Chapter 71

Uptake and Metabolism of Iron and Molybdenum

CHARLES F. EARHART

IRON ASSIMILATION

Introduction

Iron is necessary for the growth of Escherichia coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium). This iron requirement is emphasized by the variety of processes in which iron-containing proteins take part. For example, (i) heme-containing proteins participate in electron transport (cytochromes), activation of oxygen (cytochrome oxidase), and H₂O₂ reduction (catalase, peroxidase). (ii) Iron-sulfur proteins have roles in amino acid and pyrimidine biosynthesis (glutamate synthase, dihydroorotate dehydrogenase) and the tricarboxylic acid cycle (aconitase, succinate dehydrogenase), as well as in electron transport (ferredoxin). (iii) Non-heme, non-iron-sulfur proteins are required for DNA synthesis (ribonucleotide reductase), protection from superoxide radicals (superoxide dismutase), and aromatic amino acid biosynthesis (3-deoxy-d-arabinoheptulosonic acid 7-phosphate synthase). The biological utility of iron arises from several factors. Iron is able to form complexes with oxygen, sulfur, and nitrogen ligands, and these complexes are able to readily undergo acid-base and electron transfer reactions. Variations in the protein environment surrounding the complexed iron permit a wide range of Fe(III)/Fe(II) redox potentials.

Iron is the fourth most abundant element and is required in only micromolar concentrations for growth. Nonetheless, obtaining enough iron is frequently a problem. In an animal host, the iron is tightly bound by the carrier proteins transferrin and lactoferrin. In other environments, including laboratory media, the availability of iron is affected by the pH and aeration. Fe(III), the oxidation state present under aerobic conditions, forms insoluble oxy-hydroxide polymers at neutral pH. In contrast, Fe(II) is relatively soluble, and obtaining iron is a much easier task for cells growing anaerobically. One last point, the importance of which is becoming increasingly recognized, is that iron, unless tightly complexed, can promote formation of damaging hydroxyl radicals by the Fenton reaction [Fe(II) + H₂O₂ → Fe(III) + OH^– + OH·].

An understanding of iron assimilation is perhaps best approached from an evolutionary standpoint. From the solubility properties of the two stable oxidation states of iron, it can be assumed that iron uptake was not a problem of unusual difficulty until cyanobacteria appeared and began carrying out oxygenic photosynthesis. As the atmosphere became oxidizing, Fe(II) was converted to Fe(III) and insoluble hydrated ferric oxides were formed by hydrolytic polymerization reactions. Thus, perhaps a billion years after the origin of life on earth, organisms were faced with the necessity of developing efficient systems for iron assimilation. Development of many iron assimilation systems thus presumably occurred well after basic biochemical pathways had arisen; it is therefore not surprising that they show great diversity at the biochemical level. As will be discussed, E. coli and S. typhimurium have approximately five high-affinity systems each for iron transport and still other lower-affinity systems. This multiplicity and this seeming redundancy are understandable if one presumes that iron deprivation gradually increased in severity and cells had to survive in a variety of iron-deficient environments. Additional and more efficient iron uptake systems would appear; some would evolve independently from existing systems, and some might be garnered from other organisms through genetic exchange. Because these iron assimilation systems evolved comparatively late in evolution, we could reasonably anticipate that they would appropriate domains and strategies from existing enzymes and pathways. One would also predict that iron assimilation would be tightly controlled, given the multiple roles for iron-containing proteins and the potential for iron toxicity, and that some form of iron storage would be beneficial. With this as introduction, it is hoped that the multiple assimilation systems and other interwoven and complex aspects of iron metabolism to be described will fit into a coherent overview.

High-Affinity Iron Uptake by Siderophore-Utilizing Systems

To obtain iron found in insoluble complexes or bound to storage or carrier host proteins or even associated with laboratory glassware, E. coli and S. typhimurium synthesize and release into the environment small iron-chelating molecules termed siderophores (Fig. 1) (117). Ferrisiderophore complexes are subsequently taken up into the cell by specific transport components located in the cell envelope. Siderophores are generally less than 1,000 molecular weight (mol wt), are synthesized only under iron-deficient conditions, and are capable of binding Fe(III) with high affinity and specificity, Fe(III) being preferred over Fe(II). Many aerobic and facultatively anaerobic microbes synthesize siderophores, and about 100 structurally different siderophores have been described (133, 138). Despite the great variety of siderophores, they can generally be classified chemically as being either catecholates or hydroxamates, the most important functional differences among them being variations in stability and in affinity for Fe(III).

Fig. 1Some siderophores utilized by E. coli. Only aerobactin and enterobactin are bacterial products; rhodotorulic acid and the nonferrated forms of ferrichrome and coprogen are synthesized by fungi.

Source:

The catecholate siderophore enterobactin (Ent) is produced by all E. coli and S. typhimurium strains. Some strains also produce the hydroxamate siderophore aerobactin, whose biosynthetic genes can be either plasmid encoded (43) or chromosomal (145). For Ent and aerobactin, medium concentrations as high as 100 and 50 to 200 mg/liter can be attained, respectively (22), although concentrations of 10 to 20 mg/liter are more typical. Strains can use not only endogenously produced siderophores but also (because they have specific transport systems to permit their passage through the cell envelope) siderophores such as desferriferrichrome, coprogen, and rhodotorulic acid, which are produced by other organisms. Lastly, there are other molecules, such as citrate and the Ent-related molecules dihydroxybenzoic acid (DHB) and dihydroxybenzoylserine (DBS), that fail to meet the rigorous definition for siderophores but are capable of providing iron. Like siderophores, they have (i) high-affinity systems for their transport and (ii) phenolate and carboxylate oxygen ligands which have the low polarizability and high electronegativity preferred by Fe(III) (42).

Enterobactin.

Ent, also termed enterochelin, is the prototype catecholate siderophore and is found widely among the members of the family Enterobacteriaceae. It is synthesized from chorismic acid, the branch point for aromatic amino acid, folic acid, and ubiquinone biosynthesis, by a series of reactions requiring six enzymes, one of which (EntB/G) is bifunctional. Genes for Ent biosynthesis and transport and for release of iron from Ent are clustered at approximately 13.1 min in the E. coli chromosome (Fig. 2); Ent biosynthesis genes, designated ent, are located in three of the six Ent gene cluster operons. The Ent biosynthetic pathway is conveniently divided into two parts: (i) the conversion of chorismate to DHB and (ii) the synthesis of Ent from DHB and serine. DHB synthesis is the better understood and is described first.

