Keywords

1 Introduction: Nickel in the Environment

Nickel is the fifth most abundant element on earth, although much of it is concentrated in the earth’s core where it is second most in abundance at 8.5% (Zamble et al. 2017). The earth’s crust contains 0.009% nickel, placing it among the top 25 most prevalent elements. The nickel content of soils ranges from 3 to over 1000 ppm (μg/g) depending on the origin of the parent rock, with an average of ~16 ppm (Iyaka 2011). In an extreme natural case, nickel is present at >1 mg/g in serpentine soils. The nickel content of lakes and rivers averages about 1 ppm, but varies widely depending on the type of soil associated with the watershed of interest (Nriagu 1980). In the ocean, the nickel concentration is quite uniform at the surface (2–4 nM), increasing at greater depth to 10 nM in the Pacific and 6 nM in Atlantic (Glass and Dupont 2017). Anthropogenic sources of nickel include mining of nickel ores and certain industrial processes that can lead to local areas of very high nickel concentrations. Regardless of the source of nickel or the properties of diverse environments, the biologically relevant form of the metal typically occurs as the dication (Ni2+).

Nickel may have beneficial or toxic effects depending on the microbial species and the metal ion concentration. Importantly, the availability of nickel depends on factors other than just its concentration: this metal ion forms complexes with inorganic matrices such as sulfides and iron oxyhydroxides or with organic compounds such as humic and fulvic acids. These anionic substances reduce the effective nickel concentration by sequestration. A growth requirement for nickel was first reported in 1965 for two strains of hydrogen-oxidizing bacteria (Bartha and Ordal 1965) and this metal ion has since been shown to be essential for multiple microorganisms. Negative effects of nickel on microorganisms also have long been established (Babich and Stotzky 1983).

This chapter surveys the metabolism of nickel by microorganisms. It starts by describing homeostasis mechanisms by which microbes sense the intracellular (or periplasmic) concentration of this metal and use various mechanisms to facilitate either its import or efflux, depending on whether the available levels of this nutrient are inadequate or if in excess as a toxicant. The chapter then explains the basis for nickel toxicity and describes how microorganisms circumvent this damage. Next, the nickel-containing enzymes are highlighted including their biosynthetic pathways that sometimes use nickel storage and delivery proteins, nucleotide hydrolyzing enzymes, and other components.

2 Nickel Homeostasis

Many microorganisms contain elaborate systems to sense the cytoplasmic (or periplasmic) concentration of nickel, turn on nickel-specific uptake transporters when the metal ion is limiting, and increase levels of nickel-specific efflux systems when the metal is in excess (Table 14.1). Numerous excellent reviews covering these homeostatic process are available for those interested in more extensive treatment of this topic (Chivers 2017; Higgins 2019; Higgins et al. 2012a; Iwig and Chivers 2010; Li and Zamble 2009; Musiani et al. 2015; Rodrigue et al. 2017; Zeer-Wanklyn and Zamble 2017). The highlights of the better-studied systems are described here.

Table 14.1 Summary of selected nickel-dependent transcriptional regulators (sensors) and their associated nickel transport systems.

A few general comments about cellular nickel concentrations are worth mentioning. Unless located in a nickel-rich environment, microorganisms tend to concentrate the metal ion to levels exceeding that found in their external milieu. For example, Escherichia coli concentrates this metal ion from minimal or LB media (where nickel is present at ~10−8 M and ~10−7 M, respectively) to greater than 10−6 M in the cell (Outten and O’Halloran 2001). One can calculate that a single atom of nickel in an E. coli cell (~2 μm in length and ~0.5 μm in diameter or ~4 × 10−16 L) yields a concentration of approximately 4 nM (4 × 10−9 M). This value is of interest when considering the measured dissociation constants (Kd) for nickel-binding sensors such as NikR of E. coli (described in greater detail below) with a value of less than 1 pM (10−12 M)—nearly three orders of magnitude less than one per cell (Chivers and Sauer 2002; Wang et al. 2004). This apparent discrepancy arises because the measured concentration of nickel within cells does not equate to the level of free or available nickel ion—much of it is found dynamically associated with cytoplasmic components (thiols, polynucleotides, amino acids, etc.). The nickel sensors assure that an ideal regulatory balance of nickel import and export exists to provide optimal metal levels for cellular growth. Selected examples of nickel sensing transcriptional regulators are described in Sect. 14.2.1. That section is followed by discussions of the systems controlled by these sensors: nickel uptake (Sect. 14.2.2) and nickel efflux (Sect. 14.2.3).

2.1 Nickel-Sensing Transcription Factors

Metalloregulatory transcription factors undergo protein conformational changes upon binding their cognate metal ions, with consequent modification of their transcriptional regulatory properties. Figure 14.1 depicts four highly simplified mechanisms for transcription factor regulation by metal ions, but it is important to note that many other options exist. In the first case, a metal-repressor complex (co-repressor) binds to the operator region of a gene to prevent transcription, whereas the apoprotein form of the transcription factor is non-functional as a repressor. The opposite situation occurs when the apoprotein of the transcription factor directly represses transcription, with metal binding leading to de-repression that allows the gene to be transcribed. The third option parallels the first in that only the metal-bound transcription factor binds to DNA, but in this case the complex (co-activator) interacts with RNA polymerase to enhance gene expression. A final option occurs when the apoprotein of the transcription factor functions as the activator, with metal binding leading to loss of activation capability. As described below, all of these mechanisms, at least in modified forms, apply to nickel-sensing transcription factors.

Fig. 14.1
figure 1

Four potential roles for nickel-dependent transcriptional regulatory factors. (a) Co-repression of a gene by the complex of nickel bound to the transcription factor (Ni-TF). (b) Loss of repressor function induced by the binding of nickel. (c) Co-activation of a gene by the nickel-transcription factor complex. (d) Loss of activator function caused by the binding of nickel

2.1.1 NikR

E. coli NikR nicely illustrates the metalloregulatory mechanism shown in Fig. 14.1a. When nickel binds to this protein (subunit size of 15-kDa) it functions as a repressor of the nikABCDE operon encoding an ATP-binding cassette (ABC) nickel-uptake system located in the cytoplasmic membrane (de Pina et al. 1999). Mutants affecting nikR lead to constitutive expression of the nik operon. The nickel-bound repressor binds to an operator site consisting of dyad-symmetric 6-base pair recognition sequences (GTATGA) separated by 16 base pairs (Chivers and Sauer 2000). A high affinity nickel-binding site (Kd measured in the pM range except for one report of a sub-μM value) must be occupied for NikR to bind DNA, but tighter interaction with DNA (Kd ~ 20 pM) occurs when additional nickel is bound to a poorly characterized lower affinity site on NikR (Kd ranging from 30 nM to 30 μM) (Bloom and Zamble 2004; Chivers and Sauer 2002; Diederix et al. 2008; Wang et al. 2010a). Potassium also must bind to NikR for effective interaction with DNA (Wang et al. 2010b).

A series of crystal structures provides important insights into how E. coli NikR functions as a transcription factor. The tetrameric apoprotein possesses two dimeric ribbon-helix-helix DNA-binding regions separated by a nickel-binding regulatory domain (Schreiter et al. 2003). The structurally-characterized C-terminal domain is a tetramer that binds four nickel atoms at high-affinity square-planar sites using His87, His89, and Cys95 from one subunit along with His76′ from another subunit (Schreiter et al. 2003). The structure of the full-length NikR holoprotein reveals that nickel addition results in a protein conformational change affecting the relative orientations of the DNA-binding regions (Schreiter et al. 2006). A further change occurs when the nickel-NikR complex binds to operator DNA (Fig. 14.2a) (Schreiter et al. 2006). In particular, the DNA-binding regions rotate to allow both to bind to the DNA simultaneously, leading to a bend in the operator region. Co-repressor binding to this region prevents its interaction with RNA polymerase, thus leading to cessation of transcription. This structure also identifies how potassium is coordinated; i.e. via bidentate Glu30, bidentate Asp34, and backbone carbonyls of Ile116, Gln118, and Val121. Additional structural studies have probed the basis for the metal-binding specificity of E. coli NikR by identifying sites for binding other metals (Phillips et al. 2008) and analyzed low-affinity binding sites for nickel (Phillips et al. 2010). Despite its ability to bind other metals, the cellular response of NikR as a transcriptional regulator is specific to nickel.

Fig. 14.2
figure 2

Selected structures of nickel-dependent transcription regulatory factors. (a) Escherichia coli NikR homotetramer in complex with nickel and bound to DNA (PDB ID 2hzv). The inset reveals one of the four nickel-binding sites with its ligands. (b) Synechocystis sp. PCC 6803 InrS homotetrameric transcription factor in its apoprotein form (PDB ID 5fmn) with the four residues likely to coordinate one of the four metal-binding sites indicated. (c) Homodimeric Mycobacterium tuberculosis NmtR transcription factor apoprotein (PDB ID 2lkp). Four likely coordinating residues, with two others not yet identified, are shown for one of the two sites. (d) Homodimeric Streptomyces coelicolor Nur transcription factor (PDB ID 3eyy). Two types of metal sites are shown (gray and green spheres). (e) Cupriavidus metallidurans CH34 CnrX-CnrY-CnrH tripartite nickel sensor with its periplasmic CnrX homodimer (cyan and green; residues 39–148 are from PDB ID 2y39 and the membrane-bound amino-terminal region is modeled) and a complex (PDB ID 4cxf) between the transmembrane CnrY (red; residues 2–30 are from the structure and a membrane-localized helix is modeled) and cytoplasmic CnrH (yellow). The metal sites in CnrX are indicated. Nickel binding to CnrX results in release of the sigma factor CnrH for interaction with RNA polymerase. (f) NiCo riboswitch from Erysipelotrichaceae bacterium. The riboswitch RNA structure (PDB ID 4rum) contains four cobalt (brown spheres, likely equivalent to the nickel-binding sites), two magnesium (green spheres), and seven potassium atoms (purple spheres). For all panels, the proteins are illustrated in cartoon mode with partially transparent surfaces, and individual subunits are in different colors. DNA and RNA are shown as sticks with RNA also depicted as wires. Residues of interest are shown in stick mode with orange carbon atoms. When bound, nickel is shown as green spheres

NikR in Helicobacter pylori exhibits several intriguing differences from the protein described above. Like E. coli NikR, the nickel-bound H. pylori NikR functions as a co-repressor (Fig. 14.1a) of a cytoplasmic membrane-associated nickel import system; however, this NikR-dependent regulation primarily involves nixA encoding a single-component secondary or permease-type transporter (Contreras et al. 2003; Ernst et al. 2005; Wolfram et al. 2006). Another NikR target is fecDE (Jones et al. 2015), a cluster that was later expanded, renamed, and shown to encode the NiuBDE ABC-type nickel uptake system (Fischer et al. 2016). Various strains of H. pylori contain periplasmic components of the ABC-type uptake system (comparable to NikA) known as NiuB1, NiuB2, and CeuE, with the latter also regulated by NikR (Jones et al. 2015). Of further interest, ceuE and fecDE encode an ABC-type transporter in Helicobacter mustelae (Stoof et al. 2010a). NikR of H. pylori also regulates production of a system for nickel transport across the outer membrane (OM) involving the OM protein HP1512 and the ExbB-ExbD-TonB cross-periplasm energizer (Davis et al. 2006; Ernst et al. 2006). In addition, the nickel-bound form of H. pylori NikR functions as a co-activator (Fig. 14.1c) of the gene cluster encoding urease, a nickel-containing enzyme (van Vliet et al. 2002), and it modulates several other nickel-repressed or nickel-activated genes (Contreras et al. 2003; Ernst et al. 2006; Vannini et al. 2017). The operator region recognized by H. pylori NikR differs from that bound by the E. coli protein and has poorly conserved inverted repeats (Delany et al. 2005). Additional changes in regulatory properties relate to pH-dependent effects (Bury-Moné et al. 2004; Jones and Zamble 2018; van Vliet et al. 2004) and hierarchical effects that provide an order to gene regulation (Dosanjh et al. 2009; Muller et al. 2011). Nickel binding to the H. pylori tetrameric protein also differs from that for the E. coli protein; two nickel are bound with greater affinity (Kd ~ 10 nM), two with less affinity (Kd ~ 100 nM), and others with weak affinity (Kd ~ μM) (Zambelli et al. 2007a). Crystal structures are reported for the H. pylori NikR apoprotein and several holoprotein forms (Benini et al. 2011; Dian et al. 2006; West et al. 2010). The overall structures closely resemble that of the E. coli protein with some differences noted for nickel sites in the holoproteins, but all showing at least one of the familiar square-planar site (His99, His101, and Cys107 from one subunit with His88′ from a second subunit).

Several other NikR proteins are characterized to various extents, with four examples cited here. In Helicobacter hepaticus, NikR-dependent regulation resembles features of the H. pylori and E. coli systems. This NikR controls the levels of a putative inner-membrane ABC-type nickel transporter (encoded by nikABDE) and a possible OM nickel transporter (using hh0418), along with affecting levels of nickel-dependent urease and [NiFe] hydrogenase activities (Benoit et al. 2013). The situation in Brucella abortus is comparable to that in E. coli, with NikR serving as a transcriptional regulator of nikABCDE (Budnick et al. 2018). NikR from Geobacter uraniireducens binds cooperatively to promoter regions of nik(MN)1 and nik(MN)2, components of ABC-type transporters (Benanti and Chivers 2010). Finally, NikR from Pyrococcus horikoshii, for which the transcriptional target has not been reported, is notable because crystal structures were determined for its apoprotein and four nickel-bound conformations (Chivers and Tahirov 2005). The overall tetrameric structures closely resemble those cited above, with the DNA-binding domains undergoing bending and rotational changes in conformation. The high affinity nickel-binding site is square planar and utilizes His89, His91, and Cys97 from one subunit along with His78′ from a second subunit, thus creating a situation essentially identical to the other NikR nickel sites.

2.1.2 RcnR/InrS/DmeR/NcrB

The E. coli RcnR nickel and cobalt sensor operates by metal-dependent de-repression or release of the regulatory protein from the DNA (Fig. 14.1b), resulting in up-regulation of rcnA encoding a nickel/cobalt efflux system (Iwig et al. 2006; Koch et al. 2007). The designation rcn is named for resistance to cobalt and nickel, with increased sensitivity to these metals noted when rcnA is deleted (Rodrigue et al. 2005) and decreased metal sensitivity for the case of rcnR deletion (Koch et al. 2007). The 40-kDa homotetrameric RcnR sensor apoprotein binds to a TACT-G6-N-AGTA motif between the divergently transcribed rcnR and rcnA genes (Iwig et al. 2006; Iwig and Chivers 2009). Dissociation from the DNA occurs upon binding one cobalt or nickel ion per 89-residue subunit, with the metals exhibiting nM affinity (Iwig et al. 2008). X-ray absorption spectroscopy (XAS) provides evidence for partial thiolate ligation of the metals, which are 6-coordinate (Iwig et al. 2008). This technique was combined with mutagenesis methods and modeling studies (using the structure of the copper-sensor CsoR as a template) to propose the nickel is bound by the amino-terminal amine, Glu63, and His64 of one subunit along with Cys35′ of another subunit (Carr et al. 2017b; Higgins et al. 2012b; Iwig et al. 2008; Musiani et al. 2015). Crystals of the RcnR-DNA complex were reported (Li et al. 2020), but no structure is yet available.

