Copper Homeostasis in Escherichia coli and Other Enterobacteriaceae

CHRISTOPHER RENSING* AND SYLVIA FRANKE

[Section Editor: Charles Earhart]

Posted January 11, 2007

Department of Soil, Water, and Environmental Science, Shantz Bldg. #38, Rm. 429, University of Arizona, Tucson, AZ 85721

*Corresponding author. Phone: (520) 626-8482, Fax: (520) 621-1647, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it

Microbial evolution appears to be a continuous process of deletions as well as additions to the genome, with the latter often accomplished by horizontal gene transfer. These genomic changes are later fine-tuned by mutations and the different outcomes selected by the environment.

An interesting model for studying environmental influences shaping microbial evolution is provided by a multitude of copper resistance and copper homeostasis determinants in enteric bacteria. At one extreme there are symbiotic enteric bacteria, such as Buchnera spp. and Wigglesworthia glossinidia, which contain no known system to protect them from copper-mediated toxicity (2, 65). In contrast, there is a vast collection of copper resistance elements available and present in the other enteric bacteria. Some of these are located on plasmids or other mobile genetic elements, but most are encoded on the chromosome. In this chapter, we will describe these determinants and then try to relate their presence to the habitat of the respective organism, as a current hypothesis predicts that the environment should determine an organism’s genetic makeup (83).

Enteric bacteria inhabit diverse habitats, as they not only colonize the gut but can also be found as free-living organisms in soil such as Klebsiella pneumoniae (formerly also called Klebsiella aerogenes), as plant pathogens such as Erwinia carotovora subsp. atroseptica, as insect pathogens such as Photorhabdus luminescens, and as human pathogens such as Salmonella, Yersinia, and Shigella spp. These environments present the organisms with various challenges including both metal excesses and scarcities and oxidative stress. It is therefore not surprising that there are multiple genetic determinants on the chromosomes or the mobile genetic elements of these organisms to combat these challenges. There are various genetic compositions of copper homeostasis and resistance determinants among closely related species (Table 1). Unexpectedly, not much is known about copper uptake mechanisms in Escherichia coli or related bacteria. Perhaps the reason for this lack of knowledge is that there is no absolute requirement for copper since E. coli contains two alternative cytochrome oxidases, only one being a copper enzyme (Fig. 1). Most aerobic organisms need trace amounts of copper, but excess copper can be deleterious to cells. Therefore, the intracellular levels of copper need to be tightly controlled, a process termed copper homeostasis. The ability to withstand extremely high concentrations of copper is often termed copper resistance. However, the proteins involved in "copper homeostasis" and "copper resistance" are often related, so a clear-cut distinction of what constitutes a "homeostatic" protein versus a "resistance" protein cannot be made easily. Most often, chromosomally encoded regulons are considered to be involved in copper homeostasis and plasmid-carried determinants are considered to be responsible for copper resistance.

Fig. 1Interplay of the chromosomally encoded copper resistance systems in E. coli. E. coli contains three major chromosomally encoded copper homeostasis mechanisms. Copper is transported across the cytoplasmic membrane (CM) as Cu(I) via CopA, a P-type ATPase, into the periplasmic space. However, in the periplasm copper can cause damage to proteins and lipids by undergoing redox cycling. Therefore, Cu(I) is oxidized into Cu(II) via the multicopper oxidase CueO under aerobic conditions. In addition, CueO oxidizes catechol siderophores such as enterobactin and thus prevents enterobactin-mediated reduction of Cu(II) to Cu(I). Oxidized enterobactin is able to bind copper and reduce the concentration of free copper that can undergo the redox cycling. Under anaerobic conditions or in the absence of CueO, copper [probably as Cu(I)] is translocated from the periplasm into the surrounding medium via the CusC(F)BA complex. The CusCBA components form a protein complex stretching from the cytoplasmic membrane to the outer membrane (OM) as described for the TolC-AcrAB complex (48, 54). Cu(I) is thought to enter the transport complex from the periplasmic side (either by itself or transported to CusCBA via CusF), where it is then transported across the outer membrane. Copper ions, both Cu(I) and Cu(II), are drawn as blue circles with transparent blue indicating copper bound to proteins or enterobactin.