Fig. 2Major gene clusters specifically involved in Fe(III) uptake. Symbols: , iron box with direction of transcription indicated; shaded arrows, transport genes; solid arrows, biosynthetic genes; open arrows, gene for iron release; , regulatory genes; , iron-regulated open reading frame; , insertion sequence.

Source:

The biosynthetic pathway by which chorismate is converted to DHB was proposed in 1965 (119). The pathway was soon confirmed (Fig. 3), and the biosynthetic intermediates were purified (207, 208). Three enzymes (EntC, EntB, and EntA) are required, and these have been purified and characterized (199). EntC (isochorismate synthase) is a 43-kDa polypeptide (55, 144) that is active as a monomer (122). It shows strong homologies both at the nucleotide level and in amino acid sequence to two other E. coli proteins that bind chorismic acid, TrpE (component I of anthranilate synthase) and PabB (component I of p-aminobenzoate synthase). Isochorismate, the product of EntC, is also a precursor of menaquinones (16); cells with entC mutations therefore fail to grow anaerobically on glucose as well as in low-iron environments. EntB (isochorismatase) has a predicted mol wt of 32,554 (121, 131) and apparently is active as a pentamer in the isochorismatase reaction (162). Evidence that it is bifunctional, participating also in the Ent synthetase reaction, is presented below. The native 2,3-dihydro-2,3-dihydroxybenzoate dehydrogenase (EntA) is an octamer composed of 26,249-Da subunits (121, 131). Its amino acid sequence places it in the family of short alcohol-polyol-sugar dehydrogenases. In summary, the initial Ent biosynthetic reactions and the enzymes catalyzing these reactions are well characterized. The two major complications in these studies were genetic: (i) the type strain for the entC gene, AN191, actually harbored an entA mutation (193), and (ii) entB also encoded EntG activity (179).

Fig. 3Enterobactin biosynthesis and degradation (modified from reference 52 with permission). Enterobactin esterase catalyzes the complete hydrolysis of Ent to DBS.

Source:

In contrast to DHB biosynthesis, many of the details of the latter steps in Ent biosynthesis remain obscure. The reaction 3DHB + 3l-Ser + 6ATP → Ent + 6AMP + 6PP_i is thought to be carried out by a multienzyme complex (Ent synthetase) consisting of the entD, entE, entF, and entB/G gene products. EntE and EntF are activating enzymes for DHB and l-Ser, respectively, but no specific activities for EntD and EntG have been determined. A general overview of how Ent might be synthesized is described first, and then facts regarding relevant proteins are presented.

On the basis of DNA sequence homologies and enzymological data, it appears that, as originally proposed by Bryce and Brot (30), Ent synthesis is similar to that of peptide antibiotics. Ent contains three amide bonds (Fig. 1) that are made nonribosomally; these bonds are most likely made by the protein thiotemplate mechanism (110). That is, substrates (Ser, DHB) are activated as acyladenylates, bound to a peptide synthetase as thioesters, and sequentially polymerized in the order determined by their synthetase-binding domains. For Ent biosynthesis, it is likely that EntF is a peptide synthetase, activating and binding Ser to itself via a thioester linkage to bound 4'-phosphopantetheine (4PP). EntE, an adenylate-forming acyl-activating enzyme, activates DHB, and then, by a transamidation reaction with the EntF-bound Ser, DBS is formed. Auxiliary factors may be necessary for DBS formation (30, 159). Subsequently, DBS units are linked by esterification, in what diagrammatically is a reversal of the Fes-catalyzed degradation reactions (Fig. 3), released, and excreted as Ent. EntB/G and EntD presumably participate in these processes. Peptide synthetase operons also commonly encode thioesterases, either as a small separate protein or as a domain present in a larger protein. Two Ent cluster proteins (EntF and P15) have thioesterase domains; whether or not these are functional is not known, but mutants defective in P15 have no phenotype (131).

The Ser-activating enzyme EntF (mol wt, 142,000) (159, 163) catalyzes the formation of an l-seryl-AMP-EntF complex. It contains 4PP and, by amino acid sequence homology, is a member of the superfamily of adenylate-forming enzymes (194) and is related to antibiotic synthetases (73). EntE (2,3-DHB-AMP ligase) is a homodimer with a predicted subunit molecular mass of 59,299 Da (161, 180). The acyladenylate remains stoichiometrically bound to the enzyme, presumably in preparation for amide bond formation. By sequence homology, it is in the same superfamily as EntF (194) but in a distinct subfamily (73); unlike EntF, it has no thioesterase or 4PP domains and no interdomain regions (M. Ammerlaan, Ph.D. thesis, University of Texas, Austin, 1994). The predicted mol wt for EntD is approximately 23,600 (11, 35). Unlike the other three Ent synthetase polypeptides, detection of EntD is difficult; its activity may be labile (67), and the entD gene is unusual in that it contains a high frequency of rare codons and is downstream of two repetitive extragenic palindromic sequences. When overexpressed, EntD is found in the inner leaflet of the cytoplasmic membrane (11). EntD shows homologies to the products of the Bacillus subtilis sfp ⁰ (72) and Bacillus brevis gsp (20) genes, which are hypothesized to be involved in secretion of the peptide antibiotics surfactin and gramicidin S, respectively (72). The sfp ⁰ gene complements entD mutations, but unlike EntD, which is required for Ent synthesis, the sfp ⁰ product does not appear to be necessary for surfactin biosynthesis. The EntG activity was the last Ent synthetase component to be detected. The defining Mu phage-induced mutation in strain AN462 was also EntB^– and EntA^– (204), and nucleotide sequencing determined there was no separate open reading frame that could correspond to entG in that region (121, 131). Subsequently, genetic, biochemical, and immunological approaches showed that the EntG activity is encoded by the entB 3' terminus and that EntB is in all likelihood a bifunctional protein (179).