Synechocystis sp. PCC 6803 contains the nickel-responsive regulatory protein InrS (named for internal nickel responsive sensor) that belongs to the same protein family as RcnR and CsoR (Foster et al. 2012). The tetrameric cyanobacterial protein is a repressor of nrsD (Fig. 14.1b), encoding a nickel efflux pump, and the repression is relieved by nickel binding (one metal ion per subunit) with an estimated Kd of 2 × 10−14 M (Foster et al. 2012). Mutagenesis studies indicate that nickel binds to the protein using Cys53, His78, and Cys82 ligands, perhaps also including His21 (Foster et al. 2014). A crystal structure of the InrS apoprotein (Fig. 14.2b) is consistent with this same set of coordinating residues, where Cys53 from one subunit partners with the other three residues from a second subunit (Foster et al. 2017). XAS analysis of nickel-bound InrS was consistent with a square-planar Ni(His)2(Cys)2 site (Carr et al. 2017a). The protein can also bind copper or cobalt, but its physiological function is specific to nickel.

Rhizobium leguminosarum also possesses an RcnR-like protein that is denoted DmeR for divalent metal efflux repressor (Rubio-Sanz et al. 2013). The apoprotein form of DmeR represses dmeF, encoding a cation diffusion facilitator (CDF) efflux transporter, and repression is alleviated by nickel or cobalt (Fig. 14.1b). A similar dmeRF gene cluster is responsible for nickel (or cobalt)-dependent regulation and export in Agrobacterium tumefaciens (Dokpikul et al. 2016). Structural characterization has not been reported for any DmeR.

Another sequence-related family member is NcrB from Leptospirillum ferriphilum strain UBK03, an iron-oxidizing, metal-resistant bacterium containing the ncrABC gene cluster that confers resistance to nickel (Tian et al. 2007). NcrB is a histidine-rich, 89-residue protein that functions as a repressor of the gene cluster in its apoprotein form, but the regulation is derepressed by nickel binding to the protein (Fig. 14.1b) leading to synthesis of the NcrAC efflux pump (Zhu et al. 2011). The structure of NcrB has not been determined.

2.1.3 NmtR/KmtR

Examples of repressors that dissociate from the DNA upon binding nickel (Fig. 14.1b), yet are unrelated to those just described, include NmtR and KmtR from Mycobacterium tuberculosis (Campbell et al. 2007; Cavet et al. 2002). These proteins are members of the ArsR/SmtB family of transcriptional regulatory proteins (Busenlehner et al. 2003). The apoprotein forms of NmtR and KmtR bind to operator-promoter sites upstream of nmtA and cdf, encoding a P-type ATPase efflux pump and a CDF protein, respectively. Spectroscopic and mutagenesis studies suggest that nickel binds to NmtR in octahedral coordination, with the likely involvement of Asp91, His93, His104, and His107 along with residues near the amino-terminus (Cavet et al. 2002; Pennella et al. 2003; Reyes-Caballero et al. 2011). These residues are not conserved in KmtR and its higher affinity nickel binding site was suggested to involve His88, Glu101, His102, His110, and His111 (Campbell et al. 2007). The dimeric NmtR protein binds two nickel with Kd ~ 10−10 M (Reyes-Caballero et al. 2011). The solution structure for the NmtR apoprotein, determined by NMR methods (Lee et al. 2012), revealed the winged-helix appearance that is typical for this family of proteins (Fig. 14.2c).

Related nickel sensor proteins have been characterized in two other microorganisms. An NmtR sensor exists in Streptomyces coelicolor where it regulates production of a P-type ATPase efflux transporter (Kim et al. 2015). Streptomyces griseus contains another example of the ArsR/SmtB family named SrnR. This protein forms a complex with SrnQ that regulates sodF, encoding an iron-containing superoxide dismutase, by binding to the inverted repeat sequence TTGCA-N7-TGCAA. As originally proposed, neither protein binds to DNA in the absence of nickel whereas the nickel-bound hetero-octamer functions as a repressor (Fig. 14.1a) (Kim et al. 2003b). More recent efforts suggested that SrnR does bind DNA and activates transcription, whereas further binding of nickel-bound SrnQ leads to dissociation and reduced levels of transcription (Fig. 14.1d) (Beniamino et al. 2020). SrnR binds nickel with moderate affinity (Kd ~ 0.65 μM) (Beniamino et al. 2020), but the primary nickel-binding regulatory site is associated with SrnQ (one nickel per protein subunit) (Kim et al. 2003b). The structure of SrnRQ and its nickel-binding site(s) are unknown.

2.1.4 Nur

The S. coelicolor Nur (nickel uptake regulator) protein, a member of the Fur family of transcriptional regulators, binds nickel to form a complex that represses transcription of some genes (Fig. 14.1a) while activating transcription of another gene (Fig. 14.1c). Specifically, nickel is a co-repressor of the nikABCDE and nikMNOQ operons for nickel uptake and of sodF encoding an iron-dependent superoxide dismutase, while the metal is a co-activator of sodN coding for a nickel-dependent superoxide dismutase (Ahn et al. 2006; An et al. 2009). The nickel-Nur complex binds to the “Nur box” palindromic sequence tTGCaa-N5-ttGCAA. The structure of the 16-kDa Nur protein (Fig. 14.2d) reveals a homodimeric winged-helix motif that wraps around DNA and binds four metals per dimer (An et al. 2009). Two sites near the subunit interface coordinate nickel in octahedral geometry with three side chain residues (His70, His72, and His126) along with three oxygen atoms from buffer components. Two other sites, initially suggested to not be responsive to nickel, coordinate the metals in square-planar geometry using His33, His86, His88, and His90 side chains. More recent studies using a combination of mutagenesis, metal titration, and fluorescence anisotropy methods suggest that nickel binding to this site is the critical regulatory process used for protein binding to the DNA, with the other site having a fine-tuning role (Manley et al. 2020). Isothermal titration calorimetry (ITC) studies confirm the dimeric protein binds four nickel ions in pairs with Kd values of 10 nM and 280 nM (Musiani et al. 2015).

2.1.5 CnrX

Cupriavidus metallidurans (formerly named Ralstonia metallidurans) CH34 contains a nickel sensor, CnrX, that operates by a very different process. This organism tolerates high concentrations of several types of heavy metals by using a set of efflux pumps. In particular, this microorganism contains the cnrYXH-cnrCBA gene cluster (located on the pMOL28 plasmid) that is associated with, and named for, cobalt and nickel resistance (Grass et al. 2000; Tibazarwa et al. 2000). CnrCBA is the nickel exporter, CnrH is a sigma factor regulating both halves of the gene cluster, CnrY is a trans-membrane connector, and CnrX is a periplasmic nickel-sensing protein (Fig. 14.2e). The cytoplasmic face of CnrY binds CnrH while its periplasmic face binds CnrX. These three proteins work together in what is termed an extracellular function-type sigma factor; nickel coordination to CnrX (Kd ~ pM) leads to a conformational change that propagates through CnrY causing the release of CnrH, which then binds to the appropriate DNA sequence to stimulate transcription by RNA polymerase. This mechanism is akin to, but more complicated than, the co-activator mechanism described earlier (Fig. 14.1c); in particular, the nickel sensor resides in the periplasm.

The X-ray crystal structure of dimeric CnrX is available for residues 31–148 of the apoprotein and the copper-bound forms (Pompidor et al. 2008). The copper is coordinated by His42, His46, Glu63, and His119, with Met123 (numbering based on the full-length protein) also being nearby. Follow-up studies reveal structures of the nickel- (Fig. 14.2e), cobalt-, and zinc-bound forms; the nickel and cobalt species exhibit octahedral coordination involving these same five residues with bidentate Glu63 accounting for the metal-specific conformational changes needed for signaling (Trepreau et al. 2011b). The structure of the nickel-bound species includes residues 39–148, and the amino terminal region is membrane embedded. A crystal structure of the complex between the cytosolic domain of CnrY (residues 2–30) and much of CnrH (residues 5–190 except for 92–122) is known (Maillard et al. 2014). The conformational changes associated with nickel binding to CnrX were explored in greater detail by metal titration analysis and high resolution (1.1 Å) structural studies of selected variant forms of the protein (Maillard et al. 2015). The spectroscopic properties of cobalt-bound CnrX are defined and two additional cobalt binding sites are demonstrated to exist at the amino terminus of the protein (Trepreau et al. 2011a). A structure also is available for the related NccX protein from strain 31A of the same species; significant structural differences are noted when comparing to the strain CH34-derived CnrX protein, but these variations are attributed to the redistribution of hydrophobic surfaces during solubilization of the protein by detergent (Ziani et al. 2014).

2.1.6 NrsRS

A microorganism already mentioned, Synechocystis sp. PCC 6803, contains a second nickel sensor that operates by a very different mechanism from that detailed earlier (Sect. 14.2.1.2). According to the current model (Garcia-Dominguez et al. 2000; Lopez-Maury et al. 2002), NrsS has an amino-terminal domain localized in the periplasm, two transmembrane helices, and a carboxyl-terminal kinase domain. Nickel binding to the periplasmic region stimulates the kinase domain to transfer phosphate to the cytoplasmic membrane-localized NrsR, which activates transcription by binding at up to four sites in the intergenic region of the divergent nrsBACD and nrsRS gene clusters. No structural studies have been described for the NrsRS system. The behavior of this two-component system is reminiscent of the co-activator mechanism detailed earlier (Fig. 14.1c). Note that the nrsD mentioned here is the same gene as that regulated more directly by InrS, which uses a promoter binding site that immediately precedes the gene.

2.1.7 NimR

Haemophilus influenzae possesses a distinct nickel sensor belonging to the Mer family of transcriptional regulatory proteins, typically associated with regulatory responses to toxic metals such as mercury, cadmium, and lead. This protein, NimR, regulates the nikKLMQO operon encoding an ABC transporter for nickel uptake by a combination of two mechanisms shown in Fig. 14.1 (Kidd et al. 2011). Specifically, the NimR apoprotein binds to a site of dyad symmetry in the operator/promoter region of the nickel transport operon resulting in transcriptional activation (Fig. 14.1d); upon binding nickel, NimR ceases to activate and rather the complex functions as a co-repressor (Fig. 14.1b). The dimeric protein (15.6-kDa subunit) binds a single nickel at a site that has not been further characterized and the structure of the protein has not been reported.

2.1.8 Other Nickel Sensors

Nickel-sensitive transcriptional regulatory proteins exist in many microorganisms beyond those mentioned above. These regulators often are related in sequence to the proteins already described, but novel types of sensor proteins also are known. Detailed structural studies and careful analyses of the corresponding nickel-binding sites generally are not available. When functionally characterized, these additional nickel-sensor proteins often have been found to control nickel-transport systems; however, other roles also are known such as regulating the production of nickel-containing enzymes. Such non-homeostatic regulatory functions of nickel sensors were noted earlier: the control of urease expression by NikR in H. pylori (van Vliet et al. 2002) and regulation of superoxide dismutases by Nur in S. coelicolor (Ahn et al. 2006) and SrnRQ in S. griseus (Beniamino et al. 2020; Kim et al. 2003b). Two additional examples of nickel sensors not involved in nickel transport are summarized below, followed by a brief discussion of a nickel-sensing RNA.

YqjI of E. coli is a nickel-binding regulatory protein that associates with inverted-repeat sequences in the region between the divergently transcribed yqjH and yqjI genes (Wang et al. 2011). The complex regulation pattern of this system requires that YqjI apoprotein bind to both promoter regions in a cooperative manner for full repression (Wang et al. 2014). Binding of nickel reduces the DNA-binding activity of the regulatory protein, allowing for yqjH expression (Wang et al. 2011). YqjH is a flavin-containing enzyme that functions as an iron-siderophore reductase so it was renamed NfeF for nickel-responsive Fe uptake flavoprotein; consequently, YqjI was renamed NfeR for nickel-response Fe-uptake regulator (Blahut et al. 2018). In vivo transcriptional studies were used to assess mRNA levels of the wild-type protein and site-substituted variants, while purified YqjI/NfeR and several of its variants were analyzed by ITC, XAS, and other studies to better define the nickel-binding site. Although the metal exhibits 5-coordinate geometry with a sulfur ligand, none of the Cys-substituted variants affect nickel binding. Nevertheless, Cys42 and Cys43 are required for the in vivo response and these residues are suspected to form a vicinal disulfide (Blahut et al. 2018). Structural studies are needed to better define the NfeR nickel-binding site.

The H. pylori HP0868 gene product is a nickel-binding protein that affects the activity level of urease; thus, it was named Mua for modulator of urease activity (Benoit and Maier 2011). Dimeric Mua binds two nickel ions, and high nickel concentrations lead to repression of urease activity by this protein. Mua does not bind to the urease promoter region, or other DNA promoter sequences, and the mechanism of regulation is unknown. The nickel-coordinating residues are not identified and the structure of this nickel sensor has not been elucidated.

Comparative sequence analysis of genomes in the order Clostridiales reveals a novel class of structured RNA often residing upstream of czcD genes that encode CDF proteins (Furukawa et al. 2015). The corresponding RNA from Clostridium botulinum binds nickel and cobalt with a Kd of 12 and 6 μM, respectively, leading to structural changes. The metal-bound form of the NiCo riboswitch is thought to prevent formation of a terminator loop in the RNA, thus allowing the cdf gene to be transcribed. Subsequent in vitro and in vivo studies demonstrated the riboswitch also responds to ferrous ion (Xu and Cotruvo Jr. 2020). The crystal structure is available for the cobalt-bound RNA from Erysipelotrichaceae bacterium (Fig. 14.2f), revealing two cobalt (and presumably nickel) sites, two magnesium atoms, and seven bound potassium ions (Furukawa et al. 2015).

