Though an essential trace element, manganese is generally accorded little importance in biology other than as a cofactor for some free radical detoxifying enzymes and in the photosynthetic photosystem II. Only a handful of other Mn2+-dependent enzymes are known. Recent data, primarily in bacteria, suggest that Mn2+-dependent processes may have significantly greater physiological importance. Two major classes of prokaryotic Mn2+ uptake systems have now been described, one homologous to eukaryotic Nramp transporters and one a member of the ABC-type ATPase superfamily. Each is highly selective for Mn2+ over Fe2+ or other transition metal divalent cations, and each can accumulate millimolar amounts of intracellular Mn2+ even when environmental Mn2+ is scarce. In Salmonella enterica serovar Typhimurium, simultaneous mutation of both types of transporter results in avirulence, implying that one or more Mn2+-dependent enzymes is essential for pathogenesis. This review summarizes current literature on Mn2+ transport, primarily in the Bacteria but with relevant comparisons to the Archaea and Eukaryota. Mn2+-dependent enzymes are then discussed along with some speculations as to their role(s) in cellular physiology, again primarily in Bacteria. It is of particular interest that most of the enzymes which interconvert phosphoglycerate, pyruvate, and oxaloacetate intermediates are either strictly Mn2+-dependent or highly stimulated by Mn2+. This suggests that Mn2+ may play an important role in central carbon metabolism. Further studies will be required, however, to determine whether these or other actions of Mn2+ within the cell are the relevant factors in pathogenesis.
Though acknowledged as an essential trace element, manganese is generally accorded no more than marginal importance in the cellular physiology of prokaryotes (or eukaryotes). The only two roles commonly attributed to manganese are as a redox-active cofactor for certain free radical detoxifying enzymes and as the redox-active cofactor for a component of the photosynthetic water oxidizing complex. Otherwise, manganese appears only occasionally in the literature as cofactor for a handful of seemingly unrelated enzymes. Even here, though, the Mn2+-dependent version of an enzyme is generally reported only in an eclectic handful of taxa, the only rationale for which is evolutionary happenstance.
In bacteria that situation is changing, perhaps profoundly. The discovery that the vertebrate atural esistance- ssociated acrophage rotein (Nramp) is a divalent transition metal cation transporter and the concurrent realization that a substantial fraction of all bacterial genomes contain Nramp1 homologs has sparked a resurgence in the study of bacterial transition metal cation transporters. Many though probably not all bacterial Nramp transporters are highly selective for Mn2+. Further, it has become apparent that Mn2+-specific ABC-type transporters are much more widespread than hitherto appreciated. At the same time, enthusiasm over the eccentric physiologies of extremophiles has revealed an increasing number of taxonomically diverse bacteria which survive and prosper apparently without the use of iron. Instead they substitute constitutively high levels of cytoplasmic Mn2+ in its place. The blossoming of microbial genomics has also uncovered additional Mn2+-dependent enzymes. And finally, now that transporter studies indicate many bacteria can vary cytoplasmic manganese levels over several logs of concentration, manganese stimulation of enzymes which are normally considered to employ other cations (such as magnesium) may in fact be a significant everyday phenomenon in vivo.
The emphasis in this review is on ‘emerging themes’ because prokaryotic manganese biology is still young. While recent data engender enthusiastic speculation, solid molecular and/or mechanistic verification is lacking in many cases. Despite increasing breadth and relevance, manganese biology is still eclectic. This review necessarily reflects that situation, in both content and approach. Though our main focus will be Mn2+ transport and regulation in Enterobacteria, we will offer interpretations of existing literature on the transport, biochemistry and physiology of manganese in prokaryotes in broad terms. If nothing else, we hope this will convince readers that the field is vigorous and offers a worthy venue for research.
2 Manganese transporters in bacteria
2.1 Historical perspective on bacterial Mn2+ transport
Widespread interest in bacterial Mn2+ transport is a recent phenomenon. As recently as 1996, that peripatetic surveyor of metal transporters, Simon Silver, was able to write: ‘The presumed single gene for [the Escherichia coli Mn2+ transporter] has not been identified. My colleagues and I earlier summarized the literature on bacterial manganese transport [in 1977 and 1987], and no additional work on E. coli Mn2+ transport … has been reported in many years.’ . Contrast this with assertion a mere 5 years later by Jakubovics and Jenkinson: ‘Determining the relative significance of multiple Mn2+ import systems in homeostasis in individual species presents a major challenge for future research. Significantly, the observation that Mn2+ acquisition by bacteria is linked to virulence in the host suggests that Mn2+ transport is a potential new therapeutic target for control of bacterial infection .’
Though it had been known since the 1960s, largely through pioneering studies by Silver and colleagues, that bacteria possess high affinity Mn2+ uptake activity [3, 4], this unsurprising fact engendered little interest even well into the molecular era after innumerable other bacterial enzyme and transport activities had been linked to cloned genes and biochemically characterized gene products. Manganese is one of the metals involved in the water oxidizing complex of photosynthetic organisms [5–9], and one of the metals involved in the enzymatic dismutation of superoxide radical anion [10–13], but the average bacterial requirement for Mn2+ appeared so modest that the transport question hardly seemed worth revisiting.
Within the last half decade all this has changed. The 1996 discovery that the well known mammalian pathogen resistance locus bcg/lsh/ity encodes the transition metal cation transporter Nramp1 , the realization that a substantial number of bacterial pathogens have Nramp1 homologs of their own, and the subsequent demonstration by both indirect and direct means that many bacterial Nramp1 homologs transport Mn2+ as their primary substrate has rekindled interest in this cation and led to consideration of its possible role in pathogenesis [15–17].
2.2 Manganese transporters in Salmonella
2.2.1 Characterization of MntH as a Mn2+ transporter
The sole Nramp1 homolog in the Enterobacteria is encoded by the ‘yfeP’ gene, in Blattner E. coli genome notation. The E. coli yfeP locus was shown primarily by genetic means to play some role in iron and/or Mn2+ transport and sensitivity . Shortly thereafter the YfeP proteins of both E. coli and Salmonella enterica serovar Typhimurium were shown by direct transport assays to be highly selective Mn2+ uptake systems (Fig. 1) apparently energized by proton symport , similar to the mammalian Nramp1 homolog DCT1 or Nramp2 . The designation ‘YfeP’ was accordingly replaced by ‘MntH’ in both organisms, following the nomenclature championed by Helmann (see Section 2.3.1).
S. Typhimurium MntH and SitABCD are both Mn2+-selective uptake systems.Cloned mntH or sitABCD genes were individually expressed from plasmids in a strain lacking both transporters (SL1344 ΔmntH:kan ΔsitA::cam). Uptake of 10 nM 54Mn2+ in the presence of non-radioactive Mn2+ over 3 min (during which accumulation is linear with time) indicates that each transporter has a KT for Mn2+ uptake of approximately 100 nM. Dose–response curves for inhibition of 100 nM 54Mn2+ uptake by selected heterologous cations are also shown. The affinity of each transporter for ferrous iron, Fe2+, is much poorer, between 10 μM and 30 μM, consistent with KT values determined in independent 55Fe2+ uptake assays (not shown). By contrast ferric iron, Fe3+, is a very poor inhibitor of both transporters. Cd2+ is the most potent inhibitory cation for both MntH and SitABCD (not shown). Zn2+ is a much more potent inhibitor of SitABCD than of MntH (not shown), allowing decomposition of overall Mn2+ uptake by wild-type S. Typhimurium into MntH-mediated and SitABCD-mediated components. Adapted from Kehres et al. [17, 26].
With Nramp1, identification of the physiologically relevant cation(s) was (and remains) controversial. Direct comparison of selectivity for transition metal divalent cations has not yet been performed, although indirect measurements have implicated both Fe2+ and Mn2+. Regardless, because orthologous transporters are present in both pathogenic bacteria and their hosts and because the Nramp1 transporter is a resistance factor for the host the favored hypothesis has been that the pathogen and host transporters compete for cation, especially Fe2+ [19–21]. However, the demonstration that Bacterial Nramp1 homologs from diverse pathogenic as well as non-pathogenic species were highly Mn2+-selective suggested that Mn2+ transport and homeostasis might be relevant to virulence in Salmonella and to infectious disease more generally [15–17]. Subsequently, Jabado and colleagues  demonstrated that Nramp1 could act as a proton-dependent Mn2+ transporter in the host cell phagosomal membrane, emphasizing the relevance of Mn2+.
2.2.2 Characterization of SitABCD as a second Mn2+ transporter
The notion that MntH and Nramp1 might compete directly for divalent cations following macrophage engulfment of a pathogen is an attractive premise. However, ΔmntH mutant S. enterica Typhimurium turned out to be essentially as virulent as wild-type upon oral inoculation of the standard ‘susceptible’ mouse strain BALB/c, the only difference being a modest 1 or 2 days delay in the onset of morbidity and mortality compared to wild-type bacteria. Initial oral virulence studies in the C3H mouse, which unlike BALB/c is genotypically Nramp1+/+, were more suggestive of a role for MntH but not conclusive (Fig. 2 and D.G. Kehres and M.E. Maguire, unpublished observations, see Section 4.3). One of many possible interpretations for this result was that S. enterica Typhimurium carried another Mn2+ transporter capable of complementing the ΔmntH mutation. MntH proteins had already been shown to be accompanied by binding protein-dependent (ABC-type) Mn2+ transporters in both a closely related organism, YfeABCD of Yersinia pestis, and a Gram-positive organism, MntABCD of Bacillus subtilis. Because the sitABCD promoter had a palindromic repeat similar to that present in the mntH promoter, Kehres et al.  therefore characterized the cation selectivity of the S. enterica Typhimurium transporter SitABCD, previously considered a ferrous iron transport system [24, 25]. Although SitABCD can indeed transport ferrous iron, it has a much greater affinity for Mn2+, almost completely mirroring MntH in its Mn2+ affinity (∼100 nM), uptake capacity, and inhibition profile versus other transition metal divalent cations . Experiments with a ΔmntHΔsitABCD double mutant grown overnight in M9 plus glucose medium have demonstrated that these two transport systems are responsible for essentially all Mn2+ uptake but no Fe2+ uptake in wild-type S. enterica Typhimurium.
MntH and Nramp1 genotype both influence virulence.Wild-type and ΔmntH mutant S. Typhimurium SL1344 were inoculated orally into naturally Nramp1−/− BALB/c mice at a dose of ∼5×105 bacteria per mouse (open symbols) or into naturally Nramp1+/+ C3H mice at a dose of ∼5×108 bacteria per mouse (closed symbols). In animals lacking a functional Nramp1, ΔmntH bacteria were essentially as virulent as wild-type. In animals with a functional Nramp1, ΔmntH bacteria were attenuated. D.G. Kehres and M.E. Maguire, unpublished observations.
2.2.3 Additional Mn2+ transport activities in Salmonella
The ΔmntHΔsitABCD double mutant strain still exhibits a small residual Mn2+ uptake activity during overnight growth. Furthermore, a ΔmntHΔsitABCDΔgpmA triple mutant, which is totally dependent on the Mn2+-cofactored phosphoglycerate mutase GpmM for growth (see Section 184.108.40.206), fails to grow in unsupplemented medium but does grow when supplemented with 1 μM Mn2+ (D.G. Kehres and M.E. Maguire, unpublished observations). Thus a residual Mn2+ uptake activity is likely present, the transporter for which remains to be identified. Since complementation of a growth defect would not necessarily require a high transport rate (Vmax), such transport could be through one or more transporters with other or broader cation selectivity. A plausible candidate is the PitA cation-phosphate transporter of E. coli [27, 28], which symports phosphate, proton, and any of several divalent cations, with Mn2+ having the best affinity. Similar transporters have been described in Acinetobacter johnsonii and probably Yarrowia lipolytica and B. subtilis. Alternatively, since there appears to be no evidence to the contrary in the literature, one or more of the ferric Fe3+ iron siderophore systems [32–34] might be able to transport the isoelectronic (high spin 3d5) Mn2+ ion as well. (It appears to be a characteristic of many iron transport studies to ignore all other cations, assuming that iron is the only cation transported and therefore the only cation of physiological relevance.)
