Chapter 72

Transport of Inorganic Cations

SIMON SILVER

INTRODUCTION

Inorganic cations are required at three levels for the metabolism of both prokaryotes and eukaryotes. First, there is a general, relatively nonspecific need for moderately high intracellular osmolarity (see chapter 77 of this volume). K⁺ is found in the cytoplasm of all organisms, including Escherichia coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium), at approximately 200 mM, a level that varies with the osmotic strength of the extracellular growth medium. K⁺ is generally used for osmoregulation, and Na⁺ is generally found at much lower levels within cells than outside of them. Highly specific membrane transport systems are used for both K⁺ uptake and Na⁺ efflux (3, 4, 5; see below). Second, inorganic cations are needed as free and diffusible cofactors for enzymes. Mg²⁺ is the primary example of such a cofactor, but Mn²⁺ may also have this role with many enzymes and oxyanion (frequently phosphate-containing) enzyme substrates. Highly specific Mg²⁺ transport systems bring in Mg²⁺ and maintain the level of it needed for growth (see below). Third, metalloenzymes contain the inorganic cations as stably bound components of the enzyme complex. Examples of metalloenzymes include iron-proteins and proteins with bound Mn²⁺, Zn²⁺, Cu²⁺, and Ni²⁺ (48). In each case, E. coli and essentially all other bacterial cells have highly specific membrane-protein uptake systems designed to provide low yet stable levels of micronutrient cations, specifically when extracellular levels vary from starvation limitation to toxic excess (Fig. 1). Bacterial ion transport systems were reviewed in a monograph edited by Rosen and Silver (78); more recently, Hughes and Poole (43) considered transport in the more general context of bacterial metal metabolism.

Fig. 1Summary of some known and predicted E. coli membrane cation transport systems. Outer circles represent the outer and inner cell membranes. Small ovals represent individual polypeptide components in some examples studied. The hexagon represents the ferric-iron-containing siderophore.

Source:

POTASSIUM TRANSPORT SYSTEMS

Potassium is the major intracellular cation for all cell types: plants, animals, and microbes. Intracellular K⁺ concentrations are kept in the 0.1 to 0.5 M range (4), even when extracellular concentrations drop to starvation (submicromolar) levels. Maintaining the intracellular K⁺ level requires parallel potassium transport systems: a constitutive low-affinity, high-rate system(s) for times of K⁺ abundance, and an inducible high-affinity, high-specificity, low-rate system(s) for times of K⁺ starvation (Fig. 1 and 2). The number of separate K⁺ transport systems thought to occur in E. coli has recently grown (23). Figure 2 summarizes recent genetic evidence that indicates four separate K⁺ uptake systems and two K⁺ efflux systems.

Fig. 2Potassium uptake and efflux systems of E. coli. The two three-component systems, TrkG and TrkH, and the one-component Kup system are constitutively expressed, while the three-component Kdp system is made under conditions of potassium starvation or osmotic upshock. Trk is regulated by NAD, while Kdp is an ATPase. The proposed membrane-spanning patterns of Kup and the Kdp polypeptides are indicated.

Source:

The high-affinity potassium transporter of E. coli is the Kdp (K⁺-dependent growth) system, which consists of three polypeptides: KdpB (a membrane protein ATPase), KdpA (a very hydrophobic protein that determines the affinity of the system for K⁺), and KdpC (a smaller membrane protein of unclear function) (Fig. 2). Expression of the high-affinity Kdp system (less than 2 μM K⁺ transport constant) is induced under conditions of potassium starvation, by means of a decrease in membrane turgor (i.e., osmotic upshock).

Although all three components of the Kdp ATPase are membrane proteins (Fig. 2), their roles are separate. KdpA appears to provide the initial recognition of K⁺ (since "K_m mutations" almost always occur in kdpA [19]). A detailed analysis of 15 different kdpA mutants with reduced K⁺ affinities for transport showed essentially full retention of V _max for K⁺ uptake and a reduced discrimination between K⁺ and Rb⁺ that is consistent with a primary role for KdpA in initial K⁺ binding. The mutational pattern agrees with the predicted topology of the KdpA polypeptide in the cell membrane that was determined on the basis of protein fusions to alkaline phosphatase and β-galactosidase and led Buurman et al. (19) to conclude that there are two sites for K⁺ binding on KdpA: an initial high-affinity site on the periplasmic surface of the membrane and an intracellular site from which K⁺ is released into the cytoplasm. Both binding sites contribute to Kdp discrimination between K⁺ and Rb⁺ (19). The reduced-affinity mutations occur in four regions of the kdpA gene, which correspond to three (of the four) periplasmic loops and one large cytoplasmic loop (19). The altered-K_m mutations do not affect the residues found within the membrane.

KdpA determines K⁺ association for transport; KdpB is the membrane-embedded ATPase subunit, homologous in sequence and function to the more familiar animal Na⁺/K⁺ and sarcoplasmic reticulum Ca²⁺ ATPases (27, 28, 84). KdpB is a typical P-type ATPase (Fig. 3), so called because a phospho-protein intermediate can be isolated in vitro (unlike other transport ATPases that lack phosphorylated intermediates) (28, 90). Phosphorylation, where measured, is found on a conserved aspartate residue (within a highly conserved heptapeptide sequence, Asp-307–Lys–Thr–Gly–Thr–Ile–Thr, shown in Fig. 3). Forty-two positions upstream from the aspartate is an invariant proline residue, Pro-265, which for divalent-cation ATPases (see below) is frequently flanked by vicinal cysteines, though not in KdpB. This proline is found in the fourth of six proposed transmembrane α-helical regions which are common to all P-type ATPases (see, e.g., references 90 and 91). The aspartate phosphorylation site of the ATPase is part of a large cytoplasmic "kinase" domain (about 350 amino acids in length), which also includes the highly conserved Gly-Asp-Gly-Thr-Asn-Asp-Ala-Pro octapeptide sequence some 204 amino acids down from the phosphorylation site (Fig. 3) that is involved in ATP binding. This ATP-binding region is in fact the most conserved region in the sequences shown in Fig. 3 and in other P-type ATPases in bacteria and humans. Upstream from the invariant proline by 100 positions and in a separate cytoplasmic domain is the Thr-159–Gly–Glu–Ser–Ala–Pro hexapeptide that is thought to mark a "phosphatase" motif common to P-type ATPases (91) (Fig. 3). In the transport-coupling cycle of KdpB and other related ATPases, the phosphorylation site is "autokinased" from [γ-³²P]ATP subsequent to cation binding, and the phosphate is later removed by the phosphatase (after K⁺ transport) in order to recycle the protein to a high-affinity form. Because of the close homology between the known bacterial P-type ATPases and the more thoroughly studied mammalian Na⁺/K⁺ and Ca²⁺ enzymes (see, e.g., reference 54), detailed structural (91) and kinetic models can be made for the E. coli KdpB K⁺ enzyme in the absence of direct experimental evidence. In other P-type ATPases, the cation transport channel is considered to be part of the ATPase polypeptide that is homologous to KdpB, but the Kdp system has KdpA, which is believed to be a separate high-affinity cation-binding membrane transport polypeptide (19, 28, 84).