The final stages of Ent synthesis are routinely referred to as being carried out by a complex or aggregate of EntD, E, F, and B/G, but firm evidence supporting this idea is meager. Greenwood and Luke (67, 69) first proposed the existence of a complex on the basis of gel filtration and DEAE-Sephadex chromatography results, which identified complexes consisting of EntGD, EntFD, and EntFDG. These authors also proposed that the complex was associated with the cytoplasmic membrane through EntD, an idea recently supported by the localization of EntD to the cytoplasmic membrane (11). Recent attempts to demonstrate that EntD, E, F, and B/G exist as a membrane-associated complex in vivo have failed, however (Ammerlaan, Ph.D. thesis). In immunoprecipitation experiments, polyclonal antibodies against EntE, F, or B/G brought down only their cognate polypeptide. High-speed extracts presumably devoid of membrane were fully active in Ent synthetase assays, and this activity was maintained in the presence of neutral detergents (5% Triton X-100, Nonidet P-40, or Tween 20) and inhibited only by high sodium chloride concentrations (K_i = 150 mM). Similarly, experiments with N-hydroxysuccinimide ester cross-linking agents failed to provide any evidence of a complex. Evidence for an unusual intracellular location was obtained from osmotic shock experiments, which demonstrated that the majority of EntE and B/G proteins and some of the EntF proteins are released from shocked cells (F. Hantash and C. F. Earhart, Abstr. 94th Gen. Meet. Am. Soc. Microbiol., abstr. K137, p. 299, 1994). These proteins are not released by spheroplasting, so they can be classified as group D proteins (14) along with such proteins as thymidine phosphorylase, elongation factor Tu, and thioredoxin (126). No evidence indicative of complex formation was obtained, since absence of any of the synthetase proteins, including EntD, had no effect on the shockability of the remaining proteins. It appears that Ent synthetase has an unusual cellular location and that association of its several different polypeptides is transient.

Aerobactin.

Aerobactin, a dihydroxamate siderophore whose trivial name stems from the fact that it was originally isolated from Aerobacter (Enterobacter) aerogenes (66), consists of two modified lysines (N ⁶-hydroxy-N ⁶-acetyl-l-Lys) linked by amide bonds to citrate (Fig. 1). The aerobactin system has been intensively studied for two reasons: (i) regulation of aerobactin biosynthesis is relatively simple, and this has made it the system of choice for studying iron control mechanisms (12); and (ii) aerobactin is an important virulence factor, particularly in invasive diseases (145). That aerobactin is superior to Ent in providing iron to microbes in a host was at first surprising, as Ent has a higher affinity for Fe(III) than aerobactin (K_f = 10⁵² versus 10²³ for aerobactin) (133). However, this difference is almost negated at neutral pH (12, 52), where the hydroxyls of Ent are partially protonated. Also, aerobactin (i) can be recycled (23); (ii) can be secreted more readily than Ent in some (50) but not all E. coli strains (Ammerlaan, Ph.D. thesis); (iii) is more stable than Ent (free Ent has a t _1/2 of approximately 30 min at room temperature at pH 7.0 [Ammerlaan, Ph.D. thesis]); (iv) unlike Ent, does not form stable complexes with serum albumin (112, 156); and (v) apparently does not induce naturally occurring anti-aerobactin antibodies, as none have been reported. Naturally occurring immunoglobulin A antibodies against Ent which have growth-inhibiting effects that are not diminished by bovine serum albumin (50 mg/ml) (L. Wen, M.S. thesis, University of Texas, Austin, 1984) can be detected (128, 129).

Biosynthesis of aerobactin requires four genes (iucABCD), which are also designated aerDBCA, respectively (25); the genes are identical, whether plasmid borne or chromosomal (Fig. 2). Biosynthesis is initiated by the hydroxylation and acetylation of lysine; two molecules of N ⁶-acetyl-N ⁶-hydroxylysine are then condensed with the primary carboxyls of citrate to yield aerobactin (Fig. 4). Surprisingly, only iucD has been sequenced, and the only native enzyme to be purified to homogeneity is the iucB product. IucD (lysine: N ⁶-hydroxylase) is a cytoplasmic membrane protein; soluble forms have been obtained as LacZ-IucD hybrids (190), purified, and characterized. The iucD sequence (90) has homologies to genes encoding oxygenases involved in siderophore biosynthesis in Pseudomonas aeruginosa (pvd-1) and Ustilago maydis (sid1) and a flavin-containing monooxygenase of rabbit. The predicted monomer mol wt is 48,968, and a native mol wt of approximately 200,000 has been reported (190). The active enzyme requires a high-ionic-strength environment and is specific for lysine and the cofactor flavin adenine dinucleotide (FAD). The electron donor may depend on the local environment of IucD. Experiments with particulate IucD of Enterobacter aerogenes indicated that pyruvate provided electrons, presumably through FAD, for N hydroxylation and that NADPH was inhibiting (197); the inverse results were obtained with the soluble enzyme (190). IucB (N ⁶-hydroxylserine: acetyl coenzyme A N ⁶-transacetylase) completes the lysine modification. It is a stable, soluble, multimeric protein (33-kDa monomer, 150,000 to 200,000 native mol wt) that was isolated in a single fractionation step (41). N ⁶-Hydroxylysine is the preferred substrate, and l-lysine is not used. In Enterobacter aerogenes there is evidence that pyruvate, acting as a substrate for a membrane-bound pyruvate oxidase, is the source of the acetyl moiety (187, 197). Aerobactin synthetase contains two subunits, IucA (63 kDa) and IucC (62 kDa) (48, 71); IucA is required for production of N ²-citryl-N ⁶-acetyl-N ⁶-hydroxylysine, and IucC adds the final modified lysine. The enzyme is soluble and labile and in native form is found in the void volume of Sephadex G-100 columns (10). ATP is required for the activation of both citrate and N ⁶-acetyl-N ⁶-hydroxylysine. As with Ent synthesis (above), a nonribosomal peptide synthesis mechanism using protein thiotemplates has been proposed. Sequencing of iucA and iucC and searching for consensus sequences for 4PP attachment, thioesterase activity, and adenylate formation would be informative.

Fig. 4Aerobactin biosynthesis.

Source:

Excretion of Siderophores into the Periplasm and Environment.

Little is known regarding the way in which Ent and aerobactin leave cells; no genes or proteins necessary for release have been identified. It is thought that concomitant or rapid excretion of siderophores occurs after synthesis because of the possible damage such molecules could cause in the cytoplasm, even though their affinity for Fe(II) is less than that for Fe(III). There is evidence that both Ent synthetase (see above) and aerobactin synthetase (197) occupy unusual intracellular locations.

Transport of Ferrisiderophores.

  Outer membrane receptor proteins. The outer membrane permits ready passage of (i) small hydrophilic molecules of M _r less than 600 by means of porins and (ii) highly hydrophobic molecules such as steroids, apparently by lipid bilayer diffusion (149). Ferrisiderophore complexes are unable to use either pathway, requiring instead specific outer membrane receptor proteins. These receptors share several properties, including molecular masses of 74,000 to 83,000 Da, iron-regulated synthesis, and the ability to interact with the TonB protein (see below) (22, 134, 137). Their appearance in outer membrane is diagnostic for iron starvation; it is likely that all are gated pores, and they are probably the rate-limiting factor in ferrisiderophore uptake. They display regions of amino acid sequence homology, particularly near their amino termini (99).