Table 1Overview of copper and silver resistance systems among Enterobacteriaceae

In E. coli there are four regulons that are induced in the presence of copper (33, 84). Two, the CueR and the CusR regulons, are described in detail later. At least 10 copper-induced genes were found to be under the control of the CpxAR two-component system (84). However, the CpxR regulon is clearly a general stress operon that is not specific to copper and will therefore not be discussed in this chapter. Finally, the physiological function of the fourth regulon, YedW, is not known at present.

The previously identified cut genes (25) are also not copper specific and have been discussed by Rensing and Grass (58). For example, a mutation in cutF (nlpE) probably decreases induction of the CpxR regulon (84). However, an interesting recent paper clearly demonstrates the necessity of various systems for periplasmic disulfide bond formation to protect against copper-induced oxidative stress (28). A deletion of not only dsbD (which is cutA2) but also dsbC makes cells more copper sensitive. In addition, a second deletion in dsbA makes a ΔdsbC strain even more copper sensitive. Together, these findings show how interconnected the general stress response is to the more specific copper response.

PROTEIN FAMILIES INVOLVED IN CONFERRING COPPER HOMEOSTASIS AND RESISTANCE IN E.COLI

P-Type ATPases

A central component regulating intracellular copper levels, present in all free-living enteric bacteria whose genomes have so far been sequenced, is a Cu(I)-translocating P-type ATPase (57). Named CopA in E. coli, this transporter is required for intrinsic copper resistance under both aerobic and anaerobic conditions and catalyzes copper efflux across the cytoplasmic membrane into the periplasm. Expression of copA is regulated by CueR (51, 72), which is activated by intracellular Cu(I) and Ag(I) levels and belongs to the MerR family of regulators. In addition, the expression of copA may be influenced by CpxR in response to cell envelope stress (84). The expression patterns of copA are similar under both aerobic and anaerobic conditions; however, the basal level of transcription is higher under anaerobic conditions, indicating that cytoplasmic copper concentrations may be higher under anaerobic conditions than under aerobic conditions (50).

The P-type ATPase superfamily is a ubiquitous group of proteins involved in the transport of charged substrates across biological membranes. All P-type ATPases contain a conserved aspartate residue that is phosphorylated during the catalytic cycle. P-type ATPases are organized into five subfamilies based upon ion specificity, proposed topology, and sequence comparisons (6). The transport of transition or heavy-metal ions is catalyzed by a phylogenetically related subgroup of P-type ATPases that have previously been described as the CPX-type ATPases (because of the highly conserved cysteine and proline residues in transmembrane domain 6 [TM6]), soft-metal-transporting P-type ATPases, and P_IB-type ATPases (6, 57, 68). These ATPases have eight TMs, with a large cytoplasmic loop located between TM6 and TM7. In addition to the typical inhibition by vanadate that is observed in all P-type ATPases, these CPX-type ATPases show metal-dependent ATPase activity and subsequent transport through a common catalytic cycle.

In vitro, E. coli CopA transports copper only in the presence of dithiothreitol, consistent with the transport of Cu(I). Dithiothreitol is needed for reduction of Cu(II) to Cu(I), not for reducing CopA (15, 56). In addition, CopA forms an acylphosphate intermediate with [γ-³²P]ATP only in the presence of Cu(I) or Ag(I) but not Cu(II), Zn(II), or Co(II). The ATPase activity of purified CopA is stimulated only by Cu(I) and Ag(I) (15). Similar results were obtained with CopA from Archaeoglobus fulgidus, indicating that both proteins transport Cu(I) by using similar mechanisms involving conserved residues (4).

In studies using Na,K ATPases and Ca ATPases, it was shown that in P-type ATPases the catalytic cycle includes an "occluded cation" enzyme conformation. Both the Na,K ATPase and the Ca ATPase coordinate the occluded cation in TM4, TM5, and TM6, which correspond to TM6, TM7, and TM8 in P_IB-type ATPases (Fig. 2). Many studies have already shown that a so-called CPX motif in TM6 is necessary for the function of the transporter. Most often the sequence of this motif is CPC, as in E. coli CopA; however, the sequence can also be CPH, TPC, or SPC in other P_IB-type ATPases (3). This motif is thought to be part of the coordinating site required for metal transport. Sequence alignment of the region encompassing TM6, TM7, and TM8 allowed the P_IB-type ATPases to be sorted into six subgroups (3). One large subgroup named IB-1 includes all known Cu(I) ATPases, such as E. coli CopA, A. fulgidus CopA, Saccharomyces cerevisiae CCC2, and Menkes and Wilson disease proteins. All members of this group possess an invariant CPC sequence in TM6, but since other P_IB-type ATPase subgroups, such as the Zn(II)-Cd(II)-Pb(II)-transporting ATPases (named subgroup IB-2), contain a similar CPC-type motif, it is clear that other residues in this region are needed to confer metal specificity (3). With CopA from the thermophile A. fulgidus, Arguello and coworkers demonstrated that four conserved amino acids (Y682, N683, M711, and S715) in TM7 and TM8 participate in metal binding in addition to the two conserved cysteines from the CPC motif in TM6 (4). It is thought that these amino acids are responsible for metal specificity, formation of the occluded cation, and the subsequent change in protein conformation to a state favoring phosphorylation by ATP (4). The corresponding amino acids in E. coli CopA are Y784, N785, M813, and S917 (Fig. 2), but there is nothing known about if and how they might be involved in the function of the protein.

