silver but not copper SilP and CopB are the only two Ag(I) resistance pumps thus far

identified, and both have histidine-rich (albeit unrelated) N-terminal metal-bind-ing domains This may indicate that Ag(I) recognition is via histidine residues rather than the CXXC motifs of the majority of Cu(I) pumps

The 727-residue CopA protein has a single CXXC sequence in the N-termi-nal region, and a CPC sequence in the putative sixth transmembrane segment (17) Sequences for other bacterial CopA homologs have been identified, and the list is growing daily However, the physiological function of only a few of these sequences has been described, and even fewer have been investigated

biochemi-cally Two of the best characterized are the CopA pumps of Helicobacter pylori (23,24) and E coli (25) As mentioned above, disruption of the E hirae copA results in a copper requirement In contrast, disruption of either of the H pylori

or E coli pumps results in copper sensitivity This reflects the fact that the

E hirae CopA is an uptake system, while the other two homologs are efflux

pumps Again, it is quite remarkable that the direction of ion transport can be inward for one protein and outward for others that are close homologs The expla-nation cannot be that the proteins have the opposite orientation in the membrane because each uses ATP, which is found only in the cytosol; the catalytic domains

of all three must be exposed in the cytosol On the other hand, the topological orientation of the protein in the membrane has been determined only for the

H pylori enzyme (24) The protein has been shown to have eight transmembrane

segments, with cytosolic N- and C-termini It is reasonable to assume that all soft metal P-type ATPases will have a similar topology

CopA from E coli is an 834-residue protein with high similarity to copper pumps such as E hirae CopA, the human Menkes and Wilson disease proteins (25) While the CopA homologs from E hirae and H pylori have only a single N-terminal CXXC sequence, E coli CopA has two, G11LSCGHC and

G107MSCASC The presence of multiple metal-binding domains in the E coli

protein may make it a better model for the human copper pumps, which have six N-terminal CXXC sequences CopA can be predicted to have eight transmem-brane segments, including C479PC in predicted transmembrane helix 6 In addi-tion, there is a conserved HP motif in the soft metal P-type ATPases that corre-sponds to H562P in CopA Regulation and resistance exhibit different metal ion

specificities The E coli copA gene is inducible by addition of either copper or silver salts to the medium In contrast, disruption of copA resulted in sensitivity

to copper salts but not Ag(I) Thus there must be an as-yet-unidentified regulatory

protein that controls copA expression, and that regulator recognizes either Ag(I)

or Cu(I), while CopA recognizes only Cu(I) Everted membrane vesicles from

cells expressing copA accumulated64

Cu ATP and DTT were both required, and vanadate inhibited transport Even though64Cu(II) was added to the uptake assay,

it would be reduced to Cu(I) by the strong reductant DTT The fact that no trans-port of copper ion was observed without DTT strongly indicates that Cu(I) is a substrate of the pump (25)

3 P-TYPE ATPases FOR DIVALENT SOFT METALS:

PUMPS FOR Zn(II), Pb(II), AND Cd(II)

The second branch of the soft metal P-type ATPases are those for divalent soft metal ions, including Zn(II), Pb(II), and Cd(II) They can be further subgrouped into the CadA ATPases, which are found mainly in gram-positive bacteria, and ZntA ATPases, which are mainly in gram-negative bacteria (1) The first gene

for a divalent soft metal P-type ATPase to be identified was cadA, a cadmium-resistance determinant on Staphylococcus aureus plasmid pI258 (26) The cadA

gene encodes a 727-residue P-type ATPase that exhibits considerable sequence

similarity to the CopA Cu(I)-translocating ATPases When the cadA gene was expressed in Bacillus subtilis, the CadA protein was produced as a membrane

protein that could be visualized on sodium dodecyl sulfate polyacrylamide gel electrophoresis (27) CadA was shown to form a phosphorylated intermediate during the catalytic cycle (28) The phosphoenzyme intermediate was formed only in the presence of Cd(II) andγ-[32P]ATP and was sensitive to hydroxylamine treatment, which is diagnostic of an acylphosphate bond Presumably phosphory-lation occurs at Asp415, which corresponds to the conserved aspartate residue

in all P-type ATPases Everted membrane vesicles prepared from B subtilis ex-pressing cadA accumulate109Cd(II) in an ATP-dependent manner (27) The

topol-ogy of a CadA homolog from H pylori has been determined Like the H pylori

CopA, it has eight transmembrane segments (29)