The outer membrane receptor for ferrienterobactin (FeEnt) is FepA; mature FepA is a 79,908-Da (125) polypeptide that is also used as a receptor by colicins B and D. FepA, the best characterized of the iron-chelate receptors, has been purified (60), and its surface topology has been studied (130). The current model indicates that FepA contains 29 amphiphilic transmembrane β-strands, and unlike the majority of outer membrane proteins, the carboxy terminus of FepA is proposed to face the environment rather than the periplasm. FepA is specific for Δ-cis FeEnt (136), and binding occurs to a large surface-exposed loop connecting transmembrane β-strand segments (130). FepA is a gated pore; removal of the ligand-binding loop converts FepA into an energy-independent nonspecific channel (123, 164) with an internal diameter (20 Å [2 nm]) considerably larger than that of any other known E. coli porin. The FeEnt-binding loop is thought to normally block a β-barrel channel; upon activation by TonB (see below), conformational changes affecting this plug may occur such that FeEnt enters the periplasm. Native FepA probably exists as a trimer (164).

The aerobactin outer membrane receptor (IutA) is a 725-amino-acid proprotein; the mature protein contains 700 amino acids (77,345 Da) (115). Interestingly for a protein of a class that can be present in 100,000 copies per cell (25), its gene is the last one in the five-gene aerobactin operon.

Ferrichrome and the closely related serine-containing compounds ferricrocin and ferrichrysin can be assimilated by enteric bacteria but are synthesized only by certain fungi. The receptor protein, FhuA (formerly TonA), is multifunctional, also being involved in binding of microcin 25 (168), colicin M, the antibiotic albomycin (a desferriferrichrome analog), and phages T1, T5, UC-1, and φ80. It has a 33-amino-acid signal sequence, and the mature protein has a predicted mol wt of 78,992 (38). Characterization of FhuA surface derivatives (31, 111) has led to a topological model with 32 transmembrane (β-sheet) segments (111). A surface loop (residues 316 to 356, connecting proposed transmembrane segments 15 and 16) is part of the FhuA active site. Deletion of residues 322 to 355 converts FhuA into a nonspecific diffusional channel whose permeability properties are independent of TonB (107). These results, which are consistent with those obtained with FepA, provide additional evidence that these outer membrane receptor proteins are β-barrel channels which are closed (gated) by a TonB-controlled surface loop.

Coprogen and rhodotorulic acid, which are also fungal hydroxamate siderophores, utilize the FhuE receptor (M _r, 77,453; 36-amino-acid signal sequence) to provide iron to bacteria (80, 169). Absence of FhuE activity also results in inability to use ferrioxamine B and its derivative, ferrioxamine D1 (80). There is conflicting evidence regarding the receptor for these latter two compounds, however, as work with photoreactive analogs demonstrated differential inactivation of ferrioxamine B transport compared to coprogen uptake (139). The alternative proposed ferrioxamine receptor (FoxB) is remarkable in that it requires two outer membrane proteins, FoxB (66 kDa) and FoxB₂ (26 kDa), and functions in cells with tonB mutations (140).

FecA, the 81,718-Da outer membrane receptor for ferridicitrate uptake (155, 195, 198), is unique in requiring at least 0.1 mM citrate and low levels of iron for induction. Remarkably, FecA can bind FeEnt (209, 210), although its binding constant for ferridicitrate is an order of magnitude lower (D. van der Helm, personal communication).

Two additional iron-regulated outer membrane proteins are known: Cir (74 kDa) (132, 133) and Fiu (83 kDa). Fiu is not observed in E. coli B. Both serve as colicin receptors (colicins I and V bind to Cir, and colicins G, H, and E492 bind to Fiu) (25) and as receptors for the Fe(III) complexes of the Ent-related compounds DHB and DBS (84). Specifically, ferriDBS crosses the outer membrane in a TonB-dependent manner using Fiu, FepA, and to a lesser extent Cir, whereas ferriDHB primarily uses Fiu and Cir. Fiu and Cir have broad specificity, as they can also function in transport of catechol-substituted cephalosporins (44, 142).

  Requirement for TonB. Ferrisiderophores bind to their outer membrane receptors regardless of whether these proteins are purified, in outer membrane fractions, in energy-poisoned cells, or present in normal cells. The next step in ferrisiderophore uptake, passage through the outer membrane, requires the TonB protein and energized cells. TonB is also required for vitamin B₁₂ uptake (99) and for sensitivity to many phages (e.g., T1, φ80, UC-1) and group B colicins (B, D, G, H, Ia, Ib, M, Q, S1, and V) (24, 45, 152). TonB is not the receptor for any of these comparatively large entities and functions only to provide the energy which allows them to enter the periplasm. The remarkable ability of TonB to provide energy to the outer membrane, an activity first proposed by Hancock and Braun (77), has stimulated extensive research that is ably summarized in recent reviews (24, 109, 152).

The tonB genes of E. coli (153) and S. typhimurium (78) have been sequenced and shown to encode similar proteins of approximately 26 kDa. TonB proteins are unusual in that they have a high proline content (17%), amino termini with properties similar to a signal sequence but which lack a signal peptidase cleavage site and are therefore not cleaved, and a proline-rich region in their amino-terminal third. In sodium dodecyl sulfate-polyacrylamide gels, they migrate aberrantly as 36- to 40-kDa polypeptides because of this rigid proline region (118). TonB proteins of E. coli and S. typhimurium are hydrophilic with hydrophobic amino and carboxy termini, but not all TonB proteins have hydrophobic C termini (29).

TonB is located in the cytoplasmic membrane (154) and apparently requires the Sec machinery to reach this destination (152). Topology studies using TonB-PhoA and TonB-Bla fusions, anti-TonB antibodies, and protease accessibility assays (58, 78, 154, 160) indicate that the protein is anchored in the cytoplasmic membrane by its hydrophobic N terminus and that the remainder of TonB is in the periplasm. Several lines of evidence indicate that TonB spans the periplasmic space: (i) TonB can be chemically cross-linked to FepA (177); (ii) FhuA can stabilize TonB (24); and (iii) inactivating point mutations in genes for outer membrane receptors are suppressed by mutations in tonB. These last experiments are based on the finding that all TonB-dependent receptors and, surprisingly, group B colicins have a consensus pentapeptide (TonB box) close to their amino termini (172). The TonB box is thought to be the site of receptor interactions with TonB; receptors with altered TonB boxes display normal binding but are unable to transport their ligand (15, 24, 88), and mutations in tonB restore activity (15, 24, 89). Also, addition of synthetic TonB box pentapeptide (Glu-Thr-Val-Ile-Val) to growing cells inhibits TonB-dependent processes (192).