Fig. 3Alignment of KdpB and other E. coli and S. typhimurium P-type ATPases, showing overall sequences starting with (i) (sometimes) metal-binding motifs and then (ii) phosphatase regions, (iii) the Cys-Pro-Cys transmembrane presumed cation channel region, (iv) the phosphoaspartate site, (v) sites crucial for ATP binding, and (vi) the most conserved region "hinge" ending the kinase domain. Staph, S. aureus.

Source:

The sequences of additional P-type ATPases of E. coli and S. typhimurium that are considered separately below are aligned in Fig. 3. Figure 4 (see below) is an evolutionary tree of these sequences that was created by using the algorithm of Feng and Doolittle (30). KdpB of E. coli, the first P-type ATPase of a prokaryote that was sequenced, remains quite different from the more than 20 such sequences from other prokaryotes with different cation transport specificities (see below). Although KdpB homologs have been detected in other bacterial species by immunoblotting (84) and DNA-DNA Southern blotting analysis, no other potassium P-type ATPase sequences are currently available in GenBank.

Fig. 4Magnesium uptake and efflux systems of S. typhimurium. The four-component Cor system is constitutively synthesized and functions chemiosmotically for both uptake and efflux. The one-component MgtA and MgtB systems are synthesized under conditions of magnesium starvation and are P-type ATPases.

Source:

The kdpD and kdpE gene products regulate Kdp function. KdpD and KdpE are members of the two-component (sensor kinase plus DNA-binding responder) family of transcriptional regulators (103, 110). KdpD is anchored in the membrane and senses the osmotic pressure across the membrane (103; see also chapter 71 of this volume). First, KdpD probably autophosphorylates itself (at an invariant histidine residue, His-673); then KdpE is thought to be transphosphorylated (at an invariant aspartate residue, Asp-52) by KdpD. The phosphorylated KdpE is expected to bind to the DNA operator region, stimulating transcription of kdpABC. In addition to transcriptional regulation of kdpABC operon expression, the physiological functioning of the Kdp ATPase (and also of the constitutively synthesized TrkH and TrkG systems [see below]) is regulated at the physiological level such that at high turgor pressure, the Trk system has a V _max for K⁺ activity that is only 1 to 2% of that of K⁺-depleted cells (5), thus avoiding the futile use of energy for K⁺-K⁺ exchange.

In E. coli, the low-affinity high-rate systems are called Trk (for transport of K⁺) (Fig. 2). The Trk systems are expressed constitutively and not transcriptionally regulated in response to external potassium levels. The four genes trkA, trkE, trkG, and trkH are not linked on the E. coli chromosome, and their products appear to form two separate Trk systems with different inner membrane proteins, TrkG and TrkH. The trkG gene entered E. coli on the "rac prophage" and has the low value of 37% G+C (in trkG) compared with the normal (for E. coli) value of 51% G+C for trkH (5, 23). Therefore, the TrkG system is considered "extra" and not one of the usual series of pathways for K⁺ uptake. The TrkH sequence was recently corrected, and TrkH and TrkG are now known to be 41% identical at the amino acid level (83). TrkH and TrkG contain 483 and 485 amino acids, respectively, and both are thought to be almost entirely membrane embedded (Fig. 2), forming 10 transmembrane α-helical regions (5). The Trk systems share the membrane-associated TrkA and TrkE proteins (23).

TrkA is a peripheral-membrane protein that is anchored to the cytoplasmic membrane by TrkG or TrkH. Under conditions of overexpression, TrkA is found in the cytoplasm and by itself does not function in K⁺ transport. The TrkA sequence shows a protein of 458 amino acids that consists of homologous half sequences, each of which contains a recognizable nucleotide-binding motif (82). The TrkA sequence is closely homologous to NAD⁺-dependent dehydrogenase sequences for entire domains, from residue 2 to 233 and from 234 to 409. The purified TrkA protein binds NAD⁺ and NADH but not ATP (82). TrkE, the third protein (of unknown function) that is required for the TrkH system, is needed for activity of the TrkH system or for full activity of the TrkG system (23). The S. typhimurium trkA gene was isolated and sequenced by Parra-Lopez et al. (68) as a "virulence gene" that affects the resistance of the bacterial cell to mammalian-cell-produced cationic peptides. The E. coli and Salmonella TrkA polypeptides differ in only 3 of 458 amino acids, but the Salmonella TrkA may regulate the activity of a multicomponent membrane ATPase that directly provides resistance to the toxic peptides rather than acting as a component of a K⁺ transport system (Fig. 2). However, mutations in the genes governing the other components of the Salmonella Trk transport system, TrkH and TrkE, also result in the loss of pathogenicity (68). A trkA homolog (33% predicted amino acid identities) was identified in the methanogenic archaeobacterium Methanosarcina mazei (56), so Trk system K⁺ transporters may occur widely in nature. Nevertheless, how the Trk potassium transport system fulfills different roles in transport and virulence is unclear. Parra-Lopez et al. (68) also identified cytoplasmic regions of homology between TrkA and the KefC K⁺ efflux carrier (see below), as well as a shorter region of homology between TrkA and Kch1, the proposed K⁺ channel protein of Milkman (57) (see below). Whether these significant sequence homologies underlie related functions for these polypeptide regions remains to be seen.

The Trk systems are energized by the membrane potential ΔμΨ and are regulated by ATP (5). The Trk systems do not seem to couple ATP hydrolysis directly as the energy source driving K⁺ transport. It is possible that the ATP requirement is secondary via NAD binding to the TrkA protein (82) (Fig. 2). Alternatively, ATP may act by binding to the TrkE polypeptide. Unlike the K⁺-specific Kdp ATPase, the TrkH and TrkG systems recognize Rb⁺ as well as K⁺; the K_ms for both cations are about 1 mM, but the V _max for K⁺ is about 15 times that for Rb⁺ for the TrkG system and 100 times that for Rb⁺ for the TrkH system (83).