The zntA gene was first identified from sequencing of the E coli genome From its sequence, the zntA gene product could not be differentiated from copper P-type ATPases However, disruption of zntA resulted in sensitivity of E coli cells to Zn(II), not Cu(II) (30,31) The fact that zntA is a chromosomal gene suggests that it has a function in normal growth of E coli, probably in zinc homeostasis In everted membrane vesicles of E coli, ZntA catalyzes

ATP-cou-pled accumulation of65Zn(II) or109Cd(II) in a vanadate-sensitive reaction (30)

Vesicles made from a strain with a zntA disruption did not accumulate65Zn(II)

When zntA was expressed on a plasmid, vesicles from the disrupted strain

accu-mulated 65Zn(II) Although no isotope of Pb(II) is available, the data suggest

that ZntA transports Pb(II) A zntA-disrupted strain was nearly three orders of

magnitude more sensitive to Pb(II) than the wild-type strain (32) Moreover,

65Zn(II) transport in vesicles was inhibited by either Pb(II) or Cd(II) Similar

results were obtained when the zntA-disrupted strain expressed the S aureus cadA

gene on a plasmid Neither conferred resistance to Cu(II) nor catalyzed uptake

of64

Cu in vesicles Thus both ZntA and CadA are extrusion pumps for the diva-lent soft metal ions Zn(II), Pb(II), and Cd(II)

ZntA has been purified and shown to exhibit ATPase activity at rates equiv-alent to those for the hard metal P-type ATPases (33) ATP hydrolysis required

a divalent soft metal ion, the first metal-ion-dependent ATPase activity by a soft

metal P-type ATPase to be clearly demonstrated The activity was stimulated by (in order of effectiveness) Pb(II)⬎ Cd(II) ⬃ Zn(II) ⬃ Hg(II) Although the free metal ions stimulated ATP hydrolysis, the rates were higher when the soft metal ions were complexed with thiolates of cysteine or glutathione In fact, free Cd(II)

or Hg(II) inhibited activity at neutral and alkaline pHs In vivo the concentration

of glutathione is in the millimolar range (34), and it is likely that soft metal ions

do not exist free These results raise the interesting possibility that the soft metal P-type ATPases recognize metal-glutathione conjugates in vivo

ZntA and CadA homologs appear to be widespread in nature, although they have not yet been found in animals It is likely that these evolved for zinc homeostasis: zinc is required for a number of enzymes and transcription factors but is toxic in excess, so cells must have mechanisms to prevent overaccumula-tion Although most of the zinc pumps identified to date are prokaryotic, genes for homologs have been identified in other kingdoms, for example in genome of

the archeaon Methanobacterium thermoautotrophicum and in the genome of the plant Arabidopsis thaliana It would be of interest to know whether humans have

a ZntA homolog and, if so, whether there are inheritable diseases related to muta-tions in zinc pumps Since these P-type ATPases are also the first translo-cating pumps to be identified, it is possible that differential expression in Pb(II)-exposed individuals may produce variations in Pb(II) sensitivity in humans The answers to questions such as these will become clear as genome projects are completed

4 THE ArsAB ATPase: AN As(III)/Sb(III) EFFLUX PUMP

In E coli high-level resistance to As(V), As(III), and Sb(III) is conferred by the arsRDABC operon of plasmid R773 (35) The arsC gene encodes an arsenate

reductase that converts As(V) to As(III) As(III) and Sb(III) are the substrates

of the ArsAB efflux pump, which is an As(III)/Sb(III)-translocating ATPase The 429-residue ArsB subunit is an integral membrane protein located in the inner

membrane of E coli The results from construction of a series of gene fusions between arsB and phoA, lacZ, or blaM demonstrate that ArsB has 12

membrane-spanning segments, with the N- and C-termini in the cytosol (36) ArsB can function even in the absence of the ArsA subunit to catalyze translocation of the arsenite anion, with energy supplied in the form of a membrane potential (37) (Fig 3A) The ArsAB complex is able to couple ATP hydrolysis to the transport

of arsenite, making it a more efficient efflux system than ArsB alone (Fig 3B)