TonB is believed to have three functional domains: the hydrophobic N-terminal domain, a hydrophilic, elongated, rigid central domain, and a hydrophobic C-terminal domain containing amphiphilic potential transmembrane structures (109). The N terminus has several functions, including serving as an export signal, as a cytoplasmic membrane anchor, and as a site of interaction with ExbB, a cytoplasmic membrane protein required for full TonB activity (see below) (78, 97, 105, 154). The central domain provides a means of spanning the periplasm; deletions in it of greater than 30 amino acids are still TonB⁺, and such mutated cells have a phenotype only when the periplasmic space is enlarged (118). Hydrophobicity of the C-terminus is not required for E. coli TonB function (29), and deletions extending into the 3' end of tonB do not affect the membrane localization of the corresponding TonB proteins. However, deletion of the C-terminal 15, but not 8, amino acids results in inactive TonB (9).

Two additional cytoplasmic membrane proteins, ExbB (26.1 kDa) and ExbD (15.5 kDa), assist TonB. ExbB has an unusual topology (103, 106); its amino terminus is in the periplasm, there are three transmembrane segments, and the bulk of ExbB is in the cytoplasm as a large loop and a C-terminal tail. Most of ExbD is in the periplasm; there is one transmembrane segment and a long, periplasmically located C-terminal tail (102). Mutations in exbB or exbD cause cells to have a leaky TonB phenotype (54). This leakiness arises from strong similarities in amino acid sequence and topology (104) between the ExbB and ExbD pair and TolQ and TolR, respectively. TolQ and TolR are encoded by the tol locus (tolQRAB), whose products are necessary for sensitivity to group A colicins (A, E1-3, K, L, and N) and filamentous phages with single-stranded DNA genomes (186). TolQR can partially substitute for ExbBD in TonB-dependent outer membrane passage, and ExbBD can likewise compensate for TolQR for TolA-mediated translocation (27); the two uptake systems for biopolymers (TonB-ExbB-ExbD and TolA-TolQ-TolR) are related functionally and evolutionarily.

TonB has a short functional half-life (100) and, under some conditions, is chemically unstable (154, 178). The basis for the brief functional period is unknown, but proteolysis is now understood to occur when TonB is overexpressed relative to ExbB (58); OmpT is the primary protein involved in TonB degradation (177). The ability of ExbB to protect TonB argues that the two proteins interact, as does evidence suggesting that ExbB and TonB can be cross-linked (97, 177). ExbD does not stabilize TonB, but is itself stabilized by ExbB (58). The data argue for a cytoplasmic membrane complex consisting of, minimally, TonB and ExbB (27, 178). The complex may also contain ExbD as well as unidentified proteins that cross-link to TonB (177). The energy required for TonB activity is provided by the proton motive force (21).

There is no generally agreed-upon model regarding how TonB acts or the role of ExbBD. A conceptual difficulty of models in which TonB spans the periplasm and directly interacts with outer membrane receptors is the "stoichiometry quandary" (109). TonB synthesis is iron regulated (151), but even when maximally expressed, the amount of TonB would be at least an order of magnitude less than that of the number of receptors to be serviced. One TonB complex presumably must interact with many receptors, but how this could occur is unclear. Receptors do not quickly laterally diffuse in the outer membrane, and it is difficult to envision the TonB periplasmic domain moving easily through the peptidoglycan layer.

  Entry into cytoplasm. Ferrisiderophore complexes enter the cytoplasm via transport systems belonging to the ABC (ATP-binding cassette) transporter superfamily (92); these are also termed traffic ATPases (4). This superfamily has been recently reviewed (3, 57, 91). ABC transporters are characterized by four membrane-associated domains, two of which contain a defining highly conserved ATP-binding motif and two of which are very hydrophobic, usually with six transmembrane segments each. Systems which take up ferrisiderophore complexes belong to the ABC importer (periplasmic permease) subfamily (1). These not only have the four membrane domains, typically as four individual proteins, but also a periplasmic protein. The latter component is essential; it is substrate specific, binding the substrate and presenting it to the cytoplasmic membrane complex. Osmotically shocking intact cells results in loss of the periplasmic proteins and therefore inactivation of these systems. These shock-sensitive transporters are reviewed in chapter 76 of this volume. Ferrisiderophore and the vitamin B₁₂ transporters form a specific subset of the periplasmic permease subfamily. They uniquely require an outer membrane receptor protein and a functional TonB protein (see above), and they share amino acid sequence homologies in both their hydrophobic components (33, 175, 181) and their periplasmic binding proteins (113, 188) as well as in their ATP-binding peripheral membrane components.

Periplasmic binding proteins are stable monomers that undergo large conformational changes upon binding their ligand (1, 157). The binding protein for FeEnt is FepB, ferridicitrate is recognized by FecB, and, remarkably, FhuD binds Fe(III) complexed to a great variety of hydroxamate compounds including aerobactin, desferriferrichrome, coprogen, rhodotorulic acid, and ferrioxamine B. FhuD was shown to bind ferrihydroxamate complexes in a whole-cell binding procedure and by noting that such complexes protected FhuD from protease digestion (114). That FepB binds FeEnt was shown by placing it in outer membrane as part of a tribrid protein and then demonstrating binding activity by this membrane (D. Stephens and M. Choe, personal communication). FecB binds ferridicitrate (25), and this is also assumed given its location and homologies to ferrisiderophore-binding proteins.

The cytoplasmic membrane complex for FeEnt consists of FepC (147), which has the ATP-binding motif, and FepD and FepG, which are two highly hydrophobic proteins with multiple membrane-spanning α-helical segments (33, 34, 175). For ferridicitrate uptake the peripheral ATP-binding component is FecE, and FecC and FecD are the two very hydrophobic integral membrane proteins (181). Just as the periplasmic binding protein FhuD can accommodate ferrisiderophores with a great variety of structures, the corresponding membrane complex, composed of FhuB and FhuC, also accepts all ferric hydroxamates (113). FhuC is the polar peripheral ATP-binding protein, and FhuB, which is twice the size (70 kDa) of most hydrophobic permease components, provides the required two hydrophobic domains. A generic model for ferrisiderophore uptake is shown in Fig. 5. Transport is coupled to ATP hydrolysis (2).

Fig. 5General model for ferrisiderophore uptake (modified from reference 25 with permission). The TonB, ExbB, and ExbD proteins are used with every member of the iron/B12 subset of ABC importers.

