Module 5.4.4.3

Transition Metal Homeostasis

DIETRICH H. NIES* AND GREGOR GRASS†

[SECTION EDITOR: JOHN FOSTER]

Posted October 01, 2009

Molecular Microbiology, Martin-Luther-University Halle-Wittenberg, Kurt-Mothes-Str. 3, 06099 Halle/Saale, Germany

*Corresponding author. Mailing address: Molecular Microbiology, Martin-Luther-University Halle-Wittenberg, Kurt-Mothes-Str. 3, 06099 Halle/Saale, Germany. Phone: 49–345-5526352, Fax: 49–345-5527010, E-mail: This e-mail address is being protected from spambots. You need JavaScript enabled to view it .

†Present address: School of Biological Sciences, E141 Beadle Center, University of Nebraska—Lincoln, Lincoln, NE 68588–0666.

While 96% of the dry mass of an Escherichia coli cell is composed of the nonmetals C, O, H, N, P, and S, the remaining 4% is metals that are essential nevertheless. Metal ions are needed to neutralize the negative charges of the nonmetallic acids, to bridge compounds intra- and intermolecularly, and to form complex compounds, some redox active and some not. Of the 90 natural chemical elements (not counting Tc and Pm), the majority, or 68, are metals, divided into 16 light metals with densities of ≤5 g/cm³ (6 alkali metals, 6 earth alkali metals, and Al, Sc, Ti, and Y) and 52 heavy metals with greater densities. Eight elements along the borderline between metals and nonmetals (Al/Si, Ge/As, Sb/Te, and Po/At) may show some characteristics of both, being classed as semimetals (138). Since metals are characterized by low electrical resistivity, the ranking of these eight elements by their electrical resistivities yields the following: Al < As, Sb, Po << Si, Te < Ge (342; data for At not available). The ratio between the resistivities of As and Si is 10⁵, defining a clear border between metal and semimetal. Therefore, Ge actually has fewer metallic properties than As, which should be included in the list of metals instead of Ge.

Metals may also be sorted into four groups based on the configurations of their electron shells. Group 1 includes 12 main-group metals (6 alkali and 6 earth alkali metals) that fill up s-orbitals, group 2 includes 10 main-group metals (Al, Ga, As, In, Sn, Sb, Tl, Pb, Bi, and Po) that fill up p-orbitals, group 3 encompasses 30 transition metals (again, not counting Tc) that fill up d-orbitals, and group 4 includes 13 lanthanides (minus Pm), as well as 3 naturally occurring actinides (Th, Pa, and U), that fill up f-orbitals (138). These variations in the electronic configurations lead to chemical differences, which are also important for cellular biochemistry. Heavy metals are usually “soft” metals that prefer sulfur over oxygen as a binding partner, while light metals are “hard,” preferring oxygen or nitrogen over sulfur (96). Most importantly, transition metals are able to form complex compounds that are a prerequisite for any sophisticated cellular biochemistry.

Not all of these 68 metals are important for E. coli, Salmonella spp., or other bacteria. To be used in cellular biochemistry, a metal must have been available during evolution. Availability is the result of three factors, (i) element synthesis in ancient stars, (ii) sorting events during the formation of the earth's crust, and (iii) the solubility of a metal ion in water (222). As a result, elements with low atomic numbers are, in general, more available than elements with high atomic numbers. Most tri- and tetravalent cations that form insoluble hydroxides are typically not bioelements. Exceptions include Li and Be, which were consumed during stellar element synthesis, and Fe³⁺ because Fe²⁺ was available when the earth was still anoxic. Cr³⁺ may be available for cells because of the high solubility of the Cr(VI) oxyanion chromate. This leaves us with Na, K, Mg, and Ca as major metallic bioelements and with most soluble metals from the first transition period as minor bio- or trace elements. With the exceptions of Mo and W, which are essential oxyanions, all metals with an atomic number larger than 30 (that of Zn) are of interest only because of their toxicity (224). This chapter will focus on transition metals, the essential ones of the first transition period, and some “toxic-only” metals with high atomic numbers.

THE STAGE: CELLULAR METAL HOMEOSTASIS AS INTERPLAY OF TRANSPORT-MEDIATED FLOW EQUILIBRIUM AND BINDING REACTIONS

Because metal ions cannot simply traverse biological membranes, they must be transported across the outer membrane and the cytoplasmic membrane. The simple uptake of metals, with some exceptions (iron siderophore complexes, Co in B12), across the outer membrane is probably facilitated by outer membrane porins (164). In contrast, various transport protein families catalyze the specific uptake of metal ions across the cytoplasmic membrane from the periplasm to the cytoplasm.

All heavy metal cations, even the essential ones, can become toxic. To deal with this challenge, metals may be excreted from the cytoplasm back into the periplasm by cytoplasmic membrane efflux systems and even from the periplasm to the outside by outer membrane efflux systems (197, 222). Both processes, uptake into and export from the cytoplasm, cause the cytoplasmic metal concentration to be in kinetic flow equilibrium (179). A second pattern of kinetic flow equilibrium, in the periplasm, can occur if additional outer membrane efflux systems for a given metal cation exist. The periplasmic flow equilibrium will govern the cytoplasmic flow equilibrium by determining the rate of uptake into the cytoplasm. This system may be called the transport-mediated flow equilibrium of cellular metal homeostasis (222).

Transport-mediated flow equilibrium has an impact on binding processes that occur in the cytoplasm and in the periplasm. These processes include binding to the correct ligands in physiological complexes; binding to the incorrect ligands, resulting in toxic reactions; binding to regulatory sites, causing the alteration of gene expression and transport activity; and finally, binding to substrate-binding sites of transport systems. The latter process links metal binding back to transport-mediated flow equilibrium. If all these binding reactions (12, 78) were known and calculated quantitatively, we would be able to understand metalloproteomics (18) on the system level. As will be shown in the following sections, this process of unraveling the metalloproteome has just started, with the listing of the proteins involved in metal transport, binding, and regulation.

BORDERLINE TRANSITION METALS: FROM COBALT TO ZINC

Competition for Ligands in Complex Compounds

The affinities of the divalent cations Mn²⁺, Fe²⁺, Co²⁺, Ni²⁺, and Zn²⁺ for oxygen and sulfur increase in a parallel fashion in the order from manganese to zinc (222). Thus, Zn²⁺ should be able to remove Mn²⁺, Fe²⁺, Co²⁺, and Ni²⁺ from their physiological complex compounds; Ni²⁺ can replace Mn²⁺, Fe²⁺, and Co²⁺; the Co²⁺ cation can substitute for Mn²⁺ or Fe²⁺; and finally, Fe²⁺ can replace only Mn²⁺ (141). However, zinc with a completely filled d-orbital is able to form only stable tetrahedral complexes, accommodating four ligands, while each of the other five metal cations can be the center of octahedral complexes with six ligands. This property opens the possibility of distinguishing the more strongly binding Zn²⁺ cation by using octahedral complexes to select for the other metal cations. For other transition metals, such as copper and cadmium, the affinity for sulfur increases much more strongly than the affinity for oxygen. Therefore, the five metal cations of Mn, Fe, Co, Ni, and Zn can be categorized as borderline transition metals, while copper and cadmium are soft metals. On the other hand, Cu²⁺, Fe²⁺, and Mn²⁺ are redox active under physiological conditions, while Zn²⁺ is not and Ni²⁺ or Co²⁺ are redox active only in complex compounds. Thus, we will deal with zinc, cobalt, and nickel in one section and with copper, iron, and manganese in another section.

Since iron is highly redox active, the concentration of free or nonchelated iron in the cytoplasm has to be kept low, which would enable Co²⁺ and Ni²⁺ to compete efficiently with Fe²⁺ for insertion into complexes. As will be discussed below, the toxicity of Co²⁺ is based mainly on its effect on iron homeostasis. The same is interestingly true for Zn²⁺. Cd²⁺ interferes with sulfides, which also has an effect on iron homeostasis. Thus, transition metal cation toxicity in general may result at least partially from a disturbance of cellular iron homeostasis. Using different pathways for the biochemical processing of iron, cobalt, and nickel solves this problem.

Zinc

General Considerations.

Zinc and iron are the most important transition metals for all forms of life. While iron is redox active, zinc is not. Thus, iron homeostasis is a very complicated interplay of Fe²⁺- and Fe³⁺-related processes, including oxidation and reduction events, whereas Zn²⁺ homeostasis is comparatively simple. This situation is true particularly for E. coli, because this bacterium does not possess an outer membrane Zn²⁺ efflux system such as the CzcCBA efflux pump in the metal-resistant betaproteobacterium Cupriavidus metallidurans (223). The transport-mediated flow equilibrium of Zn²⁺ concentrations in E. coli results from uptake and efflux across the cytoplasmic membrane, binding to tetrahedral complexes at physiological binding sites, and a few regulatory events. Zinc toxicity in E. coli may be a result of disturbed iron homeostasis, as shown by gene array data.

Transport-Mediated Flow Equilibrium: Zinc Import.

The Zn²⁺ cation has an ionic radius of 74 pm. The other transition metal cations of the first period have ionic radii between 72 pm (Ni²⁺) and 80 pm (Mn²⁺), which are between those of the two major divalent metals, Mg²⁺ (65 pm) and Ca²⁺ (90 pm) (342). Therefore, transition metal cations are easily coimported by fast (high-transport-rate) and nonspecific systems, which are driven mostly by the proton motive force, for the uptake of major divalent biometal cations. This efficient uptake mode is complemented on the one hand by inducible, highly specific but slow or energy-expensive uptake ATPases and on the other hand by inducible efflux systems that specifically deal with surplus metals in the cell.

Protein families containing fast and rather nonspecific transport systems for divalent metal cations may or may not include zinc in their substrate range and may or may not be present in E. coli or Salmonella (Table 1). Such systems include magnesium transporters from the CorA (transport category [TC] 1.A.35 [283]) or MgtE (TC 9.A.19) protein family; the natural resistance-associated macrophage protein (NRAMP) family (TC 2.A.55), predominantly for manganese import; the Zrt/Irt-like protein (ZIP) family (TC 2.A.5) of zinc/iron transporters; the nickel/cobalt transporter (NiCoT) family (TC 2.A.52); and the Pit uptake systems (TC 2.A.20) for metal phosphate complexes.