2.2 Nickel Import Systems

Microorganisms possess nickel uptake systems to supply the requirements for nickel-containing enzymes and they use nickel efflux systems (Sect. 14.2.3) to limit the toxicity of this cation. Regulation of these transporters often is facilitated by using the nickel sensors described in Sect. 14.2.1. The following subsections describe mechanisms of nickel uptake that fall into two general classes: ABC transporters and permeases (or secondary transporters). The ABC transporter family is subdivided into those with a soluble substrate-binding protein or domain (Sect. 14.2.2.1) and those of the energy-coupling factor (ECF) subfamily (Sect. 14.2.2.2). The permease family (Sect. 14.2.2.3) includes the so-called NiCoT, UreH, and HupE/UreJ subfamilies (Chivers 2015; Eitinger and Mandrand-Berthelot 2000; Nies et al. 2017; Rodionov et al. 2006; Rodrigue et al. 2017). In addition to describing each of these systems, this section summarizes the process used to facilitate import across the OM of Gram-negative cells (Sect. 14.2.2.4). The described examples are those with high specificity for nickel, although some other metals such as cobalt are often recognized. In addition to these nickel-targeted import processes, this metal also enters some cells via transporters that function in uptake of magnesium (Snavely et al. 1991), other metals (Degen et al. 1999), metal-citrate complexes (Willecke et al. 1973), or other systems that are not detailed.

2.2.1 Nickel Uptake by ABC Transporters Possessing a Substrate-Binding Protein/Domain

The most intensively studied ABC transporter system for nickel uptake is the NikABCDE system of E. coli (Navarro et al. 1993). NikA is a periplasmic substrate (nutrient)-binding protein, NikB and NikC are integral membrane proteins, and NikD and NikE are ATPases that drive the uptake process. Nickel binds to NikA with a Kd of less than 0.1 μM (de Pina et al. 1995), but this value is shifted to ~75 nM for the complex of metal bound to l-histidine (Chivers et al. 2012). Analysis of the structure of NikA provides a fascinating glimpse into the power and limitations of crystallography. The initial apoprotein and holoprotein structures revealed the same overall fold with two domains connected by a hinge used to enclose the substrate binding site; however, the bound pentahydrate nickel lacked any direct interaction with the protein (Heddle et al. 2003)—a coordination environment incompatible with earlier XAS analysis (Allan et al. 1998). A subsequent structure of NikA complexed with EDTA-bound iron implicated the binding of a metallophore, rather than the free metal ion (Cherrier et al. 2005). Another NikA structure was proposed to resolve the disparate findings; it used a variant protein and identified a second, distant nickel-binding site with His54 and His442 as coordinating ligands (Addy et al. 2007). The notion of NikA binding a metallophore was later supported by another structure in which NikA binds a complex of nickel and an endogenous ligand, tentatively identified as butane-1,2,4-tricarboxylate (Cherrier et al. 2008). Finally, after evidence had indicated NikA actually binds the Ni-(l-histidine)2 complex (Chivers et al. 2012), the structure of this form of the protein was obtained and confirmed the metal coordination by His416 (Lebrette et al. 2013). In contrast to these extensive structural studies of NikA, no structure is reported for the other components of the Nik system. Because the NikBCDE components are likely to resemble other ABC-type metal transporters, Fig. 14.3a provides the structure of the transmembrane and ATPase components of a putative metal-chelate ABC transporter from H. influenzae (Pinkett et al. 2007) as a mimic of the nickel-specific proteins, along with that of the NikA-Ni-(l-His)2 complex (Lebrette et al. 2013).

Fig. 14.3
figure 3

Structures of nickel transporters. (a) Escherichia coli NikA and a NikBCDE-like complex used for nickel import. The periplasmic NikA substrate-binding protein (red, PDB ID 4i8c) includes an inset illustrating the metal-binding site with His416 in orange sticks, two molecules of l-histidine in blue sticks, and nickel as a green sphere. The structure of NikBCDE is unknown, but likely resembles that depicted for the metal-chelate ABC transporter of Haemophilus influenzae (PDB ID 2nq2) with transmembrane (green/cyan) and ATPase (magenta/yellow) components. (b) Thermoanaerobacter tengcongensis NikM2 (green, PDB ID 4m58), the S component of a multi-component nickel ECF-type ABC importer, with the nickel-coordinating sidechains (Met1, His2, and His67) shown as orange sticks. The full importer structure is not established, but likely resembles the four-component folate transporter shown from Lactobacillus brevis (PDB ID 4huq; S in green, T in yellow, and A in magenta and cyan). (c) Structural mimics of CDF, MFS, and P-type ATPase nickel exporters are the homodimeric, zinc-specific, YiiP from E. coli (PDB ID 3h90, green and magenta), the monomeric, putative drug exporting, YajR from E. coli (PDB ID 3wdo, cyan), and the four-domain, single-component, copper-transporting, CopB from Legionella pneumophila (PDB ID 3rfu, orange). The latter includes a transmembrane region, actuator domain (a), phosphorylation domain (p), and nucleotide binding-domain (n). (d) A structural mimic of an RND export pump for nickel is CusABC that exports copper primarily from the periplasm of E. coli (PDB ID 3ne5 for CusAB and 3pik for CusC). The RND pump is gray, except for single subunits of CusA (red), CusB (blue), and CusC (dark green). All components are shown in cartoon mode with partially transparent surfaces

Many other microorganisms possess Nik-type ABC-transporters with a soluble periplasmic nickel-binding protein (or for Gram-positive bacteria, an attached extracellular domain), although in some cases different terminology is used for the components. Examples with experimental evidence for a nickel transport function include proteins encoded by niuBDE of H. pylori (Fischer et al. 2016), nikABCDE of Brucella suis (Jubier-Maurin et al. 2001), yntABCDE of Yersinia pseudotuberculosis (Sebbane et al. 2002) and Proteus mirabilis (Brauer et al. 2020), nikZYXWV of Campylobacter jejuni (Howlett et al. 2012), and cntABCDE from Staphylococcus aureus (Remy et al. 2013). The latter Gram-positive microorganism lacks a periplasm so its NikA-like protein, CntA, is tethered to the membrane by an amino-terminal extension. Several NikA-related structures are known including CntA from S. aureus (Song et al. 2018), NikA from the same microorganism (Lebrette et al. 2015), CeuE from H. pylori (Shaik et al. 2014), NikA from B. suis, YntA from Yersinia pestis, NikZ from C. jejuni (Lebrette et al. 2014), and the analogous protein from Sinorhizobium meliloti (unpublished, PDB ID 4web). Although the overall structures of these nickel-binding proteins are related, the ligand binding sites are distinct. This finding is consistent with diversity in the substrate being transported (i.e., other than the Ni(l-His)2 complex), as exemplified by CntA’s interaction with the complex of nickel and staphylopine—a broad-spectrum nicotianamine-like metallophore (Ghssein et al. 2016; Song et al. 2018).

2.2.2 Nickel Uptake by ECF-Type ABC Transporters

The second type of ABC transporter used for nickel uptake lacks a soluble periplasmic protein or extracellular domain for binding the metal ion or metallophore. Rather, CbiMNQO, NikMNQO, and related ECF-type importers possess a small (~20 kDa), integral-membrane protein for binding nickel, termed the S component, that interacts with another integral-membrane T component and two similar or identical cytosolic ATPases known as A components (Finkenwirth and Eitinger 2019; Rodionov et al. 2006; Slotboom 2014). Early examples of such systems are those reported in Actinobacillus pleuropneumoniae (Bossè et al. 2001) and Streptococcus salivarius 57.I (Chen and Burne 2003), where the nickel import genes are located adjacent to urease genes and shown to be important for nickel incorporation into this enzyme. Among other examples, nickel transport is demonstrated for the products of nik(MN)QO (where nikM and nikN are fused) of Rhodopseudomonas capsulatus (Rodionov et al. 2006), nikLMNQO of H. influenzae (Kidd et al. 2011), and lar(MN)QO (where lar indicates association with genes encoding the nickel-dependent lactate racemase) of Lactobacillus plantarum (Desguin et al. 2014). The structure of the S component for the nickel importer of Thermoanaerobacter tengcongensis, NikM, is shown in Fig. 14.3b (Yu et al. 2014). Nickel is 4-coordinated with the Met1 amine, His2 imidazole and amide, and His67 imidazole providing square-planar ligation. This figure also illustrates a four-component Lactobacillus brevis ECF-type transporter for folic acid as a mimic of the nickel ECF-type transporter.

2.2.3 Nickel Uptake by Permeases/Secondary Transporters

Many microorganisms possess single-protein nickel uptake systems that operate either by facilitated diffusion or are driven by the proton motive force (Eitinger and Mandrand-Berthelot 2000; Eitinger et al. 2005; Rodionov et al. 2006; Rodrigue et al. 2017). The nickel/cobalt transporter (NiCoT) subgroup of these proteins is probably the best characterized. The archetype member of this family is HoxN from Cupriavidus necator (formerly named Alcaligenes eutrophus and Ralstonia eutropha). Early studies identified the gene encoding this protein to be in the hydrogenase gene cluster (Eberz et al. 1989) and showed it participates in nickel transport (Eitinger and Friedrich 1991). A topological model indicated eight trans-membrane domains (Eitinger and Friedrich 1994) and the probable nickel-binding residues have been identified by mutagenesis studies (Eitinger et al. 1997). Another NiCoT family member used to import nickel for [NiFe] hydrogenase activation is HupN of Bradyrhizobium japonicum (Fu et al. 1994). By contrast, NixA of H. pylori is a NiCoT family member that was first identified by its relationship to synthesis of active urease (Mobley et al. 1995a). As in the case of HoxN, membrane topology and extensive mutagenesis studies defined the potential nickel-binding motifs (Fulkerson Jr. et al. 1998; Fulkerson Jr. and Mobley 2000). A second subgroup of nickel permeases with only six trans-membrane domains is named for the first representative identified: UreH encoded within the urease gene cluster of the thermophile Bacillus sp. TB-90 (Maeda et al. 1994). The corresponding family member in Y. pseudotuberculosis facilitates high-affinity, nickel-specific transport (Sebbane et al. 2002). HupE and UreJ are included in the third subgroup of nickel permeases (Eitinger et al. 2005). The R. leguminosarum HupE facilitator protein, encoded by a gene co-localizing with hydrogenase genes, was investigated by mutagenesis methods that identified critical residues for nickel uptake (Albareda et al. 2015; Brito et al. 2010). UreJ of Bordetella bronchiseptica possesses 37% identity with the R. leguminosarum protein, but is encoded within the urease gene cluster (McMillan et al. 1998). Synechococcus sp. strain WH8102 also contains a hupE gene; however, the function of the encoded protein is not defined and nickel transport in this organism is associated with a different permease named SodT (Dupont et al. 2012). Of further interest, a eukaryotic example of a nickel permease, Nic1p, occurs in Schizosaccharomyces pombe (Eitinger et al. 2000). Despite their importance for nickel transport in a wide range of microorganisms, no structures have been reported for any uptake nickel permease.

2.2.4 Nickel Uptake Through the OM

Although the OM of Gram-negative bacteria possesses porin proteins that allow passage across this barrier by small molecules (< ~600 Da), it is known that uptake of iron siderophores, heme, some sugars, and other compounds is promoted by making use of OM receptors that are energized by TonB, ExbB, and ExbD proteins, with periplasmic substrate-binding components delivering the nutrient to inner membrane transporters (Braun and Hantke 2011). Evidence has accumulated that nickel is taken up in a similar way. For example, NikR and nickel regulate H. pylori genes (hp1512, hp1339–1340, and hp1561) thought to encode a receptor (FrpB4), energizer components (ExbB/ExbD), and a periplasmic protein (CeuE) (Contreras et al. 2003). Other groups confirmed the finding that FrpB4 is an outer membrane protein whose synthesis is regulated by NikR, and they identified the NikR-regulated FecA3 (encoded by HP1400) as a second likely OM receptor (Davis et al. 2006; Ernst et al. 2006). Mutagenesis studies demonstrated that H. pylori uses the TonB/ExbB/ExbD machinery to drive nickel uptake via FrpB4 (Schauer et al. 2007). H. hepaticus appears to have a similar system with NikR-regulated expression of a gene (hh0418) encoding an OM receptor (Benoit et al. 2013). Furthermore, H. mustelae possesses nixA, fecDE, nikH, and ceuE (encoding a permease, inner membrane ABC transporter, OM receptor, and periplasmic substrate-binding protein) involved in nickel transport, along with two tonB genes; TonB1 is needed for heme uptake and TonB2 is used for nickel uptake (Stoof et al. 2010a, b). Structural characterization of an outer membrane system for nickel uptake is lacking.

2.3 Nickel Export Systems

This section describes five systems by which microorganisms export nickel from the cell: RcnA and related proteins (Sect. 14.2.3.1), cation diffusion facilitators (CDFs, Sect. 14.2.3.2), the major facilitator protein superfamily (MFS, Sect. 14.2.3.3), P-type ATPases (Sect. 14.2.3.4), and resistance nodulation cell division (RND)-type efflux systems (Sect. 14.2.3.5). As in the case of the importer systems, these exporter systems often are not specific to nickel; however, each system emphasized here promotes nickel efflux.

2.3.1 Nickel Export via RcnAB and Related Systems

The nickel efflux pump of E. coli is the inner membrane-bound protein RcnA as shown by the cellular decrease in nickel resistance when the corresponding gene is mutated (Rodrigue et al. 2005). An increase in cellular nickel concentration leads to dissociation of the RcnR regulatory protein from the DNA (Sect. 14.2.1.2), leading to enhanced rcnA transcription (Iwig et al. 2006). RcnA is distantly related to the NiCoT family (Sect. 14.2.2.3). RcnB, encoded by a gene located immediately downstream of rcnA, is a periplasmic, copper-binding protein that modulates the nickel export activity of RcnA by an unclear process (Blériot et al. 2011, 2015). Genes related to rcnA, sometimes paired with rcnB-like genes, have been identified in several other bacteria and typically have alternative names. A few examples for which an involvement in nickel resistance is demonstrated are mrdH in Pseudomonas putida KT2440 (Haritha et al. 2009), nirC of Klebsiella oxytoca (Park et al. 2008), and nrcC from three species: Hafnia alvei 5-5 (Park et al. 2004), L. ferriphilum (mentioned in Sect. 14.2.1.2 as being regulated by NcrB) (Tian et al. 2007), and Serratia marscescens (Marrero et al. 2007). No structure is available for any RcnA protein, but modeling studies suggest the presence of six trans-membrane regions (Rodrigue et al. 2017).

2.3.2 Export of Nickel by Cation Diffusion Facilitators (CDF)

Nickel export via proton antiporters is carried out by CDF family members. FieF and DmeF of C. metallidurans export nickel, although they are more effective at transporting other metals (Munkelt et al. 2004). DmeF proteins from R. leguminosarum and A. tumefaciens are efficient nickel exporters (Dokpikul et al. 2016; Rubio-Sanz et al. 2013). The nickel-efflux CDF protein of M. tuberculosis is called Rv2025c and regulated by KmtR (Campbell et al. 2007), as described in Sect. 14.2.1.3. Rhizobium etli possesses a CDF transporter of nickel termed NepA (Cubillas et al. 2013). Although no structure is reported for a nickel-specific CDF, the architecture of the zinc-specific CDF transporter YiiP from E. coli is available by crystallography (Lu et al. 2009). This protein (Fig. 14.3c, left) provides a reasonable facsimile of what one might expect for a nickel-transporting CDF. The homodimer is Y-shaped with each subunit having six transmembrane regions and a C-terminal cytoplasmic region.