Fig. 2Topology of the P-type ATPase CopA. CopA of E. coli contains eight transmembrane helices (labeled I to VIII) with N and C termini located in the cytoplasm. The N terminus contains two metal binding domains (CX₂C) consisting of cysteines C14, C17, C110, and C113. Also in the cytoplasm are the functionally important beta-loop (T374 to P377), the phosphorylation domain (p-domain; D523 to G526), and the N domain (G719 to P726). Residues thought to be involved in substrate specificity besides the CPC motif in transmembrane VI (C479 to C481) are residues Y784 and N785 located in transmembrane VII as well as M813 and S817 located in transmembrane VIII (3).

Many of these P_IB-type ATPases contain cytoplasmic metal binding domains, CXXC motifs, on their N termini. In CopA, there are two of these potential metal binding sites; however, amino acid substitutions by site-directed mutagenesis showed that these motifs are not required for function and do not confer metal specificity (14). While these metal binding motifs may not be required for enzymatic function, they may still have a regulatory role (14, 79, 81).

All nonsymbiotic enterobacteria contain homologues of CopA (Table 1). Levels of identity to CopA of E. coli K-12 range from 99% (Shigella sonnei) to 62.2% (Yersinia pestis). With very few exceptions, no intracellular copper(I) chaperone has been found in enteric bacteria which may deliver Cu(I) to the cytosolic N-terminal CXXC motif as occurs in eukaryotes and some prokaryotes. However, in all sequenced strains of Salmonella there is an open reading frame encoding a 64-amino-acid protein (accession number NP 459350; Salmonella enterica serovar Typhimurium LT2) with extensive homology to other bacterial copper chaperones that could be involved in copper trafficking of cytoplasmic Cu(I) to CopA for efflux since the protein appears to correspond to the same operon. Similar putative copper chaperones are encoded in various pseudomonads, again in close proximity to a putative Cu(I) ATPase, indicating that these putative chaperones may deliver copper to the Cu(I) ATPase.

MCOs

Other known components of copper homeostasis and resistance all appear to be involved in periplasmic copper handling. Interestingly, in all cases studied, Cu(I) is the toxic species that needs to be removed, sequestered, or oxidized. Whereas some components involved in copper homeostasis can be found in both anaerobes and aerobes, multicopper oxidases (MCOs) implicated in copper tolerance in E. coli, such as CueO and the plasmid-based PcoA (PcoA and Pco-dependent copper resistance are described in a separate paragraph), can be found only in aerobic organisms. Their role in protecting cells from copper toxicity is beginning to be understood, and several features indicate that CueO, PcoA, and other related MCOs are specifically adapted to combat copper-mediated oxidative damage. Well-known MCOs include human ceruloplasmin, fungal laccases, and ascorbate oxidase from plants (70).

MCOs couple four one-electron substrate oxidation steps to the four-electron reduction of dioxygen to water (70). This reaction has been best studied with laccase, ascorbate oxidase, ceruloplasmin, and Fet3 from S. cerevisiae. MCOs typically contain four copper atoms: one Type I (T1), or "blue," copper site and three other atoms in a trinuclear cluster consisting of one T2, or "normal," copper and two T3, or "binuclear," coppers (70) (Fig. 3A). Molecular oxygen binds at the trinuclear site and is reduced to water, accepting four electrons that are shuttled through the T1 copper site, located about 13 Å away (Fig. 3B). Some MCOs, such as laccase, have little specificity and can oxidize a large variety of substrates, whereas others, such as ascorbate oxidase, are more specific. The mechanism by which oxygen is converted into water by MCOs has been the subject of intense study for some time (69, 70). Recent spectroscopic studies on laccase and Fet3 (37, 42, 52) have identified two intermediates containing oxygen species bound at the trinuclear copper center. One has a bound peroxide, and the other, called the "native" intermediate, apparently contains one or more oxygen atoms bound centrally among the copper atoms of the trinuclear copper cluster.