Zn²⁺ is imported into E. coli cells with an overall K_m of 20 μM (34). There is some interrelationship between Mg²⁺ and Zn²⁺ uptake processes in E. coli (343), but it is not exactly clear which transport systems are responsible for these physiological processes. MgtE-like magnesium transport systems do not occur in E. coli or Salmonella (331). However, Salmonella possesses three Mg²⁺ uptake systems (134), a CorA protein and two members (MgtA and MgtB) of the P-type ATPases (TC 3.A.3), a family of uptake or efflux proteins mostly for inorganic cations and other positively charged particles, from protons to Pb²⁺ (84). Mg²⁺ uptake into Salmonella cells by CorA or MgtB is not inhibited by Zn²⁺, while that by MgtA is, indicating that MgtA may be involved in Zn²⁺ uptake into Salmonella (306). E. coli does not contain an MgtB ortholog but does possess orthologs of CorA and MgtA. Since the previously described corB and mgt/mtg mutations in E. coli are in fact mutations in the mgtA gene (249), CorA and MgtA are presently the only known Mg²⁺ uptake systems in E. coli. Analogous to the system in Salmonella, MgtA may also import Zn²⁺ into E. coli cells.

PitA is a nonspecific and fast uptake system for inorganic phosphate that also imports zinc phosphate into E. coli at relatively high (>0.5 mM) Zn²⁺ concentrations (16), opening another entry route for zinc cations. NiCoTs (81) and NRAMP transporters (101) do not prefer Zn²⁺ as a substrate; a discussion of zinc uptake by these systems can be neglected given the current state of our knowledge. The ZIP transporters (80) are another known group of fast and nonspecific zinc uptake systems. ZupT is the only ZIP in E. coli or Salmonella (112). It is a constitutively expressed uptake system with broad substrate specificity for divalent metal cations (110). ZupT substrates include Zn²⁺, Co²⁺, Mn²⁺, Fe²⁺, and Cd²⁺. Thus, relatively fast and nonspecific import of Zn²⁺ into cells should be mediated by ZupT, maybe by the P-type ATPase MgtA, and at higher concentrations of zinc in the form of zinc phosphate complexes, by PitA.

In times of starvation, additional high-affinity zinc uptake systems may be induced to maintain the cellular supply of this essential trace element. As a member of the ATP-binding cassette (ABC) transport protein family (TC 3.A.1), the ZnuABC uptake system in E. coli is composed of the dimeric ZnuB membrane-bound passageway permease subunit, the dimeric ZnuC ATPase subunit, and the ZnuA protein (251), which binds Zn²⁺ in the periplasm. The binding of Zn²⁺ to ZnuA is mediated by three histidine residues (His78, His161, and His225) and one glutamate residue (Glu77) (186). The ZnuA Zn²⁺ chaperone shuttles the metal to the ZnuBC protein complex for subsequent uptake. This “outsourcing” of the actual binding process within the periplasm, combined with the high energy input dedicated to accumulation, enables the cellular zinc levels to be sustained even at low environmental Zn²⁺ concentrations. At least one molecule of ATP is hydrolyzed (producing a change of −50 kJ/mol in the energy level), and probably two positive charges are imported in the direction of the charge gradient of the proton motive force ΔΨ of 100 mV. This process results in the use of 70 kJ of energy/mol and a theoretical accumulation factor of 10¹² at 303 K.

E. coli may even posses a second import route under conditions of zinc starvation, but the data are inclusive at the moment. The ZinT (YodA) protein is expressed under cadmium stress and translocated into the periplasm (260). It binds Zn²⁺, Cd²⁺, Ni²⁺, and Hg²⁺ (159) to a lipocalin-like domain (61). Lipocalins are a heterologous family of small, mostly extracellular proteins that bind a variety of molecules (91). ZinT may function as a periplasmic zinc chaperone (159), serving as the zinc receptor in a yet uncharacterized import route.

Transport-Mediated Flow Equilibrium: Zinc Export.

Up to four efflux systems counteract the activity of the zinc uptake systems and maintain an ambient zinc concentration in the periplasm: the P-type ATPase ZntA (268), the cation diffusion facilitator (CDF; TC 2.A.4) protein ZitB (109), the CorA-like protein ZntB (350), and maybe the CDF protein FieF (YiiP) (344). Although ZntB is a CorA-like protein, it seems to export (rather than import) Zn²⁺ in Salmonella (350). Its ortholog in E. coli, B1342/YdaN, has not yet been characterized, but it is probably not involved in zinc homeostasis (350).

ZntA (268) is the most important zinc detoxification system in E. coli (79). It detoxifies Zn²⁺ by efflux to the periplasm, with a K_m of 9 μM Zn²⁺ (268). P-type ATPases such as ZntA that detoxify soft metals may even be able to export metal ions attached to thiol compounds (266), which explains their important role in metal ion homeostasis.

Less important than ZntA is the CDF family protein ZitB (YbgR) (109), which is induced specifically by Zn²⁺. E. coli possesses a second CDF protein, FieF (YiiP), which seems to produce Fe²⁺ efflux in vivo (111) and is induced by zinc (109) but is not required for zinc resistance. On the other hand, FieF binds to and transports Zn²⁺ in vitro (45, 344). The structure of FieF/YiiP indicates a structural zinc-binding site (198). Zn²⁺ may bind to this site and hold the wobbly peptide chains in the correct conformation, as in other proteins. The data presently available do not unambiguously determine whether FieF transports only Fe²⁺ or can export Zn²⁺ as well.

Transport-Mediated Flow Equilibrium for Zinc: Summary.

E. coli and Salmonella contain a set of fast and nonspecific uptake systems that cotransport Zn²⁺ and a major substrate cation like Mg²⁺ or Fe²⁺. Under conditions of zinc starvation, high-affinity uptake systems can be induced to supply sufficient zinc to the cell. Surplus cytoplasmic zinc can be removed by inducible efflux systems. The expression of inducible uptake and efflux systems is regulated at the level of transcription. Additionally, metal cation transport systems may be controlled at the level of transport activity, leading to efficient flux control of metal cation homeostasis (225).

Regulatory Circuits Involving Zinc.

The expression of the ZnuABC uptake system is under the control of the Zur protein (251), a paralog of the main iron homeostasis regulator Fur. Zur represses transcription by binding to an operator located between the divergently transcribed znuA and znuCB operons (236, 251). Zur binds two Zn²⁺ cations per monomer. Site A sequesters the cation very tightly and may be a structurally important site. Site B binds zinc more loosely and uses a tetrahedral ligand environment to favor zinc over iron binding (236). In vitro, the affinity of Zur for Zn²⁺ is in the femtomolar concentration range (236).

The expression of the main zinc efflux pump, ZntA, is under the control of the MerR-type regulator ZntR (32, 237). The sigmoidal activity functions of Zur and ZntR overlap, keeping the cytoplasmic free Zn²⁺ concentration in a narrow window between 2 × 10⁻¹⁶ M (the half-maximal induction point of Zur regulation) and 10⁻¹⁵ M (the half-maximal induction point of ZntR regulation) (237). If the cytoplasmic glutathione (GSH) content, in the millimolar concentration range, and the zinc-binding constants of GSH (88) are taken into consideration, these values agree with a zinc quota of 200 μM Zn²⁺ in the cytoplasm (236). Thus, the Znu uptake system is synthesized when the cytoplasmic Zn²⁺ concentration decreases below the homeostasis concentration window. Conversely, the ZntA efflux pump will be present when the cytoplasmic Zn²⁺ concentration rises above this window.

There may be additional fine-tuning of the cytoplasmic Mg²⁺/Zn²⁺ ratio by flux control. Each of the solution structures of the magnesium uptake systems CorA and MgtE (124, 199) indicates the presence of regulatory magnesium-binding sites. High cytoplasmic Mg²⁺ concentrations would lead to the binding of magnesium to these sites and the inhibition of transport activity. The structure of YiiP/FieF also suggests the presence of such sites for metal efflux (198). If YiiP is a Zn²⁺ efflux system, high cytoplasmic zinc concentrations may lead to the binding of zinc to these allosteric sites, resulting in up-regulation of the transport activity of the efflux system. Although structures for other CDF proteins, uptake systems, or P-type export ATPases for soft metal cations are not yet available, a multitude of predicted or experimentally verified metal-binding sites in the cytoplasmic domains of these proteins suggest that the proposed flux control of CorA, MgtE, and YiiP follows a general principle (225). Thus, a change in cytoplasmic zinc or magnesium concentrations may not only regulate the synthesis of transport systems on the expression level but also influence the activities of these systems.

Physiological Functions of Zinc.

Currently, close to 100 zinc enzymes or zinc-binding proteins in E. coli (Table 2), binding mostly one or two zinc cations per polypeptide, are known. Table 2 gives estimates of the abundance of zinc proteins from transcript quantification of gene array data (79). To this end, total mRNA was isolated from cells cultivated in the presence of trace concentrations of transition metals. The RNA polymerase of E. coli has a copy number of roughly 5,000 ± 3,000 per cell (142, 294). If the polymerase subunit RpoB represents 0.5% of all zinc enzymes (Table 2), E. coli cells will contain 1,000,000 copies of zinc enzymes (constituting about 13% of all proteins), harboring at least 1,000,000 Zn²⁺ ions, per cell. A zinc quota of 200 μM was determined previously by inductively coupled plasma mass spectrometry (236). Assuming a cellular volume of 1.8 fl, this value corresponds to 217,000 Zn²⁺ ions per cell. This is a remarkably close match, considering the rather blunt approach.

Ribosomal proteins (about 25% of the total protein) and tRNA synthetases are among the most abundant zinc proteins. Bacillus subtilis contains a paralogous pair of ribosomal proteins, one of which possesses a zinc-binding site and may serve as a cellular zinc buffer (4, 246). Under zinc limitation, the non-zinc-binding paralog is produced and replaces its zinc-storing paralog from the ribosome. The degradation of the latter should release Zn²⁺ for use by other proteins, e.g., the tRNA synthetases and RNA polymerase. A similar process operates in Streptomyces coelicolor (240, 298) and may also operate, albeit in a different fashion, in E. coli (Table 3 indicates the level of down-regulation of the rpsU gene for S21 in zinc-shocked cells). However, this issue requires further experimental attention.

Zinc as a Toxic Metal.