2.3.3 Nickel Export Involving the Major Facilitator Protein Superfamily (MFS)

As expected for the term superfamily, MFS is a large group of membrane secondary transporters that act on a vast number of substances (Saier Jr. et al. 1999). The best example of a nickel-specific exporter is NreB from Achromobacter xylosoxidans 31A, which confers nickel resistance in the native host or when synthesized in E. coli (Grass et al. 2001). NreB has a histidine-rich C-terminal region, but this sequence is not essential to its function. A related example, NrsD, exists in Synechocystis sp. strain PCC 6803 (Garcia-Dominguez et al. 2000). This protein was mentioned earlier (Sect. 14.2.1.2) because it is regulated by the nickel-sensor InrS. Although structures are not available for NreB or NrsD, several MFS transporters have been structurally characterized including the putative drug exporter YajR from E. coli (Jiang et al. 2013). The MRS proteins are thought to use a rocker-switch mechanism requiring conformational changes involving their 12 transmembrane helices. YajR, shown in its outward-facing conformation, exemplifies the likely structure of the nickel exporters (Fig. 14.3c, middle).

2.3.4 P-Type ATPases Used for Nickel Efflux

ATP hydrolysis is used to drive the export of a wide range of substances via single-component P-type ATPases (Palmgren and Nissen 2011). NmtA from M. tuberculosis or S. coelicolor, mentioned earlier as being regulated by NmtR (Sect. 14.2.1.3), is likely to be such a protein (Cavet et al. 2002; Kim et al. 2015). CtpD of Mycobacterium smegmatis exports nickel, though it has a greater preference for cobalt, and deletion of the corresponding gene leads to accumulation and sensitivity to both metals (Raimunda et al. 2012). Another member of this family of proteins is S. meliloti Nia, named for being a nickel-iron ATPase (Zielazinski et al. 2013). Nia synthesis is induced by the presence of either metal and deletion of the corresponding gene leads to the accumulation of both cations. An extra domain near the C-terminus resembles hemerythrin and binds a dinuclear iron center, suggesting an additional role for this protein may exist. While none of these proteins have been structurally characterized, they are likely to exhibit extensive similarity to the copper-transporting P-type ATPase CopB from Legionella pneumophila (Gourdon et al. 2011). This protein, illustrated as a mimic for a nickel P-type ATPase (Fig. 14.3c, right), possesses a transmembrane portion, and the actuator (a), phosphorylation (p), and nucleotide binding (n) domains.

2.3.5 RND Pumps for Export from the Periplasm

RND pumps are complicated systems for expelling a compound from the periplasm, although they are also capable of less efficient export from the cytoplasm. These systems are used as multidrug transporters and for transporting various other compounds. They consist of three parts: (1) the trimeric inner membrane RND pump that is driven by the proton motive force, (2) the periplasmic adaptor proteins that form a multimeric ring which spans between the inner and outer membranes, and (3) the trimeric OM factor (Symmonds et al. 2015). Nickel export via an RND pump was first described for C. metallidurans CH34 (Liesegang et al. 1993). The cnrCBA genes encoding this system are regulated by the products of cnrYXH (Sect. 14.2.1.5), where CnrX is a periplasmic nickel sensor protein (Maillard et al. 2015). Additional RND exporters of nickel include those encoded by cznABC of H. pylori (Stähler et al. 2006), nrsAB of Synechocystis (regulated by NrsRS, Sect. 14.2.1.6) (Garcia-Dominguez et al. 2000), and cnrA of Bradyrhizobium strains associated with a legume that grows in a high-nickel environment (Chaintreuil et al. 2007). No structure is available for any of these nickel export proteins, but the copper exporter CusABC (Fig. 14.3d) from E. coli provides a reasonable illustration of what one might expect. The CusA component of this RND pump forms a homotrimer in the cytoplasmic membrane, with each subunit possessing 12 transmembrane helices (Long et al. 2010). Six molecules of the CusB adaptor attach to the CusA homotrimer, forming a tunnel through part of the periplasm (Su et al. 2011). The distal end of this tunnel attaches to the trimeric OM component CusC that was structurally characterized as an independent protein (Kulathila et al. 2011). This homotrimer completes the tunnel through the periplasm and forms a β-barrel in the OM that is anchored by a triacylated N-terminus.

3 Other Aspects of Nickel Toxicity and Resistance

Microorganisms have long been investigated with regard to nickel toxicity and nickel resistance. In 1967, for example, R factors were shown to mediate nickel tolerance of E. coli and Salmonella strains (Smith 1967). Early reviews of nickel toxicity documented the sensitivities of various microorganisms to this metal when growing under different conditions (Babich and Stotzky 1983; Gadd and Griffiths 1978), but in general the basis for the sensitivity to this metal ion remained enigmatic. This section begins by describing studies that define the primary site of nickel toxicity in E. coli (Sect. 14.3.1), then it briefly surveys other established mechanisms of nickel stress (Sect. 14.3.2). Following a similar sequence, initial microbial screening methods identified nickel-tolerant strains of microorganisms and by the mid-1980s genetic studies [e.g. (Mergeay et al. 1985; Siddiqui and Schlegel 1987)] began to identify some of the resistance determinants. More recent studies have identified nickel efflux, described in Sect. 14.2.3, as the most widely used system for allowing microorganisms to tolerate high concentrations of this metal ion. Aspects of these exporters will not be repeated here; rather, alternative resistance mechanisms will be summarized (Sect. 14.3.3).

3.1 The Primary Site of Nickel Toxicity in E. coli: Fructose-1,6-Bisphosphate Aldolase

Growth of E. coli on glucose or fructose minimal medium is inhibited by the addition of 8 μM nickel, but no growth inhibition was observed for this environmentally-relevant concentration of nickel when using succinate, lactate, or glycerol as carbon sources (Macomber et al. 2011). As expected, the inhibitory effect was exacerbated for rcnA mutant cells, which are inhibited by only 4 μM of this metal ion. These results are consistent with a direct effect of nickel on fructose-1,6-bisphosphate aldolase (FbaA)—a critical enzyme for the metabolism of glucose and fructose, but not required for degradation of the other nutrients. The loss of cellular FbaA activity following nickel stress and the restoration of growth in the rcnA mutant cells by overexpression of fbaA confirmed this enzyme as the target of nickel toxicity. Prior structural studies had demonstrated that FbaA, a class II aldolase, possesses zinc at its active site (coordinated by His110, His226, and His264) with another zinc (coordinated by Asp144, Glu174, and Glu181) bound at a nearby non-catalytic metal-binding site (Hall et al. 1999). Nickel-treated FbaA lost approximately one of these zinc atoms while binding about one nickel atom. Significantly, the D144A and E174A FbaA variants, affecting the non-catalytic site, are resistant to nickel inhibition. These studies provide clear evidence that the primary site of nickel toxicity for glucose- or fructose-grown E. coli is the non-catalytic zinc site of FbaA (Macomber et al. 2011). Cells grown on succinate, lactate, or glycerol are inhibited by high concentrations of nickel, but the target of toxicity under these conditions remains unknown.

3.2 Other Sites of Nickel Toxicity in Microorganisms

In contrast to the detailed mechanistic understanding of nickel toxicity for glucose- or fructose-grown E. coli, described above, few studies have investigated the basis of nickel’s deleterious effects in other microorganisms (Macomber and Hausinger 2011). This section summarizes other potential sites of nickel toxicity in microbes (Fig. 14.4): oxidative stress effects, substitution of active site metal ions in important metalloenzymes (with redox-active or non-redox sites), and inhibitory binding to non-active sites of essential enzymes.

Fig. 14.4
figure 4

Mechanisms of nickel toxicity. (a) Substitution of nickel for a protein-bound redox-active metal ion with subsequent generation of ROS (reactive oxygen species). (b) Replacement of a redox-active metallocenter in an enzyme by nickel. (c) Substitution of a non-redox metal ion in an essential enzyme by nickel. (d) Other types of inactivation of an enzyme by nickel binding

Proteomic studies of nickel-challenged Burkholderia vietnamiensis and P. putida UW4 showed increased production of proteins related to oxidative stress (Cheng et al. 2009; Van Nostrand et al. 2008), consistent with nickel increasing the cellular formation of reactive oxygen species (ROS). Furthermore, E. coli cells lacking both its manganese- and iron-containing superoxide dismutases exhibited greater sensitivity towards nickel, consistent with formation of superoxide in nickel-stressed cells (Geslin et al. 2001). On the other hand, direct observations indicate that oxidative stress was not induced when treating cells with low concentrations of nickel that still inhibited growth (Kumar et al. 2017). Rather, these metal ions led to double-strand breaks in the DNA and blockage of the SOS repair pathways. These results are consistent with the known properties of free nickel ions that function poorly in generating reactive oxygen species compared to iron and copper (Valko et al. 2005). A plausible explanation for the connection between oxidative stress and nickel toxicosis is metal substitution, with the released iron and copper reacting with oxygen and leading to ROS with its consequent deleterious effects (Fig. 14.4a). Thus, nickel interaction with bacteria is unlike the situation with eukaryotes where the metal ion has been shown to directly cause oxidative stress with accompanying damage to DNA (Guo et al. 2019; Kasprzak and Salnikow 2007), alteration of the thiol/disulfide ratio (Randhawa et al. 2001), or enhanced lipid peroxidation (Niki et al. 2005; Paraszkiewicz et al. 2010).

Substitution of the relatively redox-stable nickel ion for a redox-active metal can inactivate a metalloenzyme, potentially leading to inhibition of cell metabolism (Fig. 14.4b). For example, nickel replaces the ferrous ion in taurine hydroxylase of E. coli by a slow-binding inhibition mechanism thus preventing its reactivity with oxygen (Kalliri et al. 2005); such inactivation of the enzyme would hinder growth of cells using taurine as the sole sulfur source. Similar inhibitory results are expected for other 2-oxoglutarate dependent oxygenases that play diverse roles in various bacteria (Herr and Hausinger 2018; Jia et al. 2017). Other examples of bacterial iron-containing enzymes inhibited by nickel include protocatechuate 3,4-dioxygenase and catechol 1,2-dioxygenase of R. trifolii (Chen et al. 1984, 1985), ATP:corrinoid adenosyltransferase of Salmonella enterica (Buan and Escalante-Semerena 2006), atrazine chlorohydrolase from a Pseudomonas species (Seffernick et al. 2002), and extracellular uricase from a strain of Sphingobacterium thalpophilum (Ravichandran et al. 2015). Similarly, nickel inhibits some copper and cobalt enzymes such as nitrous oxide reductase of Rhodobacter sphaeroides (Sato et al. 1998) and cobaltochelatase of Pseudomonas denitrificans (Debussche et al. 1992). Depending on nutrient availability, the decreases in these enzyme activities could interfere with cell growth.

Nickel also can be a toxicant by substituting for a non-redox metal ion, such as zinc, magnesium, or calcium, in an essential enzyme (Fig. 14.4c). In these situations, the change in metal specificity may lead to subtle changes in protein structure or affect the coordination geometry of the metal. A structurally-characterized example of the latter situation was reported for glyoxalase I, which in some species is a zinc-dependent enzyme and in others requires nickel (Honek 2015). Whereas the native forms of the enzymes exhibit octahedral coordination of the appropriate metal using four protein ligands and two waters (replaced by substrate), an altered coordination geometry is observed for enzymes with the inactive metal bound (He et al. 2000). Partial or nearly complete inhibition was reported for a range of different non-redox metal enzymes such as the magnesium-dependent DNA polymerase I (Klenow fragment) of E. coli (Snow et al. 1993), zinc- or calcium-dependent aminopeptidases from different species of Streptomyces (Arima et al. 2004), and a zinc-containing tripeptidase from Pediococcus pentosaceus (Simitsopoulou et al. 1997).

Finally, it is important to note that nickel can be inhibitory to non-metalloenzymes or by binding to sites distant from a metallocenter (Fig. 14.4d). A few such examples include cytosine 5-methyltransferase and uracil 5-methyltransferase of E. coli (Hurwitz et al. 1964), N-carbamoyl d-amino acid amidohydrolase of an Agrobacterium species (Louwrier and Knowles 1996), and mercuric ion reductase of Azotobacter chroococcum (Ghosh et al. 1999). The detailed inhibition mechanism generally has not been investigated, but the propensity of nickel to bind to histidine and cysteine side chains offers a reasonable hypothesis for enzyme inactivation. For example, N-carbamoyl d-amino acid amidohydrolase from Agrobacterium radiobacter contains a cluster of three histidine residues required for activity (Wang et al. 2001) and mercuric reductases possess a pair of cysteine residues that function in catalysis (Barkay et al. 2003); binding of nickel to these side chains could readily lead to loss of enzyme activity.

Although deleterious effects of nickel ions are proposed to include the production of reactive oxygen species and the inhibition of various types of microbial enzymes, the critical site of nickel toxicity in any particular microorganism is almost never known. Clearly, the basis of nickel stress in microbes requires further investigation.

3.3 Non-Efflux Mechanisms of Nickel Tolerance

Although nickel efflux is the most typical approach used by microorganisms for dealing with elevated concentrations of this metal ion, particular microbes handle nickel ion stress by avoidance, reduction, sequestration, compartmentalization, and several other mechanisms (Macomber and Hausinger 2011).

One straightforward technique to circumvent nickel stress is to simply avoid high concentrations of this toxic metal ion by responding to it as a chemotactic repellent, as demonstrated in E. coli (Englert et al. 2010). The mechanism of this response is not established, but it does not require nickel transport or the periplasmic NikA nickel-binding protein and it does utilize the Tar methyl-accepting chemotaxis protein.

When nickel ions cannot be avoided, some microbes minimize its toxicity by reducing the cation to the harmless elemental form. Precipitation of elemental nickel in the periplasm and on the cytoplasmic membrane was reported for Pseudomonas aeruginosa (Sar et al. 2001); the deposited metal was identified using energy-dispersive X-ray analysis. It is currently unclear whether this microorganism uses the reductive process as a means of nickel resistance. Reduction as a resistance mechanism was demonstrated for Pseudomonas sp. MBR that was grown aerobically with a wide range of electron donors (Zhan et al. 2012). Transmission electron microscopy identified nickel deposits in the periplasm and cytoplasmic membrane of this highly nickel-tolerant microorganism. Similarly, reduction of nickel ions to elemental nickel in the cytoplasm was demonstrated using Thiocapsa roseopersicina and shown to be associated with nickel resistance (Zadvornyy et al. 2010). This microorganism produces dense nanoparticles of nickel metal in the cytoplasm when treated with nickel, but only in the presence of hydrogen gas. The purified hyn-encoded [NiFe] hydrogenase from these cells recapitulates the catalytic production of the reduced metal particles.