Fig. 3Structure and catalysis scheme of the MCO CueO. (A) Ribbon diagram of CueO. Shown is the three-dimensional structure of CueO, including the copper centers T1 ("blue copper"), T2, and T3 ("trinuclear cluster"). A fifth copper atom is bound at the base of the methionine-rich helix and is essential for the oxidase activity of CueO. (B) Scheme of catalysis for the multicopper oxidase CueO. Electrons are transferred from the substrate (S), which is thereby oxidized to the product (P) via the regulatory copper (rCu) to the T1 ("blue") copper center and from there to the trinuclear cluster (T2, T3), where oxygen is reduced to H₂O. Whereas the electrons are first transferred to the regulatory copper in CueO, the electron transfer in other MCOs (laccases, ascorbate oxidase, ceruloplasmin) is directly from the substrate to the copper in the T1 copper center (data not shown).

CueO of E. coli is a 53-kDa periplasmic MCO that also possesses laccase-like activity (20, 34) and confers copper tolerance in E. coli under aerobic conditions. Expression of cueO is regulated by CueR and CpxR as is that of copA (51, 84). CueO oxidizes a wide variety of substrates including 2,6-dimethylphenol, enterobactin (a catecholate siderophore found in E. coli), and Fe(II) (20, 34). This activity is oxygen dependent. CueO is not a copper transport protein but rather performs a catalytic function to protect the cell from copper-induced damage. Deletion of cueO led to decreased activity of alkaline phosphatase (PhoA) in E. coli under copper stress. PhoA was used as an indicator to examine if CueO can protect proteins from copper-mediated toxicity in the periplasm (20).

Intriguingly, very little oxidase activity occurs without excess copper. In our initial hypothesis (21), we and others proposed that CueO converts periplasmicCu(I) to the less toxic Cu(II)(16, 20, 50, 58). This mode of copper protection was recently confirmed since CueO of E. coli and the related MCOs Fet3 in S. cerevisiae and ceruloplasmin in humans were all shown to possess cuprous oxidase activity (67, 71). Furthermore, in S. cerevisiae this activity was necessary for protection from copper toxicity only if a copper reductase was also present (64). In addition to this oxidase activity, CueO appears to protect cells from the interaction of enterobactin with copper by oxidation of enterobactin under copper stress.

Copper slowly oxidizes catechols in the presence and/or absence of MCOs. One electron is transferred to molecular oxygen in the copper-mediated oxidation of catechols, thereby forming superoxide, which is subsequently reduced to hydrogen peroxide and hydroxyl radicals. In other words, enterobactin and other catechols reduce Cu(II) to Cu(I) and thereby change the steady-state level of Cu(I) (32, 63). Increased redox-cycling of copper could lead to both increased production of reactive oxygen species and increased Cu(I) uptake. CueO can therefore perform several beneficial reactions: (I) reduction of the amount of Cu(I), (II) decrease of unoxidized enterobactin, and (III) prevention of accumulation of free intermediate reactive oxygen species by coupling the oxidation of substrates to the complete reduction of oxygen to water. These hypotheses were recently strengthened by Tree and coworkers (76).

Homologues of CueO are present in all nonsymbiotic enterobacteria with the exception of Erwinia carotovora. Levels of identity to CueO of E. coli K-12 range from 99.2% (Shigella flexneri) to 62.2% (Erwinia chrysanthemi). In E. coli, both copA and cueO are regulated mainly by CueR. This also seems to be the case in other enterobacteria since genes encoding CueR can be found in all genomic sequences of enterobacteria (except those of Buchnera aphidicola, Wigglesworthia glossinidia, Candidatus Blochmannia, and Sodalis glossinidius). The degrees of identity of the different homologues of CopA and CueO to their E. coli counterparts correlate, indicating that copA and cueO were acquired simultaneously (Table 1).