A third, minor, constitutively synthesized K⁺ uptake system is called Kup (previously called TrkD). The Kup system has a relatively low rate and a low alkaline earth ion specificity. Kup is the sole saturable pathway in E. coli for the uptake of Cs⁺ (13, 23). The DNA sequence of the kup determinant (5, 81) shows a single open reading frame capable of encoding a large, 622-amino-acid membrane protein with a highly hydrophobic amino-terminal 440-amino-acid domain that is predicted to run across the membrane 12 times (as frequently is the case for chemiosmotic membrane transporters). However, this region of Kup is not homologous to other known membrane protein sequences. The Kup sequence shows no sign of an ATP-binding motif, so the Kup system is probably chemiosmotic (5, 81). The C-terminal region of Kup contains a domain of about 180 amino acids that are hydrophilic and predicted to be intracellular. Deletion of this domain results in reduction but not total loss of Kup activity (81). TrkG, TrkH, and Kup are considered to be three distinct K⁺ uptake pathways, although they were earlier thought to be a single system. Even when all four primary K⁺ transport systems (TrkG, TrkH, Kup, and Kdp) have been removed by mutational deletions, residual K⁺ transport activity remains. Unsaturable at reasonable external K⁺ concentrations, this (presumedly additional low-affinity) pathway shows a rate about 1/250 that of Trk at 30 mM external K⁺ (83).

In addition to K⁺ uptake pathways, there are two well-studied transport pathways for K⁺ efflux by the KefB and KefC proteins (formerly called TrkB and TrkC) (6, 31, 58) (Fig. 2). Efflux by KefB and KefC is negatively regulated by glutathione and activated by adducts of glutathione (26) (Fig. 2). The kefC gene encodes a protein of 620 amino acids with a highly hydrophobic amino-terminal 380 amino acids and a hydrophilic carboxyl-terminal 240 amino acids (58). Deletion of the carboxyl-terminal region, including a short probable NAD-binding domain (positions 401 to 432, which is closely homologous to the proposed NAD-binding site of TrkA), results in inactivation of KefC activity. KefC also has a hydrophilic region homologous to glutathione-binding protein sequences (58). Overall the KefB protein is closely homologous to KefC, having a similar predicted secondary structure and 31% identical amino acid residues (I. R. Booth, personal communication). Whether KefB and KefC function as K⁺/H⁺ antiporters (as earlier thought) or by a glutathione-gated "channel" mechanism that allows rapid one-directional cation flux is currently being tested. Studies with mutants have led to the suggestion that KefC may occur in the membrane as an oligomeric aggregate (24). The Kef systems seem less related to potassium metabolism than to survival of "stress," perhaps resulting from damage caused by electrophilic sulfhydryl reagents such as N-ethylmaleimide (31). In addition to the already-listed systems shown in Fig. 2, two additional potassium efflux systems, being called KefA (involved in turgor-regulated K⁺ efflux and required for growth at high osmolarity in the presence of high K⁺) and KefD (involved in K⁺ efflux at high transient internal pH values, perhaps by an H⁺ _in/K⁺ _out exchange process, and required for growth at high pH), apparently exist (I. R. Booth et al., personal communication). Mutations in both the kefA and the kefD genes have been isolated. kefA mutants are growth inhibited at high osmolarity in the presence of high K⁺ but not Na⁺. kefB mutants of a strain also lacking Na⁺/H⁺ antiporter activity (see below) cannot grow at high pH. The kefA gene is currently being sequenced (Booth, personal communication).

Recently, Milkman (57) sequenced the DNA that encodes still another possible potassium channel, Kch, which shows sequence homology to the K⁺ channel proteins of plants and animals. The term "channel protein" implies a protein that opens a pore, allowing relatively rapid and nonspecific passage of a group of cations in a single event, unlike the "transporters" considered elsewhere in this chapter, which are considered to transfer one cation at a time across the membrane. This distinction is particularly important for a cell as small as E. coli, in which a single open channel might deplete the cell of potassium; in larger eukaryotic cells, a transiently opened channel has a smaller effect. The region of homology between the Kch sequence and sequences of channel proteins of eukaryotes includes the six putative membrane-spanning regions plus the predicted 21 amino acids involved in the K⁺-selective pore site (57). The finding of this possible channel determinant raises the question of its purpose. Together with results measuring stretch-activated K⁺ channel activity in E. coli membranes (9, 80, 104), the existence of this gene that may encode a K⁺ channel suggests a possible role for the gene in transiently protecting the E. coli cell from bursting upon exposure to hypo-osmotic shock. To fulfill this role, the Kch system needs to be made in only one or a few copies per cell and to function only rarely, when E. coli is released from the gut into dilute waters, or not at all, when the cell is growing on either hyper- or hypo-osmotic medium or when a high or low level of K⁺ is available.

MAGNESIUM TRANSPORT SYSTEMS

After potassium, magnesium is the next most abundant intracellular inorganic cation. Mg²⁺ has roles quite different from those of K⁺: Mg²⁺ functions primarily as a catalytic cofactor for many enzymes. Unlike K⁺, most intracellular Mg²⁺ is not osmotically free but is bound (in kinetically active complexes) with cellular polyanions such as nucleic acids and lipids (89). Thus, Mg²⁺ does not contribute to osmotic processes. There are three Mg²⁺ transport systems in S. typhimurium, the best-studied microbe for magnesium transport (Fig. 4). Whereas S. typhimurium with any one of these systems will grow on micromolar Mg²⁺ concentrations, cells defective in all three systems require 100 mM Mg²⁺ for growth. Two of these systems (mgtA and mgtB) are inducibly regulated and expressed only under conditions of magnesium starvation (99, 106).

The MgtA and MgtB proteins are members of the P-type class of ATPases (99, 106; Fig. 5), as described above for KdpB. MgtA and MgtB are, however, slightly longer than 900 amino acids (Fig. 3), which is much longer than KdpB, with most of the extra length occurring at the carboxyl end (106). In length and sequence homology, the Mg²⁺ ATPases are more similar to the Ca²⁺ ATPases of bacteria and eukaryotes than to the bacterial P-type ATPases for other cation substrates (91). Similarly, the two known calcium P-type ATPases in prokaryotes (one of which is listed in Fig. 5) are closer in sequence homology to the Mg²⁺ ATPases than to other P-type ATPases from bacteria. The MgtB ATPase is 49% identical at the amino acid level to the MgtA ATPase and also may consist of a single polypeptide. However, in the same operon as mgtB is a second gene, mgtC, that encodes a 22.5-kDa polypeptide of unknown function but that is not required for Mg²⁺ transport (99). Of all bacterial cation-translocating P-type ATPases, only MgtB has had its membrane topology carefully delineated by gene fusions with periplasmic functional and cytoplasmic functional reporter genes (95). The structure contains six transmembrane segments placed similarly to those of KdpB (Fig. 2; see also text above) plus four transmembrane segments in the carboxyl-terminal 140 amino acids, whose predicted positions are similar to those in P-type ATPases of eukaryotic cell membranes but are missing in most shorter P-type ATPase sequences from prokaryotes.