Only a few investigations deal with zinc toxicity in E. coli or Salmonella. It is difficult to derive any general principle about the molecular reasons behind zinc toxicity from this information. Global transcript studies have examined the cellular response of E. coli to toxic levels of Zn²⁺. One experiment (79) challenged cells in batch culture with 100 μM Zn²⁺ for 10 min, while the others used various zinc concentrations (359) or investigated the chronic effect of 200 μM Zn²⁺ (178) in continuous cultures without inorganic phosphate in the growth media.

Table 3 provides results from a reexamination of the zinc-shocked cells, as taken from the supplemental material of reference 79. Beside a regulatory effect on ribonucleoside diphosphate reductase and a few gene products with unknown functions, the regulated genes encode mainly factors involved in zinc homeostasis, iron metabolism, or cell wall integrity (Table 3). High zinc concentrations disturb iron metabolism in Saccharomyces cerevisiae cells also (244). Additionally, high zinc concentrations influence cysteine biosynthesis in E. coli (359). Transcripts for the periplasmic metal-binding protein ZraP (Table 3) showed the strongest up-regulation in one study (231) and were strongly up-regulated in a second study (178) but not in a third (359). Increased expression of ZraP may lead to efficient zinc buffering in the periplasm, capturing the metal cation before it can reach the cytoplasm.

Nickel and Cobalt

Transport-Mediated Flow Equilibrium.

All three magnesium uptake systems, CorA, MgtA, and MgtB, in Salmonella (and probably also the CorA and MgtA orthologs in E. coli) are inhibited by Ni²⁺ and Co²⁺, albeit to different degrees (306). CorA is probably the main route for the uptake of Ni²⁺ and Co²⁺ into E. coli cells (355), since corA mutants are cobalt tolerant (343). Moreover, NRAMP systems like MntH (3, 101) and the ZIP family transporter ZupT (110) may provide additional entry routes (Table 1). The uptake of Ni²⁺ or Co²⁺ phosphate complexes by PitA, however, has not been shown (16). Thus, a variety of nonspecific uptake systems for Ni²⁺ and Co²⁺ in E. coli and Salmonella seem to exist.

E. coli does not contain a NiCoT uptake system for Ni²⁺ or Co²⁺ but does possess an inducible ABC uptake system, NikABCDE (270, 356). NikA binds Ni²⁺ in the periplasm with a K_d of <0.1 μM. Binding to each of the approximately 23,000 NikA proteins in the periplasm of anaerobically growing E. coli cells takes place on at least six carboxylate side chains of aspartate or glutamate residues at a one-to-one ratio (5, 72). NikA has a 10-fold preference for binding to Ni²⁺ over other divalent metal cations (72). Additionally, NikA binds heme and is involved in periplasmic heme assembly (293). NikA-bound Ni²⁺ is transported into the cytoplasm by the integral membrane proteins NikBC and the ATPase subunits NikDE of the Nik ABC uptake system (220). There is no known cobalt-specific ABC uptake system in E. coli or Salmonella (125), and there is probably no need for such a system, as discussed below.

So, reminiscent of the Zn²⁺ uptake systems, various nonspecific and fast uptake systems are responsible for the uptake of Ni²⁺ and Co²⁺ into Escherichia or Salmonella cells while an inducible, slow, specific, and expensive ABC transporter supplies Ni²⁺ under conditions of depletion. The counterpart of the Ni²⁺ and Co²⁺ uptake systems is the efflux protein RncA (YohM) of the NiCoT protein family (272). Other efflux pumps in E. coli or Salmonella are not known (223). Therefore, the transport-mediated flow equilibrium system for Ni²⁺ and Co²⁺ is composed of (i) nonspecific and fast uptake of both cations (by CorA, MgtA, MgtB, ZupT, and MntH), (ii) inducible and highly specific uptake of Ni²⁺ only (by NikABCDE), and (iii) RcnA-mediated efflux of both cations.

Physiological Actions of Nickel and Cobalt.

There are two prominent nickel-dependent enzymes in E. coli: glyoxylase I and hydrogenase (125). In addition, the fructoselysine-3-epimerase FrlC is activated by 100 μM Ni²⁺ (345). Other enzymes such as the argininosuccinate lyase ArgH and aminopeptidases PepB, YpdE, and YpdF are also activated by Ni²⁺, but other divalent cations (at least Co²⁺in the case of ArgH) may substitute for this cation. Finally, many pathogenic E. coli strains may contain genes for the classical nickel-containing enzyme urease. However, due to a stop codon in the ureD gene, most of these strains do not express urease activity (219).

Methylglyoxal, the aldehyde form of pyruvate, is generated in small amounts during various biochemical processes, including glycolysis and fatty acid and protein metabolism. The highly toxic substance is degraded via four different pathways in E. coli (153). Pathway I comprises the formation of S-lactoyl-GSH from methylglyoxal and GSH by glyoxylase I (GloA), followed by the hydrolysis of this compound by glyoxlase II (GloB) or the S-lactoyl-GSH hydrolase activity of YeiG to form D-lactate and GSH. Since glyoxylases in mammals depend on zinc, the nickel dependence of GloA was a surprise (52). Ni²⁺ is bound to the enzyme in an octahedral complex composed of two histidine residues, two glutamate residues, and two water molecules (62). Ni²⁺ but also Co²⁺, Zn²⁺, Cd²⁺, and to a lesser degree, Mn²⁺ bind tightly to GloA. All but the Zn²⁺ form of GloA are active (53). Thus, GloA activity is not Ni²⁺ specific.

The only exclusively nickel-dependent metabolic pathway in E. coli known today involves hydrogen evolution and consumption by hydrogenases. E. coli harbors gene clusters for four different hydrogenases. Hydrogenases I and II are involved primarily in consuming molecular hydrogen as a source of reducing power (15), while hydrogenase III is part of the formate hydrogenlyase complex. The hyf cluster for hydrogenase IV seems to be a silent operon. The active hydrogenases contain Ni²⁺ in a binuclear metal center ligated to four cysteine residues, two of them bridging to the second metal, iron (125). This center is assembled in several steps that begin with iron insertion (22) and continue with GTP-dependent (203) nickel insertion by HypC, HypA (alternatively HybG or HybF), and SlyD proteins (10, 177, 363), the cleavage of a carboxy-terminal metal insertion signal sequence (202, 275), and if required, transport into the periplasm (271) by the twin-arginine protein export pathway.

Hydrogenases are needed only under anaerobic conditions, when E. coli conserves energy by fermentation. The absence of hydrogenases retards but does not eliminate anaerobic growth (355). Thus, nickel is not an essential metal cation for E. coli, under either anaerobic or aerobic conditions. Interestingly, the same is true for free cobalt cations: all enzymes listed as cobalt enzymes need a divalent metal cation but not exclusively Co²⁺ (125, 153), except in all B₁₂-dependent reactions. The cofactor B₁₂ and its derivatives contain cobalt as a kinetically stable Co(III) complex. E. coli is not able to synthesize B₁₂ and has to take up the compound or precursors (175) from the environment by using the Btu ABC uptake system. Salmonella enterica serovar Typhimurium strains are B₁₂ prototrophs (85, 310). Thus, E. coli depends exclusively on Co(III) complexes in B₁₂ but does not need free Co²⁺ cations.

Toxic Actions of Cobalt and Nickel.

Based on their binding affinities, Co²⁺ and Ni²⁺ should be effective competitors for Fe²⁺ (141, 222). This situation has indeed been demonstrated. In S. enterica, Co²⁺ interferes heavily with iron homeostasis and iron-sulfur cluster assembly. By doing so, it affects siroheme production, leading to a decline of sulfide reductase activity and sulfur assimilation (327) in Salmonella. Similar results were obtained for E. coli (263). In addition, Co²⁺ produces reactive oxygen species when bound to GSH (181). Ni²⁺ also interferes with iron homeostasis (63) in different organisms (194).

Regulatory Circuits Involving Nickel and Cobalt.

In E. coli, Co²⁺ is not essential (except as B₁₂) and Ni²⁺ is needed only to enhance anaerobic growth as an essential cofactor of hydrogenases. However, each metal, when present in surplus, interferes with iron homeostasis. Crucial control points include the regulation of the Nik ABC uptake and RncA efflux systems.

The expression of the Nik ABC uptake system is under triple control. The Fnr regulator prevents nik expression under aerobic conditions, the NarL response regulator links nik expression to nitrate respiration, and the nickel-responsive protein NikR represses nik when sufficient cytoplasmic Ni²⁺ is available (276). The nikR gene is located directly downstream of the nikABCDE operon and is expressed by read-through or induction by its own constitutive promoter (71). NikR is a member of the ribbon-helix-helix family of transcription factors (50). It is arranged as a quaternary protein consisting of two dimeric DNA-binding domains separated by a tetrameric regulatory domain that binds four nickel ions (286). Two binding sites are high affinity (131) and, when occupied, cause NikR binding to the nik operator site (180). Additional occupation of the low-affinity site enhances operator binding (24); increases footprinting and regulates the rearrangement of the NikR conformation and nickel coordination (42); and improves the stability of the metal-protein complex (51). Other transition metal cations can bind to NikR and even outcompete Ni²⁺, but the resulting proteins are less stable than the Ni²⁺-containing forms (341). Thus, metal cation binding to NikR leads to DNA binding and mild repression of nik, but only the nickel-containing NikR proteins remain stable at the nik operator. In a second step, even higher nickel concentrations rearrange the repressor-operator complex, leading to tight repression of nik.

The second regulatory circuit affecting Ni²⁺/Co²⁺ homeostasis involves regulation of the expression of the rcnA gene for the Ni²⁺/Co²⁺ efflux pump RcnA by RcnR (272). This control circuit is very complex and has not been understood in full detail yet. The expression of rcnA is induced by Ni²⁺ and Co²⁺. Overproduction of rcnR inhibits the induction of rcnA by metal cations. When rcnR or fur, the gene of the global repressor of iron homeostasis, is deleted, the expression of rcnA is also induced by iron (167). RcnR and Fur bind to the common promoter region of the divergently transcribed rcnA and rcnR genes (143, 167). Moreover, there is an interconnection with NikR control (143).

Summary of Cobalt and Nickel Homeostasis.