In addition to Mn2+ uptake activities, S. enterica Typhimurium also expresses a robust Mn2+ efflux activity, of undetermined KT but with an efflux rate probably comparable to the uptake capacities of MntH and SitABCD (D.G. Kehres and M.E. Maguire, unpublished observations). Efflux appears normal in the ΔmntH or the ΔsitABCD single mutant, suggesting that neither of these transporters is the primary efflux system. One intriguing candidate in both E. coli and S. enterica Typhimurium is the ybiR open reading frame (ORF) whose start codon overlaps the stop codon of the MntR manganese repressor (see Sections 2.5 and 3.3.2) in an ATGA sequence. The closest BLAST hits to YbiR with assigned functions are the Mycobacterium tuberculosis ArsB and ArsA arsenate efflux proteins (Z96072). There is a marginally closer hit to a putative Lactococcus lactis cation transporter (AE006407) (D.G. Kehres and M.E. Maguire, unpublished observations). The YbiR ORF in the S. Typhimurium LT2 genome (AE008734) is currently annotated as homologous to unspecified di- and tri-carboxylate transporters.
2.3 Bacterial Mn2+ transporters in general – the broader picture
2.3.1 Classes of Mn2+ transporters
Primary Mn2+ uptake activities that have been linked to specific gene products across phylogenetically diverse Bacteria and Archaea predominantly involve members of two transporter families – the Nramp class (MntH) and a cation-transporting ABC permease class – Transport Commission numbers TC 2.A.55 and TC 3.A.1.15, respectively, in Saier's terminology [35, 36].
A GenBank™ search shows that MntH class transporters are very common though not ubiquitous in the Bacteria while there are a few examples of this class in the Archaea, including Ferroplasma acidarmanus, Thermoplasma volcanium and acidophilum, Sulfolobus solfataricus, and Natronomonas pharaonis. Other than MntH from E. coli and S. enterica Typhimurium, no Archaeal and only two other Bacterial members of the Nramp family have been characterized to date. The B. subtilis homolog MntH comprises a major Mn2+ uptake activity in that organism . In contrast, the M. tuberculosis homolog Mramp probably transports Mn2+ but clearly also transports significant amounts of both Fe2+ and Zn2+. Since relative cation affinities of transported metals have not been determined in either case, it is not possible to determine which cation transport activities are physiologically relevant.
ABC-type transporters with Mn2+ uptake activity have been characterized in a variety of both Gram-positive and Gram-negative organisms . A GenBank search shows this class to be extremely widespread with about twice as many examples as for the MntH class. A few Archaea also have close homologs, including Methanosarcina and Halobacterium sp., Methanothermobacter thermoautotrophicusΔH, Archeoglobus fulgidus, and Pyrococcus abyssi. It may be noteworthy that, unlike the Bacteria, there is no example as yet of an Archaeal species having both an ABC class and an MntH class of transporter. As with the MntH class, the physiological relevance of Mn2+ uptake compared to that of other transition metals has not in general been determined for any of these ABC class transporters.
As the overall transport selectivity situation clarifies, we suggest that the most reasonable nomenclature is that proposed by Helmann , whereby Mn2+-selective, proton symporting Nramp-type transporters should be named ‘MntH’ and Mn2+-selective ABC-type transporters named ‘MntABC(D)’.
The only other Mn2+ transporter known at present is a P-type ATPase, the MntA transporter of Lactobacillus plantarum (TC 3.A.3.3.3). An apparently similar transporter has recently been reported in the nematode Caenorhabditis elegans. In the interest of unambiguous nomenclature, we suggest that Mn2+-selective P-type ATPases be named ‘MntP’.
2.3.2 Phylogenetic distribution of each class
Though a complete taxonomical enumeration of bacterial Mn2+ transporters is beyond the scope of this review – and would be premature and speculative in any event given that the transport selectivity of most bacterial Nramp and cation-transporting ABC-type systems remains conjectural – a few observations are appropriate. First, multiple Nramp isoforms can exist in a single bacterial genome. Pseudomonas aeruginosa has two distinct paralogs MntH1 and MntH2, both of which transport Mn2+ with affinities of 0.1–1 μM although their complete cation profile has not been determined (D.G. Kehres and M.E. Maguire, unpublished observations). Genes for four distinct Nramp isoforms can be found in the genus Burkholderia among the sequence reads of the incomplete B. mallei, B. pseudomallei, and B. cepacia genomes, though at present no more than three have been found in any one species. The transport specificities of all these paralogs remain to be determined.
A partial phylogeny of Bacterial Nramp homologs has been presented by Cellier and colleagues [15, 41]. It should further be noted that there are two other paralogous protein families of unknown transport function that are more closely related to Nramp transporters than any other known proteins. Homology extends throughout the length of the protein for all three families, and predicted membrane topologies are equivalent. Each of the other two families consist of at least 10 members among currently released microbial genomes. Gros and colleagues  and Liu and Culotta  have identified two conserved motifs in ‘authentic’ Nramps (Table 1). These motifs are necessary for activity, but it is not known if they are involved in cation binding. The additional two families of related proteins all have conserved motifs at the same relative location within the sequence, but the motifs are distinct for each family. There is precedent for each new family co-existing in the same species with an authentic Nramp.
Representative species (GenBank protein designation)
Family I (Nramp/MntH)
S. enterica Typhimurium (MntH)
E. coli (MntH)
B. subtilis (MntH)
M. tuberculosis (Mramp)
M. leprae (Lramp)
Clostridium acetobutylicum (CAP0063)
N. meningitidis (NMB0227)
Haemophilus influenzae (HI1728)
B. subtilis (YcsG)
S. aureus (SA1432)
C. glutamicum (NCgl0963)
C. acetobutylicum (CAC3329)
C. acetobutylicum (CAC0685)
M. leprae (ML2667)
aUpper case letters represent highly conserved residues while lower case letters reflect the most frequently found amino acid at less conserved positions. See text for discussion.
The cation-selective ABC-type transporters are difficult to distinguish from other ABC-type transporter subclasses by BLAST homology, and individual bacterial genomes often encode several dozen ABC-type transporters. Thus, the true distribution of Mn2+-selective ABC-type transporters is hard to estimate. However, as noted above, both the ABC class and the Nramp class are broadly distributed among bacterial taxa, each appearing in one third to two thirds of all species. There are also numerous precedents for both Nramp-type and ABC-type Mn2+ transport activities existing in a single organism –B. subtilis (MntH and MntABCD) and S. enterica Typhimurium (MntH and the now inaptly named SitABCD) are two phylogenetically diverse examples already mentioned.
Conversely, several completed bacterial genomes (including among the Proteobacteria the pathogens Helicobacter pylori and Vibrio cholerae) appear not to encode members of any of the three known classes. Since these organisms do encode Mn2+-dependent enzymes, novel Mn2+ transporter types likely remain to be discovered.
2.3.3 Mechanistic and inherent chemical aspects of Salmonella Mn2+ transport
The affinities of S. enterica Typhimurium MntH and SitABCD for Mn2+ and for inhibitory cations follow surprisingly similar patterns [17, 26]. These selectivities appear to be reflected in the behavior of other bacterial Mn2+ transporters to the extent that they have been characterized. Though not fully understood yet, these patterns are worth noting because they reflect the underlying thermodynamics and kinetics of transition metal–protein interactions and thus hold clues to the transport mechanism. We will focus on MntH and SitABCD, referring to other transporters when relevant.
Since phagosome acidification is characteristic of the S. enterica Typhimurium–macrophage interaction, and since the default assumption about Nramp transporters is that they are energized by proton symport, Kehres et al. determined the pH dependence of Mn2+ uptake and cation inhibition for both MntH  and SitABCD . For Mn2+ itself, KT was strikingly invariant between pH 8.2 and pH 6.0 for both transporters. In contrast Vmax increased for MntH as pH decreased while Vmax for SitABCD showed the opposite response, being high at pH 8.2 and virtually absent below pH 6.5. Thus the two transporters may be optimized by evolution for different cellular environments or tasks.
One feature shared with the great majority of Mn2+ transporters is the high affinity of both MntH and SitABCD for the closed-shell 4d10 cation Cd2+. The Ki of Cd2+ inhibition for both transporters is ≤1 μM, only 3–10-fold poorer than the KT for Mn2+. In Synechocystis sp. PCC 6803 this same ratio applies although the affinities are about 10-fold poorer for each cation . A S. enterica Typhimurium strain overexpressing MntH is more susceptible to Cd2+ toxicity than wild-type cells suggesting that Cd2+ is actively transported  although this has not been directly measured. By contrast, the closed-shell 3d10 cation Zn2+ inhibits with a rather poor Ki of around 3–10 μM for SitABCD and 300 μM for MntH (and can be used to distinguish each transporter in a cell expressing both ). Such selective sensitivity to Cd2+ is also found in Nramp from M. tuberculosis and in many ABC-type Mn2+ transporters [38, 44], although again for the latter class whether Cd2+ is transported or merely an inhibitor has not generally been resolved. The P-type ATPase MntA of L. plantarum can also transport Cd2+. As these three classes of transporter bear no ancestral or structural resemblance to one another, yet none has successfully evolved a means of discriminating toxic Cd2+ from essential Mn2+, some interesting cation–ligand chemistry awaits discovery.
For ferrous iron, Fe2+, the KT for transport (or equivalently, Ki for inhibition of Mn2+ transport) strikingly decreases from pH 8.2 to pH 6.0 for both MntH and SitABCD. Clearly Fe2+ is interacting differently from Mn2+ in at least some of the liganding environments present during a transport cycle, and this different mode of interaction results in the identical trend of pH dependence for two unrelated transporters. Another aspect of cation–ligand chemistry shared by MntH and SitABCD is that the Ki for ferric iron, Fe3+, shows the same pH invariance that Mn2+ exhibits, although Fe3+ is a relatively weak inhibitor due to its trivalent charge. Interestingly, with a 3d5 configuration, Fe3+ is isoelectronic with Mn2+ whereas Fe2+ is not. A connection between d electron configuration and pH dependence of Ki is supported by the behavior of the remaining inhibitors. Co2+, Ni2+, and Cu2+, which like Fe2+ have non-symmetric 3d shells, also show decreases in Ki with decreasing pH, while Zn2+, which although not isoelectronic with Mn2+ does have a symmetric 3d10 shell, has a Ki almost unaffected by pH. Clearly the chemistry of Mn2+ selectivity owes something to d electron symmetry in both of these structurally unrelated classes of transporters.
2.4 Regulation of Salmonella Mn2+ transporters
Transcriptional regulation of Salmonella Mn2+ transporters has only been partially clarified to date. MntH and SitABCD are both repressed in the presence of Mn2+ or iron. In addition, MntH (but apparently not SitABCD) is activated in the presence of H2O2. Beyond these relatively unsurprising features, though, both MntH and SitABCD appear to be regulated by growth conditions, in a complex fashion which still awaits molecular characterization. Further, MntH and SitABCD both respond transcriptionally upon invasion of Salmonella into macrophages and the bacterium's establishment in a ‘Salmonella containing vesicle’ (SCV) [45–47]. Whether these responses involve more than the interaction of metal, peroxide, and growth regulators is a question just beginning to be addressed.
2.4.1 Regulation of MntH and SitABCD by cations
Both MntH  and SitABCD  are subject to repression by the metal-dependent transcription factors Fur and MntR, each of which can in principle use either Fe2+ or Mn2+ as cofactor as described below (Section 3.3.2). In the case of MntH, regulation exhibits the expected pattern (Fig. 3). That is, Fur repression primarily involves Fe2+ and MntR repression primarily involves Mn2+. SitABCD regulation appears to be similar (A. Janakiraman and J.M. Slauch, unpublished observations). It is not clear why transporters with high selectivity for Mn2+ should also be repressed by iron. Given that both classes of transporter are repressed by both cations, it seems unlikely that it is adventitious. Possibly the simultaneous elevation of cytoplasmic Fe2+ and Mn2+ levels is harmful for reasons not yet fathomed. Conceivably there could be competition for sites designed specifically for one or the other metal, or perhaps there are undesirable redox interactions between complexes of the two metals, or perhaps the total combined concentration of the two must be kept within the overall buffering capacity of the cell.