Fig. 5Tree of alignments of the seven known E. coli and S. typhimurium P-type ATPases: KdpB (E. coli); MgtA and MgtB (S. typhimurium); MgtA (E. coli ECOUW92); and the three new E. coli sequences of unknown cation specificity, ORF732 (found in GenBank accession number U00039; ECOUW76) and the His-1 and His-2 (histidine-rich metal-binding motif) ATPases (GenBank accession numbers U16658 and U16659). In addition, three non-E. coli P-type ATPases (from H. pylori [possibly Zn2+ or Cu2+ specificity; Clancy, personal communication], S. aureus CadA [Cd2+ specificity; 61], and Synechococcus PacL [Ca2+ specificity; 47]) are shown to place the sequences in context.

Source:

Recently, part of the E. coli genome sequence, covering min 92.8 to 0.1 on the traditional map, was released. In this sequence (accession numbers U14003 and ECOUW93), there is an open reading frame (after correcting for two frameshifts easily seen by comparison) that is 80% identical to the S. typhimurium mgtA gene and that potentially encodes a protein 898 amino acids long that is 90% identical in sequence to S. typhimurium MgtA. It is listed in Fig. 5 as MgtA of E. coli.

Figure 3 shows detailed sequence comparisons of overall length and specific functional motifs for four E. coli and S. typhimurium P-type ATPases (KdpB, MgtA, MgtB, and the new ORF732 [of unknown cation specificity, but possibly copper translocating; 100] together with two additional ATPase sequences, one from Helicobacter pylori CopA [J. Clancy, personal communication] and one from Staphylococcus aureus CadA) in order to place these sequences in context. Unlike Kdp, which consists of three polypeptides (Fig. 2), the other systems all appear to consist of a single known polypeptide. Whether additional polypeptides are involved is not currently clear.

The third S. typhimurium Mg²⁺ transport system is called Cor (for cobalt resistance, since Co²⁺ is an alternative substrate for this system) (37, 94). The Cor system is the major Mg²⁺ transport system (with the highest V _max) in other gram-negative bacteria under most growth conditions (94) (Fig. 4) and can function in either direction both for uptake and for efflux by cation-cation exchange. Cor is expressed constitutively in S. typhimurium. Ni²⁺ is also a substrate for the Cor system. Cor consists of four polypeptides, which are determined by the corA, corB, corC, and corD genes. CorA appears to be the primary inner membrane protein responsible for uptake via the Cor system. Mutations in corA result in loss of Cor-specific cation uptake; mutations in corB, corC, or corD do not affect the K_m or the V _max for Mg²⁺, Ni²⁺, or Co²⁺ uptake but result in lowered Mg²⁺ efflux, so the triple mutant CorB^– CorC^– CorD^– totally lacks Mg²⁺ efflux. Normally, Mg²⁺ efflux does not occur at low extracellular levels such as 100 μM Mg²⁺ but does occur at higher extracellular levels, perhaps by cation-cation exchange (37).

Elegant gene fusion studies have elaborated the membrane topology of CorA, indicating that this 316-amino-acid polypeptide, whose sequences have been determined for both E. coli and S. typhimurium (96), starts with a large periplasmic N-terminal domain 231 amino acids in length that is followed by three closely spaced hydrophobic (presumedly membrane α-helical) regions, which take up most of the remaining 85 amino acids (Fig. 4). β-Lactamase fusions to positions in the first 231 amino acids retain full enzyme activity, indicating positions in the periplasm (96). β-Galactosidase fusions only in the short (presumedly intracellular) regions between the first and second transmembrane span and after the third transmembrane span allow growth on lactose (96), providing the basis for the model of CorA. The predicted periplasmic domain contains an unusual amount (about 30%) of charged amino acids, with both positively charged (Arg and Lys; 12%) and negatively charged (Glu and Asp; 17%) residues abundant. The short intracellular loops contain an abundance (5 of 10) of Arg or Lys residues, which presumably serve to insert the protein into the membrane following the "positive inside rule." In addition to the E. coli and S. typhimurium CorA proteins that are predicted to be 96% identical in sequence, CorA homologs are in the protein sequence banks from Bacillus subtilis (SwissProt accession no. P40948) and Mycobacterium leprae (part of a 39-kbp cosmid sequence in GenBank protein accession no. U15180).

IRON SIDEROPHORE TRANSPORT SYSTEMS

Iron is the best known and perhaps the most important inorganic cation micronutrient. Essentially all bacterial cells contain Fe²⁺ (incorporated in heme-containing proteins such as cytochromes and in nonheme iron proteins). The only known exception is lactic acid bacteria (1), which do not contain heme proteins and substitute Mn²⁺ in some intracellular roles more commonly played by Fe²⁺. In E. coli, six Fe³⁺ transport systems can function in aerobic conditions (14), and another, for Fe²⁺ transport (feoAB), functions under anaerobic conditions (when Fe²⁺ is more stable) (45). These systems are considered in depth in chapter 71 and are only briefly described here. Also see chapter 71 for details and structures of the iron transport chelates, which are called siderophores (from the Greek words for iron bearing). Fe³⁺ transport systems are named for the siderophores involved. The five siderophore systems for E. coli are (i) the enterochelin system, which uses a trimeric catechol synthesized by E. coli; (ii) the aerobactin system, which uses a hydroxamic acid chelate consisting of two modified lysine residues coupled to a citric acid residue and is encoded by some ColV plasmids in E. coli; (iii) the chromosomally encoded citrate-Fe³⁺ transport system, which depends on exogenously (i.e., environmentally) available citrate; (iv) the ferrichrome uptake (Fhu) system, which takes up Fe³⁺ chelated with hydroxamate siderophores synthesized by fungal cells; and (v) the ferrioxamine B hydroxamate system, which transports a different range of bacterial hydroxamate-Fe³⁺ complexes synthesized by nonenteric bacteria. These five systems have separate outer membrane receptors but share a dependency on TonB (and ExbB and ExbD) for functioning of the outer membrane receptor proteins (see the example of the iron-citrate system below; Fig. 6). In addition, there is a low-affinity system that functions at high concentrations of Fe³⁺, apparently without needing a siderophore transport cofactor. The genes governing aerobactin uptake may be on the chromosome in some organisms, such as Klebsiella (Aerobacter) aerogenes, in which aerobactin was first found, or on a plasmid, such as the genes for the aerobactin system that functions as a virulence factor on the ColV-class plasmids in enteric bacteria. Once accumulated within the cell, the iron cations are either incorporated into iron-specific proteins (both heme- and non-heme-containing) or stored in either of two E. coli iron storage proteins, ferritin and bacterioferritin (42).