Resistance-Nodulation-Cell Division Protein (RND)-Containing Transporters

The second system responsible for periplasmic copper handling in E. coli is the four-part CusCFBA complex. Especially under anaerobic conditions, CusCFBA is important for Cu(I) translocation from the periplasm across the outer membrane (50). Copper is probably more toxic under anaerobic conditions because copper is taken up as Cu(I), which is much more prevalent under anaerobic conditions. Increased presence of Cu(I) under anaerobic conditions is also reflected in the transcriptional response (50). Cu(I) has a much higher affinity for sulfhydryl groups than Cu(II) and thus could potentially replace other metals in proteins as well as deplete the glutathione pool.

In E. coli, the cus operon encoding the structural gene products CusC, CusF, CusB, and CusA is induced by copper and to a lesser extent by silver and regulated by the adjacent two-component system CusRS (17, 47). Moreover, using microarray analysis, cusF was shown to be the gene most strongly induced by copper, followed by cusC (12). CusCBA are homologues of the CBA family of transport systems found exclusively in gram-negative bacteria. CBA-type transporters are often composed of three proteins encoded by a single operon and are involved in the export of metal ions, xenobiotics, and drugs (77).

The three conserved components of this system include a proton-substrate antiporter of the RND family (61, 77), a membrane fusion protein (61), and an outer membrane factor (53). The resultant assembly is thought to form a periplasm-spanning complex connecting the cytoplasm to the extracellular space. Recently determined crystal structures of the outer membrane factor TolC (35) and the antiporter AcrB (48, 54) (both proteins from E. coli) and the membrane fusion protein MexA of Pseudomonas aeruginosa (1, 27) provide our first structural look at the components of this complex. AcrB and TolC form trimers creating a channel that is modeled as connecting the cytoplasm to the extracellular space. In contrast, MexA seems to form a tridecamer. However, in the proposed models of the CBA efflux complex, a ring formed by 9 (27) or 12 (1) molecules of MexA is proposed.

It is likely that these systems also transport substrates from the periplasm to the extracellular space since other transporters of the RND superfamily such as AcrB and MexB can cause efflux of substrates that do not cross the cytoplasmic membrane (49). It was therefore suggested that the binding of the substrate may occur on the periplasmic side of the transporter (85). The structure of the antiporter AcrB shows a channel leading from the periplasmic space to the central cavity that is a possible mode of entry of substrates (41, 48). Furthermore, more and more evidence points to the periplasmic domains of RNDs as conferring substrate specificity (13, 43, 46).

Genetic and phenotypic analyses strongly suggest that the products of the cusCFBA genes form a copper-extruding complex (18, 21, 47, 50). CusA is thought to serve the role of a proton-substrate antiporter, similar to AcrB; CusB is a putative membrane fusion protein like MexA and AcrA; and CusC is thought to be an outer membrane factor, structurally similar to TolC. CusF is a novel periplasmic metallochaperone with a unique fold compared to those of other small copper binding proteins (Fig. 4) (40). Genetic analysis of different E. coli deletion strains suggests periplasmic rather than cytoplasmic copper as the substrate for the Cus efflux complex (21).

Fig. 4Structure of the periplasmic copper binding protein CusF. CusF is shown as a ribbon diagram. Residues undergoing a significant change in chemical shift in nuclear magnetic resonance experiments after addition of Cu(I) are labeled in red. The side chains of the residues most likely involved in coordinating Cu(I), H36, M47, and M49, are shown.

In contrast to the CueO oxygen-dependent MCO, the CusCBA efflux complex works under both anaerobic and aerobic conditions. Recent results showed that Cus-mediated copper detoxification is especially important under anaerobic conditions (29). Under aerobic conditions, copper sensitivity is observed only when both cueO and cusCFBA are deleted (21).

Surprisingly, the Cus system is not present in many enteric bacteria. It is absent in all strains of Salmonella, Y. pestis, and Yersinia pseudotuberculosis but can be found in Yersinia enterocolitica, where it is present as CusCBA. S. flexneri, which is closely related to E. coli, has cusCFBA, but there is an insertion sequence in the cusB gene. In addition, homologous systems have often been identified on plasmids such as the large virulence plasmid pLVPK of K. pneumoniae CG43 (10) and pMG101 of S. enterica serovar Typhimurium (22). These systems are often referred to as silver resistance, or sil, determinants in the literature and are significantly different from the cus system (see percent identities in Table 1). However, since it could be demonstrated that CusCFBA of E. coli K-12 also confers a slight silver resistance, all of these determinants can probably confer both silver and copper resistance, albeit at different specificities and different levels of silver resistance (17, 23). The level of resistance and apparent specificity may also be dependent on the presence of other genes involved in copper and silver resistance on the chromosome. An illustration of the proposed functions of these chromosomally encoded copper homoeostasis systems is given in Fig. 1.