Another approach to overcome nickel stress is to reduce its concentration by sequestering it externally, in the periplasm, or within the cell. An external process is used by a nickel-resistant, sulfate-reducing strain of Desulfotomaculum that generates a soluble dark-brown nickel sulfide product (Fortin et al. 1994). External binding of nickel to anionic capsular polysaccharides is another way to potentially reduce the available concentration of this metal ion. The poly-γ-glutamate capsule of Bacillus licheniformis and the cell wall of Bacillus subtilis are known to bind nickel; however, no studies have examined whether such binding confers nickel resistance to the cells (Beveridge and Murray 1976; McLean et al. 1990). Periplasmic sequestration of nickel is an option suggested for an isolate of P. putida S4 (Tripathi and Srivastava 2006). When these cells were exposed to nickel they produced an 18-kDa protein that may tightly bind the metal; however, detailed characterization of nickel binding by the peptide was not reported. Similarly, nickel sequestration in the periplasm was proposed for Cupriavidus pauculus KPS 201 that had been isolated from a nickel-rich ultramafric ecosystem; however, in this case two larger proteins of 74- and 66-kDa were suggested to be involved in binding the metal (Pal and Paul 2010). Again, no analysis of the proteins’ interactions with nickel was reported. Cellular sequestration of nickel ions by an 82-residue peptide (SCO4226) in S. coelicolor was proposed (Lu et al. 2014), though clear evidence for participation in nickel resistance has not appeared. The structure of the holoprotein reveals a homodimer with ferredoxin-like folds containing four weakly bound (Kd of 10 μM) nickel per dimer at distinct surface sites. Other cytoplasmic proteins used to sequester nickel are Hpn-1 and Hpn-2 (Sect. 14.4.12), which are abundant proteins of H. pylori and related microorganisms. These proteins are characterized for their nickel-binding abilities and are known to confer nickel resistance to these cells (Ge et al. 2006a; Seshadri et al. 2007; Vinella et al. 2015).

An alternative version of the sequestration mechanism is compartmentalization. An illustration of this approach involves nickel-resistant strains of Saccharomyces cerevisiae that place the metal into a vacuole as a complex with the amino acid histidine (Joho et al. 1990, 1992; Nishimura et al. 1998; Pearce and Sherman 1999). Such import of nickel into a subcellular organelle is not available to bacteria.

A poorly characterized system is operant in B. vietnamiensis PR1301 that exhibits greater resistance to nickel when grown at pH 5 than at pH 7 (Van Nostrand et al. 2008). Two-dimensional gel electrophoresis coupled with mass spectrometric methods identified several proteins that are produced at greater abundance using the nickel-resistant conditions; surprisingly, these proteins are associated with cell shape and membrane composition. Fatty acid methyl ester analysis revealed a change in membrane structure for cells capable of growth at higher nickel concentrations. The precise mechanism of resistance in this microorganism remains to be defined.

A genome-wide functional screening approach identified seven genes associated with nickel resistance in Acidiphilium sp. MP, a microbe isolated from the metal-contaminated Rio Tinto river in Spain (San Martin-Uriz et al. 2014). Two genes (hslVU) encode an ATP-dependent protease that, when heterologously expressed, confer nickel resistance to E. coli by an unknown mechanism. The other genes encode enzymes involved in lipopolysaccharide or branched chain amino acid biosynthesis; in these cases, it is unclear whether these molecules are directly involved in nickel resistance.

A proteomic analysis comparing the effects of 2 mM nickel treatment on P. putida UW4 revealed significant increases in 82 proteins and significant decreases of 81 proteins (Cheng et al. 2009). The proteins amplified during nickel stress include those for general stress adaptation, oxidative stress, and heavy metal efflux.

A role for superoxide dismutase in nickel resistance was suggested from studies using E. coli (Geslin et al. 2001). The nickel resistance of individual sodA or sodB mutants, encoding the manganese- or iron-dependent superoxide dismutase enzymes, was comparable to the wild type cells; however, the double mutant which lacked both activities exhibited increased sensitivity to this metal. These results suggest that excess levels of cellular nickel can promote the generation of ROS with deleterious consequences, and that superoxide dismutase alleviates these issues.

In summary, a wide range of mechanisms beyond the use of nickel efflux systems are likely to play a role in allowing microorganisms to tolerate this metal ion. Unfortunately, the relevance of many of these systems to nickel stress is not convincingly demonstrated and much work on alternative resistance mechanisms remains to be done.

4 Nickel Utilization by Microorganisms

In the past decade, nickel-containing enzymes have been the subject of several general review articles (Alfano and Cavazza 2020; Boer et al. 2014; Cheng et al. 2016; Kaluarachchi et al. 2010; Maroney and Ciurli 2014) and a book (Zamble et al. 2017). Here, the biochemical reactions of these enzymes are described (Fig. 14.5), their structures are highlighted (Fig. 14.6), and in many cases the pathways for biosynthesis of their metallocenters are summarized. The order of presentation is approximately in order of the complexity of their active site metallocenters, with glyoxalase I (Glo1), acireductone dioxygenase (ARD), quercetin 2,4-dioxygenase (QueD), and superoxide dismutase (SOD) each containing mononuclear nickel centers, urease using a dinuclear nickel site, [NiFe] hydrogenases possessing a heterodinuclear center along with auxiliary [FeS] clusters, carbon monoxide dehydrogenase (CODH) possessing an inorganic [Ni-4Fe-4S] complex plus other [FeS] clusters, acetyl-coenzyme A (CoA) synthase/decarbonylase (ACS) having both the CODH components and a [4Fe-4S]-Ni-Ni cluster, and finally 2-hydroxyacid racemases/epimerases (HAR/E) and methyl-S-coenzyme M reductase (MCR) possessing organometallic cofactors [the nickel-pincer nucleotide (NPN) and nickel-tetrapyrrole cofactor (F430), respectively] with nickel coordinated to an organic framework. In addition, other examples of possible nickel-containing enzymes are mentioned. Finally, cellular nickel storage is briefly addressed.

Fig. 14.5
figure 5

Reactions of nickel-dependent enzymes. (a) Glyoxalase (Glo1). (b) Two distinct acireductone dioxygenase (ARD) reactions depending on whether iron or nickel is bound. (c) Quercetin 2,4-dioxygenase (QueD). (d) Superoxide dismutase (SOD). (e) Urease. (f) Hydrogenase. (g) Carbon monoxide dehydrogenase (CODH). (h) Acetyl-CoA synthase/decarbonylase (ACS). (i) 2-Hydroxyacid racemase/epimerase (HAR/E). (j) Methyl coenzyme M reductase (MCR)

Fig. 14.6
figure 6

Structures of nickel-dependent enzymes and expanded views of their active sites. (a) E. coli glyoxalase homodimer (PDB ID 1f9z) showing a bridging nickel site bound to four side chains and two waters. (b) Acireductone dioxygenase of K. oxytoca monomer (PDB ID 1zrr) showing the nickel atom bound to four side chains and two waters. (c) Quercetin 2,4-dioxygenase from Streptomyces sp. strain FLA (PDB ID 5flh) with nickel bound to four side chains and two waters. (d) Homohexameric superoxide dismutase of Streptomyces coelicolor (PDB ID 1t6u) with nickel bound to the N-terminal amine, an amide, and three side chains. (e) Urease from Klebsiella aerogenes (PDB ID 1fwj) with one copy of each of three types of subunits colored and the other two copies shown in light and dark gray (see text for a description of non-(UreABC)3 aggregation states) along with its dinuclear nickel site, coordinating residues, and three waters. (f) Heterodimeric [NiFe] hydrogenase from Desulfovibrio vulgaris Miyazaki F (PDB ID 4u9h) with its [NiFe] cluster in the H2-reduced state, four coordinating side chains, three diatomic iron ligands, and a bridging hydride atom. (g) Rhodospirillum rubrum carbon monoxide dehydrogenase homodimer (PDB ID 1jqk) containing an exposed [4Fe-4S] D-cluster bridging the subunits, a [4Fe-4S] B-cluster within each subunit, and the two [1Ni-4Fe-4S] C-clusters. (h) The α2β2 acetyl-CoA synthase/decarbonylase from Clostridium autoethanogenum (PDB ID 6ytt) with a central CODH component shown as in (g) and the terminal ACS components containing the [4Fe-4S]-Ni-Ni A-cluster. (i) Monomeric lactate racemase from Lactobacillus plantarum (PDB ID 5huq) as an example of a 2-hydroxyacid racemase/epimerase with its covalently tethered nickel-pincer nucleotide cofactor. (j) The (αβγ)2 methyl coenzyme M reductase from Methanothermobacter marburgensis (PDB ID 1mro) with its coenzyme F430 cofactor. For each protein, the different subunits are shown in cartoon view in different colors and with semi-transparent surfaces. Key residues and diatomic ligands are shown as sticks. Spheres are shown for nickel (green), water molecules (red), iron (brown), and inorganic sulfur (yellow)

4.1 Glyoxalase I (Glo1)

Methylglyoxal and other 2-ketoaldehydes are naturally produced compounds, but they are toxic due to their electrophilic reactivity with biological molecules leading to advanced glycation end-products. The glyoxalase enzyme system converts these molecules into d-hydroxyacids in a three-step sequence: the substrate initially reacts reversibly and non-enzymatically with glutathione (γ-glutamyl-l-cysteinylglycine, GSH) or other intracellular thiol, glyoxalase I (abbreviated Glo1 here, but sometimes referred to as GlxI) isomerizes the hemithioacetal intermediate to form the corresponding thioester, such a S-d-lactoylglutathione, via a cis-enediolate intermediate (Fig. 14.5a), and glyoxalase II (Glo2) catalyzes its hydrolysis to obtain the product and regenerate the thiol (Honek 2015). Of interest for this chapter, Glo1 is a nickel-containing enzyme in some microorganisms (Honek 2017).

Glo1 from humans, yeast, and some bacteria is a zinc-dependent enzyme (Honek 2015, 2017), thus it came as a surprise to find the zinc-bound E. coli protein is inactive while the nickel-bound enzyme exhibits maximal activity (Clugston et al. 1998). Structural studies of E. coli Glo1 revealed a homodimer with its two active sites positioned at the interface of the subunits (Fig. 14.6a), closely resembling the structure of the human enzyme (He et al. 2000). Each nickel of E. coli Glo1 coordinates His5 and Glu56 from one subunit, His74′ and Glu122′ from the other subunit, and two water molecules, mimicking the octahedral zinc site of the human enzyme. The lack of activity for zinc-bound E. coli Glo1 arises from a switch to trigonal bipyramidal geometry for the metal, as shown by the crystal structure and XAS analysis (Davidson et al. 2000, 2001; He et al. 2000). These findings are consistent with the need for substrate to coordinate the metal binding site in a bidentate manner for the reaction to occur. Several other bacterial Glo1 enzymes utilize nickel including those from the important pathogens Y. pestis, P. aeruginosa, and Neisseria meningitis (Sukdeo et al. 2004), the solvent-producing bacterium Clostridium acetobutylicum (Suttisansanee et al. 2011), and the soil-dwelling actinomycete S. coelicolor (Suttisansanee and Honek 2019). The structure of the C. acetobutylicum enzyme demonstrated the conformational flexibility of the enzyme, with domain swapping allowing for metal coordination by side chains that all derive from the same subunit (Suttisansanee et al. 2011). S. coelicolor Glo1 is of special interest because it uses mycothiol, 1-d-myo-inositol-2-(N-acetyl-l-cysteinyl)amido-2-deoxy-α-d-glucopyranoside, instead of glutathione in the reaction. An alternate thiol known as trypanothione, N1,N8-bis(l-γ-glutamyl-l-hemicysteinyl-l-glycyl)spermidine, is used by the enzymes from the eukaryotic protozoa Leishmania major, Leishmania donovani, and Trypanosoma cruzi, which also depend on nickel for activity (Ariza et al. 2006; Greig et al. 2006; Padmanabhan et al. 2005; Vickers et al. 2004). The mechanism by which nickel is specifically incorporated into Glo1 is unknown, and no auxiliary proteins have been shown to facilitate its biosynthesis.

4.2 Acireductone Dioxygenase (ARD)

ARD catalyzes a key step in the methionine salvage cycle by converting 1,2-dihydroxy-3-keto-5-(thiomethyl)pent-1-ene (acireductone) to formate and 2-keto-4-methylthiobutyrate (KMTB) (Fig. 14.5b). KMTB is subsequently transaminated to form methionine, an amino acid and a precursor of S-adenosylmethionine (AdoMet). Among several cellular metabolic transformations that utilize AdoMet, polyamine biosynthesis produces methylthioadenosine (MTA). MTA is transformed by a series of reactions to regenerate acireductone, thus conserving the valuable sulfur atom in these metabolites. In 1993, a bacterial enzyme was purified that uses acireductone as substrate, but rather than producing KMTB it generates carbon monoxide, formate, and 3-methylthiopropionate (MTP) (Wray and Abeles 1993). Six years later this enzyme from K. oxytoca was shown to be a metal-substituted ARD (Dai et al. 1999). Thus, ARD catalyzes two distinct reactions depending on the identity of the metal that is bound: it produces KMTB when possessing iron and generates MTP when coordinating nickel. The purpose of the off-cycle pathway is unknown, but the observed partial nickel occupancy of the protein from wild-type cells suggests this reaction is physiologically relevant (Deshpande et al. 2017).

The nickel in K. oxytoca ARD, as probed by XAS analysis, is octahedral with O/N ligands that include 3–4 His residues (Al-Mjeni et al. 2002). The solution structure of nickel-ARD from this microorganism was obtained by NMR spectroscopy using special methods to overcome issues related to the metal ion paramagnetism (Pochapsky et al. 2002, 2008). Nickel coordinates His96, His98, Glu102, and His140, along with two water molecules (Fig. 14.6b). A crystal structure for cadmium-bound ARD from Bacillus anthracis indicated an essentially identical structure. NMR, XAS, and mutagenesis studies of K. oxytoca ARD revealed that iron binds to the same site of the protein as nickel (Chai et al. 2008; Ju et al. 2006), but ferrous ion resulted in subtle structural changes in the C-terminal region, a loop, and two helices so the two isoforms can be chromatographically separated. An extensive series of biomimetic studies have investigated the distinct chemistries in models of the active site with bound nickel versus iron [reviewed in (Deshpande et al. 2017)]. In all active forms, substrate displaces two waters from the six-coordinate metal and binds with bidentate coordination. Importantly, this binding likely results in a 6-membered ring for nickel and a 5-membered ring for iron. Subsequent reactions with oxygen generate distinct intermediates that decompose differentially to the observed products. Nothing is known about how the cell controls the metal specificity of the enzyme; e.g., no accessory proteins are known to be required for metal delivery.