Cytoplasmic levels of both cobalt and nickel cations are maintained by regulating the transport-mediated flow equilibrium, encompassing nonspecific uptake and RcnR-controlled expression of the RcnA efflux system. When more nickel is needed for hydrogenase biosynthesis under anaerobic conditions, NikR activates the expression of the Nik ABC uptake system. Nickel chaperones channel the cation into the hydrogenase assembly complexes. Cobalt is essential only as Co(III) in B₁₂. All these processes maintain cellular Co²⁺ and Ni²⁺ quotas that are one to two orders of magnitude lower than the Zn²⁺ quota (238), which prevents cobalt and nickel interference with iron homeostasis.

MAIN REDOX-ACTIVE TRANSITION METALS: MANGANESE, IRON, AND COPPER

General Considerations

Under physiological conditions, zinc, cobalt, and nickel are maintained as free divalent cations, although other oxidation states may exist in complex compounds. In contrast, the oxidation state of iron may vary between Fe(II) and Fe(III) [and Fe(IV)], that of copper may vary between Cu(I) and Cu(II), and manganese may adopt various oxidation states besides Mn(II) (342). A change in the oxidation state of iron or copper may lead to the production of radicals, especially when molecular oxygen is present. This makes the borderline transition metal ferrous iron, as Fe²⁺, dangerous and the soft metal cuprous copper, as Cu⁺, even more so. Copper homeostasis in E. coli and Salmonella is described in the chapter by Rensing and Franke (Chapter 5.4.4.1) and will not be covered here. Manganese, on the other hand, is able to quench radicals and is the transition metal with the lowest toxicity for E. coli cells (222, 224).

Manganese

General Considerations: Toxicity versus Beneficial Effects.

All oxidation states of manganese, from Mn(II) to the strongly oxidizing Mn(VII), are quite stable, especially Mn(II) and Mn(IV) (138). The most stable forms of iron are Fe(II) and Fe(III), and those of copper are Cu(I) and Cu(II). These properties define a major difference between copper/iron and manganese: the former prefer one-electron redox reactions leading to the formation of radicals, while one- and two-electron redox reactions are possible for manganese. This characteristic leads not only to the low toxicity of Mn²⁺ but also to beneficial effects of inorganic Mn²⁺-containing complexes (137). As first observed in a lactic acid bacterium (9), Mn²⁺ appears to be able to detoxify superoxide radicals in numerous bacteria, not only as an active part of superoxide dismutase SodA, but also as a component in inorganic complexes or as an element loosely bound to the surfaces of proteins (64, 137, 332). In this capacity, Mn²⁺ is also able to prevent radiation damage (59). Since superoxide dismutase deletion mutants can be complemented by high manganese concentrations in the growth medium, this protecting effect is also important for E. coli (6).

Transport of Manganese.

Because Mn²⁺ has such low toxicity, bacteria have not evolved efflux systems for this metal cation. The only example for such a system is a manganese-detoxifying CDF protein from plants (68). Therefore, manganese homeostasis in bacteria seems to rely solely on regulated uptake (Table 1).

Mn²⁺ is actively transported into E. coli and Salmonella by the NRAMP MntH (158). The K_m values for Mn²⁺ uptake are 0.2 and 0.1 μM for E. coli and Salmonella, respectively. The V_max allows up to 25 single cations per cell per s to be taken up (301). Other substrates for MntH are Fe²⁺ and the toxic metal cation Cd²⁺. The transport of manganese is inhibited by Ni²⁺ and Co²⁺, indicating that these cations may be additional substrates for MntH. Conversely, calcium and magnesium do not inhibit manganese uptake and are probably not transported by MntH (302).

Other possible entry routes for Mn²⁺ into E. coli cells are provided by the ZIP transporter ZupT (110) and the CorA magnesium uptake system (249). So while magnesium does not inhibit manganese uptake, manganese interferes with magnesium transport (299, 304). An FeoB-related protein transports Mn²⁺ in Porphyromonas gingivalis (60). FeoB is involved in iron transport in E. coli, but a role in manganese uptake has not been investigated (281). The Mn²⁺-transporting P-type ATPases found in lactobacilli are not found in E. coli (113).

However, a highly specific Mn²⁺ ABC uptake system designated SitABCD occurs in Salmonella (157). This system is also commonly found in pathogenic E. coli strains (282) but is missing in K-12-like strains. The reason for this difference may be that Mn²⁺ transporters are essential for the survival of pathogenic bacteria in mammalian hosts that possess functional NRAMP transporters (247).

Control of Manganese Homeostasis.

As noted above, there are no known Mn²⁺ efflux systems in bacteria. Thus, the intracellular manganese concentration is controlled solely by regulating uptake. This regulation has to integrate not only the information on the cytoplasmic manganese concentration but also that on the iron concentration and the oxidative-stress situation (137). As a result, the expression of the mntH gene is under the control of the Fur iron uptake repressor and the MntR manganese uptake repressor in E. coli (250). In Salmonella, mntH expression is also regulated by OxyR (156), a LysR-like regulator active only in its oxidized form (318).

Like that of mntH, the expression of the SitABCD uptake system in Salmonella is under MntR and Fur control, and OxyR may also play a role (139). Parallel regulation of mntH and sit prevents excess Mn²⁺ accumulation at sufficient cytoplasmic Fe²⁺ and Mn²⁺ concentrations and enhances Mn²⁺ uptake when efficient detoxification of reactive oxygen species is needed.

Manganese-Dependent Enzymes.

Currently, about 50 E. coli proteins show some dependence on Mn²⁺ (Table 4). Most, however, can function with other divalent metal cations (Mg²⁺, Fe²⁺, Co²⁺, and Zn²⁺, alone or in combination with one another and other divalent metal cations) as well. The list of Mn²⁺/(any divalent cation)-dependent enzymes probably reflects some in vitro artifacts allowing the functional reconstitution of an Mn²⁺-depeleted enzyme by another divalent metal cation. These artifacts may not signal any importance of that cation for the respective enzyme in vivo.

Manganese has an important role in the detoxification of superoxide radicals, as inorganic manganese compounds and as a constituent of the manganese-containing cytoplasmic superoxide dismutase SodA (19, 155). Besides SodA, E. coli possesses a periplasmic copper-zinc superoxide dismutase (Salmonella has two of these enzymes) and the iron-containing cytoplasmic superoxide dismutase SodB, which is present under anaerobic and aerobic conditions. SodA is synthesized only under oxic conditions and seems to defend especially DNA against oxidative damage (136, 330). For SodA and a few other enzymes, no substitutes for Mn²⁺ have been found or tested (Table 4).

All these data mark Mn²⁺ as a unique transition metal cation in bacteria. In contrast to the other metals discussed here, Mn²⁺ does not bind too strongly to thiols, thus avoiding damage to proteins; it probably does not displace other transition metal cations from their physiological complexes; and it is unlikely to produce radicals but quenches them instead. Manganese may thus be the most benign cation among the heavy metals.

Homeostasis of Iron in Escherichia and Salmonella

Overview of Iron.

Similar to zinc, iron plays a critical role in biological systems. Due to its redox properties, iron is utilized by countless iron-containing enzymes involved in electron transfer reactions in which electron transport is facilitated by changes in the oxidation state of the metal. These reactions encompass, but are not limited to, aerobic and anaerobic respiration, photosynthesis, and general metabolic pathways such as the tricarboxylic acid cycle and CO₂ and N₂ fixation, as well as DNA biosynthesis.

In contrast to the relative ease by which zinc is acquired by bacterial cells, sophisticated means for seeking and acquiring iron have developed as dictated by the physicochemical properties of this crucial transition metal. The first problem is that under conditions of aerobiosis and neutral pH, iron in its predominant trivalent ferric oxidation state (Fe³⁺) is rather insoluble and precipitates in the form of hydroxide complexes. This property limits available iron in aqueous solutions at pH 7.0 to about 10⁻¹⁸ M (221). However, concentrations of at least 10⁻⁷ to 10⁻⁵ M are needed to sustain optimum life functions (8). The difference in solubility between zinc and iron may explain the evolutionary pressure leading to the great variety of iron uptake mechanisms (Fig. 1) compared to the relatively few zinc transport systems. In the following section, we will explore the multitude of transporters involved in iron homeostasis in Escherichia and Salmonella.

Fig. 1Transporters involved in iron homeostasis in Escherichia and Salmonella. The uptake of organic ferric iron chelates or ferric siderophores is accomplished by a two-step mechanism involving specific receptors of the outer membrane (OM) and cognate cytoplasmic membrane (CM) ABC transporters. Iron chelates are shuttled within the periplasm by specific binding proteins. Siderophores such as enterobactin, salmochelin, yersiniabactin, and aerobactin are biosynthesized by various Escherichia and Salmonella strains. These aposiderophores are secreted by cytoplasmic membrane permeases alone or in concert with outer membrane efflux duct proteins such as TolC. Elemental, nonchelate iron cations are taken up by cytoplasmic membrane permeases. Ferric iron uses ABC transporters similar to ferric siderophores, whereas ferrous iron is transported by members of different transport families, which may use GTP hydrolysis or the proton motive force to drive iron uptake. Under conditions of iron overload, cells may use efflux permeases for the rapid detoxification of potentially toxic ferrous iron. Protein structures were generated using the Protein Workshop software (216). PP, periplasm; CP, cytoplasm.

Moving Iron across the Outer Membrane: Outer Membrane Receptors and Siderophores.

When one thinks of bacterial iron acquisition, siderophores and their dedicated outer membrane receptors are the first mechanisms that come to mind. Siderophores, small organic iron-chelating compounds, are excellent tools for iron chelation, thus counteracting the low solubility of ferric iron. Escherichia and Salmonella produce one of the most efficient siderophores we know of, enterobactin (or enterochelin) (232, 256). This cyclic triester of 2,3-dihydroxy-N-benzoyl-L-serine (DHBS) exhibits extremely high affinity for ferric iron (K_m = 10⁵² M⁻¹) (41). This high-affinity siderophore enables cells to thrive even in environments that are nearly devoid of iron.

How does E. coli harvest those siderophore-iron complexes and wrest the tightly bound iron from them? The first task is accomplished by specific outer membrane ferric siderophore receptors. These β-barrel proteins are also named IROMPs (iron-regulated outer membrane proteins) because their expression is repressed in an iron-dependent manner. The pores of the β-barrel IROMPs are tightly closed by a globular domain, called the plug, cork, or hatch. Regulated transport through these ligand-gated porin receptors requires the activity of the TonB-ExbBD complex, which transduces energy from the proton motive force of the cytoplasmic membrane to the outer membrane receptors (28, 133). This energy relay system promotes efficient ligand transport mediated by the movement of the plug domain.