Plasmid-Carried Copper Resistance and Periplasmic Copper Binding Proteins

In addition to the previously described chromosomally encoded genes, there are determinants on mobile genetic elements that can confer additional resistance to copper in E. coli. The copper resistance mechanisms encoded on the chromosome enable most E. coli strains to tolerate copper concentrations of up to 4 mM on Luria-Bertani agar, and other E. coli strains, isolated from copper-enriched environments, are able to tolerate up to 20 mM copper. The best-studied copper resistance determinant responsible for the latter case was found on plasmid pRJ1004 (74) and named pco (for plasmid copper resistance). The pco copper resistance determinant has been found frequently on plasmids such as pRJ1004 from E. coli strains isolated from pigs fed on copper as a growth promoter (8). Sequence analyses identified six open reading frames organized as an operon, pcoABCDRS, and a seventh gene further downstream, pcoE, regulated by a separate copper-inducible promoter (8). Other pco-homologous systems have been identified in copper-resistant enterobacteria by using pco as a DNA probe (82). Restriction analyses demonstrated that these strains do not contain pRJ1004 but other unrelated plasmids. More recently, sequencing projects for various enterobacterial plasmids identified the pco determinant on plasmids of K. pneumoniae, Serratia marcescens (PcoABCD), and Enterobacter spp. (PcoA) (Table 1). In fact, all other known homologous pco determinants were found exclusively on plasmids isolated from enterobacteria. The sole exception to this rule may be pcoAB, which may be localized on the chromosome of K. pneumoniae; however, at this stage of genome sequencing it is difficult to determine whether this determinant is plasmid carried or chromosomally encoded (Table 1). Interestingly, homologues of pcoE were encountered much more frequently than the pco determinant. In E. coli the presence of CopA, transporting Cu(I) across the cytoplasmic membrane into the periplasm, is required for pco to confer additional copper tolerance (38).

The pcoRS genes encode a two-component regulatory system responsible for copper-dependent expression of pcoABCD (60). As mentioned previously, PcoA is a periplasmic MCO (29) that can functionally substitute for CueO although these enzymes are only distantly related (38). CopA, a homologue of PcoA found in Pseudomonas syringae, binds up to 11 copper atoms per molecule, which is more than necessary for enzyme activity.

PcoB is predicted to be an outer membrane protein. Expression of pcoAB confers a higher level of copper tolerance to a copper-sensitive E. coli strain than expression of pcoA alone, even at a much lower expression level (38). This finding indicates a possible interaction of these two proteins. PcoB may funnel Cu(I) directly to PcoA for oxidation; however, the other proteins, PcoCD, are necessary for full pco-mediated copper resistance, as was also observed for the homologous CopABCD-system from P. syringae (45).

PcoC was shown to be a periplasmic protein able to bind one copper atom per molecule (29, 38). Spectroscopic analysis (electron nuclear double resonance and electron spin-echo envelope modulation) indicates that at least one histidine ligand and the sulfur of a nearby methionine may function as another ligand ( 29). Methionine residues have been shown to be ligands for copper in several proteins, for example, in E. coli CusF (18, 40) and in the eukaryotic copper transporter Ctr (36, 55). However, mutation of a hypothetical copper binding motif in PcoC did not result in a loss of copper binding ability but changed the copper binding constant (38). The recently determined PcoC crystal structure displayed an interface between methionine-rich regions in adjacent molecules in the crystal. This led the authors of the study to suggest that the role of such regions may be in protein-protein recognition, possibly between two PcoC molecules in a dimer or between PcoC and PcoA, rather than in metal binding (80). CopC of P. syringae, a homologue of PcoC, was able to bind both Cu(I) and Cu(II), on different sites of the protein. The Cu(I) binding site of CopC contains four methionines and a histidine as possible copper ligands in the motif MX₂MX₂MXHX₂M. However, the third methionine has been excluded from involvement in Cu(I) binding. The Cu(II) binding site of CopC is composed of two histidine residues, a glutamine, and an aspartate (5). Since protein-protein interaction depends on metal binding, and probably also oxidation state, the methionine-rich region of PcoC may be involved in both.