4.3 Quercetin 2,4-Dioxygenase (QueD)

Many soil fungi and bacteria decompose flavonoid plant metabolites such as quercetin (3,5,7,3′,4′-pentahydroxyflavone) via oxidative metabolism that initiates by the action of quercetin 2,4-dioxygenase or quercetinase, releasing carbon monoxide and forming 2-protocatechuoylphloroglucinolcarboxylic acid (Fig. 14.5c) (QueD). The fungal enzyme uses copper as a cofactor, whereas bacterial forms of the protein are more promiscuous in their metal ion requirements (Fetzner 2012). For example, Bacillus subtilis quercetinase purifies with stoichiometric manganese, but is also active with cobalt and copper (Schaab et al. 2006). Of particular interest here is QueD from Streptomyces sp. FLA. This dioxygenase contains nickel when isolated from the wild-type microorganism and is most active using this metal ion (Nianios et al. 2015). Streptomyces QueD is dimeric, with each subunit exhibiting the typical 6-stranded β-barrel cupin fold (Fig. 14.6c) (Jeoung et al. 2017). Nickel binds to three histidine and one glutamic acid, with additional coordination by two water molecules. The C3 hydroxyl group of the substrate displaces one of the waters and binds to the metal ion. The second water molecule is lost upon binding of oxygen which exhibits side-on coordination (Jeoung et al. 2017). No information is available regarding how the protein specifically binds nickel.

4.4 Superoxide Dismutase (SOD)

The superoxide anion, O2, is both a reactive oxygen species and a cellular signaling compound (Wang et al. 2018). To limit the internal levels of this compound, cells utilize one or more superoxide dismutases (Fig. 14.5d) or a superoxide reductase (an enzyme that reduces this molecule to H2O2 and is not discussed further) (Abreu and Cabelli 2010; Sheng et al. 2014). Biology has evolved three types of SODs with independent sequences, structures, and metallocenters; i.e., the copper/zinc SODs, the manganese- or iron-containing SODs, and the nickel-containing SODs. This section describes the discovery, properties, structure, and activation of the latter enzymes (Campeciño and Maroney 2017).

In 1996, the first examples of nickel-containing SODs were purified from Streptomyces seoulensis IMSNU-1 and S. coelicolor ATCC 10147 (Kim et al. 1996; Youn et al. 1996a). These enzymes possess near stoichiometric amounts of nickel (0.74 equivalents per 13.4-kDa subunit) and, unlike Glo1 or ARD, exhibit distinct spectral properties associated with their metallocenters. For example, the ultraviolet-visible (UV-vis) spectrum of nickel SOD possesses an absorption maximum near 380 nm, a broad peak near 530 nm, and the typical protein-derived electronic transitions. More remarkable, the nickel-containing SOD exhibits an electron paramagnetic resonance (EPR) spectrum with g values of 2.304, 2.248, and 2.012, where the latter component is split into a triplet (Youn et al. 1996a). This spectrum was attributed to a low spin (S = ½) Ni3+ with an axial N-ligand leading to hyperfine splitting of the high-field feature. S. griseus possesses both a nickel-containing SOD, with properties essentially identical to that just described (0.89 Ni/13.0-kDa subunit, analogous UV-vis and EPR spectra), and an iron-containing SOD (Youn et al. 1996b). S. coelicolor also possesses both types of SOD, and their synthesis was found to be regulated by nickel (Kim et al. 1996). Analysis of a variety of Streptomyces isolates from clinical and soil samples demonstrates the universality of nickel-dependent SODs in this genus, with many strains also having iron-dependent activity (Leclere et al. 1999). Furthermore, a phylogenetic analysis demonstrates that sodN, the gene encoding these enzymes, is distributed widely in actinomycetes and cyanobacteria, along with being present in some gammaproteobacteria, bacteriodetes, planctomycetes, deltaproteobacteria, and even two marine eukaryotes (Dupont et al. 2008). Curiously, a Mycobacterium sp. has two copies of sodN, though it is unknown if either is functional.

Analysis of S. coelicolor sodN indicates ninefold enhanced transcription in the presence of nickel and shows that it encodes a 14-residue-extended precursor protein lacking activity (Kim et al. 1998b). Expression of sodN in E. coli provides only the longer version of the protein with no activity, whereas expression of a truncated gene yields small levels of SOD activity (Kim et al. 1998b). Transcription of the gene encoding the iron-containing SOD, sodF, also is regulated at the transcriptional level by nickel, but in this case the transcript levels are reduced by the presence of this metal (Chung et al. 1999; Kim et al. 1998a). The repression of sodF in S. griseus is accomplished by a complex of SrnR, a nickel sensor, and SrnQ (Sect. 14.2.1.3) (Kim et al. 2003b). In addition, the nickel-bound sensor Nur acts as a repressor of sodF and an activator of sodN in S. coelicolor (Sect. 14.2.1.4) (Ahn et al. 2006; An et al. 2009).

Initial structural studies of the nickel-containing SOD focused on XAS of the S. seoulensis enzyme, revealing a five-coordinate metal site with about three sulfur donors, one nitrogen donor, and a nitrogen or oxygen donor (Choudhury et al. 1999). Reduction of nickel by dithionite results in loss of a nitrogen or oxygen donor, suggesting ligand dissociation. EPR spectra for the recombinant protein purified from E. coli with truncated S. coelicolor sodN has the same appearance as the wild-type enzyme, whereas the H1Q variant protein spectrum is altered; this result is consistent with His1 serving as a nickel ligand (Bryngelson et al. 2004). Studies comparing the XAS and magnetic circular dichroism spectra of the wild-type protein with variants involving Cys2 and Cys6 confirmed these residues also coordinate the metal (Johnson et al. 2010; Ryan et al. 2010). A detailed understanding of the nickel SOD metallocenter and the overall protein architecture is derived from crystal structures for the S. seoulensis (Wuerges et al. 2004) and S. coelicolor (Barondeau et al. 2004) enzymes (Fig. 14.6d). The proteins from both sources reveal nearly identical hexamers of 4-helix bundle monomers, as expected for the almost identical sequences. Nickel binds to what has been termed a “nickel-hook motif”, which requires protein truncation. Planar coordination of the metal is provided via the N-terminal amine, the backbone amide of the second residue, and the side chains of Cys2 and Cys6. This coordination environment is thought to lower the reduction potential of the metal by over 1 V, allowing it to be capable of catalyzing the dismutation reaction. The oxidized form of the protein also includes His1 as an axial ligand, but this ligand dissociates upon reduction (Barondeau et al. 2004; Wuerges et al. 2004). Several other nearby residues help to stabilize these interactions, including Pro5 that is present in cis configuration, but the region is unfolded in the absence of the metal.

Genomic sequence analyses were used to identify a gene adjacent to sodN in Prochlorococcus marinus MIT9313 that was hypothesized, then shown, to encode a protease specific for the N-terminus of SOD, as needed for enzyme activation (Eitinger 2004). The product of sodX is an SOD-specific peptidase that allows for generation of active SOD activity in E. coli cells co-expressing the P. marinus sodN. Bioinformatics studies have shown that most, but not all, sodN genes are adjacent to sodX (Dupont et al. 2008). Another protein, CbiXhp, had been proposed to function during activation of nickel-containing SOD in S. seoulensis (Kim et al. 2003a). This protein, named on the basis of its sequence similarity to cobaltochelatase (CbiX) of Bacillus megaterium, was purified by its ability to bind to nickel-nitrilotriacetic acid resin using a histidine-rich sequence at its C-terminus. Co-expression of the S. seoulensis genes encoding CbiXhp and SodN in Streptococcus lividans leads to greater levels of SOD activity than for expression of only sodN. It is possible this protein functions in nickel storage or as a nickel metallochaperone; however, no further analysis of S. seoulensis CbiXhp has been reported and homologs to this gene are not associated with sodN of other microorganisms. Thus, it is likely the enhanced SOD activity noted in those studies was fortuitous.

4.5 Urease

Urease and its substrate urea hold a special place in the history of science. The Dutch botanist Herman Boerhaave first isolated urea from urine in ~1727, although the French chemist Hilaire Marin Rouelle is often credited with the discovery of this substance in 1773 (Kurzer and Sanderson 1956). The name urea was coined by Antoine Francis comte de Fourcroy and Louis Vauquelin, who studied the compound in 1799 (Fourcroy and Vauguelin 1799). Urea is the first organic molecule ever synthesized from inorganic materials, an accomplishment of the German chemist Friedrich Wöhler in 1828 (Wöhler 1828). Others had shown that urea in urine transforms into carbonate and ammonia, but Louis Pasteur reported in 1860 that a living organism is responsible for this alkaline fermentation (Pasteur 1860). Shortly thereafter Philippe van Tieghem isolated a bacterium, later called Micrococcus ureae, that catalyzes this decomposition reaction (van Tieghem 1984), and in 1890 Pierre Miguel described several additional “ureolytic” microorganisms and termed the transforming agent “urease” (Miquel 1890). The next few decades included intense discussions of what physical components confer enzyme activities. Key to answering this question was the discovery by Takeuchi in 1909 that soybean, among other plants, possesses large amounts of urease activity (Takeuchi 1909). Working with seeds of the jack bean plant in 1929, James B. Sumner purified urease by crystallization, the first enzyme ever crystallized, hence demonstrating that enzymes are proteins (Sumner 1926). Nearly 50 years later, Bert Zerner and colleagues showed that jack bean urease contains nickel (Dixon et al. 1975), therefore becoming the first identified nickel-dependent enzyme. Bacterial ureases also contain nickel, as first demonstrated for the enzyme from Arthrobacter oxydans (Schneider and Kaltwasser 1984). The following paragraphs describe the reaction, distribution, structures, and biosynthesis of microbial ureases.

4.5.1 Urease Functional Significance

Urea is an important metabolite in the global nitrogen cycle and its hydrolysis by urease (Fig. 14.5e) has far-reaching consequences. The substrate urea is produced industrially at a vast scale (>200 million metric tons per year) for use as a nitrogen-based fertilizer. When such fertilizers are released into the environment, urea hydrolysis by plant and microbial ureases produces ammonium ions for use by crops and raises the pH due to consumption of protons. Rapid metabolism of fertilizer urea may damage agricultural crops due to ammonia toxicity or alkalinity, and these conditions volatilize the ammonia causing loss of this valuable nitrogen source (Mobley and Hausinger 1989). Several biological sources of urea exist, with the most prominent being the arginase-catalyzed hydrolysis of arginine. Arginase is found in many microorganisms and plays an essential role in mammals, where it functions to detoxify ammonia via the urea cycle that produces the urea found in urine. Urea metabolism by microorganisms can have detrimental medical consequences (Burne and Chen 2000; McLean et al. 1988; Mobley and Hausinger 1989; Mobley et al. 1995b). Ureolytic pathogens of the urinary tract are responsible for a substantial fraction (10–15%) of urinary stones due to high pH-induced precipitation of struvite and apatite salts (Flannigan et al. 2014; Griffith et al. 1976). The same process accounts for encrustation of catheters (Norsworthy and Pearson 2017). Infections of the urinary tract also can lead to pyelonephritis, the acute inflammation of the kidney and its pelvis (Nielubowicz and Mobley 2010). Peptic ulcers, gastritis, and even stomach cancers are associated with colonization of the stomach lining by H. pylori, a pathogen containing very high levels of urease (Atherton 2006; Covacci et al. 1999; Kusters et al. 2006; Wroblewski et al. 2010). Enzymatic metabolism of urea diffusing into the stomach from the bloodstream consumes protons immediately around the microorganism, thus acting as a buffer against the low pH environment (Sachs et al. 2002). Also of interest, high concentrations of ammonia produced by microbial metabolism of urea is implicated in cases of ammonia encephalopathy, hepatic encephalopathy, and hepatic coma (Mobley and Hausinger 1989; Mobley et al. 1995b; Rai et al. 2015).

Microbial urease activity is widespread in the environment, with the enzyme distributed in selected bacteria, archaea, yeasts, filamentous fungi, and algae (Mobley and Hausinger 1989). Its regulation varies with the source (Collins and D’Orazio 1993). For example, many microorganisms increase urease production during times of nitrogen starvation so as to utilize urea as a nitrogen source. Some strains that infect the urinary tract induce urease synthesis in the presence of elevated concentrations of urea. Furthermore, still other species exhibit generally constant levels of the enzyme. Urease typically is localized in the cytoplasm; however, this protein is quite stable and maintains its activity when released into the environment upon cell death, leading to the incorrect notion that urease is secreted by microbes (Marcus and Scott 2001).

4.5.2 Urease Structure

The structure of urease exhibits one of four architectures, depending on the source. Most bacterial forms of the enzyme contain three types of subunits (UreA, UreB, and UreC), each present in three copies, i.e. (UreABC)3, with the entire structure containing three dinuclear nickel active sites as first elucidated for urease from Klebsiella aerogenes (Fig. 14.6e) (Jabri et al. 1995; Pearson et al. 1997). The extensively studied enzyme from Sporosarcina (formerly Bacillus) pasteurii has an analogous structure (Benini et al. 1999). Yersinia enterocolitica urease also has three similar types of subunits, but structures obtained by crystallography (unpublished) and cryo-electron microscopy (Righetto et al. 2020) have shown these subunits assemble into a [(UreABC)3]4 supermolecular spherical structure containing 12 active sites. Similarly, crystallography and cryo-electron microscopy studies indicate a dodecamer spherical structure is present in H. pylori and other Helicobacter species (Cunha et al. 2021; Ha et al. 2001); however, in these cases only two types of subunits are present with the smaller subunit (UreA) corresponding to a fusion of UreA plus UreB of most bacteria, and the larger subunit (UreB) corresponding to UreC in other genera. This [(UreAB)3]4 situation also occurs for a dinuclear iron-containing urease from Helicobacter mustelae (Carter et al. 2011). In eukaryotes, a single type of subunit contains sequences homologous to each of the bacterial subunits. Three copies of this fused subunit assemble into a trimer, and two trimers stack together back-to-back forming a hexamer with six active dinuclear nickel active sites. Although no microbial example of such a structure has been reported, the architecture of fungal ureases probably resembles that of the jack bean enzyme (Balasubramanian and Ponnuraj 2010). The active sites of all ureases are identical, with their two nickel metal ions bridged by a carbamylated lysine residue, two histidine residues binding one of those nickel ions, and two histidine residues plus an aspartic acid residue coordinated to the second nickel ion. Each metal ion has a terminally-bound water and a third water bridges the two nickel ions, thus providing 5-coordinate and 6-coordinate geometries for the dinuclear site. The reaction mechanism of the enzyme metallocenter was recently reviewed (Mazzei et al. 2020).