Source:

  Release of iron from siderophores. The intracellular fate of ferrisiderophore complexes is not well understood and seems to vary depending on the siderophore. Both Ent and desferriferrichrome appear to function just once before being modified to less active forms, whereas aerobactin can be reused an average of three times under some conditions (23). Concomitant with ferrichrome transport, iron is reduced and the resulting desferriferrichrome moiety is acetylated, thereby becoming less able to function as a siderophore, and secreted into the medium (85). In a cell-free system that mimicked the in vivo processes, both the reduction and modification activities were associated with membrane (171). Ent is enzymatically digested during delivery of its iron ion; the fes gene product, Ent esterase, hydrolyzes the ester bonds of Ent, yielding DBS as the final degradation product. The ability of DBS to function as a secondary siderophore is described above.

Because Fe(II) is bound less strongly than Fe(III) by siderophores and because iron exists as Fe(II) in the cytoplasm, enzymatic reduction of iron as a release mechanism has been repeatedly proposed. Ferrisiderophore reductases have been detected in fungi and gram-positive and gram-negative organisms (42). Fischer et al. (59) observed a soluble and a membrane-bound reductase in E. coli; NADH and NADPH functioned equally well as sources of reducing power, and FAD and flavin mononucleotide (FMN) both increased enzyme activity. In general, these enzymes have almost no substrate specificity, and this certainly holds for E. coli. The soluble 26,000-Da reductase purified using ferrichrome as the substrate also reduced carrier-free Fe(III) and every ferrisiderophore tested, including eight other hydroxamate compounds, two synthetic Ent analogs, and ferridicitrate. Two of the ferrisiderophores that were reduced do not support E. coli growth. The membrane-bound reductase, unlike the soluble enzyme, was sensitive to amytal and less likely to use oxygen as a competing substrate. Reduction of Fe(III) was the only apparent physiological role for these enzymes. They were not regulated by the Fur repressor (see below) and had similar activities whether cells were grown under aerobic or anaerobic conditions. The soluble enzyme might be identical to a ferridicitrate reductase described by Williams and Poole (202) and to a 26-kDa flavin reductase (flavin reductase enzyme, Fre) (6, 39). Additional soluble reductase activity in E. coli has been detected and attributed to HMP (hemoglobinlike protein). C-terminal sequence homologies indicate that both Fre and HMP have NAD(P)⁺-dependent FAD-binding oxidoreductase domains typical of the ferredoxin NADP⁺ reductase family (6). Studies on ferripyoverdine reductase of P. aeruginosa led Halle and Meyer (75, 76) to suggest that the ferrisiderophore reductase was in fact an NADH:FMN oxidoreductase, whose FMNH₂ product was able to reduce iron complexed to siderophores. The P. aeruginosa enzyme, functioning under anaerobic conditions, released iron from six siderophores, including Ent. Fontecave et al. (62) have reviewed evidence that the ferrisiderophore reductases are in fact flavin reductases, producing free reduced flavins able to transfer electrons to ferric complexes of all kinds. In this recent view, specific ferrisiderophore reductases may not exist.

Fes is the only protein known with a specific role in iron release from a siderophore (116, 148); in its absence, iron is not removed from FeEnt (52). Early siderophore electrochemistry experiments (37, 143) indicated that the reduction potential of FeEnt, unlike that of other ferrisiderophores or complexes of Fe(III) with Ent hydrolytic products, was so low that it was well outside the range of physiological agents. Therefore, the role of Fes was postulated to be to hydrolyze Ent, making it possible for the Fe(III) bound to DBS and DBS dimers and trimers to be reduced. However, instances where iron was provided to Fes⁺ but not Fes– cells by Ent analogs lacking ester bonds (87, 94, 196) were soon reported. At least some of the Fe(III)-Ent analogs used the normal binding-protein-dependent system to cross the cytoplasmic membrane and receptor FepA for outer membrane passage, but there are some conflicting data on the latter point (53, 87, 196). The growth promotion ability of nonhydrolyzable analogs of Ent led to the suggestion that Fes is actually a reductase and that labile ester bonds, if present, are only coincidentally cleaved (87, 97, 196).

In vitro studies with Fes show that it is a single component (68) and that both partially purified and purified Fes have severalfold greater esterase activity on Ent than on FeEnt (28, 68). Reduction is not necessary for Ent hydrolysis, as esterase activity occurred with Al(III) Ent as substrate. Reductase and esterase rates were similar with FeEnt as substrate, but no detectable reduction occurred when Fe(III) was complexed to an analog with no ester bonds. The in vitro data argue that esterase activity is a prerequisite for iron reduction (28). That Ent analogs devoid of ester bonds can provide sufficient iron for bacterial growth by a Fes-dependent process is unexplained. Some possible considerations are that the growth assays are more sensitive than in vitro reductase assays and that in vivo iron release occurs in an environment far different from that of the test tube.

Low-Affinity Iron Transport

High-affinity iron uptake systems are repressed when cells are grown in medium containing at least 5 to 10 μM iron (200). Under these iron-replete conditions, the low-affinity system for iron assimilation functions. This system is poorly defined, and no mutants defective in this system have been reported. Iron might be transported in several ways. The presence of ascorbic acid facilitates iron assimilation by Ent^– cells, so a reductive step, possibly in conjunction with the feo system (82) (see below), could be involved (137, 150). Also, it is now recognized that a number of small molecules found in culture supernatants can function as siderophores. Monocatecholates can be true siderophores (146), although in E. coli and S. typhimurium synthesis of DBS and DHB would be repressed under iron-rich conditions. For the tribe Proteeae, α-keto acids and α-hydroxy acids have siderophore activity (51, 191); evidence concerning whether or not such compounds can similarly provide iron to E. coli and Salmonella spp. is contradictory (51; R. Reissbrodt, W. Beer, and V. Braun, personal communication).

Ferrous Iron Transport

An Fe(II) transport system has been described for E. coli (82). Two genes are present at the feo (ferrous iron uptake) locus, which maps at 74.9 min (101). The two-gene operon, feoAB, is regulated both by FNR, the transcriptional activator for anaerobic repiratory genes, and Fur, the aporepressor for all E. coli transport systems (see below). The feoA gene may encode a small (75 residues) protein, and FeoB is a large (M _r, 84,473) cytoplasmic membrane protein with homologies to ATPases, suggesting that ATP provides the energy for ferrous iron uptake. In the mouse intestine, an anaerobic environment, the feo system appears to be more important than siderophore-based systems for E. coli colonization (185).