Fig. 6Structure and regulation of the E. coli citrate-iron transport system (modified and updated from Silver and Walderhaug [93]). FecR is the cytoplasmic membrane iron-citrate sensor, FecI is the positive activator (component of the RNA polymerase), and Fur is a negatively acting transcriptional repressor protein.

Source:

Each of the five pathways for Fe³⁺ uptake in E. coli (and the comparable systems in other bacteria) involves a series of gene products for (i) regulation, (ii) transport, and (iii) energy coupling. I describe the iron dicitrate system (Fig. 6) in detail as just one example. Three regulatory proteins are involved. FecR and FecI couple in order to sense the presence of ferric citrate bound to the outer membrane FecA protein and to transmit and to positively activate the DNA operator-promoter region of the fecABCDE operon (89). We are beginning to understand the molecular mechanism of this signal transduction. Extracellular ferric citrate binds to FecA, and by a series of conformational changes (involving the FecA, TonB, ExbB, ExbD, FecR, and FecI proteins), transcription is activated. The energy-dependent conformation changes in the first four of these proteins may be the same as those involved in iron uptake from citrate (Fig. 6).

FecR is a cytoplasmic membrane protein (as shown in Fig. 6) that when activated by TonB undergoes a conformational change (111) that is in turn transmitted to the cytoplasmic protein FecI (15), which is a positively acting σ ⁷⁰-type RNA polymerase subunit. FecI in turn binds to the promoter region upstream from the fecABCDE operon (73, 101; Fig. 6). Only the cytoplasmic first 59 (of 317) amino acid residues of FecR are required for fecABCDE transcription (62). Proteolytic cleavage patterns of mutant constitutive forms of FecR indicate that conformational changes occur in the mutant constitutive forms of the protein (111).

In addition, the absence of intracellular Fe²⁺ (sensed by the global iron regulatory protein Fur [2, 15, 22], which functions as a repressor of fec, as it also does for the alternative iron transport systems) is needed for transcription of the fec genes. The fec operon encodes five proteins. The first, FecA, is a large outer membrane protein (Fig. 6) and a member of the TonB class of energy-dependent outer membrane proteins (38, 40). The inner membrane protein TonB stretches across the periplasmic space to make physical contact with a conserved "TonB box" on the FecA protein and to transmit the energized state of the inner membrane to open the outer membrane channel for iron dicitrate. There is no evidence that energy expenditure is required directly for transport of iron dicitrate through FecA, in contrast to a role for energy in "gating" the FecA channel. With the ferrichrome system equivalent, FhuA, the deletion of a 34-amino-acid supposedly outer-surface loop results in conversion of this TonB-dependent energy-gated channel to an open nonspecific outer membrane diffusion channel (51). The TonB protein is stabilized by and is dependent on still other proteins, ExbB and ExbD, as with other members of the TonB family (62, 92).

Once in the periplasmic space, the iron dicitrate associates with the Fe³⁺ dicitrate-binding protein FecB (Fig. 6). Citrate is removed and released extracellularly. Fe³⁺ moves into the inner membrane proteins FecC and FecD, which form the Fe³⁺ transport channel. In this aspect, the Fe³⁺-citrate system differs from most siderophore-dependent transport pathways with which the Fe³⁺ enters the cell along with the siderophore. The inner membrane proteins FecC and FecD are homologous in sequence (101) and presumably have functional regions similar to those of other TonB-dependent systems. These systems all have inner membrane protein pairs and a membrane-associated ATP-binding protein (FecE in this case) that couples ATP hydrolysis to the transport of Fe³⁺ into the cell. Intracellular Fe³⁺ is rapidly reduced to Fe2+ and incorporated either into heme groups (by the enzyme ferrochelatase) or into nonheme iron proteins. If one considers the example of iron dicitrate transport in E. coli and how much is known about the details of its molecular mechanisms (Fig. 6), one gains respect for the exactness of microbial ion transport mechanisms and their regulation. We expect to find similar preciseness for the other iron-siderophore transport systems in E. coli and other bacteria and also for non-iron cation transport systems.

TRACE NUTRIENT TRANSPORT: NICKEL, MANGANESE, ZINC, AND COPPER

Nickel

Nickel is required by many microorganisms for enzymatic activities (41). E. coli, in particular, has several nickel-containing membrane-bound hydrogenases that are required for anaerobic growth (113). The discovery of a highly specific nickel transport system in E. coli (112, 113) established nickel uptake by a high-affinity transporter in addition to transport by low affinity through the nonspecific magnesium carrier, Cor. The Nik nickel transport system is specific for Ni²⁺ and is determined by an operon of five genes (59). The deduced amino acid sequences of components of the Nik transport system are highly homologous to those of the periplasmic binding protein systems for oligo- and dipeptides and very likely constitute an ATP-binding cassette transport ATPase (Fig. 7 [59]). Other bacteria use single-polypeptide high-affinity nickel transporters (25) rather than multicomponent ATPases. The deduced gene products include NikA (524 amino acids long; probably a periplasmic nickel-binding protein), NikB and NikC (314 and 277 amino acids long, respectively; both are predicted to be integral membrane proteins, each has six membrane-spanning regions, and the two together form the transport apparatus), and NikD and NikE (253 and 268 amino acids long, respectively; both contain canonical ATP-binding motifs and are assumed to provide energy and/or regulation for the transport system; Fig. 7). The Nik transport system, along with an additional protein required for insertion of nickel into nickel-containing enzymes, is required for metabolism of this essential trace element (113). Nik^– mutant cells lack ⁶³Ni²⁺ uptake activity (59) and hydrogenase activity and have 10% or less of normal cellular nickel levels (59). Hydrogenase activity can be restored by adding high (0.5 mM) Ni²⁺ to the growth medium (112). The Nik system is regulated by the product of the anaerobic regulatory gene fnr, which is consistent with the initial discovery of nik mutants during a study of hydrogenase genes (112, 113). In addition to being transported by the nickel-specific Nik system, Ni²⁺ is transported into E. coli and S. typhimurium cells by the low-affinity primarily Mg²⁺ carrier Cor (97, 113).