4.5.3 Urease Maturation

The bacterial genes encoding the urease subunits typically are clustered with genes encoding accessory proteins that are required for urease activation (Mobley et al. 1995b). For example, ureDABCEFG of K. aerogenes encodes the three-subunit urease and four proteins (UreD, UreE, UreF, and UreG) needed to obtain a mature enzyme (Lee et al. 1992; Mulrooney and Hausinger 1990). Similarly, ureABIEFGH of H. pylori codes for the two-subunit enzyme, a proton-gated urea channel (UreI) (Weeks et al. 2000), and the same four maturation components (note: Helicobacter UreH is homologous to UreD of other bacteria) (Cussac et al. 1992; Labigne et al. 1991). In other microorganisms, the order of these genes may vary and additional genes may interrupt or flank the cluster such as those encoding ABC-type nickel transporters (Sect. 14.2.2.2) in A. pleuropneumoniae (Bossè et al. 2001) and S. salivarius 57.1 (Chen and Burne 2003) or encoding nickel permeases (Sect. 14.2.2.3) such as UreH and UreJ in Bacillus TB-90 (Maeda et al. 1994) and B. bronchiseptica (McMillan et al. 1998), respectively. In addition, the products of non-contiguous genes may facilitate urease activation in some species by assisting in nickel uptake, such as NixA in H. pylori (Wolfram and Bauerfeind 2002) (Sect. 14.2.2.3), or nickel delivery, such as HypA and HypB in the same organism (Olson et al. 2001). The biosynthesis of active urease almost universally requires the participation of UreD/H (two designations for the same protein), UreE, UreF, and UreG (Farrugia et al. 2013b; Nim and Wong 2019). One dramatic exception to this statement is urease of B. subtilis which possesses three types of subunits closely related to those of other bacterial ureases yet lacks genes related to the four urease maturation genes (Kim et al. 2005).

The urease maturation pathway (Fig. 14.7) begins with the metallochaperone or metal-delivery protein UreE. This protein from Klebsiella aerogenes possesses a carboxyl terminal His-rich sequence (10 of the last 15 residues) that binds multiple nickel ions (5–6 nickel per dimer, Kd ~ 10 μM) (Grossoehme et al. 2007; Lee et al. 1993). By inserting a stop codon into the corresponding gene, one obtains a truncated protein that lacks the His-rich region yet it still functions in urease activation and binds ~2 nickel ions per dimer (Colpas et al. 1998, 1999; Colpas and Hausinger 2000). The His-rich region is not a universal component of UreE as demonstrated by the H. pylori protein (Musiani et al. 2004). Crystal structures of UreE (Fig. 14.7a) are known for the proteins from K. aerogenes (Song et al. 2001), S. pasteurii (Remaut et al. 2001; Zambelli et al. 2013), and H. pylori (Banaszak et al. 2012; Shi et al. 2010). In all cases, the protein exists as a homodimer with each subunit possessing a metal-binding domain and a separate domain that is disposable for urease maturation (Mulrooney et al. 2005). The essential nickel-binding site is located at the interface of the UreE subunits (Colpas et al. 1999; Colpas and Hausinger 2000) where it is bound by complementary histidine residues and additional ligands in a disordered region near the carboxyl terminus (Banaszak et al. 2012; Zambelli et al. 2013). Secondary metal-binding sites may “feed” the metal into the essential UreE interfacial site for subsequent transfer to urease (Colpas et al. 1999; Grossoehme et al. 2007; Lee et al. 1993). In H. pylori, HypA·UreE is known from chemical cross-linking of the proteins (Benoit et al. 2007). This complex presumably reflects the previously mentioned H. pylori-specific enhancement of urease activity by HypA (Olson et al. 2001). Evidence indicates that nickel is transferred from HypA to UreE via a HypA·(UreE)2 complex (Hu et al. 2018; Yang et al. 2014), with HypA and UreG competing for the UreE binding site (Benoit et al. 2012).

Fig. 14.7
figure 7

Urease maturation pathway. (a) UreE is a homodimeric nickel metallochaperone, as shown for the protein from Sporosarcina pasteurii (PDB ID 4l3k). (b) UreE forms complexes with and transfers nickel to its cognate UreG, as depicted for the homodimeric UreG from Klebsiella pneumoniae with bound nickel and GMPNP (PDB ID 5xkt). (c) UreG binds to a complex of UreF and UreD/H, as illustrated for the (UreF·UreD)2 complex of Helicobacter pylori (PDB ID 3sf5). (d) The resulting complex of three accessory proteins is illustrated by the GDP-bound (UreG·UreF·UreD)2 from H. pylori (PDB ID 4hi0). (e) The accessory proteins bind to urease apoprotein, shown for (UreA·UreB·UreC)3 of Klebsiella aerogenes (PDB ID 1kra). (f) A model of the resulting complex has (UreG·UreF·UreD/H) bound via the UreD/H component to UreB of the urease apoprotein. GTP hydrolysis at UreG is coupled to release of nickel from that component, with the metal traversing a molecular tunnel through UreF and UreD (see inset) to be delivered to the nascent active site in UreC. Proteins are depicted in distinct colors using cartoon mode with semi-transparent surfaces, nickel atoms are shown as green spheres, and selected side chains along with nucleotides are indicated by sticks

The second critical component in urease maturation is UreG. This protein, a nickel-binding GTPase, has been purified and characterized from a wide range of organisms, including the bacteria K. aerogenes (Boer et al. 2010; Martin-Diaconescu et al. 2017; Moncrief and Hausinger 1997), Klebsiella pneumoniae (Yuen et al. 2017), S. pasteurii (D’Urzo et al. 2014; Zambelli et al. 2005), H. pylori (Fong et al. 2013; Yang et al. 2015; Zambelli et al. 2009), M. tuberculosis (Zambelli et al. 2007b), and the archaea Methanocaldococcus jannaschii and Metallosphaera sedula (Miraula et al. 2015). Whereas some UreG proteins are monomeric in dilute solution (Boer et al. 2010; Moncrief and Hausinger 1997), others are dimers (Zambelli et al. 2005, 2007b), and still others exhibit a monomer-to-dimer equilibrium with the latter species stabilized by addition of zinc (D’Urzo et al. 2014; Miraula et al. 2015; Zambelli et al. 2009). All UreG species bind zinc or nickel, but the number of equivalents bound and the measured affinities differ for each protein [reviewed in Martin-Diaconescu et al. (2017)]. The structure of an N-terminally truncated UreG from K. pneumoniae in the presence of nickel and the GTP analogue GMPPNP reveals a dimer, with the critical nickel ion coordinated symmetrically by the conserved Cys-X-His motifs at the interface of the subunits (Fig. 14.7b) (Yuen et al. 2017). Additional nickel ions associate with each of the two GMPPNP molecules, corresponding to the more typical Mg∙GTP interaction. Despite the ability of isolated K. pneumoniae UreG to crystallize into a distinct structure (Yuen et al. 2017) and the reported structure of a protein complex including H. pylori UreG (Fong et al. 2013), extensive evidence suggests that the S. pasteurii and archaeal proteins are intrinsically disordered (Miraula et al. 2015; Palombo et al. 2017; Zambelli et al. 2005, 2012). The significance of this finding is unclear.

Several studies have demonstrated an interaction between UreG and UreE, consistent with the two proteins transiently sharing coordination of the nickel. For example, nickel stimulates the interaction between the K. aerogenes proteins, forming a UreG∙Ni∙(UreE)2 complex (Boer et al. 2010). In contrast, a (UreG)2∙Ni∙(UreE)2 complex was identified for the S. pasteurii proteins (Merloni et al. 2014). Using the H. pylori components, (UreG)2∙Zn∙(UreE)2 (Bellucci et al. 2009) along with UreG∙Ni∙(UreE)2 and (UreG)2∙Ni∙(UreE)2 were observed, with the latter found to predominate when Mg∙GTP was present (Yang et al. 2015). No structure is available for any complex of UreE and UreG, but a model was proposed for the H. pylori species (Bellucci et al. 2009). A consensus hypothesis has emerged that nickel-loaded UreE transfers nickel to UreG by direct interaction of these protein components.

In the next step, UreG interacts with UreF and UreD/H to form a complex that has been characterized for components from H. pylori and K. aerogenes. Most notably, the crystal structure of H. pylori (UreG·UreF·UreD)2 is available with GDP bound to UreG (Fig. 14.7d) (Fong et al. 2013), as well as structures of the precursors (UreF·UreD)2 (Fig. 14.7c) (Fong et al. 2011) and (UreF)2 (Lam et al. 2010). In addition, chemical cross-linking and ion mobility-mass spectrometric analysis of K. aerogenes proteins identified (UreG·UreF·UreD-MBP)2, where UreD-MBP is a fusion of UreD with the maltose-binding protein (MBP, used to enhance solubility) (Carter and Hausinger 2010). This K. aerogenes complex binds to the cognate urease apoprotein, (UreA·UreB·UreC)3 (Fig. 14.7e), to form (UreG·UreF·UreD-MBP)3·(UreA·UreB·UreC)3 (Fig. 14.7f) (Eschweiler et al. 2018; Farrugia et al. 2013a). The K. aerogenes apoprotein is identical in structure to the holoprotein, except for the absence of nickel and the lack of carbamylation at the active site lysine (Jabri and Karplus 1996). The connection between the accessory protein heterotrimer and urease occurs via UreD and UreB as shown by production and characterization of a (UreD)3·(UreA·UreB·UreC)3 complex (Park et al. 1994). Similarly, co-expression of ureF, ureD, and ureABC allows for production of (UreF·UreD)3·(UreA·UreB·UreC)3 (Moncrief and Hausinger 1996). Significantly, incubation of (UreF·UreD)3·(UreA·UreB·UreC)3 with UreG, UreE, and nickel resulted in GTP-dependent generation of near fully active urease (Soriano and Hausinger 1999; Soriano et al. 2000). No structure is reported for any accessory protein bound to the urease apoprotein.

A current model for activation of bacterial urease (Fig. 14.7e) is derived by merging studies of the K. aerogenes and H. pylori urease systems. From ion mobility-mass spectrometry studies (Eschweiler et al. 2018; Farrugia et al. 2013a), it appears that the dimeric (UreG·UreF·UreD)2 complex dissociates to monomeric UreG·UreF·UreD units when forming the (UreG·UreF·UreD)3·(UreA·UreB·UreC)3 apoprotein complex. Chemical cross-linking studies indicate specific interactions between UreD and UreB, requiring the urease subunit to undergo a hinge-like motion that provides better access to the nascent active site, and small angle X-ray scattering studies are consistent with the overall structure of this species (Chang et al. 2004; Quiroz-Valenzuela et al. 2008). Computational studies using the H. pylori (UreH·UreF·UreG)2 crystal structure (Fong et al. 2013) and a homology model for K. aerogenes UreG·UreF·UreD (Farrugia et al. 2015) indicated the likely existence of a molecular channel for nickel transfer through UreF and UreH/D. Support for such a tunnel was obtained for the K. aerogenes components; UreD variants predicted to occlude the tunnel abolished the ability to activate the urease apoprotein (Farrugia et al. 2015). Although the role of GTP hydrolysis during activation remains unclear, a reasonable hypothesis is that nucleotide hydrolysis is coupled to the release of nickel into the tunnel. Diffusion of the metal ion through the molecular channel provides access to the nascent active site that otherwise is buried within UreC. Coordination of nickel to the active site residues leads to conformational changes resulting in release of the accessory proteins.

4.6 [NiFe] Hydrogenase

Hydrogenases catalyze the reversible oxidation of dihydrogen gas to form two protons and two electrons (Fig. 14.5f) (Adams et al. 1981; Peters et al. 2015; Vignais et al. 2001; Vignais and Billoud 2007). In some microorganisms, the electrons derived from hydrogen are used to reduce oxygen, sulfate, carbon dioxide, nitrate, or other electron acceptors, and this coupling serves as a means to provide energy to the cell. In particular microbes, the reverse reaction is used to eliminate excess cellular reducing potential derived by fermentation or photosynthesis. Thus, hydrogenases play central roles in the energy metabolism of many microorganisms. Three main enzyme classes are differentiated on the basis of their metallocenters: [NiFe] hydrogenases, [FeFe] hydrogenases, and iron-only hydrogenases. Only the [NiFe] hydrogenases are relevant to this chapter.

4.6.1 Types and Structures of [NiFe] Hydrogenases

All of the [NiFe] hydrogenases possess at least two types of subunits: a large subunit (~60 kDa) containing nickel plus iron at a dinuclear [NiFe] active site cluster and a small subunit (~30 kDa) typically containing three iron-sulfur clusters that shuttle electrons between the [NiFe] active site and an external electron carrier. The names of the genes encoding the subunits are quite variable in different microorganisms with further nomenclature complications arising in microorganisms possessing multiple [NiFe] hydrogenases. Four distinct functional groupings and 22 subgroups of these enzymes have been identified on the basis of sequence analyses (Greening et al. 2016; Peters et al. 2015; Vignais and Billoud 2007).

Group 1 [NiFe] hydrogenases are periplasmic or membrane-associated and utilize H2 for energy generation. Many periplasmic, heterodimeric enzymes have been characterized; e.g., the first [NiFe] hydrogenase to be structurally elucidated was the heterodimeric enzyme from Desulfovibrio gigas (Volbeda et al. 1995). The large subunit contains the deeply buried [NiFe] center (this site will be described separately, below), and the small subunit contains a linear arrangement of [4Fe-4S], [3Fe-4S], and [4Fe-4S] clusters for electron transfer with a partner electron carrier located in the membrane. Similar heterodimeric structures are characterized for [NiFe] hydrogenases from other sulfate-reducing bacteria including Desulfovibrio desulfuricans, Desulfovibrio vulgaris Miyazaki F (Fig. 14.6f), Desulfomicrobium norwegicum (formerly D. baculatum), D. vulgaris Hildenborough, and Desulfovibrio fructosovorans (Garcin et al. 1999; Higuchi et al. 1997; Marques et al. 2010; Matias et al. 2001; Volbeda et al. 2005). Duplication of this architecture is seen in several non-sulfate-reducing bacteria such as Hydrogenovibrio marinus, the photosynthetic purple-sulfur bacterium Allochromatium vinosum, and an enzyme with very high affinity for H2 from the actinobacterium C. necator (Ogata et al. 2010; Schäfer et al. 2016; Shomura et al. 2011). Additional enzymes that operate using low concentrations of H2 are found in soil bacteria such as Streptomyces (Constant et al. 2010, 2011) and may have similar structures.