Mn2+ and iron regulate MntH expression.Mn2+ and iron are able to repress expression of β-galactosidase from a mntH::lacZ reporter. The reporter is carried on a single-copy plasmid and introduced into a wild-type background to avoid artifacts of either copy number or absence of functional MntH. Experiments in strains lacking either the Fur (A) or MntR (B) regulatory circuits still show cation repression. Mn2+ repression is mediated through MntR since mutation of the MntR binding site (‘MntR motif’) in the mntH promoter abolishes Mn2+ repression. In contrast, iron can act through either or both of Fur and MntR since it represses even in the absence of either Fur or the Fur binding site in the mntH promoter (Fur motif). In practice however, the iron concentration needed to repress via MntR is significantly higher so that physiologically, Mn2+ acts through MntR while iron acts through Fur .
2.4.2 Regulation of MntH by hydrogen peroxide
Though SitABCD and MntH exhibit virtually indistinguishable metal repression, MntH but not SitABCD expression is activated by reactive oxygen. Transcription of mntH is activated only by hydrogen peroxide, acting through OxyR [17, 48], superoxide has no effect at all [15, 17]. Both of these activation features are puzzling. Since the expression patterns, substrate affinities, and transport capacities of MntH and SitABCD are otherwise similar, there is no obvious physiological rationale why only one should respond to reactive oxygen stress. MntH is more active than SitABCD in acidic culture media , but peroxide activation shows no apparent pH dependence (D.G. Kehres and M.E. Maguire, unpublished observations). Nor does the distinction that MntH is energized by the proton gradient and SitABCD by ATP seem consequential. In rapidly growing cells ATP levels are generally more than adequate for transport purposes, and even under relatively ATP-depleted conditions, the energetic cost of transcribing and translating either transporter would likely outweigh any differential energy economy in their function.
The fact that only hydrogen peroxide activates MntH suggests that newly taken up Mn2+ may have some other primary purpose than cofactoring the Mn2+-dependent SodA. Superoxide damage is not irreversible. Its primary toxicity is to inactivate 4Fe–4S cluster proteins, which inhibits growth since most respiratory enzymes contain 4Fe–4S clusters; however, inhibition is reversible and need not ultimately be fatal . By contrast, hydrogen peroxide can be deadly even on short time scales due to processes like the Fenton reaction (in the presence of Fe2+) which generate hydroxyl radicals. Ways in which Mn2+ might mitigate acute peroxide damage will be discussed in Section 3.2 below.
2.4.3 Regulation of MntH and SitABCD by growth state
In addition to metal and peroxide regulation, transcription of mntH and sitABCD is influenced by various aspects of cellular growth. As with metal repression, both Mn2+ transporters exhibit broadly equivalent growth dependence. Basal expression is low during initial growth, then increases as cells progress through the mid-exponential to stationary phase (Fig. 4). None of the demonstrated regulators Fur, MntR, or OxyR, is individually required for mntH upregulation during growth . RpoS, the stationary phase sigma factor, also has no effect on mntH transcription  but sitABCD upregulation is at least partially RpoS-dependent (A. Janakiraman and J.M. Slauch, personal communication). Beyond this, growth phase dependence has yet to be linked to any of the numerous other transcription factors known to play a role in the complex transition from exponential to stationary phase.
MntH expression depends on growth state.Wild-type S. Typhimurium SL1344 carrying a mntH::lacZ reporter on a single-copy plasmid was grown in M9 salts supplemented with 0.2% glucose starting at OD600=0.025 (∼2.5×107 cells ml−1). As inferred from β-galactosidase activity, total MntH levels in the culture are low during early exponential phase, increase markedly during the final 1–2 generations of growth, then no longer increase once the cells enter stationary phase. D.G. Kehres and M.E. Maguire, unpublished observations.
Both transporters share the additional property that their expression is lower in complex media (i.e. LB) than in minimal media. The magnitude of upregulation in minimal media is greater in faster growing cultures (those with richer carbon sources, e.g. glucose or casamino acids) than in slower growing cultures (those with poorer carbon sources, e.g. acetate or succinate) (D.G. Kehres and M.E. Maguire, unpublished observations). With mntH, strains lacking the fur regulatory circuit show elevated transporter expression even in LB, but individual mntR and oxyR mutants show essentially wild-type media dependence . Comparable studies have not yet been performed with SitABCD.
Overall, these growth-related phenomena are predictable based on known properties of the mntH and sitABCD promoters. Complex media probably have higher trace Mn2+ and Fe2+ levels than minimal media, and faster growing cells probably produce more endogenous reactive oxygen than slower growing cells. Whether the three known regulatory circuits suffice to explain growth-dependent transcription, though, is far from clear at the present moment. However, since deletion of fur but not mntR affects basal regulation of mntH, iron but not Mn2+ or superoxide appears to be the dominant repressor, which is somewhat of a puzzle but another indication that the cellular regulation of iron and Mn2+ are intimately connected.
2.4.4 Regulation of MntH and SitABCD by the phagosomal environment
Two nearly universal features of infectious disease are that the host attempts to restrict the growth of pathogens by depriving them of essential trace nutrients, notably iron, and that the host attempts to kill pathogens with toxic compounds, notably hydrogen peroxide. Thus it is not surprising that both mntH expression  and sitABCD expression (M.L. Zaharik, D.G. Kehres, M.E. Maguire and B.B. Finlay, unpublished observations) are upregulated when S. enterica Typhimurium invades a murine macrophage. Further, since Nramp1 almost certainly serves to deplete the phagosomal compartment of transition metals, it is not surprising that expression of the bacterial Nramp1 homolog mntH is higher in Nramp1+/+ macrophages than in Nramp1−/− macrophages (Fig. 5) . While the compartmental complexity of the Salmonella-macrophage system makes it difficult to manipulate conditions with anything approaching the control obtainable in bacterial culture, clearly one of the major goals of bacterial Mn2+ research over the next few years will be to explain in vivo mntH and sitABCD expression in light of the in vitro defined molecular circuitry.
MntH expression depends on the Nramp1 status of an invaded macrophage.Wild-type S. Typhimurium SL1344 carrying a mntH::lacZ reporter on a single-copy plasmid was allowed to invade RAW264.7 macrophages which either contained or lacked a transfected murine Nramp1 gene. Expression of mntH is induced within 1–2 h in both cases, but induction is much more potent in the presence of the host mntH homolog Nramp1 than in its absence. β-Galactosidase was measured by luminometry using a fluorescent cell-permeable substrate. Adapted from Kehres et al. .
2.4.5 Regulation of bacterial Mn2+ transporters in other bacteria
As in Salmonella, Mn2+ transporters in other bacteria are generally subject to transcriptional control by substrate metals and by reactive oxygen. The molecular details are similar in some cases, intriguingly different in others. In many bacteria, a correlation between growth state and cytoplasmic Mn2+ levels as seen in Salmonella is also apparent. Rapidly growing cells tend to contain less Mn2+ than slowly growing or stationary cells, though this has yet to be related to transcriptional regulation in most cases. However, since data on these other forms of regulation are scant, we will confine our discussion to metal regulation.
Regulation of bacterial Mn2+ transporters by substrate metals was first demonstrated in B. subtilis, where the Mn2+ binding MntR protein represses the Nramp-type transporter MntH under high Mn2+ conditions, but still allows the ABC-type transporter MntABCD to be active under low Mn2+ conditions . MntR is a member of the DtxR superfamily of transcription factors, whose main physiological function previously had been thought to be environmental regulation of diphtheria toxin and similar agents in response to transition metals like iron. However, the rapid discovery of additional Mn2+-responsive DtxR homologs in other bacteria, among them the TroR repressor of TroABCD in Treponema pallidum, the ScaR repressor of ScaACB in Streptococcus gordonii, and the Enterobacterial MntR repressors first in E. coli and then in Salmonella, suggests that the MntR family is clearly a second major branch within the DtxR superfamily. Based on present evidence, MntR homologs are the most widespread class of Mn2+-dependent Mn2+ transporter regulators in bacteria [38, 53].
While their phylogenetic distribution does not appear as extensive as that of the MntR family, two other classes of metal-dependent transcription factors can regulate Mn2+ transporters. Though it responds more strongly to Fe2+ than to Mn2+, Fur can clearly regulate Mn2+ transport loci in E. coli and S. enterica Typhimurium . Fur also regulates the Mn2+- and Fe2+-transporting YfeABCD system in Y. pestis. Since Fur homologs and ABC-type putative cation transporters of unknown substrate selectivity are found in most Proteobacteria, regardless of whether they contain an Nramp homolog, Fur regulation of Mn2+ transporters may be a more general phenomenon than currently appreciated. Finally, the recent discovery that ManR–ManS regulates mntCAB in Synechocystis 6803  provides the first example – but likely not the last – that Mn2+ transport can also be regulated by two-component phosphorelay systems.
2.5 Current perspectives on bacterial Mn2+ transport regulation
Now that attention is being paid to it, Mn2+ transport in bacteria turns out – not surprisingly – to be as complex and redundant as iron transport has long been renowned for being. Beyond the realization that most if not all bacteria express multiple Mn2+ uptake and efflux systems, another significant fact is now apparent. The impression endemic in the literature is that intracellular Mn2+ levels in most cell types are low, on the order of 10 μM. But when Mn2+ uptake systems are expressed, de novo 54Mn2+ uptake alone (which ignores pre-existing Mn2+) routinely yields cytoplasmic 54Mn2+ levels of ≥300 μM in both E. coli and S. enterica Typhimurium [17, 26]. Levels in the mM range can readily be achieved under some conditions (D.G. Kehres and M.E. Maguire, unpublished observations) – far higher, incidentally, than any iron concentration reported under a variety of conditions.
Thus Enterobacteria, which have never been expected to require Mn2+ for much more than the SodA superoxide dismutase, can vary [Mn2+]cytoplasmic over at least two orders of magnitude (3–10 μM up to 1–3 mM), and can do so on a time scale as short as a minute based on documented uptake and efflux rates . Therefore, they would almost certainly do so purposefully under appropriate environmental conditions – and so by extension might the majority of bacteria. Explaining this remarkably wide dynamic range is the most significant challenge in the field of prokaryotic manganese biology today.
Control of Mn2+ transporter expression by metal-cofactored regulators is, not surprisingly, the most general pattern observed among bacteria as a whole. Activation of specific transporters in response to reactive oxygen intermediates is also common and may be nearly universal given that mutation of Mn2+ transporters confers increased sensitivity to peroxides or superoxide in many bacteria whose transcriptional circuitry has yet to be examined. Beyond that, though, there is a pervasive trend among bacteria that cytoplasmic Mn2+ content varies approximately inversely with growth rate. Rapidly growing cells have a relatively low (tens of μM range) Mn2+ content, uptake increases as growth slows, and stationary cells have typically at least an order of magnitude more Mn2+, if not much more (hundreds of μM to mM range). Whether this growth state regulation involves more than an intertwining of responses to metal depletion and redox stress is unknown. However, it seems probable that at least some additional transcriptional regulators will eventually be ascribed to growth regulation, be they widespread or peculiar to individual species.
Mn2+ homeostasis involves multiple transporters and is tightly regulated by multiple inputs in bacteria. That alone implies that Mn2+ must play specific roles in bacterial physiology. Further, a correlation between Mn2+ content and growth state implies that Mn2+ transporters do more than simply supply steady housekeeping levels of Mn2+ to cofactor a small number of redox and other incidental enzymes. This will be the subject of the remaining two sections of this review.
3 Biochemistry of manganese
3.1 Physical chemistry of manganese
Most of the biochemistry of Mn2+ can be explained by two properties: it is redox-active, and it is a close but not exact analog of Mg2+.
3.1.1 Redox biochemistry of manganese
Like iron, Mn2+ can cycle readily in vivo between the 2+ and 3+ oxidation states . Two redox chemistry points are important in this context. The first and more general point is that the reduction potential of any molecule depends on its ligand environment. The second, more specific point is that manganese is ‘less reducing’ than iron under most biological conditions.