Fig. 7Ni2+ uptake ATP-binding cassette traffic ATPase of E. coli.

Source:

Manganese

In contrast to that for nickel, the energy source for the highly specific E. coli Mn²⁺ transporter (87, 88) is the chemiosmotic membrane potential (10). The evidence for chemiosmotic transport is that this highly specific Mn²⁺ transporter functions in right-side-out membrane vesicles that are energized by the membrane potential but lack ATP. Mn²⁺ transport in membrane vesicles has been studied with E. coli (10), Bacillus subtilis (11), and S. aureus (69), indicating generality. Nevertheless, the presumed single gene for this chemiosmotic, potential-driven, electrogenic uniporter has not been identified. Tight regulation of the synthesis and maintenance of the Mn²⁺ transport in response to manganese starvation or abundance has been demonstrated in B. subtilis (32) but not in E. coli. My colleagues and I earlier summarized the literature on bacterial manganese transport (86, 89), and no additional work on E. coli Mn²⁺ transport (and little work with other bacteria) has been reported in many years.

Zinc

Highly specific zinc transport has not been satisfactorily demonstrated in E. coli or any other prokaryote (see reference 85 for earlier literature). However, the existence of zinc metalloenzymes (48) makes it likely that high-affinity zinc transport systems do exist and are expressed during zinc starvation. It is possible (but not tested) that the ORF732 P-type ATPase of the E. coli chromosome (Fig. 3 and 5) (100) will be the first example of a Zn²⁺-specific transporter. The only reason to predict this specificity is the closer sequence homology of this ATPase to the Cd²⁺ ATPase of staphylococci (Fig. 5) and to the H. pylori enzyme, for which an effect on cellular Zn²⁺ content has been found (Clancy, personal communication). Under conditions of adequate zinc, low-affinity magnesium carriers can also take up Zn²⁺ (60).

Copper

More emphasis has been placed on understanding copper transport (summarized by Brown et al. [17]) than on understanding that of zinc. Nevertheless, there is little definitive information concerning copper uptake, efflux, and intracellular regulation in E. coli. Rouch (79) reported preliminary measurements of transport of ⁶⁴Cu²⁺ by E. coli and obtained evidence for both energy-dependent uptake and efflux. These results have been repeatedly summarized (most recently in reference 17), but unfortunately, the primary data have not been published, repeated, or confirmed elsewhere. Mutations that lead to copper sensitivity were obtained at six loci, called cutA through cutF (79), and these were assumed to be the determinants of proteins involved in copper uptake, intracellular sequestration (intracellular copper binding is presumed to be important, since copper is toxic, and the level of free intracellular Cu²⁺ level is probably quite low), and efflux (to maintain a homeostatic control over cell levels). From the phenotypes of the mutant strains, preliminary hypotheses were made for each gene function (17, 79). CutA and CutB were thought to be components of a copper uptake system(s), CutC and CutD were considered components of a copper efflux system, and CutE and CutF (for which the mutant strains are not only sensitive to high copper levels but also dependent on added copper for growth) were thought to control intracellular copper-binding proteins.

The genes corresponding to two of these mutations, cutE (75) and cutA (33), were cloned and sequenced. Recently, two additional genes, cutC and cutF, were also identified and sequenced (39a). Unfortunately, there are uncertainties about whether the cut gene products have direct effects on copper transport or affect copper sensitivity secondarily by determining cell surface components. Secondary effects on copper sensitivity resulting from mutant surface components are quite reasonable since, for example, the respiratory chain cytochrome oxidase contains two well-defined copper centers, one within the membrane and the other at the periplasmic surface (43a).

The mutant cutE cells do, however, overaccumulate radioactive ⁶⁴Cu²⁺, and the overaccumulation is corrected by a plasmid containing cutE (75). CutE is a 512-amino-acid apparently soluble protein with a single potential copper-binding motif (His-Phe-Gln-Met-Ala-Arg-Met) that is similar to metal-binding motifs found in other proteins (75). However, Gupta et al. (39) reported that the product of the cutE gene in S. typhimurium is an apolipoprotein N-acyl-transferase that is located on the periplasmic surface of the inner membrane. Failure to acylate lipoprotein in the outer membrane might increase copper sensitivity by altering outer membrane permeability. The other gene, cutA, was independently isolated and sequenced by Fong et al. (33) by complementing the cutA mutation and by Crooke and Cole (21), who sought genes affecting anaerobic nitrite reductase activity. The primary function of CutA remains unclear. CutA is predicted to be a large 62-kDa membrane protein with potentially eight membrane-spanning α-helices and a large, hydrophilic, carboxyl-terminal region. On the basis of sequence homologies, this region has been predicted by Crooke and Cole (21) to function as a protein disulfide isomerase. If this hypothesis is correct, then CutA might be involved in correct folding of either a Cu²⁺ transporter or a protein indirectly affecting copper permeability.

In contrast to the unsettled situation with regard to E. coli cut genes, the products of a number of recently sequenced genes in nonenteric bacteria are thought to be Cu²⁺ transport ATPases oriented either inwardly (for uptake) or outwardly (for efflux). These products include the CopA (uptake) and CopB (efflux) ATPases of the gram-positive microbe Enterococcus hirae (63, 64), for which the genes appear to lie in a single operon, suggesting that they are carefully regulated. In the cyanobacterium Synechococcus sp. strain PCC7942, the genes for uptake (ctaA [70]) and efflux (pacS [46]) are unlinked. Remarkably, direct protein Western blot (immunoblot) analysis showed that the cyanobacterial efflux ATPase is located in the photosynthetic thylakoid membrane, not the cytoplasmic membrane (47). Continuing this litany of copper-translocating ATPases are two quite dissimilar sequences for Cu²⁺ ATPases from H. pylori, the presumed causative agent of duodenal and gastric ulcers, whose sequences were determined by Ge et al. (35) and Clancy (personal communication). The Ge et al. (35) H. pylori sequence differs importantly from the CtaA, CopA, and PacS sequences (each of which has a dithiol predicted Cu²⁺-binding motif at the amino terminus of the large ATPase protein, as shown in Fig. 3) by having this metal-binding motif determined by a separate small gene that is immediately downstream of the gene for the ATPase.