E. coli strain K-12 possesses several different outer membrane receptors (Table 5) governing ferric siderophore recognition (reviewed in references 29 and 165). These include the FepA, CirA, and Fiu receptors that recognize catecholate siderophores. FepA takes up cyclic enterobactin. Degradation products of enterobactin, which also function as siderophores, can be channeled through CirA and Fiu. The FhuA and FhuE receptors are dedicated to xenosiderophores, such as ferrioxamines, and fungal siderophores, such as ferrichromes, ferric rhodotorulic acid, and ferric coprogen. While E. coli does not typically utilize citrate as a carbon source, citrate is used by E. coli K-12 and various other strains as a small siderophore to shuttle ferric citrate complexes to the FecA receptor. For YncD, another receptor, we are still ignorant of the cognate substrate. However, orthologs of this protein can be found frequently in bacteria.

Pathogenic strains of Escherichia and Salmonella harbor an even greater variety of iron receptors than E. coli K-12 (Table 5), clearly demonstrating the important role of iron for survival in a host. The glucosylation of enterobactin through IroB extends this siderophore's value enormously when the bacteria reside within a mammalian host. Glucosylated enterobactin derivatives, named salmochelins, can evade the host's defense system: enterobactin is bound by the mammalian bacteriostatic protein neutrophil gelatinase-associated lipocalin and thereby inactivated (90), whereas salmochelin is not. Strains producing salmochelin also harbor the gene for the outer membrane receptor IroN, which is specific for salmochelin and other catecholate-type siderophores (123).

One receptor also frequently found in pathogenic strains is IutA, the receptor for the hydroxamate siderophore aerobactin. While aerobactin does not possess the same high affinity for ferric iron as enterobactin, aerobactin is more stable at an acidic pH (pH 5.6) (335).

Some pathogenic E. coli strains produce the siderophore yersiniabactin, which was first identified in Yersinia enterocolitica (127). It is a mixed polyketide-polypeptide siderophore composed of a phenolic group, one thiazolidine ring, and two thiazoline rings (102). The ferric yersinabactin outer membrane receptor FyuA is encoded by part of the ybt operon, which is responsible for the biosynthesis of yersiniabactin.

Additional outer membrane receptors in pathogenic E. coli and Salmonella strains have been identified. The Iha receptor recognizes linear DHBS derivatives (185). The FoxA receptor confers upon Salmonella the ability to utilize ferrioxamines (163). The substrates of IreA (279), FitA (239), and C2482 (119) have yet to be determined.

Some E. coli strains, in addition to employing siderophores, can also liberate bound Fe³⁺ from host carriers. The ChuA receptor, for instance, enables strains to use heme as an iron source (329). It remains to be seen whether further receptors in Escherichia or Salmonella strains can be identified in the future.

Iron Uptake Permeases of the Inner Membrane.

Once inside the periplasm, iron-siderophore complexes are recognized by specific periplasmic binding proteins, which are usually part of a tripartite ABC transporter system. The function of the binding protein is to shuttle the loaded siderophore to a specific permease that is energized by an ATP-hydrolyzing subunit. Depending on the siderophore, the dimeric permease may transport the ferric siderophore or only the metal cation to the cytoplasm. If the siderophore is being transported, specific esterases will break down the siderophore inside the cytoplasm, thus weakening the affinity for iron (30). This process is accompanied by the reduction of the ferric ion (Fe³⁺) to the soluble ferrous ion (Fe²⁺).

Although the outer membrane receptors build a functional unit with an ABC-type transport system, more outer membrane receptors are dedicated to iron acquisition than there are cognate cytoplasmic membrane iron permeases (Table 6). This arrangement stems from the fact that some permeases function with more than one receptor. So far, the following ABC systems related to ferric siderophore uptake in Escherichia and Salmonella have been identified: (i) Fep for enterobactin and other catecholate siderophores (349), (ii) Fhu for ferric ferrioxamines, ferrichromes, aerobactin, and rhodotorulic acid (86), (iii) Fec for ferric citrate (315), (iv) Fit for an unknown substrate (239), and (v) Chu (329) and (vi) the peptide permease DppABCDF for heme uptake (184). Further, YbtP and YbtQ are likely to be the uptake pumps for ferric yersiniabactin (89). As far as we know, no periplasmic binding protein is involved in the Ybt transport process.

The identity of the uptake mechanism for ferric salmochelin in E. coli or Salmonella has been controversial. Initially, uptake was thought to utilize the ATPase IroC (369), a relative of the yersiniabactin ATPases YbtP and YbtQ but twice the size of each of these “half-transporters.” Recent experimental evidence now supports the idea that IroC is involved in aposalmochelin secretion rather than uptake (see below). Instead, the periplasmic enterobactin-binding protein FepB, along with its cognate ABC transporter FepCDG, is responsible for salmochelin uptake across the inner membrane (56).

In addition to the plethora of receptor-dependent iron acquisition systems in Escherichia and Salmonella, various strains possess ABC systems dedicated to nonchelate Fe³⁺ uptake. In these systems, Fe³⁺ enters the periplasm by facilitated diffusion through outer membrane porins and encounters the ABC transport system. Such systems include EitABCD in avian pathogenic strains of E. coli and AfuABC in E. coli O157:H7, which both are distantly related to SitABCD in S. enterica (305). Similar systems can also be found in other members of the Enterobacteriaceae, such as Yersinia, and may contribute to bacterial virulence. The Sit system is involved in the abilities of various E. coli strains to colonize a mouse host. A strain with sit deleted exhibits reduced colonization of the lungs, liver, and spleen (281). However, it is not clear if the decreased virulence is due to diminished iron or manganese uptake by the bacterium.

Similar to the array of systems for Zn²⁺, Ni²⁺, and Co²⁺, one specific and a variety of rather nonspecific uptake systems contribute to the import of ferrous iron, Fe²⁺. These transporters are encoded mainly by the bacterial chromosome and are not ABC transport systems. The most prominent representative is probably the FeoB protein, an integral membrane protein with a soluble N-terminal domain that functions as a GTPase (206). FeoB proteins are widespread among bacteria and are involved in the transport of Fe²⁺ and Mn²⁺ (43). However, the transport of Fe²⁺ through FeoB has not been directly demonstrated, so FeoB may also stimulate iron and manganese uptake by an unknown transport system. Whatever FeoB does, GTPase activity is necessary for its function.

Another minor uptake route for Fe²⁺ is MntH, the general Mn(II) uptake system of Escherichia and Salmonella strains. The expression of MntH from a plasmid in an E. coli mutant with all known iron uptake systems deleted results in higher cellular growth yields than those from a plasmid-only control (23). Similar to E. coli cells expressing MntH, E. coli cells lacking all known iron transporters but expressing the ZIP transporter (see above) ZupT exhibit improved growth yields compared with those of a control (110).

Recently, a novel bacterial iron uptake permease from different E. coli strains was identified and characterized. This protein belongs to the ferric transporter (FTR) family of iron transporters, which are known best in the forms from lower eukaryotes such as yeasts and algae. The E. coli gene of this family, named efeU (elemental ferrous iron uptake), is part of a tricistronic operon. This operon is under control by the Fur repressor in response to the iron status (114) and is repressed at high pHs by CpxAR (39). Currently, we are ignorant of the contribution of EfeU to the cellular iron status, but when the efe operon is expressed in a strain lacking all known iron uptake systems, cells gain a major growth advantage (39, 114). Interestingly, in E. coli the EfeU protein probably transports ferrous iron, whereas in yeast the related Ftr1p protein transports ferric iron (114, 316). This difference may be explained by the different auxiliary proteins involved in the transport process, but further work in this area is needed.

Some E. coli strains harbor a second FTR-type iron permease (D. Koch and G. Grass, unpublished data). The respective gene (EcolF_01003470) comprises a much larger open reading frame than that of efeU, and it is part, not of a tri-, but of a dicistronic operon. Similar to the auxiliary proteins (YcdBO) for the tricistronic operon product, the unrelated protein (the EcolF_01003469 protein) accompanying the second FTR permease is localized in the periplasm. While none of these auxiliary proteins show similarity to canonic periplasmic binding proteins, they still may be involved in iron delivery to the integral iron permease. It is not unlikely, however, that the large soluble N-terminal domain of the second FTR transporter itself functions as a periplasmic collector for iron cations for further translocation into the cytoplasm. The distribution of this second FTR-type iron permease is, however, much narrower than that of EfeU (Table 6), and its physiological function has yet to be determined.

Fate of Iron in the Cytoplasm: Sequestration, Storage, or Efflux.

Iron has to be prevented from taking part in Fenton reactions, which produce reactive oxygen species that immediately damage proteins, lipids, and nucleic acids. Thus, keeping intracellular iron levels in a narrow physiological range is important for cell survival. This maintenance can be accomplished by (i) the strict regulation of iron uptake, (ii) the control of delivery to iron-containing proteins, (iii) the storage of cytoplasmic iron, or (iv) the efflux of surplus iron.

Iron homeostasis in Escherichia and Salmonella is strictly regulated through the Fur protein. Upon iron binding under conditions of iron repletion, Fur is converted into a repressor, efficiently shutting down the biosynthesis of most iron uptake proteins and siderophores (14). Moreover, the Fe-Fur protein also represses the transcription of the small regulatory mRNA RyhB. The expression of RyhB during iron starvation results in the hybridization of the RyhB transcript with mRNAs of iron-consuming or storage proteins such as iron superoxide dismutase, aconitase, or ferritin (208). In other words, the active Fe-Fur repressor indirectly enables the biosynthesis of these proteins through the repression of RyhB. This rather complex regulatory mechanism makes sure that excess iron is safely handled.

Proteins such as ferritin, bacterioferritin, and the DNA-binding protein of starved cells (Dps) can store surplus cytoplasmic iron (8). These proteins oxidize ferrous iron and accumulate the resulting ferric iron inside. For the control of delivery, iron-binding proteins such as CyaY (176), YggX (108), and YtfE and probably ApbC, as well as ApbE (11), divert a portion of the iron pool to diverse targets such as biosynthesis or the repair of iron-sulfur clusters. We do not know if Escherichia or Salmonella possesses a general iron chaperone for trafficking within the cytoplasm in addition to these specific iron chaperones. Such a protein could be another means to prevent Fenton reactions catalyzed by uncontrolled cytoplasmic iron.