PcoD is a cytoplasmic membrane protein thought to take up copper in order to fill the catalytic center of apo-PcoA before PcoA is transported into the periplasm via the TAT pathway (58). Since the P. syringae CopCD genes are discussed as a periplasmic copper binding and uptake system (9), this may also be the case for PcoCD (58). E. coli genes encoding homologues of PcoC (yobA) and PcoD (yebZ) can be found not only on the chromosome of E. coli but also in Salmonella, Yersinia, Shigella, and Photorhabdus spp. Nothing is yet known about a possible function of these hypothetical proteins. Since both proteins, YobA and YebZ, lack methionine-rich motifs and are not induced by copper, an involvement in copper handling is unlikely.

Another protein encoded by the pco determinant is PcoE. In contrast to pcoABCD, expression of pcoE is regulated by the chromosomally encoded two-component system CusRS (47). PcoE accumulates copper in the periplasm; however, expression of pcoE in the presence or absence of the other pco genes has no influence on copper resistance in E. coli (38, 58). Interestingly, homologues of PcoE are often encoded as part of so-called sil (silver resistance) determinants and are called SilE. Although silE is part of the silver resistance systems, expression of silE, like that of pcoE, occurs from its own promoter. PcoE (encoded close to a pcoABCD operon) and SilE (encoded close to silCFBA) are homologous but clearly distinct from each other. Whereas PcoE encoded on pRJ1004 is 92.4% identical to PcoE carried on plasmid R478 (S. marcescens), the level of identity to SilE carried on pMG101 (S. enterica serovar Typhimurium) is less than 50%. In contrast, the levels of identity of several SilEs carried on various plasmids range from 88 to 100%. SilE from S. enterica serovar Typhimurium was shown to bind five silver ions per polypeptide (66); however, at this stage it is still not clear how or if PcoE/SilE may be involved in copper and silver resistance.

In addition to these well-characterized resistance operons, there are numerous other genes that appear to be involved in copper binding and trafficking that have not been studied in great detail. SilE and its homologue PcoE, for example, are thought to effect the periplasmic binding and sequestration of silver and copper, respectively. There are other proteins that reduce copper sensitivity, but at this stage not much is known about their mechanisms.

HABITATS OF DIFFERENT ENTEROBACTERIACEAE AND THE PRESENCE OF COPPER RESISTANCE DETERMINANTS

One main question still remaining is how the living environments of different Enterobacteriaceae are reflected in their genetic makeup regarding copper homoeostasis. If only the plasmid-carried copper and silver resistance systems are examined, this question is relatively easy to answer. For example, copper is used as a growth promoter in livestock (26) and plasmids carrying copper resistance systems (pco related) can frequently be isolated from these animals (8, 26, 82). The same can be said for plasmid-bound silver resistance systems such as the sil determinant carried on pMG101 of S. enterica serovar Typhimurium (22, 24, 44). sil-related genes can also be found on plasmids from clinical isolates (23), and sequencing projects have identified sil determinants carried on several virulence plasmids within the Enterobacteriaceae (10, 19, 31, 59; H. C. Tsai, K. M. Wu, T. L. Liao, Y. M. Liu, H. J. Chen, Y. C. Chang, C. H. Chang, R. Kirby, C. Chen, C. W. S. Chen, H. Y. Chang, C. P. Fung, J. T. Wang, and S. F. Tsai, unpublished results [http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?val=NC_006625]). Since silver compounds are used as clinical antibacterial agents, these agents select for the presence of silver resistance determinants.

Whereas the occurrence of plasmid-carried copper and silver resistance determinants can easily be linked to the respective environments of the microorganisms, correlating the chromosomal contribution to copper (and silver) resistance to the environment is more complicated. Obligate endosymbiotic representatives (B. aphidicola, C. Blochmannia floridanus, C. Blochmania pennsylvanicus strain BPEN, and W. glossinidia) have no known copper homeostasis systems encoded on their genomes. All these organisms are obligate endosymbionts of insects and have, compared to other enterobacteria, highly reduced genomes (62, 65, 73, 78). They could afford to lose their copper homeostasis mechanisms because they are protected by the cellular defense mechanisms of their hosts. S. glossinidius strain morsitans also is an endosymbiont of the tsetse fly like W. glossinidia, but the reduction of the genome is not as extensive as that of genomes of other endosymbiontic enterobacteria (75), i.e., 2,432 chromosomally encoded proteins compared to 611 in W. glossinidia, 583 in C. Blochmania floridanus, and 564 in B. aphidicola strain APS. In contrast to those of B. aphidicola, C. Blochmannia, and W. glossinidia, the chromosome of S. glossinidius strain morsitans contains genes encoding homologues of CutC, YobA, and YebZ. However, how and if these proteins are involved in copper homeostasis has not been determined.