The relative reduction potentials of a transition metal and a potential target molecule, which is what determines whether the redox reaction can actually occur or not, depend critically on the particular liganded form each assumes in vivo. As an inorganic example, ‘free’ (solvated) Mn2+ has a reduction potential too high to reduce H2O2 in aqueous solution; however, replacing one or two inner hydration shell waters with bicarbonate results in rapid catalase activity by a Mn2+-dependent disproportionation reaction [56, 57], at least in vitro. As an enzymatic example, an important difference among the iron, manganese, and copper/zinc-cofactored superoxide dismutases is how the precise liganding environment of each protein's cation binding site ‘fine tunes’ the reduction potential of the appropriate metal with respect to that of superoxide so as to make both halves of the reaction cycle – metal oxidizing and metal reducing – thermodynamically feasible [58–60].
The second more specific redox chemistry point is that manganese has less reducing potential than iron under most biological conditions. This is due primarily to the different 3d electron occupancies of the two cations. Qualitatively, Mn2+ (3d5) is stabilized relative to Mn3+ (3d4) whereas Fe3+ (3d5) is stabilized relative to Fe2+ (3d6) because the symmetry of the half-filled d5 shell confers inherent thermodynamic stability. Quantitatively, the electrochemical consequence is that Mn3++e−→Mn2+ has a standard reduction potential of +1.51 V while Fe3++e−→Fe2+ has a standard reduction potential of +0.77 V. The more positive the reduction potential, the more likely the oxidized form of the metal is to accept an electron and thus be reduced. Therefore, starting from the relatively reduced cation, Mn2+ is less likely than Fe2+ to donate an electron to and thereby reduce some other molecule. A useful way to remember this is to recall that free ferrous iron, Fe2+, is extremely toxic in the presence of hydrogen peroxide because Fe2++H2O2→Fe3++OH−+OH free radical (the Fenton reaction), whereas the analogous reaction with Mn2+ essentially does not occur.
There are two consequences of this redox chemistry. First, the important similarity between Mn2+ and Fe2+ is that their intrinsic reduction potentials are close enough to those of many common biological compounds that each metal can be recruited, by careful choice of liganding environment, to perform biologically useful redox catalysis. Second, the critical difference between the two metals is that the higher reduction potential of Mn2+ renders the free (i.e. solvated) Mn2+ innocuous under conditions (notably in an aerobic environment) where free Fe2+ would actively generate toxic free radicals. Cells can thus tolerate very high cytoplasmic concentrations of Mn2+ with essentially no negative redox consequences. This is not the case with iron or with any other biologically relevant redox-active metal.
3.1.2 Manganese as a surrogate for magnesium
The second property of Mn2+ that bears on biochemistry is that it is a close but not exact surrogate for Mg2+. The greatest similarity is in the context of structure. Mg2+ is an ideal ‘structural’ cation for biological molecules (notably phosphorylated ones like nucleic acids and many intermediary metabolites), because it completely lacks d electrons and its 2s22p6 electron configuration confines it to a strict octahedral liganding geometry with liganding bond angles very close to 90°. This geometry is useful in organizing the conformations of complex compounds or macromolecules. At the same time, the lack of d electrons means there is very little covalent interaction between Mg2+ and its ligands. Thus Mg2+ is a ‘labile’ and rapidly exchangeable cation that does not interpose itself in the way of other close intermolecular interactions. Mn2+, with its relatively similar ionic radius and the relatively minor involvement of its stable symmetric 3d5 shell electrons in bonding, readily exchanges with Mg2+ in most ‘structural’ environments and exhibits much of the same octahedral, ionic, ‘labile’ chemistry.
The similarity is poorer in the context of catalysis however. Inherent stability notwithstanding, the 3d5 electrons of Mn2+ do interact to some extent with electrophilic ligands. Thus Mn2+–ligand bonds are in general much more flexible than Mg2+–ligand bonds, in both length and angle. When Mn2+ replaces Mg2+ in a ‘catalytic’ environment, in an enzyme, this becomes significant. Mn2+ is much better than Mg2+ at maintaining energetically stabilizing interactions with the target molecule and enzyme amino acid residues through the distortions of the original bonding geometry that occur in the progression from bound substrate to transition state to bound product. (In the classical sense Mn2+ is better at lowering the activation energy of the process being catalyzed.) Thus substituting Mn2+ in the active site of a nominally Mg2+-cofactored enzyme quite commonly results in improved enzyme efficiency (see Section 220.127.116.11).
And finally – since this is primarily a transport review – the similarity between Mg2+ and Mn2+ appears to be poorest in the context of transport. Mg2+ has by far the greatest charge density of any biologically relevant divalent cation [61–63], and as a result solvated ‘free’ Mg2+ holds onto its inner hydration shell waters avidly [64, 65]. Just as H3O+ is often written in place of H+ to more accurately reflect the predominant solvated form of a proton, we ought to write Mg(H2O)62+ in place of Mg2+ to reflect the solvated state of divalent magnesium. (A useful way to remember this is to recall how often Mg2+ is involved in ‘oriented water’ reactions, where the actual catalysis is performed by a solvent molecule which owes its precise positioning to an adjacent active site cation [66, 67].) Recognition of the fully hydrated cation, Mg(H2O)62+, is essential, for example, to the function of the major prokaryotic Mg2+ uptake transporter CorA . All other biologically relevant divalent cations, including Mn2+, have much lower charge densities and do not retain the full inner hydration shell to any significant extent. Moreover, whereas Mg2+ complexes exclusively display a hexacoordinate octahedral geometry, Mn2+ and other divalent cations can also assume coordination states with more than six ligands, necessitating a geometry other than octahedral. This may largely explain what would otherwise be a puzzle given the interchangeability of the two cations in many contexts; that is, Mn2+ transporters do not have any appreciable affinity for Mg2+, either as a substrate or as a competitive inhibitor, nor do Mg2+ transporters have any appreciable affinity for Mn2+.
3.2 Mn2+ and detoxification of free radicals
3.2.1 Mn2+-dependent enzymatic detoxification of free radicals
The two best known roles for Mn2+, which each involve its activity as a redox-active cofactor in different contexts, will not be discussed here at any length. The role of Mn2+ in the photosynthetic water oxidizing complex [5–9] is outside the scope of this review. The detoxification of superoxide by Mn2+-cofactored superoxide dismutase (SodA), the most phylogenetically widespread Mn2+-dependent enzyme [69–73], has also been extensively reviewed [12, 74–76]. We will only note that Mn2+-dependent superoxide dismutases are widespread in both the Bacteria and the Archaea [77–79]. Participation of Mn2+ in other free radical detoxifying processes has received little attention until recently, but a number of new roles, both enzymatic and non-enzymatic, now merit consideration.
A family of Mn2+-dependent enzymes selective for H2O2 have been termed ‘non-heme catalases’ since they are structurally and mechanistically unrelated to conventional catalases which are cofactored by iron bound in a prosthetic heme group. The initial non-heme Mn2+ catalase was described in L. plantarum [80, 81], and homologs were soon found in the phylogenetically distant Bacterium Thermus thermophilus and the Archaeal Pyrobaculum calidifontis. With the recent report of a homolog, KatN, in S. enterica Typhimurium , involvement of Mn2+ in non-heme-dependent detoxification of H2O2 may be more than an idiosyncrasy of a few species. However, the physiological significance of KatN in organisms like S. enterica Typhimurium that also express conventional catalases is unknown.
In addition to enzymes whose function is direct detoxification of oxygen free radicals, a family of catabolic heme enzymes called Mn2+ peroxidases couples the redox activity of H2O2 to degradation of nutrients such as lignin via oxidation of a manganese bound to a propionate side-chain of the heme group. These have thus far only been identified in lignolytic fungi, initially as the MnP series of isozymes in Phanerochaete chrysosporium [85–90]. A source of confusion is that the term Mn2+ peroxidase has also been applied to the Mn(II)→Mn(III) oxidizing activity of CpeB from the bacterium Streptomyces reticuli, a conventional iron-cofactored heme catalase, best known for its homology to the KatG catalase-peroxidase of M. tuberculosis. This enzyme contains a probably adventitious Mn2+ oxidizing activity in its C-terminus, at a site distinct from the iron–heme active center. Fungal MnP-type Mn2+ peroxidase proteins show no sequence homology to CpeB. Given the number of bacterial genomes sequenced to date, the possibility that any significant number of bacterial species have acquired catabolic MnP-type Mn2+ peroxidases seems remote.
18.104.22.168 Reactive nitrogen detoxification
A virtually unexplored question is whether Mn2+ may be involved in detoxifying reactive nitrogen as well as reactive oxygen. Salens, synthetic chelators that are derivatives of N,N′-bis(salicylidene)ethylenediamine chloride, form complexes with Mn2+ that show intriguing pharmacological efficacy as combined superoxide dismutase/catalase mimics [93, 94]. They have recently been shown, in their oxo forms, to oxidize nitrous oxide (NO) to the more benign nitric oxide (NO2), and nitrite (NO2−) to the more benign nitrate (NO3−) in vitro . While salen–Mn2+ complexes are of course not enzymes, bacteria obviously could synthesize similar compounds. This leads us to the next section.
3.2.2 Non-enzymatic detoxification of free radicals by Mn2+
The chemical basis for scavenging of superoxide radical by Mn2+ complexes in vitro was established by Fridovich's laboratory in 1982 , and the basis for bicarbonate-dependent Mn2+-catalyzed disproportionation of hydrogen peroxide in vitro was subsequently established by Stadtman and co-workers [57, 97, 98]. The latter group also found that this presumably beneficial activity could have deleterious consequences in that Mn2+–bicarbonate complexes also accelerated the oxidation of amino acids, most likely through direct Mn2+ coordination of an amino acid with two or three bicarbonates plus an H2O2 molecule in an octahedral reaction complex. The point of such detail is to demonstrate how straightforwardly this could be envisioned taking place inside a bacterial cell. Transition metals in vivo are invariably sequestered in complexes. While many doubtless involve small metabolites or large macromolecules whose presence has nothing to do with Mn2+ per se, many if not most organisms synthesize metal-specific chelating ‘metallochaperone’ proteins . Thus where ‘free’ transition metals are abundant, it is not difficult to envision that specific metal-dedicated small chelators might be synthesized as well, which would serve to control and direct that metal's redox reactivity.
While not commonplace, there is precedent for the in vivo abundance of ‘free’ transition metals, most specifically Mn2+. It has long been known that the bacterium L. plantarum is essentially devoid of iron and heme but maintains a remarkably high (35 mM) cytoplasmic concentration of Mn2+. It is believed that much of this Mn2+ is ‘free’ in a chemical sense and substitutes functionally for iron and heme in as yet chemically undefined ways. Interestingly, two mammalian pathogens have recently been shown to be chemical ‘extremophiles’ of the same sort, showing no apparent requirement for iron but maintaining cytoplasmic Mn2+ at high levels: Borrelia burgdorferi and Streptococcus suis. Although the existence of non-enzymatic Mn2+ redox chemistry is largely inferred in these ‘high-Mn2+’ bacteria, evidence that it actually plays a significant role in reactive oxygen tolerance has been suggested experimentally in the case of more usual, iron–heme containing, bacteria such as Staphylococcus aureus and Neisseria gonorrhoeae.
Unfortunately, no new mechanistic studies on the chemistry of non-enzymatic free radical detoxification have been reported in recent years, and while it seems plausible that bacteria should have learned to harness the redox chemical potential of free transition metals and to vary the nature and concentration of dedicated biogenic metal chelators in response to changing metabolic and environmental needs, how or even whether they actually do so remains to be established.
3.3 Mn2+ and general metabolism
3.3.1 Mn2+-dependent proteins
While its role in free radical detoxification is well documented, association of Mn2+ with proteins involved in other metabolic processes is more haphazard. The major impediments in understanding involvement of Mn2+ are, first, that relatively few Mn2+-dependent proteins are known, and second, that many proteins reported to be Mn2+-dependent can also be cofactored by other cations, rendering the relevance of Mn2+ to their actual physiological function unclear. Until recently Mn2+-related bacterial proteins have been so few in number and so eclectic in function that it was difficult to do much more than enumerate them. Due in part to the increasing interest in Mn2+ arising from its involvement with Nramp1, the scarcity problem is probably well on its way to being alleviated. In anticipation that discovery of Mn2+-related proteins in the more general sense will continue to accelerate, we propose classifying them into three broad categories (Table 2). Proteins involved in regulation will be discussed in Section 3.3.2, enzymes involved in a variety of as yet unrelated metabolic processes will be discussed in Section 3.3.3, and enzymes specifically involved in carbon metabolism will be highlighted in Sections 3.3.4 and 3.3.5.