Should iron overload occur despite these control measures, cells may also utilize iron efflux systems. Indeed, FieF (YiiP), a permease of the CDF protein family, probably performs iron efflux in E. coli. This protein transports Fe²⁺ in vitro, its gene is inducible by iron, and the expression of fieF results in diminished cellular iron concentrations (111). In vitro, FieF also binds to and transports Zn²⁺ (344); however, FieF does not mediate Zn²⁺ efflux in vivo (111). It remains to be seen if Zn²⁺ ions are substrates bound during the transport process or if they fulfill a structural role, with ferrous iron being the actual substrate of FieF (198). In other bacteria such as Magnetospirillum gryphiswaldense (115) and Coxiella burnetii (31), orthologs of FieF are also involved in iron transport. Each CDF protein appears to pump out cytoplasmic ferrous iron. In M. gryphiswaldense, this iron is used to load intracellular magnetosome vesicles; in C. burnetii, the role of a CDF protein seems to be iron detoxification under iron surplus conditions.

Efflux Processes Related to Iron Uptake.

Much effort has been invested in elucidating ferric siderophore uptake into the cell. Much less is known about how aposiderophores leave the cell. The siderophores enterobactin, salmochelin, aerobactin, and yersiniabactin are all produced intracellularly by different Escherichia and Salmonella strains and have to be secreted. The major facilitator EntS was the first permease discovered to translocate apoenterobactin across the cytoplasmic membrane in E. coli (99). EntS is a member of the vast major facilitator superfamily of transport proteins (TC 2.A.1) and is probably driven by the proton motive force. The entS gene is part of the approximately 21-kb genomic enterobactin cluster in E. coli, an arrangement which by itself indicates a possible function for entS in enterobactin homeostasis. Once secreted to the periplasm, apoenterobactin is too large to exit the periplasmic compartment through simple diffusion. The outer membrane factor (TC 1.B.17) TolC was identified as the efflux valve for enterobactin (23). The trimeric TolC efflux duct does not function on its own, but other transport components such as transport ATPases, resistance-nodulation-cell division (RND) proteins (TC 2.A.6), or proteins of other transporter families have to partake in efflux in concert with a connecting membrane fusion protein. However, attempts to identify the missing pieces in apoenterobactin secretion have not been successful. Deletions of genes for potential TolC-interacting transport proteins do not significantly affect apoenterobactin secretion. Multiple deletions of all RND genes of E. coli diminishes apoenterobactin secretion but not the level observed in a tolC deletion strain (Hart and Grass, unpublished results).

As noted earlier, the export of salmochelin to the periplasm most likely involves the IroC transport ATPase (56, 123, 369). The question of how aposalmochelins leave the periplasm, however, remains open. It is possible that TolC is involved in this process, too.

Three permease genes are within the genomic locus encoding yersiniabactin biosynthesis and transport. The transport ATPase YbtPQ is involved in ferric yersiniabactin uptake into the cytoplasm, while the major facilitator YbtX is probably responsible for apoyersiniabactin secretion (89). The route for apoyersiniabactin translocation across the outer membrane is unclear.

Even less is known about aerobactin secretion. Given the localization of the shiF gene within the aerobactin locus, the product of this gene may be the major facilitator permease related to apoaerobactin transport across the cytoplasmic membrane. Further, aerobactin may be small enough to leave the periplasm through nonspecific outer membrane porins.

TOXIC-ONLY SOFT HEAVY METALS: CADMIUM, LEAD, SILVER, GOLD, AND MERCURY

General Considerations

All the essential transition metal cations (with the exceptions of Mo and W) belong to the first transition period of the periodic table. The toxic-only metal cations belong to the second and third transition periods. Because these cations are bigger than those of the first transition period, their atomic orbitals fit nicely to those of the sulfur atom (138), leading to higher affinities for sulfur than for nitrogen or oxygen. Interaction with sulfur appears to be the major reason for their toxicity.

Homeostasis for the essential cations of the first transition period involves three components: (i) inducible, highly specific uptake of the respective metal at low environmental concentrations, (ii) nonspecific and fast uptake by systems with broad substrate affinities, and (iii) inducible efflux at high concentrations. The transport-mediated flow equilibrium system for the toxic-only metal cations is simpler than that for the essential metal cations and comprises only nonspecific uptake systems and inducible efflux systems.

Cadmium and Lead

Lead, which is not counted as a transition metal, is discussed with cadmium because the metals have similar sulfur affinities and, thus, toxicities to E. coli (222) and because they are detoxified by the same efflux systems. Pb₃(PO₄)₂ has a solubility product of 10⁻⁵⁴ (342), allowing for the efficient precipitation of lead phosphate, which decreases the availability of lead in bacterial environments. Pb²⁺ is also a much bigger cation (radius, 120 pm) than Cd²⁺ (radius, 97 pm), limiting uptake into the bacterial cell by transport systems for smaller essential metal cations. Both factors may reduce cytoplasmic levels of available lead compared to those of cadmium, making cadmium the more dangerous of these two metal cations.

Uptake, Regulated Efflux, and Binding.

Cadmium is not purely toxic only, since the marine diatom Thalassiosira weissflogii has evolved a carbonic anhydrase that may function with Cd²⁺ instead of Zn²⁺ in zinc-depleted environments such as the open ocean. The enzyme binds Zn²⁺ and Cd²⁺ to its active center with equal degrees of ease (358), and its gene is widespread in marine microorganisms (248). In E. coli or Salmonella, however, no cadmium- or lead-dependent physiological function is known.

The nonspecific uptake of Cd²⁺ interferes with Mn²⁺ uptake (173). The K_m for Cd²⁺ uptake is 1.8 μM, and the K_i for the competitive inhibition of Mn²⁺ uptake by Cd²⁺ is 1.2 μM (173). Definitively, the MntH Mn²⁺ uptake system is responsible for Cd²⁺ transport activity in E. coli (158). The ZIP transporter ZupT may also offer an uptake route for the toxic cation Cd²⁺ (110, 112) (Table 1).

Not much is known about Pb²⁺ uptake in E. coli (Table 1). The iron/lead transporter family (67) (TC 9.A.10) has two members in E. coli (114) (see above), but a role for them in Pb²⁺ uptake has not been observed. There is no dedicated lead uptake system of the iron/lead transporter protein family, like PbrT in C. metallidurans (25), in E. coli.

E. coli cells handle toxic cadmium concentrations through inducible (140) efflux, followed by damage repair (214). The P-type ATPase ZntA is responsible for Pb²⁺ as well as Cd²⁺ (and Zn²⁺) efflux (292). The expression of zntA, half maximal at a 19 μM concentration of Cd²⁺ in the growth medium, is under the control of the ZntR regulatory protein (230). ZntR is a MerR family protein composed of two domains, a C-terminal signal recognition domain and an N-terminal domain that contains the DNA-binding helix-turn-helix motif (32). Like other MerR-type proteins, it distorts the promoter region in response to metal binding, thus allowing the RNA polymerase to bind to a now-optimized promoter conformation (237). ZntR has high (femtomolar) affinity for Zn²⁺ (236); its affinity for Cd²⁺ has not been determined.

The two CDF proteins of E. coli, ZitB and FieF (YiiP), do not contribute or contribute only little to cadmium resistance in this bacterium (79, 109). Thus, MntH and ZntA are the main contributors to the transport-mediated flow equilibrium system for Cd²⁺ in E. coli. Additionally, ZinT (YodA) (260) may be a major cadmium-binding protein of E. coli (61, 317) and used to detoxify cadmium by sequestration. With a size of 25 kDa, however, ZinT is smaller than the previously described 39-kDa unnamed cadmium-binding protein of this bacterium (160).

Toxicity and Damage Repair.

Most of the lead bound to E. coli cells is attached to membrane lipids (170) or lipopolysaccharides (253). Thus, Pb²⁺ may replace Ca²⁺ (107). Moreover, the induction of zraP expression by Pb²⁺ via the two-component regulatory system HydHG indicates that lead can interfere with zinc metabolism (182).

Cadmium affects cells in several ways. It induces single-strand breaks in the DNA (213) and inhibits DNA repair (258, 324). Cd²⁺ also damages proteins (174) and inhibits thioredoxin activity (273) due to its strong affinities for sulfide and thiol groups (see below). Several gene products contribute to cadmium tolerance, because E. coli ΔgshA or ΔgshB (129, 130, 243), ΔdsbA or ΔdsbB (267, 313), ΔrpoE (79), ΔycfR (364), and ΔyqhD (254) mutants (besides the ΔzntA mutants) are more sensitive to cadmium than the wild type.

GshA and GshB synthesize GSH, which is probably involved in protecting the cells from Cd²⁺ by sequestering this cation, especially in the absence of ZntA (129, 130). The role of GSH as a reducing agent is not connected to cadmium tolerance in E. coli (243). The identification of the other products of the genes mentioned above suggests that the cell envelope is a target for cadmium toxicity: RpoE is an extracytoplasmic function sigma factor of E. coli essential for cell wall integrity and defense against many stressors, including transition metal cations (79). The periplasmic protein DsbA is a disulfide catalyst that introduces disulfide bonds into proteins folding within the periplasm. DsbB is required for the reoxidation of DsbA (267, 313). YcfR (BhsA) is a putative outer membrane protein that is synthesized in biofilms and under stress conditions, including acid, heat, and hydrogen peroxide treatments, as well as exposure to cadmium (364). Cadmium affects outer membrane proteins and makes the cells more hydrophobic, thus increasing cell aggregation. Finally, YqhD is part of a GSH-independent, NADPH-dependent mechanism of response to lipid peroxidation (254), suggesting that Cd²⁺ causes lipid peroxidation.

Besides these effects on the cell envelope, most of the damage exerted by Cd²⁺ stems from its strong affinity for sulfur (130, 339). Cadmium seems to sequester sulfide generated during cysteine biosynthesis in assimilatory sulfate reduction and binds to the thiol groups of proteins. Moreover, the sulfide released during the synthesis of iron-sulfur clusters may be sequestered by Cd²⁺, and some clusters may be directly damaged by this cation. Both the sequestering and damaging of clusters disturb cellular iron homeostasis, leading to superfluous free iron and to increased oxidative damage, including lipid peroxidation. Since cysteine starvation leads to a decline in the cellular thiols needed to repair this oxidative damage, a negative feedback cycle that ultimately inhibits cellular energy conservation and thus growth is formed (130).