Iron Storage

E. coli has at least two iron storage proteins, bacterioferritin (BFR) (206) and ferritin (FTN) (96). Both proteins are similar to eukaryotic ferritins; they are composed of 24 subunits which form an approximately 500,000-Da spherical shell capable of accommodating thousands of iron atoms in its central cavity (5). BFR and FTN show 21% and 22% amino acid sequence identity with human ferritin H chains, respectively; sequence conservation is particularly evident for the ferroxidase center residues (8). In eukaryotes, ferroxidase catalyzes the initial reaction in iron-core formation whereby Fe(II) is oxidized to Fe(III) in the presence of dioxygen; both BFR (7) and FTN (95) have this activity. The bfr and ftn genes map close to each other at 73 min, their protein products show 17% amino acid sequence identity, and both proteins have been purified (7, 95). BFR (subunit M _r, 18,495) contains heme units, approximately one for every two subunits, and was formerly designated cytochrome b ₁ (205). In vitro, free FTN (subunit M _r, 19,400) is degraded, apparently by iron-induced free radicals.

Andrews et al. (7) have suggested that the roles of FTN and BFR are in short-term iron flux and long-term iron storage, respectively, based on the following additional differences between the two proteins. BFR binds iron more avidly than FTN and is synthesized maximally during periods of slow growth, including stationary phase. FTN, on the other hand, has a more active ferroxidase and therefore incorporates iron faster than BFR, and the FTN content does not vary during the growth cycle.

Regulation

The Fur Regulon.

The chief protein affecting expression of iron-regulated genes is Fur (Fe uptake regulation). Fur-like proteins are found in a variety of gram-negative bacteria and probably in gram-positive organisms as well (25, 32, 184). Mutations in the cognate fur gene were first isolated fortuitously in S. typhimurium (56). All iron-regulated activities tested were expressed constitutively in the mutant strain. Hantke isolated E. coli with fur mutations by screening a strain bearing an fhuA-lacZ fusion for constitutive β-galactosidase expression (79). The fur gene was subsequently cloned, mapped (15.5 min), shown to encode a negative regulator (81), and sequenced (170). Fur is a histidine-rich (8%), 16,795-Da protein; the high histidine content permits its one-step purification by metal ion affinity chromatography (201).

The Fur protein was shown to act as an aporepressor in in vitro studies with aerobactin operon genes (13, 49), which were known to be regulated at the transcriptional level (18). Fe(II) and other divalent heavy-metal ions [Co(II), Cd(II), Cu(II), and Mn(II)] served as corepressors, indicating, as predicted (203), that intracellular pools of free iron exist and that these pools can act in regulation. Rapid oxidation of Fe(II) to Fe(III), which does not function as a corepressor (135), led to the use of Mn(II) in Fur-DNA binding experiments. This effect of Mn(II) also presumably is the basis for an efficient positive selection method for fur mutations based on manganese resistance, whereby relatively high intracellular Mn(II) apparently represses iron-regulation genes by interacting with Fur (83). A 19-bp consensus binding site (iron box, GATAATGATAATCATTATC) for the holorepressor was proposed based on DNase I protection experiments of the fur and aerobactin operon promoters (46, 49). Footprinting experiments also detected a similar sequence in the cir operator (70). This dyad symmetric element has been found in the promoter region of at least 16 iron-regulated E. coli genes, sometimes in two copies (25). New Fur-regulated genes continue to be discovered. A Fur titration assay (184) revealed nine new E. coli genes that either are Fur regulated or encode iron storage/binding proteins. For the aerobactin promoter, which has two overlapping iron boxes, only one is functional in vivo and only it binds Fur in vitro (46). The location of iron boxes in the major E. coli gene clusters concerned with siderophore synthesis and ferrisiderophore uptake is shown in Fig. 2; only the Ent and Fec clusters consist of more than one Fur-regulated operon. Fur-binding sites are generally located between the –35 and –10 regions of the promoter; that these operator-promoter regions can bind either Fur [Mn(II)] or RNA polymerase but not both simultaneously was shown by gel mobility retardation assays (201).

Fur, in vivo, exists as a multimer. Genetic evidence (negative complementation) indicates that mixed oligomers form in vivo (26), and Hill plots based on in vitro data suggest that active Fur is a dimer (12).

Structural studies on Fur are incomplete. Significant conformational changes occur upon metal binding, as indicated by increased protease sensitivity and DNA binding. Deletion of the N-terminal region abolishes DNA binding, and each monomer apparently has two nonidentical metal-binding sites (40). A monomer model based on nuclear magnetic resonance studies has been proposed in which Fur consists mainly of parallel α-helices (165, 166, 167). In this structure a general cation-binding site composed of C-terminal Glu, Asp, and Lys was identified, and the dominant Mn(II)/Fe(II)-binding site was composed primarily of three His residues at positions 31, 32, and 131. It is unclear whether or not these results can be reconciled with evidence that cysteine thiol groups are required for Fur activity or that the essential metal-binding domain is C-terminal (40).

The Iron Stimulon.

Regulation by Fur does not fully explain the complex cellular responses to changes in iron availability; at least some genes of the Fur regulon are subject to additional control mechanisms, and some iron-regulated genes may be independent of Fur control. Early indications of the complexity of iron control included the following: (i) strains with certain ompB alleles synthesized massive amounts of Ent without concomitant appearance of iron-regulated outer membrane proteins (124); (ii) different patterns of control were detected in the synthesis of iron-regulated membrane proteins (108); and (iii) fepA transcription was enhanced an additional 10-fold when strains with fur mutations were iron-starved (79). Systematic studies using two-dimensional polyacrylamide gel electrophoresis of whole-cell proteins from S. typhimurium (65) and Vibrio cholerae (120) revealed the extent of the iron stimulon and the general types of relationships that occur between iron-regulated genes and Fur. In both organisms, the largest class of iron-regulated proteins was negatively controlled by Fur. However, other classes included proteins negatively regulated by iron independently of Fur, as well as proteins positively controlled by Fur, some of which additionally required iron for induction and some which were induced by iron only in the absence of Fur. Some examples of Fur-regulated genes controlled by additional transcriptional regulators follow.