A minority (?) of phylogenetically diverse Bacteria
Transcription factor (mntR)
Repress Mn2+ uptake transporter expression
Homologs in diverse Bacteria, extent unknown
[16, 48, 50–52, 103]
ppGpp hydrolase (spoT)
Hydrolyze RNA synthesis regulator ppGpp
Practically ubiquitous in Bacteria
Protein phosphatases (prpA and prpB)
Dephosphorylate many cellular proteins, may dephosphorylate some two-component signal transduction systems in Bacteria. Overall function unknown. Physiological relationship to Mn2+-dependent protein kinases unknown.
All cells, highly conserved between prokaryotes and eukaryotes
Synthesize osmoprotectant putrescine from agmatine (decarboxylated arginine)
Enterobacteria, most other Bacteria
Aminopeptidase P (pep and pepQ)
Hydrolyze atypical X-Pro sequence
Many Gram-negative bacteria
[75, 174–178, 263]
Catalyzes interconversion of 3-phosphoglycerate and 2-phosphoglycerate
Enterobacteria, other Bacteria and plants
[191, 264, 265]
Fructose-1,6-BP phosphatase (glpX)
Convert fructose-1,6-BP to fructose-6-phosphate
Enterobacteria, extent otherwise unknown
B. Other Mn2+-dependent enzymes in Bacteria
Synthesis of cyclic AMP
Aromatic hydrocarbon metabolism
Oxidation of catechols and other aromatics
A. globiformis and similar enzymes in many soil bacteria
Intermediary metabolic enzymes that process PEP and pyruvate
Enolase, PEP carboxylase, PEP carboxykinase, pyruvate kinase, malic enzymes. All are highly Mn2+-activated, but may not be totally Mn2+-dependent.
The fact that Mn2+ can participate in a particular enzymatic reaction does not of course automatically mean that Mn2+ is the physiologically relevant cation. Such a conclusion requires careful analysis of two factors. First, activity of a protein with a variety of cations must be tested. Second, the concentration dependence of active (or inhibitory) cations must be determined.
For example, both MntH and SitABCD can clearly transport Fe2+ but not Mg2+, Ca2+, or Co2+, demonstrating that while they are not limited to a single substrate, they are highly selective for only a subset of those possible substrates. Demonstrating that any particular transporter can mediate flux of cation Xn+ does not mean that Xn+ is the physiologically relevant cation, even where homologs in other systems have previously been shown to be physiologically relevant transporters for that cation. The lack of ability to transport a substrate is clearly evidence of a lack of physiological relevance, but a positive finding of transport is not sufficient to prove physiological relevance until multiple substrates have been tested.
Once the spectrum of substrates is known, it is still necessary to determine a protein's activity as a function of cation concentration. Without such measurements it is not possible to assign physiological relevance to any particular substrate. For example, we have argued that the ability of MntH and SitABCD to transport Fe2+ is irrelevant in the physiological sense because the affinity of the transporters for Fe2+ (≥30 μM) is far above any reasonable environmental concentration of free Fe2+ that would be encountered by a living S. enterica Typhimurium [17, 26]. In contrast, an affinity of 100 nM for Mn2+ is in line with the apparent environmental concentrations of Mn2+ in biological systems. Similar analysis of the concentration dependence of the interaction between MntH or SitABCD with many other divalent cations leads to the conclusion that Mn2+ is the most physiologically relevant substrate for each [17, 26].
In the discussion below, the question of actual Mn2+dependence will be addressed on a case by case basis, as the evidence permits. In most cases however, the determination of physiological relevance awaits additional and careful biochemical analysis of each individual protein. Thus, unambiguously calling many proteins ‘Mn2+-dependent’ will likely remain problematic for some time.
3.3.2 Mn2+-dependent regulatory proteins
Mn2+ has now been implicated in the regulation of numerous cellular processes. It is important to realize that, unlike many small agents, Mn2+ could potentially regulate both ‘statically’ and ‘dynamically’. In transcriptional regulation for example, Mn2+ functions statically as a ligand that organizes the DNA binding activity of proteins like Fur and MntR (and may also modulate the structure of RNA). In signal transduction however, Mn2+ functions dynamically as a catalytic cofactor, for enzymes such as SpoT and adenylyl cyclase whose products are regulatory signals as well as for enzymes that modulate protein activity through covalent modification.
22.214.171.124 Mn2+ binding transcription factors
Currently, two different DNA binding transcriptional regulators in S. enterica Typhimurium are known to be cofactored by Mn2+, Fur and MntR. Fur has been extensively characterized, MntR only recently identified. Fur is primarily an iron-dependent transcriptional repressor of various iron homeostatic genes [53, 105, 106]; however, it has long been known that in many contexts Fur containing Mn2+ represses as effectively as Fur containing Fe2+ [107–110]. The relative potency of Fe2+ or Mn2+ apparently depends on the detailed DNA context of the binding site. The large Fur regulon now contains at least three genes –sodA, mntH, and sitABCD– whose physiological function deals with Mn2+. In addition, Fur has more functions than simple cation-mediated repression. Fur regulates some genes in response to acid pH rather than divalent cations . In addition, some Fur-regulated genes are activated rather than repressed by metal binding to apo-Fur [112–115]. Recent evidence suggests that activation by Fur can be indirect, arising from conventional Fur repression of a small regulatory RNA which, if not repressed, specifically enhances degradation of the activated gene's mRNA . There is no basis for assuming that iron is necessarily the relevant cofactor for Fur for all these non-canonical regulatory events.
The Mn2+-dependent repressor, MntR, has only recently been characterized [48, 52]. In S. enterica Typhimurium it currently is known to repress only the two Mn2+ transport loci, mntH and sitABCD (A. Janakiraman, D.G. Kehres, M.E. Maguire and J.M. Slauch, unpublished observations). Since DNA sequences resembling the MntR binding motifs in these two promoters are difficult to find in the genome as a whole, the MntR regulon is evidently smaller than that of Fur. Just as Fur can be cross-regulated by Mn2+ under some conditions, MntR can be cross-regulated to some extent by Fe2+ in place of Mn2+. In the absence of an intact MntR regulatory circuit, a significantly higher iron concentration is needed to repress the MntH Mn2+ transporter than in the presence of an intact MntR regulatory circuit . This effect presumably involves Fe2+ binding as a cofactor to MntR although this has not been directly demonstrated. Mn2+ binding transcription factors are known to regulate a small number of other genes, for example gpmA in T. pallidum, but the extent of MntR regulation of genes with functions other than Mn2+ transport remains to be determined. Further, little insight into the possible involvement of Mn2+ can be gained by comparing the properties of Fur to those of other regulators in the Fur family, or the properties of MntR to those of other regulators in the DtxR family, because in general, the detailed cation specificities of most of these homologs are poorly known.
126.96.36.199 Regulation by SpoT ppGpp phosphatase
The best characterized and likely the most phylogenetically prevalent example of regulation by catalytic Mn2+ is in the bacterial ppGpp signaling pathway. The nucleotide ppGpp is also called the ‘stringent response alarmone’. It's synthesis by RelA is stimulated by uncharged tRNAs (i.e. amino acid starvation), and the presence of ppGpp rapidly halts cell growth by inhibiting ribosomal RNA operon expression [117–122]. In addition, however, modulation of ppGpp levels in growing cells plays a key role in adapting normal growth to even transient changes in the nutritional environment by globally coordinating mRNA and stable RNA synthesis . Mn2+ is the obligatory catalytic cofactor for the only cellular ppGpp hydrolase, the N-terminal domain of the bifunctional ppGpp synthetase/hydrolase SpoT . While most of our knowledge about ppGpp physiology comes from studies in the Enterobacteria, ppGpp and SpoT synthase/hydrolase proteins are ubiquitous in bacteria [125–128] and thus Mn2+ involvement in growth regulation is also ubiquitous.
188.8.131.52 Regulation by adenylyl cyclases
In addition to ppGpp, another essential small signaling nucleotide whose metabolism involves Mn2+ in at least some bacteria is cyclic AMP. Unlike the enzyme found in many other pathogens, the Rv1625c adenylyl cyclase of M. tuberculosis H37Rv is strongly stimulated by Mn2+, suggesting that changes in the Mn2+ content of the M. tuberculosis phagosome could be keyed either to growth properties of the bacterium itself or to dysregulation of cyclic AMP-mediated signaling in the macrophage itself. Mn2+ responsive adenylyl cyclases do not appear to be common in the prokaryotic genomes sequenced to date, but their overall prevalence and significance are not yet known.
184.108.40.206 Regulation by protein phosphatases and kinases
In addition to global growth regulation, catalytic Mn2+ participates in a variety of more restricted signaling pathways or control mechanisms. The pathway for which appreciable evidence has been accumulated to date involves eukaryotic-like protein phosphatases and protein kinases [130–134]. S. enterica Typhimurium  and E. coli each have two protein phosphatases PrpA and PrpB, for which, frustratingly, few specific protein substrates have yet been identified, but which nevertheless have been shown both in S. enterica Typhimurium and E. coli to be involved in a variety of envelope stress responses. Like the homologous phosphatase of bacteriophage lambda  and several eukaryotic members of the same ‘PPP’ phosphatase family , PrpA and PrpB are activated far more potently by Mn2+ than other biologically relevant cations. Intriguingly, both prpA and prpB are located in the chromosome of S. enterica Typhimurium at sites associated with pathogenicity and horizontal gene transfer. The prpB locus is located at one boundary of the pathogenicity island SPI1, a large cluster of virulence genes whose activity is required for invasion of macrophages  while prpA is adjacent to a prophage insert (D.G. Kehres and M.E. Maguire, unpublished observations).
A role for Mn2+ has also recently been established in a family of eukaryotic-type serine/threonine kinases that includes both a Gram-negative member, Pkn2 from Myxococcus xanthus, whose autophosphorylation is markedly Mn2+-dependent, and a Gram-positive member, PknA from M. tuberculosis [141, 142], which regulates morphological changes during cell division. Neither PPP phosphatases nor PknA-type kinases are ubiquitous among prokaryotic genomes sequenced to date, thus the overall significance of Mn2+-mediated protein phosphorylation in bacterial metabolism remains to be seen.
220.127.116.11 Modulation of RNA structure/stability
All three examples of ‘catalytic’ Mn2+ regulation discussed above involve phosphotransferase enzymes, a situation discussed further in Section 3.3.5. But ‘static’ Mn2+:phosphate interactions might play a regulatory role as well. Just as binding of Mn2+ instead of Fe2+ can alter the conformation and activity of transcription regulatory proteins like Fur, binding of Mn2+ instead of Mg2+ alters the tertiary structure of tRNA and the activity of ribozymes in prokaryotes and eukaryotes [143–148]. Mn2+ but not Mg2+ activates the ATPase of the mRNA capping complex in yeast and presumably other eukaryotes .
Structural changes induced by high intracellular Mn2+ concentrations could also alter the activity of small RNAs involved in transcriptional or translational regulation, which are currently being characterized at an accelerating pace and likely contribute to gene regulation in all bacteria [150–154]. While it is currently unknown if cytoplasmic Mn2+ levels affect the activity of any regulatory small RNA in vivo, the field is so new that this feature has likely not been investigated.
3.3.3 Mn2+-dependent metabolic enzymes
Patterns are at last emerging among the regulatory roles of Mn2+, but unfortunately the known universe of Mn2+-dependent metabolic enzymes remains small and, with the exception to be discussed below in Section 3.3.4, stubbornly recalcitrant to generalization. Without attempting premature systematics, therefore, the present review will simply highlight a short list of enzyme types in which the reported involvement with Mn2+ is either too recent or too tenuous to have been featured in previous discussions of the subject. Available evidence does not yet justify strong claims for the importance of Mn2+ in overall metabolism. In many cases, it has not been determined whether a given enzyme is strictly Mn2+-dependent or whether Mn2+ simply stimulates it to a greater extent than other functional cations. In many cases as well, it has not been determined whether the Mn2+ enzyme is the sole source of a given activity in a given organism or whether it is simply one of a set of more or less redundant enzymes. Nevertheless, it is the authors’ contention that the very diversity of these enzymes betokens a broader and deeper role for ‘metabolic Mn2+’ than has hitherto been presumed.