Silver

Silver is used in medicine to treat wounds and to prevent infections (92). Ag⁺ binds strongly to sulfur and is extremely toxic to microbial cells (222), including those of E. coli (361). Before reaching the cytoplasm, Ag⁺ inhibits respiration after dissociating the components of the respiratory chain (135), probably by acting on the iron-sulfur centers (27). The presence of Ag⁺ causes the cytoplasmic membrane to detach from the cell wall (87) and affects ribosomes (360) and DNA (87). Finally, if the cells are not killed, they are forced into a “viable but not culturable” state (151).

The pathway for Ag⁺ uptake into E. coli is not known, but Ag⁺ has an ionic diameter similar to that of K⁺ (342), so it may use potassium uptake systems. Moreover, AgCl has a low solubility product, and the difference on Pauling's electronegativity scale between Ag⁺ (1.9) and Cl⁻ (3.0) is 1.1, indicating only 26% ionic character for the chemical bond between the two ions (342). Thus, AgCl may be sufficiently neutral to diffuse across the cytoplasmic membrane into the cytoplasm. Indeed, the difference between silver-resistant and -sensitive E. coli strains is more pronounced in growth media with low halide concentrations, while the sensitivities of both types of strains increase with higher halide concentrations (117)

For the cells to maintain viability, Ag⁺ has to be removed from the cytoplasm or, better, prevented from ever reaching the cytoplasmic membrane. Silver resistance in E. coli is enhanced when the uptake of the toxic cation across the outer membrane is limited (189). Cytoplasmic detoxification is performed by the P-type ATPase CopA (321), which detoxifies Cu⁺ and Ag⁺ and is induced by cytoplasmic concentrations of either metal (265). Transport from the periplasm across the outer membrane into the extracellular space is performed by the RND protein-driven transenvelope efflux system CusCFBA (93, 94) (see below and also Chapter 5.4.4.1 by Rensing and Franke). E. coli contains the Cus system for periplasmic detoxification of Cu⁺ and Ag⁺ (Table 1), but interestingly, Salmonella does not.

A sophisticated silver resistance determinant can be found on IncH1 plasmids of Salmonella (118). This sil determinant (116) encodes the SilP P-type ATPase that detoxifies the cytoplasm by the export of Ag⁺ to the periplasm, where Ag⁺ is bound by SilE. The RND protein-driven transenvelope efflux system SilCBA expels Ag⁺ from the periplasm. The periplasmic metal chaperone SilF may channel Ag⁺ to the SilCBA efflux system much like CusF seems to deliver Cu⁺ to CusCBA (13, 94). Finally, the two-component regulatory system of SilS and SilR is employed to control the expression of the sil genes in response to silver exposure (116).

Gold

Gold cations are quite toxic for microorganisms, but the concentration of soluble Au³⁺ in the environment is usually rather low. Therefore, gold toxicity was not expected to be of concern for an enteric bacterium. Escherichia and Salmonella are unlikely to come in contact with elevated gold concentrations in their habitats, the gut and elsewhere. As a consequence, it came as a surprise that a variety of Salmonella strains harbor homeostasis systems dedicated exclusively to gold cations (46, 257). Gold homeostasis mechanisms of Salmonella exhibit extensive similarities to those for copper and silver cations; thus, it is plausible that gold resistance evolved from copper resistance. All the transition metal ions from group IB of the periodic table (Cu, Ag, and Au) occur as monovalent ions in the bacterial cytoplasm. Being soft Lewis acids, they exhibit strong, nonspecific affinity for sulfhydryl groups in peptides and proteins, which contributes to their toxicity.

Similar to those for Ag⁺ and Cu⁺, two major ways for the detoxification of Au⁺ in bacteria are known. First, there are P-type ATPases that transport the metal cation from the cytoplasm to the periplasm. Second, periplasmic Au⁺ can be recognized by tripartite efflux complexes comprising an RND protein (TC 2.A.6), a membrane fusion protein (TC 8.A.1), and a TolC-like outer membrane factor protein (TC 1.B.17) similar to the Cus or Sil system mentioned above. These RND protein-driven efflux complexes are powered by the proton motive force and export their substrates from the periplasm across the outer membrane to the outside (223).

GolT (gold transporter) is a monovalent metal cation translocating P-type ATPase related to CopA and SilP. E. coli CopA is a Cu⁺ and Ag⁺, but not Au⁺, transporter (319), whereas Salmonella CopA may also recognize Au⁺ as a substrate (46). GolB (gold binding) is a cytoplasmic gold chaperone, probably shuttling the metal substrate to the efflux ATPase. Similar chaperones for Cu⁺ in other bacteria have been characterized, and the best-studied example is CopZ from Enterococcus hirae (54). All gol genes are activated strongly by gold cations but only weakly or not at all by other transition metal cations, including Ag⁺ and Cu⁺ (46). The gesABC operon is transcribed divergently from golTS, and it encodes an RND protein-driven efflux complex related to CusCBA or SilCBA but lacks a homolog of the gene encoding the periplasmic copper chaperone CusF (257). Interestingly, the RND proteins GesB and CusA exhibit only relatively low sequence conservation, grouping GesB closer to RND proteins that transport organic substances while CusA is a typical member of the metal-transporting RND protein family (223, 333). It should also be emphasized that Salmonella, in contrast to E. coli, does not possess a Cus system.

Gold homeostasis in Salmonella is governed by two regulators, GolS (gold sensing) and SctR, which is the ortholog of the copper homeostasis regulator CueR from E. coli. All three proteins are members of the MerR family of repressors. GolS shares 42% amino acid sequence identity over 126 residues with CueR (320) and SctR from S. enterica (162). The known targets of GolS are three different operator-binding sites, one upstream of the golTS operon, a second upstream of golB (46), and a third upstream of gesABC (257). SctR regulates the expression of genes encoding the P-type ATPase CopA and the multicopper oxidases CueO and CuiD (83, 320) in response to Cu(I), Ag(I), or Au(I) ions (319).

It remains to be elucidated why some strains of Salmonella harbor efflux proteins dedicated to gold detoxification. Perhaps gol and ges operons were acquired by horizontal gene transfer from bacteria thriving in metal-contaminated sediments or ores, but this idea is highly speculative. Maybe GesCBA functions mainly as a xenobiotic efflux pump (228, 257) that only adventitiously recognizes and transports gold. However, this possibility does not explain the presence of gene products that recognize, bind to, and transport only gold cations: the regulator GolS, the chaperone GolB, and the efflux ATPase GolT.

Mercury

Mercury will be discussed only briefly because mercury toxicity and resistance in E. coli or Salmonella are not much different from those in other bacteria (300). Hg²⁺ binds strongly to sulfur and is thus very toxic. Fortunately for all bacteria, Hg²⁺ can be easily reduced to metallic mercury (E₀′ = +434 mV), which has low vapor pressure, diffuses easily out of the cell, and dilutes in the environment. This reaction is performed by the NADPH-dependent mercury reductase MerA. Since this reaction is possible only in the cytoplasm, the mercury uptake system MerT (or substitute transporters such as MerC or MerF) accepts Hg²⁺ from the periplasmic mercury-binding protein MerP and delivers it to the cytoplasmic reductase MerA (288) (Table 1). Organomercurial compounds can also be detoxified by the organomercurial lyase MerB, which releases Hg²⁺ for further processing by MerA. The expression of the mer determinant, which is usually located on transposons or plasmids, is controlled in a Hg²⁺-dependent manner by the regulatory proteins MerR and MerD (300).

OXYANIONS OF Cr, Mo, W, V, AND As

Overview

Transition metals with one (Sc, Ti, La, Ac, lanthanides, and actinides) or two (Ti, Zr, and Hf) d-orbital electrons exist as tri- or tetravalent cations that readily form insoluble hydroxide complexes not available for living cells at neutral pH values. Metals of the transition group VIb (Cr, Mo, and W), in contrast, may form soluble oxyanions when the respective metal is in the +VI oxidation state. Thus, chromate, molybdate, and tungstate are of some biological importance. The discussion of arsenate is included here because arsenium can be addressed as a metal and arsenate has outstanding environmental importance.

Chromate

Chromate is transported into bacterial cells by nonspecific sulfate uptake systems (227, 234) (Table 1). Many bacteria are able to counteract this nonspecific uptake by employing chromate efflux systems that belong to the chromate transporter protein family (226), but E. coli is not among these bacteria. Since E. coli cannot detoxify cytoplasmic chromate by efflux, chromate may react with reduced cytoplasmic components like GSH to form Cr³⁺ in two- and one-electron reduction steps, causing the generation of detrimental radicals. As a defense, E. coli contains an adventitious NADH-dependent chromate reductase, YieF, a flavoprotein that reduces chromate to Cr³⁺ using two NADH molecules as electron donors (1) but may have another function in the cellular metabolism (1). While three electrons are provided to the chromium atom, the last electron is transferred to another acceptor, such as molecular oxygen, leading to reactive oxygen species that have to be detoxified further on, e.g., by superoxide dismutases. In E. coli, the fate of the Cr³⁺ ion is not known; however, other bacteria such as the metal-resistant C. metallidurans contain as part of their chromate resistance determinant a gene encoding a chromate-inducible superoxide dismutase probably dedicated to this purpose (150). Mutants of E. coli devoid of YieF, the superoxide dismutase SodB, or the catalase KatE are chromate sensitive (1), indicating that antioxidant proteins may be part of the chromate detoxification system as well. So chromate can be detoxified in E. coli by reduction rather than by efflux.

Molybdate

Although Cr³⁺ is an essential metal cation in humans (65, 66), any beneficial effect of chromium in bacteria has not been observed. Molybdenum, on the other hand, is used in many biochemical reactions, such as the fixation of molecular nitrogen and the reduction of nitrate (289). Both processes require a specific but totally different form of a molybdenum cofactor, which is assembled after the uptake of molybdate by an ABC protein family uptake system (Table 1). Since E. coli is not able to fix molecular nitrogen, it contains only molybdenum cofactors of the pteridin group, which are discussed in Chapter 3.6.3.13.

Tungstate and Vanadate

Tungstate is used instead of molybdate in many anaerobic but also some aerobic bacteria, as well as in archaea (289), but in E. coli it is only known as an inhibitor of alkaline phosphatase and a few other enzymes (314). The same is true for vanadate, which has a beneficial biological role in other bacteria but not in E. coli or Salmonella (224).