The promoter region of fur contains both an iron box and a CAP (catabolite-activator protein) site. Fur is well expressed under all conditions tested, but its synthesis is diminished by the FeFur complex and is stimulated by cyclic AMP-CAP (47). Fur^– strains are also unable to grow on certain nonfermentable carbon substrates, including succinate, fumarate, and acetate (82); in the case of succinate, the uptake rate and succinate dehydrogenase levels were greatly reduced. In wild-type strains, these could be cases where Fur acts as a positive regulator, enhancing transcription from genes required for growth on these three substrates; Fur synthesis would be stimulated by the high levels of cyclic AMP-CAP (25). The genes required for ferricitrate uptake (fecABCDE) are also subject to a second regulatory system (25, 176, 195). The genes fecI and fecR map upstream of the transport operon and encode a cytoplasmic protein and a cytoplasmic membrane protein, respectively. FecR is believed to act as a sensor of ferridicitrate; in the presence of this inducer, FecR signals FecI, which then acts as a transcriptional activator for the fecABCDE genes. Binding of FecI to this DNA has been demonstrated (V. Braun, personal communication). Therefore, transcription of fecABCDE requires the absence of FeFur and positive activation by FecI. Lastly, the gene sodA (manganese-containing superoxide dismutase [MnSOD]) is preceded by an iron box but is not directly concerned with iron uptake. Instead, MnSOD plays a crucial role in maintaining a low intracellular level of superoxide radical (O₂ ^–). Superoxide can directly cause cell damage and, in excess, can indirectly lead to hydroxyl radical formation. The reduction of iron by superoxide [Fe(III) + O₂ ^– → Fe(II) + O₂] and the presence of H₂O₂ provide the reactants for the Fenton reaction (17, 61). When too much iron is taken up, as is the case for cells with fur mutations, an increase in oxygen-dependent DNA damage can be detected. The sodA iron box binds MnFur with high affinity (141), but MnSOD synthesis is also affected by five additional global transcriptional regulators. Negative regulators influencing sodA expression include Arc (aerobic respiration control) (189), FNR (anaerobic respiratory control) (86), and IHF (integration host factor), while the products of soxRS (superoxide response) and certain soxQ alleles are activators (36). This multiplicity of regulatory molecules is apparently necessary to minimize oxygen toxicity while cells are assimilating the appropriate amount of iron for growth under either aerobic or anaerobic conditions.

A complete understanding of the iron stimulon will explain many phenomena, including why S. typhimurium fur strains are unable to mount an effective acid tolerance response, which is necessary to survive low-pH stress (63, 64, 65), and why Ent is overproduced in mutants unable to carry out an iron-dependent step in the posttranslational modification of some tRNAs (19). As is the case with fur and sodA, the key genes will presumably be members of several regulons.

A remarkable recent observation is that feo mutations confer a fur-like phenotype. It is hypothesized that the feo uptake pathway can be a major source of iron for Fur-mediated regulation (D. Touati, personal communication). Also, some E. coli strains have additional high-affinity systems for iron import. A number of clinical strains, but not K-12, can obtain iron from hemin; two distinct systems exist, based on hybridization results using an internal fragment of the Shigella dysenteriae heme utilization gene as a probe (M. Mills and S. Payne, personal communication).

MOLYBDENUM ASSIMILATION

Molybdenum is an essential trace element in E. coli and S. typhimurium, as it is for virtually all organisms. Molybdoenzymes of these bacteria play key roles in anaerobic respiration and include formate dehydrogenase-N (fdhGHI), nitrate reductase (narGHJI), trimethylamine N-oxide (TMAO)/dimethyl sulfoxide (DMSO) reductase (dmsABC), TMAO reductase (torA), and a second formate dehydrogenase (H) associated with formate-hydrogenlyase. Molybdoenzymes carry out the adventitious reaction of reducing chlorate to the toxic chlorite anion. Mutations resulting in chlorate resistance are therefore typically located in genes (chl) necessary for general steps in Mo assimilation, i.e., either in Mo uptake or Mo-cofactor synthesis. Current designations for chl genes are provided (174).

Assimilation of Mo, as molybdate ion, MoO₄ ^2–, is similar to that for iron in that an ABC importer transport system is utilized. No molecules analogous to siderophores have been reported, however, and there is also no outer membrane receptor involved. For growth of wild-type cells, medium containing 0.1 μM molybdate is sufficient (173). The high-affinity transport system is induced in medium containing less than 10 nM molybdate. Mutants defective in this system grow normally in medium supplemented to 0.1 mM molybdate, suggesting either that a low-affinity system also exists or that another oxyanion transport system can recognize molybdate when it is present in high concentrations. (In some filamentous fungi, the sulfate transport system also accommodates SeO₄ ^2– and MoO₄ ^2– [93].)

The ABC importer for molybdate is encoded by the modABCD operon (17 min). The modC (chlD) gene encodes the identifying ATPase, and modB (chlJ) specifies an integral membrane protein (98). The other requisite components, the periplasmic binding protein and the second, very hydrophobic, cytoplasmic membrane protein, are the products of modA and modD, respectively (R. Gunsalus, personal communication). Two promoters for the modABCD operon have been identified (S. Rech, U. Deppenmeyer, and R. P. Gunsalus, Abstr. 94th Gen. Meet. Am. Soc. Microbiol., abstr. H3, p. 200, 1994). Transcription from one, located upstream of modA, is regulated by Mo, thereby explaining the Mo-sensitive control of chlD noted (127). A repressor that functions when high intracellular Mo concentrations occur is probably involved. Constitutive low-level transcription occurs from the second promoter, which is located within modA.

The Mo present in molybdoenzymes is associated with molybdopterin (MPT) (93, 158) to form the active Mo cofactor (Fig. 6). The free cofactor is extremely labile; a 45,000-Da carrier protein stabilizes the cofactor and presumably donates it to apoproteins of molybdoenzymes, from which it does not normally dissociate. The Mo cofactor exists in several forms; it may or may not have a ribonucleotide present. In E. coli, MPT exists in the guanine dinucleotide form. The biosynthetic genes for MPT synthesis (moaABCDE and moeAB) are located at 17 min and 18 min, respectively, and the locus for conversion of MPT to MPT guanine dinucleotide (mob) is at 87 min.

Fig. 6Structure of molybdopterin and the cofactor form of MPT guanine dinucleotide.

Source:

Nitrate is the preferred anaerobic respiratory substrate; in its presence, transcription of the nitrate reductase operon is stimulated while that of reductase operons for the other electron acceptors (DMSO, TMAO, and fumarate) is reduced (74, 183). Nitrate also induces the synthesis of the narK product, a nitrate-nitrite antiporter, and formate dehydrogenase-N, which copurifies with nitrate reductase (182). Mo participates in this NarL-mediated regulation in an unknown manner to enhance its efficiency. Thus, under Mo-poor conditions, the ability of nitrate to either induce or repress the five transcriptional units listed above is lost. No regulatory role for Mo cofactor has been determined. A complete description of the molybdoenzymes, their control, and their relationships to other electron transport pathways is found in chapters 17 and 42.