18.104.22.168 Polyamine biosynthesis
An often cited eukaryotic Mn2+ enzyme is arginase, which functions in the mammalian nitrogen cycle to remove the guanidinium group from arginine thus generating excretable urea [155–158]. Bacteria do not in general have arginase, although some species of Bacillus are an exception [159, 160]. Instead, bacteria have a ubiquitous close homolog, the equivalently Mn2+-selective enzyme agmatinase . Agmatinase performs an essential role in osmoregulation by removing the guanidinium group from agmatine (decarboxylated arginine) thus generating putrescine, the smallest of the three common bacterial osmoprotective polyamines. It is of interest that the Mn2+-dependent mammalian agmatinase is markedly regulated by exposure to lipopolysaccharide and cytokines .
22.214.171.124 DNA biosynthesis
Reduction of ribose to deoxyribose in nucleotides is a free radical process performed by the enzyme ribonucleotide reductase, of which three distinct functional classes are widely known – one uses iron plus oxygen, one uses deoxyadenosyl cobalamin, and one uses S-adenosyl methionine for catalysis [162, 163]. Recently, species from two divergent branches of bacteria have been found to contain a fourth class of ribonucleotide reductase. This class has a di-ferric heme group but also requires oxygen and manganese. The presence of this class of ribonucleotide reductase has been demonstrated in the Firmicute B. subtilis and the Actinobacteria Corynebacterium glutamicum and ammoniagenes and Propionibacterium shermanii [164–166] with a possible example in the Archaea in Methanobacterium thermoautotrophicum.
126.96.36.199 Phospholipid biosynthesis and processing
Mn2+ has been implicated in the activity of some phosphotransferases that add or remove the polar headgroups of phospholipids . The principal phosphatidylserine synthase Pss of B. subtilis is one such enzyme . The sphingomyelin hydrolase of Pseudomonas sp. strain TK4 is another . By contrast, the diacylglycerol pyrophosphate phosphatase PgbB of E. coli, which does not require any cation for activity, is potently inhibited by Mn2+. In addition, the report that the steroid-17-20-desmolase and 20-α-hydroxysteroid dehydrogenase activities of Clostridium scindens both require Mn2+ now implicates its redox chemistry in the processing of membrane lipids themselves .
188.8.131.52 Polysaccharide biosynthesis
Production of capsular polysaccharide and of exopolysaccharide upon formation of a biofilm are almost universal behaviors among bacteria. An intriguing trend may now be emerging with reports that Mn2+ is more effective than Mg2+ in stimulating polymerization of UDP-sugars into both the type 3 capsular polysaccharide of a Gram-positive bacterium, Streptococcus pneumoniae, and the secreted hyaluronic acid exopolysaccharide of a Gram-negative bacterium, Pasturella multocida.
184.108.40.206 Protein catabolism
The importance of protein turnover and of protein catabolism, both during rapid growth in fluctuating complex environments and during starvation, is self evident. Since one of the major bottlenecks in protein catabolism is hydrolysis of the atypical peptide–proline linkage, it is interesting that the aminopeptidases PepP and PepQ of the Bacterial E. coli and the Archaeal Pyrococcus furiosus are stimulated by Mn2+ [174, 175], suggesting that they could be as Mn2+-dependent as the corresponding enzymes of Aspergillus, Drosophila and humans [176–178]. Mn2+ is involved in at least one other peptide cleavage reaction as well. Like its porcine and tomato LAP homologs, the leucine aminopeptidase activity of E. coli PepA is strongly activated by Mn2+ although most leucine aminopeptidases are Zn2+-activated. Intriguingly, although any connections with Mn2+ are conjectural at this point, E. coli PepA is also a DNA binding protein involved in plasmid ColE1 site-specific recombination , the P. aeruginosa PepA homolog PhpA is involved in transcriptional repression of alginate exopolysaccharide synthesis  and V. cholerae PepA is involved in pH regulation of virulence genes .
220.127.116.11 Small molecule catabolism
In addition to its biosynthetic role in deoxynucleotide production, the redox activity of Mn2+ also plays catabolic roles, notably in degradation of aromatic compounds as exemplified by the 3,4-dihydroxyphenylacetate 2,3-dioxygenase MndD from Arthrobacter globiformis strain CM-2 . A feel for the physiological tradeoffs between Mn2+- and Fe2+-cofactored versions of particular enzymes can be obtained by noting that unlike the more commonly found Fe2+-cofactored extradiol-cleaving catechol dioxygenases, MndD is resistant to inhibition by H2O2 and cyanide anion, though it is severely inhibited by Fe2+.
18.104.22.168 Sugar catabolism
Some but again only a small percentage of the enzymes involved in sugar catabolism are Mn2+-dependent or have Mn2+-dependent isoforms . Three recent examples are the l-fucose isomerase [186–188] and cellobiose-6-phosphate hydrolase CelF  of E. coli, and the d-glucosaminate dehydratase of Pseudomonas fluorescens. The latter enzyme also illustrates the recurring theme of tradeoffs, in this case between Mn2+ and Mg2+ cofactors; it is not the primary dehydratase reaction but rather a D-glucosaminate aldolase side-reaction, possibly adventitious but nevertheless metabolically consequential, that is stimulated in the presence of Mn2+.
3.3.4 Mn2+ and central carbon metabolism
The one area of metabolism in which an appreciable number of Mn2+ utilizing enzymes have been identified is, surprisingly, central carbon metabolism, the network of chemical processes charged with consuming whatever carbon sources happen to be available to the cell and converting them into both a balanced output of growth precursors and a sufficient supply of energy in the form of ATP and reducing equivalents to assemble these precursors into new cell mass. A claim that Mn2+ is important for growth may seem remarkable since every factor involved in such a fundamental process should already have been implicated by genetic surveys of spontaneous or induced growth deficient mutants. On the other hand, the perspective afforded by the discussion of individual enzymes above suggests that any network in which a substantial fraction of the enzymes act on phosphorylated molecules is a prime candidate to involve Mn2+.
As a caveat, no area better exemplifies the observation made in the introduction that speculation about the possible roles of Mn2+ far outpaces existing molecular and mechanistic facts. Nevertheless, it is the authors’ contention that Mn2+ is indeed important in carbon metabolism. The reasons that this fact has not already emerged from conventional genetic and biochemical studies are, first, that Mn2+ enzymes may play their most significant roles in extreme and/or rapidly changing nutrient environments which have not received much experimental attention to date, and second, that carbon metabolism and Mn2+ homeostasis are both extremely ‘robust’ in the face of individual or even multiple mutations. To illustrate the potential of such a proposition, even if not yet completely validated by experiment, Section 22.214.171.124 will consider two central metabolic reactions for which both a conventional Mn2+-independent enzyme and a novel Mn2+-dependent enzyme have now been identified in S. enterica Typhimurium and E. coli, and Section 126.96.36.199 will consider a cluster of functionally related enzymes whose activities all differ depending on whether Mn2+ is substituted as a cofactor in place of Mg2+.
188.8.131.52 Phosphoglycerate mutase ‘GpmM’
Phosphoglycerate mutase, which interconverts 3-phosphoglycerate and 2-phosphoglycerate, is one of four reversible enzymes that operate in an essentially branchless sequence in the middle of the network of intermediary metabolism and collectively convert the smallest and most oxidized glycolytic metabolite, glyceraldehyde-3-phosphate, into the smallest and least oxidized gluconeogenic metabolite, phosphoenolpyruvate (PEP). Some bacteria express a 2,3-bisphosphoglycerate-cofactored ‘dependent’ form of phosphoglycerate mutase (‘DPGM’), termed GpmA in E. coli. This form is homologous to the sole phosphoglycerate mutase found in vertebrates. Many other bacteria have only a 2,3-bisphosphoglycerate ‘independent’ form of phosphoglycerate mutase, or ‘IPGM’, structurally a strictly Mn2+-dependent protein and mechanistically a distinct enzyme from GpmA. This form is homologous to the sole phosphoglycerate mutase found in plants . It may also be found in some Archaea .
The best characterized bacterial IPGM, that of B. subtilis, is inhibited by acidification of the cytoplasm, a property which the bacterium uses to arrest carbon metabolism and stockpile 3-phosphoglycerate within the forming spore as a future carbon and energy source . Perusal of current genomic sequence data indicates that at least 40% of Bacteria, including Salmonella, E. coli and other Enterobacteria, appear to have two phosphoglycerate mutases, one of each type. The E. coli IPGM homolog, which until now has been referred to as an ORF of unknown function YibO, has only recently been characterized biochemically as a functional phosphoglycerate mutase . In parallel with Helmann's nomenclature for Mn2+ transporters, we propose that YibO and bacterial IPGM-type phosphoglycerate mutases in general henceforth be designated ‘GpmM’.
Why Enterobacteria retain two alternative types of phosphoglycerate mutase is not yet known. White and co-workers  have reported that in E. coli, GpmA contributes almost all the phosphoglycerate mutase activity detectable in exponential phase cells growing in rich broth (LB) and that GpmM activity increases to 20–30% of the total activity by the time cultures reach stationary phase. Somewhat in contrast, unpublished studies by the authors indicate that a ΔGpmM S. enterica Typhimurium strain grows substantially more poorly than a ΔGpmA S. enterica Typhimurium strain in minimal salts supplemented with any of a variety of simple carbon sources (D.G. Kehres and M.E. Maguire, unpublished observations) and that GpmM expression is maximal in the mid-exponential phase (Fig. 6). While such observations have so far provided more questions than answers, Enterobacterial GpmA and GpmM are not simply redundant in their function. Furthermore, the strictly Mn2+-dependent GpmM enzyme is clearly more than a peripheral player in metabolism, even under conventional in vitro growth conditions.
Expression of the Mn2+-dependent phosphoglyceromutase (GpmM) during growth.A gpmM::lacZ reporter construct on a low copy plasmid was introduced into a wild-type strain. Expression of β-galactosidase was measured during aerobic growth of the culture in M9 plus glucose medium. As inferred from β-galactosidase activity, total GpmM levels in the culture are low during early exponential phase, increase markedly during the final one to two generations of exponential growth, then no longer increase once the cells enter stationary phase. Interestingly, although this expression appears almost identical to that of mntH (Fig. 4), it is not controlled by either Fur or MntR. D.G. Kehres and M.E. Maguire, unpublished observations.
A similar situation obtains for at least one other step of central carbon metabolism. In addition to the well characterized Fbp enzyme which converts fructose-1,6-bisphosphate to fructose-6-phosphate during gluconeogenesis, S. enterica Typhimurium and E. coli have recently been found to express a distinct Mn2+-dependent fructose-1,6-bisphosphatase, encoded by glpX. This gene is the third and last gene in an otherwise well characterized operon whose preceding gene products are the GlpF glycerol facilitator (transporter) and the GlpK glycerol kinase . While this initial study reported no distinctive phenotype for E. coliΔGlpX mutants compared to ΔFbp mutants, it did note that the enzymes are subject to different allosteric control.
The substrate of both enzymes, fructose-1,6-bisphosphate, is an allosteric modulator of central metabolic enzymes like PEP carboxylase . Their product, fructose-6-phosphate, is a much more potent activating cofactor than fructose-1,6-bisphosphate for the pleiotropic transcriptional regulator of glycolytic, gluconeogenic, and respiratory metabolism Cra (previously designated FruR) [196, 197]. Thus the possibility arises that GlpX might be another Mn2+-dependent regulatory signal-generating protein, like SpoT and adenylyl cyclase, in addition to its unknown contribution to gluconeogenic carbon flux. Finally, Mn2+ is also an essential cofactor for a fructose-1,6-bisphosphatase activity from the Archaeal Methanococcus jannaschii; interestingly, this enzyme is bifunctional, acting as an inositol monophosphatase .
3.3.5 Perspective – Mn2+ in phosphotransferase enzymes
There is a striking prevalence of phosphotransferases among known Mn2+-dependent enzymes just outlined in Sections 3.3.2–3.3.4. A variety of other Mn2+-dependent phosphatases also exist, including an inorganic pyrophosphatase from B. subtilis [199, 200], alkaline and protein phosphatases from Archaeal Halobacterium sp. and Haloferax volcanii [201, 202].