Arsenate

Arsenate is toxic because it interferes with phosphate metabolism (21). The redox potential of the arsenate-arsenite couple (at pH 7) is 140 mV (342) and therefore within the range for cellular biochemistry. Thus, arsenate can serve as an electron acceptor in anaerobic respiration processes, while arsenite can be oxidized by chemolithoautotrophic bacteria (303). These microbial activities lead to the geological mobilization of arsenium compounds, which may contaminate drinking water (217). In addition to oxidation, the methylation of arsenite can be performed by microbes, leading to a volatile, strong (garlicky)-smelling, and toxic arsenical compound (261).

Arsenate enters E. coli mainly by the PitA phosphate uptake system (Table 1), the fast and nonspecific phosphate uptake system, while the inducible PstABC uptake system discriminates more between arsenate and phosphate (347, 348). Once inside the cell, arsenate is reduced to arsenite and removed by efflux systems (21). Arsenite itself enters E. coli through the GlpF glycerol uptake system (211).

The IncF1 plasmid R773 confers resistance to arsenate upon E. coli (126). Resistance is based on energy-dependent efflux (215). The ars determinant on plasmid R773 contains the genes arsR (encoding a regulatory protein), arsD (encoding a metal chaperone), arsC (encoding an arsenate reductase), arsB (encoding an arsenite efflux pump located in the cytoplasmic membrane), and arsA (encoding an ATPase) (47, 274). Two modes of energy coupling are possible. In the presence of ArsB alone, arsenite efflux is driven chemiosmotically. In the presence of ArsB and ArsA, efflux is energized mainly by ATP hydrolysis (74).

On the chromosome, E. coli contains a second, simpler arsRCB determinant (40, 75), which is more widespread than the high-efficiency arsRDABC determinant in bacterial genomes (284). A strain possessing both ars determinants is able to grow in the presence of 0.4% (24.4 mM) sodium arsenate or 0.05% (3.85 mM) sodium arsenite, while a plasmid-free strain, which contains the chromosomal arsRCB only, is markedly inhibited at 0.1% (6.1 mM) sodium arsenate or 0.0125% (1 mM) sodium arsenite (126). The deletion of arsRCB decreases arsenate resistance a further 5- to 10-fold and arsenite resistance 10- to 20-fold (40).

The 583-amino-acid ArsA ATPase binds to the ArsB integral membrane protein to constitute an efficient arsenite efflux pump (328). ArsA is composed of two halves, A1 and A2, which are connected by a small linker peptide (147). ArsA contains two ATP-binding sites and three binding sites for As(III) or Sb(III) (367). The two ATP-binding sites, which may not function equivalently, are located adjacent to each other at the A1-A2 interface (148). The binding sites for three As(III) or Sb(III) ions are also located at the A1-A2 interface but at a distance of 2 nm away from the ATP-binding sites. The binding of either anion to ArsA stimulates ATPase activity (367). Each ion forms complexes via three ligands, including an amino acid residue from A1 and an amino acid residue from A2; the third ligand is not an amino acid but perhaps a chloride (367) or hydroxide ion.

The specific amino acid residues involved in the binding of As(III) or Sb(III) to the metal-binding site of ArsA may differ. Two cysteine residues constitute site 1, a cysteine and a histidine residue make up site 2, and a histidine and a serine residue form site 3. Potential increases in substrate affinity from site 1 to site 2 and then to site 3 indicate different functions of the three sites. Indeed, the double mutation of the histidine and serine residues of site 3 has little effect on metal binding, while the double mutation of the cysteine and histidine residues of site 2 decreases binding significantly (277). The double mutation of the two cysteine residues from site 1, however, abolishes metal binding and metal-stimulated ATPase activity altogether (278). Cells containing an ArsA mutant derivative carrying the double mutation in site 1 extrude arsenate with higher efficiency than those expressing arsB alone, exhibiting a level of resistance between that of cells with wild-type ArsA and that of cells with ArsB only. This finding shows that the ATPase activity of ArsA alone is able to enhance metal efflux above the level mediated by ArsB alone. Moreover, metal stimulation of the ATPase activity of ArsA increases efflux efficiency even more.

ArsB, a secondary transport system driven by the chemiosmotic force, is able to export arsenite across the cytoplasmic membrane in the absence of ArsA (171). Since ArsB and ArsA transport arsenite (and antimonite) but the ars determinant also mediates resistance to arsenate, arsenate has to be reduced to arsenite prior to efflux. This arrangement makes sense because the structure of arsenate is very similar to that of phosphate. It may cause a problem to bind arsenate efficiently to the substrate-binding site of an efflux system in the presence of high concentrations of intracellular phosphate, which may thus competitively inhibit arsenate efflux. However, there is no molecule in the cytoplasm closely resembling arsenite. So the reduction of arsenate to arsenite, followed by the efflux of the resulting ion, is an efficient means to detoxify arsenate in the presence of a high background concentration of phosphate.

ArsC is responsible for the reduction of arsenate to arsenite. In E. coli, reduction depends on GSH (233) and glutaredoxin (104), with glutaredoxin 2 being the primary electron donor for the reaction (295). The proposed reaction cycle is composed of five steps (21). First, Cys12 of ArsC nucleophilically attacks arsenate, leading to a primary thioarsenate adduct (207). GSH attacks the arsenium atom of the adduct in a second nucleophilic step, yielding an ArsC-Cys12-S-As(V)-S-GSH product. Glutaredoxin reacts with this compound, leading to the reduction of As(V) to As(III) and the release of a GSH-glutaredoxin heterodisulfide. The final two steps rearrange the ArsC-bound As(III) and release As(OH)₃ at the completion of the catalytic cycle (70).

The transcription of the arsRDABC operon leads to a metastable mRNA that is rapidly degraded into more stable arsRDC- and arsA-specific messages (241). In this way, more copies of the soluble ArsC or membrane-associated ArsA protein are synthesized and the cell membrane is not compromised by too many copies of the integral membrane protein ArsB. Transcription initiation is controlled by the ArsR repressor protein (352, 354). Together with metal-dependent regulators from a cyanobacterium and from Staphylococcus aureus, ArsR was the original member of a new family of regulatory proteins (132). ArsR is a dimeric protein (357) that binds As(III) in a tricoordinate thiol complex to three cysteine residues. Two of these, Cys32 and Cys34, are important for As(III) binding and the conformational change that leads to the release of the repressor protein from the operator (297). Cys37 is not essential but nevertheless binds As(III) (296).

Genes for arsD occur only in ars determinants that contain an arsA gene. The small, 13-kDa ArsD protein is involved in the regulation of ars expression (353). It binds to the ars promoter, albeit with lower affinity than ArsR (48), and the binding of arsenite and antimonite to two independent sites (Cys12/Cys13 and Cys112/Cys113) of ArsD releases the protein from DNA (187). The ArsD dimer binds four As(III) ions cooperatively (188). However, this regulatory function seems not to be the main function of ArsD. The protein also serves as a metal chaperone that transports As(III) to ArsA. It thereby increases the affinity of ArsA for arsenite, leading to enhanced arsenite efflux at low cytoplasmic arsenite concentrations (190). For this delivery process, only the binding of As(III) to Cys12, Cys13, and the newly identified Cys18, but not the binding of As(III) to Cys112/Cys113, is important (191, 192).

Thus, E. coli contains a simple arsenate resistance system encoded by the chromosome and a more sophisticated one encoded by a plasmid. Both systems are controlled by ArsR paralogs, reduce arsenate to arsenite by using ArsC, and export the resulting arsenite by using ArsB. The plasmid-encoded system is more efficient, with the sequestration of arsenite by ArsD, the delivery of the metalloid to ArsA, and ATP-dependent export to the outside by the ArsAB protein complex.

COMMON THEMES: INTERACTING ION METABOLISMS IN E. COLI

Shades of common themes seem to emerge from the hazy fog of ion metabolism in E. coli. First, there is transport. High-rate but low-affinity uptake systems provide a variety of cations and anions to the cells. Control of the respective systems seems to be mainly through the regulation of transport activity (flux control), with the control of gene expression playing only a minor role. If these systems do not provide sufficient amounts of a needed ion to the cell, genes for ATP-hydrolyzing, high-affinity, but low-rate uptake systems, e.g., ABC transport systems and P-type ATPases, are induced. On the other hand, if the amount of an ion is in surplus, genes for efflux systems are induced. Thus, by combining different kinds of uptake and efflux systems with regulation at the levels of gene expression and transport activity, the concentration of a single ion in the cytoplasm and the composition of the cellular ion “bouquet” can be rapidly adjusted and carefully controlled. The concentrations of toxic ions are maintained at levels below the toxicity thresholds; those of essential but also toxic transition metal cations are kept at the optimum concentrations between starvation and toxicity levels.

The toxicity threshold of an ion is defined by its ability to produce radicals (for copper, iron, and chromate), to bind to sulfide and thiol groups (for copper, zinc, and all cations of the second and third transition periods), or to interfere with the metabolism of other ions (chromate interferes with sulfate, arsenate interferes with phosphate, and most transition metal cations, with the exception of manganese, interfere with iron).

Iron poses an exceptional metabolic problem due its metabolic importance and the low solubilities of Fe(III) compounds, combined with the ability to cause dangerous Fenton reactions. This dilemma for the cells led to the evolution of sophisticated multichannel iron uptake and storage pathways to prevent the occurrence of unbound iron in the cytoplasm. Nevertheless, other cations may interfere with this tight control and have to be dealt with. Toxic metals like Cd²⁺ bind to thiols and sulfide, preventing the assembly of iron complexes and releasing the metal from iron-sulfur clusters. However, these toxic-only metals can be identified by their high affinities for thiol groups and exported to the extracellular space. In the unique case of mercury, the cation can be reduced to the volatile metallic form. The interference of nickel and cobalt with iron is prevented by the low abundances of these metals in the cytoplasm and their sequestration by metal chaperones in the case of nickel or by B₁₂ and its derivatives in the case of cobalt. The most dangerous metal, copper, catalyzes Fenton-like reactions, binds to thiol groups, and interferes with iron metabolism. E. coli solves this particular problem probably by preventing copper uptake, combined with inducing rapid efflux if the metal happens to enter the cytoplasm.