As a semimetal or metalloid, arsenic can have nonmetallic properties, existing

in solution as the oxyanions arsenate and arsenite Alternatively, As(III) and Sb(III) have properties similar to those of soft metals, interacting with high affin-ity with thiolates ArsB has only a single cysteine residue that is not required for

F IGURE 3 The ArsAB pump The E coli Ars arsenite translocator has two

modes of energetics (37) (A) ArsB is an inner-membrane protein that can function as a secondary∆ω-coupled uniporter (B) The ArsAB pump is a com-plex of the ArsA ATPase and ArsB membrane carrier and functions exclu-sively as a primary ATP-driven pump ArsA has two halves, A1 and A2, which are homologous to each other and are connected by a 25-residue linker pep-tide Hydrolysis of ATP by ArsA is coupled to arsenite or antimonite transloca-tion by ArsB

ArsB catalysis; thus ArsB does not interact with As(III) as soft metals but must recognize and transport the nonmetallic, oxyanionic forms of the metalloids (38)

In contrast, the 583 residue ArsA protein, which is the catalytic subunit of the pump, is allosterically activated by the soft metals As(III) or Sb(III) (39) In the absence of its partner, ArsB, ArsA can be expressed and purified as a soluble protein ArsA has two halves, A1 and A2, connected by a 25-residue linker A1 and A2 are homologous to each other, clearly the result of an ancient gene dupli-cation and fusion Both A1 and A2 have a consensus nucleotide-binding domain (NBD), both of which are required for activity (40,41)

As(III) or Sb(III) specifically stimulates ATP hydrolysis (39) It is clear that they do so as soft metals, with the thiolates of Cys113, Cys172, and Cys422 serving as ligands to the metals (42) Those three cysteine residues are located

in different regions of the primary sequence, which implies that the protein folds

in such a way that the cysteines are brought in proximity to each other, allowing them to coordinate with the metalloid (43) From X-ray diffraction data of crystals

of As(III) or Sb(III) complexed to small molecule dithiols, the lengths of an

As-F IGURE 4 Proposed geometry of the ArsA metalloid-binding domain The metalloid-binding site ArsA is proposed to be a trigonal pyramidal structure containing the three-coordinately liganded sulfur thiolates of Cys113, Cys172, and Cys422, with either As(III) or Sb(III) at the apex The bond angles and distances are predicted from crystallographic analysis of small arsenic or an-timony thiol compounds

S bond and Sb-S bond have been shown to be 2.23 and 2.45 A˚ , respectively, with S-As-S and S-Sb-S angles of 92.7° and 84.8°, respectively (44) Extrapolat-ing from the shape of these small molecules, a model of the metal-bindExtrapolat-ing site

in ArsA has been proposed (Fig 4) (43) Filling of the allosteric site with metal results in an increase in the rate of ATP hydrolysis at the NBD This is associated with conformational changes that occur upon metalloid binding (45) A1 and A2 are tethered by a 25-residue flexible linker peptide (46), but otherwise the NBDs

do not interact in the absence of metalloid (47) When As(III) or Sb(III) coordi-nates with Cys113 and Cys172 in A1 and Cys422 in A2, this draws the two halves

of the protein closer, bringing the two nucleotide-binding sites into contact, which increases the rate of ATP hydrolysis (48,49) (Fig 5)

How activation occurs is an open question, as is whether both NBDs are catalytic Both sites bind nucleotides, even when the other is inactivated by muta-genesis (50) The two sites do not appear to be equivalent, either in sequence or

in binding properties In the absence of metalloid activator, the A2 site binds the ATP analog 5′-p-fluorosulfonylbenzoyladenosine (FSBA) but not ATP, leading

to the hypothesis that the A1 NBD is a high-affinity binding site but that the A2 site can bind ATP only when the enzyme is activated (47) Although mutations

in either NBD eliminate the high, activated rate, ArsAs with inactive A2 sites still hydrolyze ATP at the basal, unactivated rate (48) In contrast, A1 substitu-tions result in complete inactivation These results suggest that the A1 site exhib-its independent unisite catalysis in the absence of As(III) or Sb(III); participation

of the A2 site requires metalloid binding, which produces multisite catalysis (48)