We have already observed that substitution of Mn2+ for Mg2+ generally ‘improves’ the catalytic efficiency of such enzymes (Section 3.1.2). This combination of factors leads us to wonder whether the ability of bacteria to manipulate cytoplasmic Mn2+ levels over two orders of magnitude (Section 2.5), with a general trend toward higher Mn2+ concentrations in dense and/or stationary phase cultures (Section 2.4.5), correlates in any way with the activity patterns of phosphotransferase enzymes in general. Such a perspective predicts that strictly in terms of metabolism, elevated cytoplasmic Mn2+ levels should have both negative and positive consequences for the cell. This in turn suggests that metabolic considerations may be at least as important as free radical detoxification considerations, if not more so, in determining Mn2+ homeostasis. Possible examples are briefly outlined below.
184.108.40.206 Mn2+ stimulates misincorporation by DNA and RNA polymerases
From the standpoint of polynucleotide synthesis, elevated Mn2+ levels are probably undesirable during rapid growth when the bacterial chromosome is being replicated, since it is a familiar fact to every PCR user that Mn2+ is mutagenic and stimulates misincorporation. On the other hand, elevated Mn2+ levels may be desirable during periods of stress or stasis when maintaining the closed duplex integrity of the chromosome is paramount because Mn2+ has been shown to stimulate certain kinds of translesion DNA repair processes . Such considerations no doubt provide some of the rationale for why exponential growth appears to correlate with low Mn2+ levels and stationary phase with higher Mn2+ levels among bacteria in general.
220.127.116.11 Mn2+ may modulate collective activity of enzyme networks
Can a corresponding rationale for the correlation of Mn2+ levels with growth state be found among the phosphotransferases in intermediary metabolism? Perusal of the extensive biochemical literature on such enzymes suggests a surprising answer. Namely, the list of enzymes for which at least one prokaryotic, fungal, or vertebrate representative has been shown to be stimulated by Mn2+ in vitro includes a cluster of five enzymes all of which use PEP or pyruvate as substrate or product (Table 2 and Fig. 7). The increase in each enzyme's kcat/Km has always been dismissed as a test tube artifact since the concentration of Mn2+ required is high, typically >100 μM. However, our new appreciation of Mn2+ transporters indicates that actual in vivo modulation of the corresponding enzymes should be well within the Mn2+ homeostatic capability of most bacteria.
Mn2+-stimulated enzymes in intermediary carbon metabolism.In addition to the strictly Mn2+-dependent phosphoglycerate mutase GpmM, Mn2+-stimulated versions of several intermediary metabolic enzymes have been reported either in eukaryotes or prokaryotes. Genes encoding the Gram-negative S. Typhimurium or E. coli forms of all these enzymes are highlighted in the diagram. Intriguingly, these form a cluster consisting of almost all the enzymes which involve PEP or pyruvate as substrates or products. Specific examples of Mn2+-stimulated enzymes include enolases (eno) from P. furiosus and Saccharomyces cerevisiae; PEP carboxylases (ppc) from E. coli and N. gonorrhoeae; PEP carboxykinases (pck) from Vibrio costicola, Mycobacterium smegmatis, and chicken ; pyruvate kinases (pyk) from S. Typhimurium , M. smegmatis, Streptococcus lactis and S. mutans, and both the NAD-linked  and NADP-linked  malic enzymes (mez) of E. coli. Though Enterobacteria have no pyruvate carboxylase to convert oxaloacetate directly to pyruvate, Mn2+-dependent versions of this enzyme are found in P. fluorescens, M. smegmatis and also in rat , human , and other vertebrates. B. subtilis, Aspergillus niger, and many white-rot fungi including Flammulina velutipes and Collybia velutipes have Mn2+-dependent oxalate decarboxylases and/or oxalate oxidases [257–261]. Figure adapted from Kessler and Knappe .
What might be the consequence of simultaneous upregulation of an entire set of Mn2+-stimulated enzymes? Consider just the set defined in Fig. 7. Upregulation of enolase, PEP carboxylase, and PEP carboxykinase upregulation should only lead to more rapid equilibration among the cell growth precursors 2-phosphoglycerate, PEP, and oxaloacetate plus possibly some futile cycling. Upregulating pyruvate kinase and both the NADH- and NADPH-linked malic enzymes, though, should be more consequential. The net effect would likely be a substantial flux of carbon out of the three growth precursor pools into pyruvate, which bacteria primarily use to generate energy and balance reducing equivalents. On balance, this offers a strikingly consistent rationale for the observed Mn2+ level versus growth state correlation. Low Mn2+ levels are conducive to large precursor pools and occur during rapid growth. High Mn2+ levels are more conducive to energy economy and occur during slow growth.
The proposition that Mn2+ acts as a ‘switch’ between ‘growth mode’ and ‘maintenance mode’ metabolism is intended as a hypothetical illustration. Mn2+ stimulation has not yet been specifically demonstrated for the Enterobacterial versions of some of these enzymes. Conversely, this flux analysis assumes that no other Enterobacterial carbon metabolic enzymes are sensitive to Mn2+ stimulation, something which has also not yet been examined and indeed seems unlikely. The two essential points are, first, that several Mn2+-sensitive carbon metabolic enzymes exist, and second, that growth stage-dependent excursions in the [Mn2+]/[Mg2+] ratio occur in E. coli, S. enterica Typhimurium, and probably most bacteria. Thus, while enzymatic details and specific impact on growth regulation probably differ among species, redistribution of carbon fluxes in response to varying Mn2+ levels is likely a widespread phenomenon.
4 Physiologic and pathogenic implications of manganese biochemistry
In addition to biochemical characterization in a variety of contexts, Mn2+ has been implicated physiologically in several, generally idiosyncratic ways. One which merits mention is the puzzling ability of Mn2+ to induce up to three rounds of cell division, unaccompanied by any net increase in cell mass or volume, in stationary cultures of Deinococcus radiodurans. The problem is that in a field as new as prokaryotic Mn2+ biology, biochemistry and physiology seldom connect. Many of the biochemical activities that are detailed above, e.g. the presumed alternatives to iron-based chemistry in L. plantarum, B. burgdorferi, and S. suis, are as yet unaccompanied by persuasive demonstrations of physiological relevance. Conversely, physiological observations like that regarding Deinococcus or for example that regarding effects of a ΔGpmM mutation on S. enterica Typhimurium growth are unaccompanied by convincing biochemical explanations. Thus, arguments that Mn2+ plays a critical role in any particular aspect of bacterial physiology can only be made from incomplete data. Nonetheless, the biochemical themes emerging above do predict some general physiological roles, both during planktonic growth and in the specialized environment of an infected host organism, and these will now be discussed briefly.
4.1 Mn2+ and metabolic flexibility during rapid growth
The prevalence of instances where Mn2+ either activates a second messenger processing enzyme or directly modulates the efficiency of a metabolic enzyme suggests at least one physiological theme. Mn2+ may provide growing cells with a ‘rapid response’ capability, allowing them to fine tune metabolism on a time scale of seconds, much faster than even bacteria can respond with de novo gene transcription and translation. The best documented examples involve optimization of carbon catabolic networks in the face of fluctuating nutrient environments, but similar rapid adaptations may occur in response to acute stresses such as heat, radiation or acid. The efficiency of many catabolic enzymes and thus the net fluxes resulting from the concerted action of many such enzymes is almost certainly influenced by the cytoplasmic [Mn2+]/[Mg2+] ratio. Since cytoplasmic Mg2+ levels do not appear to vary greatly [62, 205, 206], regulation of Mn2+ uptake and efflux transporters is likely the cell's principal way of manipulating this important ratio. The implications for pathogenesis are obvious, since precipitous variations in both nutrient pools and cation pools available to bacteria are hallmarks of the process of infection at many stages (see Section 4.3).
Mn2+ could fine tune not just the input side of growth but also the output side. A possible rationale for the existence of alternative Mn2+-dependent isoforms of intermediary metabolic enzymes like phosphoglycerate mutase and 1,6-bisphosphofructose phosphatase could be a role in discharging excess catabolites  or ‘toxic waste recycling’. The existence of large multi-enzyme complexes in several branches of metabolism, in glycolysis, respiration and fatty acid metabolism for example, is supported by a number of individually inconclusive but collectively quite persuasive experimental findings in bacteria and in eukaryotes [208–214]. The hallmark of such multi-enzyme complexes is that each enzyme delivers its product directly to the active site of the next member of the pathway. That is, intermediate products are ‘channeled’ between adjacent enzymes in the interior complex. The only bulk diffusional encounters involve binding of the initial substrate and release of the final product.
There is evidence for multi-enzyme glycolytic complexes in E. coli [214, 215] and evidence for specific interaction between GpmA homologs and enolase in mammals [216, 217]. Suppose therefore that GpmA, consistent with its expression throughout the Enterobacterial growth cycle , is predominantly sequestered in channeling complexes. Under such conditions, it would be unable to access ‘heterologous’ 2- or 3-phosphoglycerate arising from degradation of toxic methylglyoxal  and possibly hydroxypyruvate . GpmM, thanks to its unrelated structure, would presumably not assemble into such complexes and therefore would be available to recycle dangerous heterologous phosphoglycerates and thus complete the degradation of toxins.
4.2 Mn2+ in ‘maintenance’ during slow growth or stasis
Mn2+ levels are generally highest in slow growing or non-growing bacteria. An established physiological correlate is that the manganese superoxide dismutase SodA of E. coli or S. enterica Typhimurium is more important than the iron superoxide dismutase SodB for either viability during starvation-induced stasis or outgrowth upon acquisition of fresh nutrients [219–221]. The most appealing biochemical correlate, though it has yet to be documented physiologically, is that Mn2+ can participate in non-enzymatic radical scavenging, in complexes with inorganic ligands like bicarbonate [56, 57], stationary phase-specific intracellular organo-Mn2+ compounds analogous to synthetic salens , or specific intracellular chaperone proteins . Mn2+ might also contribute to stationary phase nutrient economy by activating the Mn2+-dependent peptidases PepP and PepQ allowing efficient recycling of proteins to amino acids . Finally, although strictly Mn2+-dependent, the Ka of some enzymes for Mn2+ is relatively poor. An increase in Mn2+ content during stasis would result in activation. The protein phosphatases PrpA and PrpB are an obvious example [135, 136]. PrpB has a low Ka for Mn2+ of about 1 μM and thus is probably active throughout the cell cycle; however, PrpA has a Ka of >50 μM, much higher than the expected steady state free Mn2+ concentration of the cytosol. A marked increase in cellular Mn2+ during stationary phase would be expected to activate this phosphatase, although with unknown consequences because its substrates for dephosphorylation are not yet known.
With the previous discussion for perspective, it is possible to address whether Mn2+ is important for virulence and if so, predict some ways in which Mn2+ might be involved. In S. enterica Typhimurium at least, the data presented in Fig. 2 and the more extensive data recently published by Boyer et al.  showing that both mntH and sitABCD are important clearly provide an affirmative answer to the question of Mn2+ importance in pathogenesis. Mutation of both selective Mn2+ transporters of S. enterica Typhimurium markedly attenuates virulence, and attenuation is in part dependent on the presence of a functional Nramp1 allele in the host.
If the question of whether Mn2+ is important has been answered, the obvious next question is ‘why is it important?’ The role of MntH and SitABCD is presumably to supply Mn2+ for specific Mn2+-dependent enzymes. Therefore one or more Mn2+-dependent enzymes must be important for virulence since function of a Mn2+-dependent enzyme would be compromised in a Mn2+ transport deficient cell. Specific Mn2+-dependent enzymes have been discussed above, but it is currently unknown which enzyme(s) are most important. The following discussion is presented therefore in terms of the S. enterica Typhimurium invasion cycle, noting where in that cycle Mn2+ could be important.
For Salmonella, pathogenesis encompasses several distinct sequential processes [225–228]. At early stages, the behavior of Mn2+ transport deficient bacteria is consistent with what would be expected based on discussions above. For later stages, no evidence is yet available, but reasonable predictions can be made.