A newly sequenced open reading frame (ORF732; Fig. 3 and 5) (100) from the E. coli genome-sequencing project encodes a 732-amino-acid P-type ATPase that might be a copper-translocating ATPase (Fig. 3 and added data), but there has been no direct experimental test of this hypothesis. The ATPase gene lies in an apparent operon with a second gene determining a smaller, 208-amino-acid, probably membrane protein that lacks significant homologs in protein sequence libraries. Genes for two additional E. coli P-type ATPases that might encode copper-translocating ATPases with histidine-rich N-terminal regions (GenBank protein accession numbers U166658 and U16659)/ (5) were found recently by accident during a search for human copper ATPase homologs (108). Their functions have not been tested. In summary, there are Cu²⁺ P-type ATPases in other microorganisms, and I predict that one or more will also be found in E. coli and S. typhimurium. Furthermore, genes for two copper ATPases (mutations in which cause the lethal diseases Menkes’ syndrome and Wilson’s disease) have been identified in humans, and their functions in disease have been predicted largely on the basis of bacterial homologs (18, 91, 109).

In addition to the six (or more) cut chromosomal genes that are thought to determine Cu²⁺ transport in E. coli, a plasmid-determined copper (pco) resistance system is also thought to function by altering copper transport and sequestration (79; summarized most recently in reference 17). According to sequence analysis (53), there are four resistance-determining genes, pcoA through pcoD, whose protein products are thought to exist in the inner membrane (PcoD), outer membrane (PcoB), and periplasmic space (Cu²⁺-binding proteins PcoA and PcoC). There remains considerable uncertainty as to how these four proteins function in copper uptake and efflux. In addition, the Pco copper resistance system is regulated by two proteins, PcoR and PcoS (53), which are members of the large class of two-component membrane sensor/kinase (PcoS) and transphosphorylated DNA-binding regulatory protein (PcoR) that regulate the responses of E. coli to many environmental signals (see chapter 80).

CALCIUM AND SODIUM EFFLUX

Calcium Secondary Transporters and ATPases

All bacterial cells maintain intracellular calcium levels below that of the growth medium (85) by means of cell membrane transporters that function either as primary ATPases or as secondary Ca²⁺/H⁺ chemiosmotic exchangers (55, 77). E. coli maintains an intracellular concentration of 0.1 μM Ca²⁺ independently of extracellular concentration (34) by using secondary Ca²⁺/H⁺ and Ca²⁺HPO₄ ^2–/H⁺ chemiosmotic exchangers (55, 77). Studies of calcium efflux by intact cells (34) and uptake of calcium by everted membrane vesicles support these conclusions (77). Calcium-sensitive mutants that have reduced calcium uptake activity with inside-out subcellular membranes were isolated (16). Detailed transport studies with membrane vesicles and studies with inhibitors supported the existence of the two separate systems (55, 77).

Using selection for a Na⁺/H⁺ antiporter activity, Padan and Schuldiner (66) recently isolated a gene, chaA, mapping at 27 min on the E. coli chromosome, that encodes a 39.2-kDa membrane polypeptide (that from sequence homology appears to be related to calsequestrin). Membrane vesicles overproducing this protein show enhanced Ca²⁺/H⁺ and Na⁺/H⁺ activities, providing the first indication that the ChaA protein may be a Ca²⁺/H⁺ antiporter with a low affinity for sodium ions also (66, 77). Although the physiological need to maintain low levels of intracellular calcium in E. coli is not known, sudden increased levels of intracellular calcium cause motile E. coli cells to stop smooth swimming and begin to tumble (107). More essential reasons for maintaining low levels of intracellular Ca²⁺ may be to prevent activation of Ca²⁺-dependent enzymes such as proteases, lipases, and nucleases until they are extracellular.

Calcium P-type ATPases have not been found in enteric bacteria. However, it is now well established that a P-type ATPase (such as exists in E. coli and S. typhimurium for K⁺ and Mg²⁺ uptake) is responsible for Ca²⁺ efflux and tolerance in the cyanobacteria Synechococcus (8, 46) and Synechocystis (36) spp., and additional calcium efflux ATPases have been described in other prokaryotes (55, 77).

Sodium Efflux via Na+/H+ Antiporters

All bacteria maintain intracellular sodium at levels below that of the extracellular medium. Although some bacteria achieve low intracellular Na⁺ levels by primary pumps (either ATPases or Na⁺-translocating decarboxylases [summarized in reference 76]), sodium efflux from E. coli is due to three chemiosmotic systems: a general monovalent cation-H⁺ exchanger that functions mostly in K⁺-H⁺ exchange and two specific Na⁺(Li⁺)/H⁺ antiporters (12, 66, 67, 76). Sodium coupling is utilized to balance two processes: sodium-substrate cotransport systems are frequently used to drive amino acid uptake by cells, and sodium/proton antiporters play crucial roles in maintaining intracellular pH, especially by moving protons inward at high extracellular pH (76).

The genetic determinants of both E. coli Na⁺/H⁺exchangers have been identified and designated nhaA (37a) and nhaB (71, 72). The nhaA locus encodes (50) a membrane protein of 41.3 kDa that (when present in large amounts from the gene cloned on a plasmid) confers Li⁺ resistance on cells and determines enhanced Na⁺-H⁺ exchange activity in subcellular vesicles (37a, 105). A strain from which nhaA is deleted becomes sensitive to high salinity and alkaline pH (65). The nhaB locus was sequenced, and the NhaB protein was identified as a membrane protein, deduced from the sequence to be a 55.5-kDa protein possibly crossing the membrane in 12 helical segments (71). The predicted sequences of NhaA and NhaB are not strongly homologous. NhaB sodium-proton exchange remains constant as extracellular pH increases, in contrast to the activity of NhaA. Mutants lacking both nhaA and nhaB are very sensitive to Na⁺ and Li⁺. However, in the absence of Na⁺ or Li⁺, nhaA and nhaB are not needed by E. coli cells.

Membranes prepared from the double nhaA nhaB deletion mutant strain lack Na⁺-H⁺ exchange activity (71). Both the NhaA and the NhaB proteins have been purified and shown to function in efflux of ²²Na⁺ when they are reconstituted in proteoliposomes (66, 67, 72, 105). In both cases, single polypeptides function as antiporters. Both NhaA and NhaB are electrogenic rather than neutral antiporters when reconstituted. For NhaA, the stoichiometry was 2 H⁺/Na⁺ (66, 67). The functioning of NhaA increased 2,000-fold when the pH was increased from 6.5 to 8.5, showing that NhaA functions as a pH sensor for the E. coli cell (67). One of the eight histidine residues in NhaA (His-226; by site-directed mutagenesis) is involved in pH sensing by NhaA (36a, 67). The NhaB protein differs from NhaA in vitro; it is unaffected by pH, less sensitive to Li⁺ ions, and sensitive to amiloride, which is an inhibitor of eukaryotic antiporters (67).