Phylogenetic analysis of the enterobacteria using small-subunit rRNA to determine the evolutionary history of the different Enterobacteriaceae is complicated. Specifically, it is a challenge to elucidate how the endosymbionts with their reduced genome fit in since representatives of the genera Buchnera and Candidatus contain only one gene each encoding 16S rRNA whereas W. glossinidia encodes both rrsI and rrsII and all other enterobacteria contain seven 16S rRNA genes. Since the rrs genes in any one organism are already heterogenic, a comparison to genes from other organisms is not always easy (11). For example, Lerat and coworkers (39) demonstrated that either Wigglesworthia brevipalpis or, alternatively, B. aphidicola could group with Y. pestis or that B. aphidicola was more closely related to E. coli and S. enterica serovar Typhimurium than Y. pestis, depending on which genes were used for comparison.

All other enterobacteria sequenced so far encode at least one of the previously described copper resistance systems. They all carry the genetic information for a copper-transporting P-type ATPase and, with the exception of E. carotovora subsp. atroseptica, an MCO as well as its regulator CueR. This seems to be the minimal genetic makeup that these organisms need to survive in their environment, independent of whether they are insect, plant, or mammalian pathogens, like Photorhabdus, Erwinia, or Shigella, or part of the gut flora of birds (salmonellae) and mammals (E. coli). The most variable part is the cus system, not only in terms of its presence but also the homologies among the proteins encoded on the chromosomes of the different enterobacteria (Table 1). It is completely absent in all strains of Salmonella as well as P. luminescens. P. luminescens is a nematode symbiont and a broad-spectrum insect pathogen. It is possible that these habitats contain less copper on average than the mammalian gut, the natural environment of E. coli, where the copper content can vary widely depending on the diet. Salmonellae are also responsible for causing disease and are therefore capable of surviving in mammals. A putative RND system is encoded directly downstream of copA and a putative copper chaperone is encoded on the chromosome of S. enterica serovar Typhimurium LT2; however, since the identities to CusCFBA of E. coli K-12 are very low, less than 25% for the RND transporter, this system is probably not involved in copper handling in S. enterica serovar Typhimurium.

The human pathogen S. flexneri contains a cus determinant that is interrupted by an insertion sequence element and therefore cannot be functionally active. Since S. flexneri is very closely related to E. coli, this interruption must have been a recent event in evolutionary terms. Other strains of Shigella appear to contain the complete cus determinant, and the encoded proteins show over 95% identity to CusCFBA of E. coli K-12 (Table 1).

In different Yersinia strains, there are also differences regarding the presence of a cus system. Y. enterocolitica is the only sequenced species among the genus Yersinia containing cusCBA; however, a gene encoding CusF, which is necessary for full cus-mediated copper resistance in E. coli, is not present (18). In contrast, both Y. pestis and Y. pseudotuberculosis do not contain a cus determinant. Due to the fact that both organisms cause gastroenteritis in humans and thus can be found in the same environment, the gut, whereas Y. pestis colonizes the blood and lymphatic systems, an environmental explanation of their genetic makeup regarding copper resistance determinants is difficult to propose.

The same is true regarding the genus Erwinia; Erwinia spp. are plant pathogens, and the genomic sequences of two strains are available. Although both E. chrysanthemi and E. carotovora subsp. atroseptica contain the genetic information for the P-type ATPase CopA and a cus system, only E. chrysanthemi contains a cueO gene. E. carotovora subsp. atroseptica is the only one of the sequenced Enterobacteriaceae missing cueO. Both Erwinia strains encode a cus system that does not contain a gene encoding the outer membrane factor CusC. In addition, CusF and CusB are encoded together by a single open reading frame, with CusF being the C-terminal part of the membrane fusion protein. Therefore, looking at these admittedly anecdotal cases we have to conclude that, based on our current knowledge, it is not possible to link the copper content and bioavailability in the environment to the genetic makeup of different Enterobacteriaceae.