Following passage through the acidic environment of the stomach, S. enterica Typhimurium invades host tissues by inducing its own phagocytosis by a macrophage or epithelial cell in an intestinal Peyer's patch, a process dependent on the SPI1 pathogenicity island gene cluster. Nothing is known about Mn2+ levels inside the bacterium at this point; however, since both mntH and sitABCD are transcriptionally induced after invasion (Fig. 4), they can be presumed to be inactive during the invasion process per se . Indeed invasion of either RAW264.7 macrophages or HeLa epithelial cells is as efficient for ΔmntH and ΔsitABCD mutants as it is for wild-type S. enterica Typhimurium (M.L. Zaharik, D.G. Kehres, M.E. Maguire and B.B. Finlay, unpublished observations).
4.3.2 Persistence inside macrophages
Survival of the bacterium within a macrophage as it circulates through the lymphatic system involves a remodeling of the phagosome. The macrophage would like to turn the phagosome into a lysosome but the bacterium can subvert this process. In the case of S. enterica Typhimurium for example, a variety of effectors secreted by the Type III secretion system encoded on SPI2 (rather than SPI1) remodel the vesicle into a SCV [46, 228]. Numerous experiments in the well established cultured macrophage model for S. enterica Typhimurium show that replication is modest at this stage. Thus bacteria plausibly contain enough endogenous Mn2+ to sustain them during this stage. Further, since bacteria with an intact SPI2 are apparently able to prevent H2O2 containing peroxisomes from fusing with the developing SCV [47, 228], reactive oxygen stress may also be modest during this stage. Indeed, ΔmntH and ΔsitABCD mutants persist as well as wild-type once inside the phagosome of RAW264.7 macrophages (M.L. Zaharik, D.G. Kehres, M.E. Maguire and B.B. Finlay, unpublished observations).
4.3.3 Escape from the macrophage
For S. enterica Typhimurium the next stage of infection involves escape from a host macrophage, typically by inducing apoptosis, a process also dependent at least in part on proteins encoded within SPI2. The apoptotic cascade involves manipulation of the phosphorylation state of phosphatidylinositol second messengers. Both phosphatidylinositol synthesis and headgroup exchange in mammalian systems are known to be Mn2+-dependent [229–232]. In this context, it may be relevant that the diacylglycerol pyrophosphate phosphatase PgbB of E. coli is potently inhibited by Mn2+. Further, it is certainly relevant that Nramp1 is recruited to the SCV thus potentially manipulating local transition metal cation levels available to both the host and the pathogen [22, 233, 234]. As noted above, PrpA, which has a role in stress response, is inactive at low Mn2+ concentrations and thus would also be affected by host cell restriction of pathogen Mn2+.
4.3.4 Persistence in infectious foci
After passage through circulating macrophages in a sequence of phagocytosis/apoptosis events, S. enterica Typhimurium eventually takes up residence in the spleen or the liver. It is currently unclear whether S. enterica Typhimurium in the liver reside in macrophages, hepatocytes, or extracellular spaces, but the infectious foci that develop in this organ usually take the form of granulomas, typically rich in neutrophils [235–237] as well as macrophages . Thus both nutrient starvation and free radical stress would be much more severe at this stage than at earlier points during infection. From this point forward, unfortunately, experimental evidence of any sort is scarce, and evidence bearing on the importance of Mn2+ is non-existent. However, it would be expected that the stationary phase ‘maintenance’ functions attributed to Mn2+ in the section just above will come into play during bacterial persistence in granulomas.
4.3.5 Systemic phase
It is also currently unclear whether the explosive proliferation that converts an immunologically sequestered cluster of infectious foci into a systemic blood-borne bacteremia occurs within macrophages and/or hepatocytes or in extracellular fluids. Regardless, bacteria will experience an abrupt change in nutrient availability upon exit from the granuloma, recalling the importance of SodA in stationary phase survival or outgrowth [219, 221]. Subsequently they experience rapid fluctuations in nutrient levels and composition, precisely conditions that call for great metabolic flexibility. Concomitantly, by leaving the relative immunological shelter of the host tissue, they are also exposed to all the stresses brought to bear by the complement cascade and the rest of the blood-borne innate immune system.
4.3.6 Persistence in a carrier state
Finally, a characteristic of typhoid infection is that S. enterica Typhimurium and S. Typhi take up residence in the gallbladder, where they most likely form biofilms on gallstones, explaining their persistence in a chemotherapeutically recalcitrant ‘carrier’ state . Since biofilms almost by definition involve secreted exopolysaccharide, the involvement of Mn2+ in exopolysaccharide production by other bacteria (Section 3.3.2) suggests that it may play a role in S. enterica Typhimurium biofilms, and thus the typhoid carrier state, as well.
5 Conclusions and directions for future research
In hindsight it is not surprising that manganese has traditionally been accorded little importance in the biology of prokaryotes. There are good reasons why mutants with severe manganese-related phenotypes have not been discovered before. The most obvious reason is that mutants are hard to generate since we now know that manganese homeostasis in most bacteria is tightly controlled by multiple and redundant factors. A less obvious reason, to revisit just one speculation, is that manganese may have less to do with the growth behavior generally monitored in the laboratory than with physiology on time scales either much shorter or much longer than most existing research has typically considered, e.g. adaptation to abrupt environmental change during growth or adjustment to long term change during periods of stasis.
The situation with regard to manganese mutants is changing. The resurgence in transporter studies should soon make it feasible to experimentally alter Mn2+ homeostasis in many bacteria by genetic or environmental manipulation. At the same time, application of comparative genomics to the growing spectrum of Mn2+-dependent bacterial enzymes should make it possible to anticipate many phenotypic consequences of altered manganese homeostasis.
In the meantime, though, the field continues to offer a mix of tantalizing but often purely phenomenological results. Adequate molecular or mechanistic syntheses of the biochemical capabilities with the actual physiological uses of this under-appreciated metal remain distant. What we have attempted to do in this review, at rather greater length than a simple catalog of Enterobacterial Mn2+ transport facts would warrant, is offer one ‘template’ for organizing future studies of Mn2+ biology. Likely many of the speculative proposals above will turn out to be wrong. More likely, much of the eventual story will involve enzymes and processes that we have not yet identified or considered. Regardless, the bacterial Mn2+ story is certain to remain interesting, changeable and full of surprises for some time to come.
Silver, S. (1996) Transport of inorganic cations. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F.C., Curtiss III, R., Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., Reznikoff, W.S., Riley, M., Schaechter, M. and Umbarger, H.E., Eds.), pp. 1091–1102. American Society for Microbiology, Washington, DC.
(2000) Towards understanding the chemistry of photosynthetic oxygen evolution: dynamic structural changes, redox states and substrate water binding of the Mn cluster in photosystem II. Biochim. Biophys. Acta 1459, 481–488.
(2000) Natural resistance to intracellular infections: natural resistance-associated macrophage protein 1 (Nramp1) functions as a pH-dependent manganese transporter at the phagosomal membrane. J. Exp. Med. 192, 1237–1248.
(1991) Metallobiochemistry of magnesium, coordination complexes with biological substrates site specificity, kinetics and thermodynamics of binding, and implications for activity. Inorg. Chem. 30, 2740–2747.
(1990) Manganese(II)-bicarbonate-mediated catalytic activity for hydrogen peroxide dismutation and amino acid oxidation: detection of free radical intermediates. Proc. Natl. Acad. Sci. USA 87, 394–398.
(1990) Manganese(II) catalyzes the bicarbonate-dependent oxidation of amino acids by hydrogen peroxide and the amino acid-facilitated dismutation of hydrogen peroxide. Proc. Natl. Acad. Sci. USA 87, 389–393.
(2001) Accumulation of manganese in Neisseria gonorrhoeae correlates with resistance to oxidative killing by superoxide anion and is independent of superoxide dismutase activity. Mol. Microbiol. 40, 1175–1186.
(1993) Binding of the ferric uptake regulation repressor protein (Fur) to Mn(II), Fe(II), Co(II), and Cu(II) ions as co-repressors: electronic absorption, equilibrium, and 57Fe Mossbauer studies. J. Inorg. Biochem. 50, 193–210.
(1998) Molecular and genetic analysis of two closely linked genes that encode, respectively, a protein phosphatase 1/2A/2B homolog and a protein kinase homolog in the cyanobacterium Anabaena sp. strain PCC 7120. J. Bacteriol. 180, 2616–2622.
(1999) Crystal structures of Bacillus caldovelox arginase in complex with substrate and inhibitors reveal new insights into activation, inhibition and catalysis in the arginase superfamily. Struct. Fold. Des. 7, 435–448.
(2000) Manganese-dependent ribonucleotide reductase from Propionibacterium freudenreichii subsp. Shermanii: partial purification, characteristics and role in DNA biosynthesis. Mikrobiologiia 69, 471–477.
(1989) xerB, an Escherichia coli gene required for plasmid ColE1 site-specific recombination, is identical to pepA, encoding aminopeptidase A, a protein with substantial similarity to bovine lens leucine aminopeptidase. EMBO J. 8, 1623–1627.
Iwamoto, R, Taniki, H, Koishi, J, Nakura, S. d-glucosaminate aldolase activity of d-glucosaminate dehydratase from Pseudomonas fluorescens and its requirement for Mn2+ ion. Biosci. Biotechnol. Biochem. 59, 1995. 408–411
(1984) Phosphoenolpyruvate carboxylase of Escherichia coli. Specificity of some compounds as activators at the site for fructose 1,6-bisphosphate, one of the allosteric effectors. J. Biochem. (Tokyo) 95, 637–642.
Lin, E.C.C. (1996) Dissimilatory pathways for sugars, polyols and carboxylates. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F.C., Curtiss III, R., Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., Reznikoff, W.S., Riley, M., Schaechter, M. and Umbarger, H.E., Eds.), pp. 307–342. American Society for Microbiology, Washington, DC.
Srere, P.A., Sherry, A.D., Malloy, C.R. and Sumegi, B. (1997) Channeling and intermediary metabolism. In: Channelling and Intermediary Metabolism (Aguis, L. and Sherrat, H.S.A., Eds.), pp. 201–217. Portland Press Ltd., London.
(1997) Reconstitution, morphology and crystallization of a fatty acid beta-oxidation multienzyme complex from Pseudomonas fragi. Biochem. J. 328, 815–820.
(1997) Critical roles of neutrophils in host defense against experimental systemic infections of mice by Listeria monocytogenes, Salmonella typhimurium, and Yersinia enterocolitica. Infect. Immun. 65, 630–635.
(1985) Regulation of methionine synthesis in Escherichia coli: effect of metJ gene product and S-adenosylmethionine on the in vitro expression of the metB, metL and metJ genes. Biochem. Biophys. Res. Commun. 133, 731–739.
(1995) Effects of divalent cations and nucleotides on the 14CO2-oxaloacetate exchange catalyzed by the phosphoenolpyruvate carboxykinase from the moderate halophile, Vibrio costicola. Biochem. Mol. Biol. Int. 36, 1225–1234.
(1981) Role of metal cofactors in enzyme regulation. Differences in the regulatory properties of the Escherichia coli nicotinamide adenine dinucleotide phosphate specific malic enzyme, depending on whether magnesium ion or manganese ion serves as divalent cation. Biochemistry 20, 2503–2512.
Kessler, D. and Knappe, J. (1996) Anaerobic dissimilation of pyruvate. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F.C., Curtiss III, R., Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., Reznikoff, W.S., Riley, M., Schaechter, M. and Umbarger, H.E., Eds.), pp. 199–205. American Society for Microbiology, Washington, DC.
(1998) Spectroscopic identification of a dinuclear metal centre in manganese(II)-activated aminopeptidase P from Escherichia coli: Implications for human prolidase. J. Biol. Inorg. Chem. 3, 470–483.
Fraenkel, D.G. (1996) Glycolysis. In: Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology (Neidhardt, F.C., Curtiss III, R., Ingraham, J.L., Lin, E.C.C., Low, K.B., Magasanik, B., Reznikoff, W.S., Riley, M., Schaechter, M. and Umbarger, H.E., Eds.), American Society for Microbiology, Washington, DC.
Keech, D.B, Utter, M.F. Pyruvate carboxylase – properties. J. Biol. Chem. 238, 1963. 2609
(1966) Pyruvate carboxylase. VI. The presence of tightly bound manganese. J. Biol. Chem. 241, 3480–3487.