Synthesis of the Na⁺/H⁺ antiporter NhaA is induced by Na⁺ and Li⁺ cations (49). A regulatory gene, nhaR, which encodes a member of the LysR (positive regulatory) protein family and is needed for nhaA transcriptional regulation, has been sequenced (74). The NhaR protein binds to the nhaA promoter in Na⁺-sensitive transcriptional regulation, indicating that NhaR as well as NhaA senses Na⁺ concentration (67). A mutant of NhaR (Glu134Ala) increases the inducibility of nhaA by Na⁺ and also shows increased sensitivity to Na⁺ ions (20).

AMMONIUM UPTAKE

E. coli has been known to have a specific NH₄ ⁺ uptake pathway since the energy-dependent uptake of methylammonium (¹⁴C₃NH₃ ⁺) was reported (102). The membrane transport systems for NH₄ ⁺ in E. coli were recently reviewed by Barnes and Jayakumar (7), who are responsible for our current understanding of this topic. It is asked why a bacterial cell would have a transport system for the ammonium cation when un-ionized NH₃ can freely diffuse across biological membranes (52). The answer seems to be that with a pK_a of 9.3, very little ammonia exists at physiological pH, and with nitrogen often being growth limiting, there are selective advantages for a highly specific high-affinity transport system (52).

A ¹⁴C-labeled analog for the transport system affords easier experimentation than use of 10-min-half-life ¹³NH₄ ⁺. Quite reasonably for a system affording growth advantages at low substrate availability, the ¹⁴C₃NH₄ ⁺ transport activity is repressed by growth on high-NH₄ ⁺ medium. Accurate measurements of ¹⁴C₃NH₃ ⁺ uptake require filtration without washing, since even rapid rinsing of cells on filters removes highly membrane permeating ¹⁴C₃NH₂ (7). An estimate of the affinity for substrate gave a K_m of 36 μM ¹⁴C₃NH₃ ⁺, while a K_i of 0.5 μM NH₄ ⁺ was estimated (some 70 times higher affinity for the natural substrate than for the analog). E. coli mutant strains lacking primary NH₄ ⁺ (and ¹⁴C₃NH₃ ⁺) uptake activities were isolated (44), and these strains require higher ammonia concentrations for growth. NH₄ ⁺ appears to enter E. coli by way of a K⁺-NH₄ ⁺ exchange process (7). A high level of intracellular K⁺ is essential for ¹⁴C₃NH₃ ⁺ uptake, which is driven by the high K⁺ _in/K⁺ _out ratio (see above). However, the energy requirements for ¹⁴C₃NH₃ ⁺ uptake are complex, with a need for both a membrane potential and ATP (7). It is not clear whether these requirements are directly or indirectly related to the need for K⁺ exchange in the ¹⁴C₃NH₃ ⁺ uptake process.

Regulation of ¹⁴C₃NH₃ ⁺ uptake is also complex. Uptake is repressed by growth on high-NH₄ ⁺ media and transcriptionally repressed by the Ntr nitrogen regulon. ¹⁴C₃NH₃ ⁺ uptake is further regulated at the physiological level by glutamine feedback inhibition (7). Glutamine synthetase, which is the primary enzyme for incorporating ammonia into organic intermediates in E. coli, has a K_m of 1.8 mM NH₄ ⁺ (7).

E. coli mutants defective in ¹⁴C₃NH₃ ⁺uptake have lesions in either of two genes, amtA and amtB. Mutations in either gene lead to a >95% reduction in ¹⁴C₃NH₃ ⁺ transport (7). The amtA gene is located at position 95.8 min on the E. coli chromosome and was sequenced (7). The amtA gene encodes a 246-amino-acid polypeptide that is not homologous to other proteins in available protein sequence libraries (29). The S. typhimurium amtA gene has also been sequenced (7). The AmtA protein appears to be hydrophilic and gives no indication of being a membrane transport protein. Therefore, the role of AmtA in ammonium transport is unclear. The functions of the amtB gene and its product are also unclear (7).

CONCLUSIONS

For more than 30 years, E. coli has proved to be a paradigm organism for transport of inorganic cations as well as for the many other topics handled in these volumes. In general, there are specific physiological and biochemical roles for each cation formed from an element from atomic number 1 (hydrogen) to 30 (zinc). For each of these cations, highly specific membrane transporters are found in E. coli (generally one to three per cation) to ensure adequate uptake of needed cations, even when the levels of other related cations vary separately. Systems designed for starvation conditions are generally tightly regulated. Those used at times of nutritional abundance may be less selective and constitutively synthesized. For naturally abundant cations that are not used for intracellular functions (notably Na⁺ and Ca²⁺), cation-proton exchange systems function to maintain intracellular concentrations substantially below those in the growth milieu. No small ionic compound (organic or inorganic) passes through the bacterial cell membrane without a specific carrier protein. The "bug’s eye view" of the periodic table, including the views of E. coli and S. typhimurium, is that every element (i.e., cation and anion) is handled specifically, with its own set of membrane proteins and corresponding genes. In addition to the nutrient cations considered in this chapter, toxic heavy-metal cations are formed from elements above atomic number 30 (such as Cd and Hg), which occur frequently in environmental settings. Bacteria, including E. coli, have additional genes that confer resistances to toxic cations of heavy metals. These resistances also involve careful specificity, regulation, and membrane transport systems (e.g., see reference 92) but lie beyond the scope of this discussion.

With such broad coverage of genetic determination of cation transport systems, as reflected in this chapter, one will look whenever a new bacterial genome is totally sequenced (starting with the report by Fleischmann et al. [32a]) to see what is found that is related to systems of E. coli and S. typhimurium.

ACKNOWLEDGMENTS

This chapter summarizes 25 years of interest by our laboratory that has been supported by grants from the National Institutes of Health and National Science Foundation. The preparation of this report was supported by a grant from the Department of Energy. Many individuals whose work is cited have been generous in providing unpublished results and guidance as to the meaning of those results. In particular, I thank Evert Bakker, Ian Booth, Nigel Brown, Wolfgang Epstein, Frank Gibson, Barry Lee, Michael Maguire, and Mark Walderhaug. Mark Walderhaug prepared the figures.