Additional subunits beyond the dimeric enzyme are present in many other group 1 cases. For example, E. coli uses a membrane-bound cytochrome b to shuttle electrons from hydrogenase-1 into the quinone pool, which then transfers electrons to an external electron acceptor. This enzyme is encoded by hyaAB and the cytochrome b is encoded by hyaC, all found in the hyaABCDEF operon (Sargent 2016). The structure of E. coli hydrogenase-1 contains two copies of each hydrogenase subunit (Volbeda et al. 2012) and this (HyaAB)2 unit forms a complex with a single copy of HyaC (Volbeda et al. 2013). The C-termini of each small subunit have helical extensions that interact with a four-helix bundle of the membrane-bound cytochrome b. The three [FeS] clusters in each small subunit [including one with an extra cysteine residue replacing an inorganic sulfide (Volbeda et al. 2012)] aligns with the heme of HyaC to facilitate electron transfer. A second group 1 [NiFe] hydrogenase in E. coli, hydrogenase-2, has two standard subunits encoded by hybO and hybC plus an [FeS]-containing protein in the periplasm and a membrane-associated protein, encoded by hybA and hybB respectively, all associated with the hybOABCDEFG operon (Sargent 2016). The structure of (HybCO)2 is known and closely resembles that of (HyaAB)2, but only a predicted structure is available for the (HybCOAB)2 complex (Beaton et al. 2018).

Group 2 [NiFe] hydrogenases are cytoplasmic proteins that catalyze H2 uptake. Four subgroupings are found, but only two are briefly mentioned here. First, the cyanobacterial enzymes (with subunits often designated HupSL) recapture the H2 that is produced as a requisite side reaction of nitrogenase during its reduction of N2 (Tamagnini et al. 2007). Second, H2-sensing enzymes (commonly designated HupUV or HoxBC) participate in the transcriptional regulation of other hydrogenases. C. necator provides a paradigm for this type of [NiFe] hydrogenase system; its group 2 cytoplasmic H2-sensor HoxBC is used to regulate production of the NAD-dependent, membrane-bound, HoxFUYH group 3 enzyme (see below) (Lenz and Friedrich 1998). No group 2 [NiFe] hydrogenase is structurally characterized at this time.

The group 3 [NiFe] hydrogenases typically function as bidirectional cytoplasmic enzymes that possess more than the two standard subunits. These additional components bind and transfer electrons to cytoplasmic electron carriers such as NAD+, NADP+, or 8-hydroxy-5-deazaflavin (coenzyme F420, a cofactor found primarily found in methanogenic archaea). The structure of the F420-reducing [NiFe] hydrogenase from Methanothermobacter marburgensis is known from cryo-electron microscopy studies (Vitt et al. 2014). The large and small hydrogenase subunits, FrhA and FrhG, are relatively standard (except the first [4Fe-4S] is coordinated by Asp and three Cys instead of the usual four Cys). Also present is the F420-binding component, FrhB, that contains another [4Fe-4S] (aligned with the three clusters in FrhG) and FAD that interacts with the coenzyme F420. These three components form a spherical (FrhABG)12 complex; i.e., a dodecamer of the heterotrimer. Such a large assemblage is not found in all group 3 members. For example, a cytoplasmic, NAD+-reducing [NiFe] hydrogenase from Hydrogenophilus thermoluteolus Th-1 is a heterotetramer (Shomura et al. 2017). This protein complex includes the large subunit HoxH (with an extra glutamate ligand of nickel compared to the typical situation, see below), the small subunit HoxU (with a non-typical and non-linear [4Fe-4S], [4Fe-4S], [2Fe-2S] arrangement), HoxY (containing a [4Fe-4S] between the [NiFe] site and first [4Fe-4S] of the small subunit), and HoxF (possessing a [4Fe-4S] in line with two clusters in the small subunit and an FMN at the NAD+-binding site). Of further interest, some enzymes in this group act in electron bifurcation reactions, such as a methanogen MvhAGD enzyme that uses H2 (E′o = −414 mV) to simultaneously carry out an energetically downhill reaction (the reduction of a heterodisulfide, E′o = −140 mV) and an energetically uphill reaction (the reduction of a ferredoxin, E′o = −500 mV) (Thauer et al. 2010).

Group 4 [NiFe] hydrogenases are complex membrane-bound enzymes, typically containing at least six types of subunits, that evolve hydrogen from cells using low potential reductants, such as carbon monoxide or formate, and often use ferredoxin as an electron carrier. The classic example of this type of enzyme is E. coli hydrogenase-3, encoded by hycBCDEFGHI where hycEG corresponds to the standard subunit-encoding genes. This enzyme functions as a formate hydrogenlyase to metabolize formate to H2 and CO2 (Sargent 2016). Closely resembling the operon sequence for hydrogenase-3 is hyfABCGHIJ of E. coli, encoding hydrogenase-4 that may also function in formate metabolism; however, hydrogenase-4 activity has not been established. Also noteworthy are group 4 enzymes that function as energy-converting hydrogenases (ECHs) that couple electron transfer to ion translocation. One example of this type of enzyme is EchABCDEF of Methanosarcina barkeri that utilizes the proton motive force to facilitate the thermodynamically unfavorable reduction of ferredoxin by H2 (Meuer et al. 1999). Sequence analyses indicate the small subunits of these enzymes contain a single [4Fe-4S] cluster (Greening et al. 2016) although other components may provide further electron transfer sites. No structures of group 4 enzymes are available.

The active sites of all four groups of [NiFe] hydrogenases possess a [NiFe] dinuclear cluster. The structure of the heterodimeric, periplasmic enzyme from D. gigas first revealed this site as nickel with another metal (tentatively assigned to iron) that binds three non-protein ligands depicted at the time as water molecules (Volbeda et al. 1995). Four cysteine residues bind the nickel and two of these ligands bridge to the second metal. Follow-up studies with this protein confirmed the identity of iron and suggested the three non-protein ligands are diatomic molecules (Volbeda et al. 1996). The nature of the iron-coordinating ligands was established from studies in which Chromatium vinosum [NiFe] hydrogenase was purified from cells grown on medium containing 13C and 15N, then subjected to infrared spectroscopy (Happe et al. 1997; Pierik et al. 1999). Intense features in the 2150 to 1850 cm−1 region demonstrate the presence of one carbon monoxide and two cyanide ligands. Subsequent studies (Ash et al. 2019; Fontecilla-Camps et al. 2007; Lubitz et al. 2014; Ogata et al. 2016; Shafaat et al. 2013; Volbeda and Fontecilla-Camps 2017), clarified the structures and properties of the [NiFe] site in enzymes from different species and in different states. Of particular interest is the active site structure of the H2-reduced protein from D. vulgaris Miyazaki F at the remarkable resolution of 0.89 Å (Fig. 14.6f) revealing a hydride bridging the two metal ions (Ogata et al. 2015). In contrast to the typical four cysteines binding the dinuclear site, the D. norwegicum and D. vulgaris Hildenborough enzymes have a selenocysteine substituting for the C-terminal cysteine residue (Garcin et al. 1999; Marques et al. 2010). Another interesting structural variation is a [NiFe] hydrogenase from C. necator that binds four cyanide ligands (Van der Linden et al. 2004). The biosynthesis of the extraordinary [NiFe] site is an amazingly complicated process as described in the following paragraphs.

4.6.2 [NiFe] Hydrogenase Maturation

Our understanding of [NiFe] hydrogenase maturation is highly dependent on genetic and biochemical studies performed in E. coli where investigators have shown the hypABCDE operon is needed for activation of hydrogenases-1, -2, and -3 (Jacobi et al. 1992; Lutz et al. 1991). The HybA-HypE proteins encoded by this operon, the non-subunit gene products encoded by the hya, hyb, and hyc operons, and other E. coli proteins all work in concert to generate active forms of the organism’s three enzymes. A detailed description of these extensive studies for E. coli, and analogous biochemical and genetic work carried out with other microorganisms, is beyond the scope of this chapter, so the reader is referred to review articles describing the multiple insights generated by the laboratories of August Böck (Böck et al. 2006), Deborah Zamble (Lacasse and Zamble 2016; Leach and Zamble 2007), Gary Sawers (Forzi and Sawers 2007; Sawers and Pinske 2017), and others (Cheng et al. 2016; Peters et al. 2015; Sargent 2016; Senger et al. 2017). Here, I focus on a structural description of the elaborate [NiFe] hydrogenase activation process (Fig. 14.8), where much of this remarkable set of work was carried out using Thermococcus kodakarensis KOD1 components in the laboratory of Kunio Miki (Miki et al. 2020).

Fig. 14.8
figure 8

Hydrogenase maturation pathway. All structures are from Thermococcus kodakarensis except where indicated. (Upper left) The HypE homodimer (PDB ID 3wjr, cyan and dark blue) binds to monomeric HypF (PDB ID 4g9i, red to blue rainbow color to illustrate its four domains) forming the heterotetramer (HypF-HypE)2 (PDB ID 3vti, from Caldanaerobacter subterraneus subspecies tengcongensis MB4). (Upper middle) HypF hydrolyzes carbamoylphosphate, adenylylates the resulting carbamate, and converts the carboxyl terminal cysteine of HypE to the carbamoylated residue (PDB ID 3wjp). HypE dehydrates this modification to generate a thiocyanylated cysteine (PDB ID 3wjq). (Upper right) Monomeric HypC (PDB ID 2z1c, maroon) and monomeric HypD (PDB ID 2z1d, yellow) generate a heterodimer (PDB ID 3vyr). Two molecules of this heterodimer bind iron and form a complex with the cyanylated HypE dimer, yielding the dimer of HypC·HypD·HypE (PDB ID 3vys). (Middle left) Monomeric HypX (PDB ID 6j0p, white, from Aquifex aeolicus) is a formyl-THF decarbonylase. Carbon monoxide derived from this reaction and two cyanide from cyanylated HypE become ligands to iron bound to HypC·HypD. The iron and its diatomic ligands are transferred into the hydrogenase large subunit, HyhL. (Bottom right) Dimeric HypB (PDB ID 3vx3, orange and brown) binds two molecules of monomeric HypA (PDB ID 3a43, green) to form a complex (PDB ID 5aun) with multiple nickel-binding sites. Nucleotide hydrolysis (ATP or GTP depending on the species) by this complex is coupled to nickel insertion into HyhL (slate blue) via HyhL·HypA (PDB ID 5yy0). (Bottom left) The HybD endoprotease (PDB ID 5ija, wheat) cleaves near the carboxyl terminus of HyhL causing the [NiFe] cluster to become buried in the protein and allowing for complex formation with the small subunit HyhS. Proteins are shown in cartoon depictions with semi-transparent surfaces. Metal and inorganic sulfur atoms are shown as spheres (nickel in green, zinc in grey, iron as brown, and sulfide as yellow). Selected side chains and nucleotides are illustrated in stick mode

The cyanide ligands of the metallocenter are generated by a multi-step reaction sequence involving HypE and HypF. Structures are known for dimeric HypE from T. kodakarensis (Watanabe et al. 2007) as well as monomeric HypF proteins from this organism (Tominaga et al. 2012), Caldanaerobacter subterraneus (Shomura and Higuchi 2012), and (in truncated form) E. coli (Petkun et al. 2011). In addition, the structure is known for the transient heterotetramer (HypE·HypF)2 from C. subterraneus (Shomura and Higuchi 2012). The four-domain HypF hydrolyzes carbamoylphosphate to form carbamate that reacts with ATP to generate carbamoyladenylate, which is used to generate a thiocarboxamide at the C-terminal cysteine residue of HypE (Reissmann et al. 2003). HypE then catalyzes an ATP-dependent dehydration reaction to yield the thiocyanate derivative of this cysteine. In addition to the non-modified HypE species, structures of the thiocarboxamide and thiocyanate forms are reported (Tominaga et al. 2013).

HypC and HypD function as the scaffold for the iron atom of the [NiFe] site. Structures are known for the individual proteins and for HypC·HypD from T. kodakarensis (Watanabe et al. 2007, 2012). HypC has a β-barrel domain containing the essential Cys2 residue and an α-helix extension at its C-terminus. HypD is a larger protein containing a [4Fe-4S] bound by four cysteine residues, four other redox-active cysteine residues located immediately adjacent to the cluster, and the essential Cys38 residue. The HypC·HypD heterodimer is hypothesized to bind iron (from a still unidentified source) using HypC Cys2 and HypD Cys38. Two molecules of HypC·HypD bind to the HypE homodimer, and the (HypC·HypD·HypE)2 structure is known. The cyanide adduct of the HypE C-terminal cysteine is transferred onto the HypC·HypD-bound iron atom, presumably making use of reductive chemistry involving the [4Fe-4S] and the cysteine tetrad electron transfer components of HypD. This process is repeated twice to provide two cyanide ligands on the iron. The CO ligand is derived from formyl-tetrahydrofolate (THF) using formyl-THF decarbonylase, or HypX (Schulz et al. 2020), which was structurally defined for the protein from Aquifex aeolicus (Muraki et al. 2019). Studies with the C. necator enzyme show the formyl group is first transferred onto a tightly bound CoA molecule, with decarbonylation leading to release of CO and reformation of CoA (Schulz et al. 2020). The HypC·HypD-bound iron-(CN)2CO complex is subsequently transferred into the hydrogenase large subunit HyhL by a process that remains obscure.

Nickel is provided to HyhL by the metallochaperones HypA and HypB that are universally required for [NiFe] hydrogenase activation (Böck et al. 2006; Lacasse and Zamble 2016; Miki et al. 2020). Extensive efforts have characterized the nickel-binding properties and other features of these proteins from various organisms, and the structures of dimeric HypB and monomeric HypA from T. kodakarensis are known (Sasaki et al. 2013; Watanabe et al. 2009) as is that of their complex in the presence of ATPγS (Watanabe et al. 2015). In this microorganism, HypB does not bind nickel so it cannot deliver the metal to HypA as described for the E. coli proteins. Rather, the binding of HypB to HypA increases the affinity for nickel at the HypA site from a Kd of 4.1 μM for the free protein to 7.3 nM for HypA·HypB. Nucleotide hydrolysis (typically using GTP, but the T. kodakarensis HypB protein uses ATP) is coupled to nickel insertion into HyhL via HyhL·HypA, for which the structure is established (Kwon et al. 2018). In E. coli, HypA is used only to activate hydrogenase-3 whereas a second version named HybF activates hydrogenase-1 and hydrogenase-2 (Blokesch et al. 2004). In B. japonicum, HypB binds ~9 nickel ions per monomer (predominantly using a His-rich region at the amino terminus) and was suggested to be, but not established as, a nickel storage protein named nickelin (akin to the term ferritin used for iron storage proteins) (Olson and Maier 2000). HypB of this microorganism is critical for activation of its [NiFe] hydrogenase, and its GTPase activity promotes nickel insertion into the protein (Fu et al. 1995). In E. coli, SlyD participates in nickel delivery to [NiFe] hydrogenases and perhaps also in nickel storage (Kaluarachchi et al. 2011; Zhang et al. 2005). This protein is one of several host factors required for lysis by a bacteriophage (hence, giving rise to the sensitive to lysis D or slyD name) (Maratea et al. 1985) and it is a common contaminant of poly-histidine-tagged proteins purified by immobilized metal affinity chromatograph (Wülfing et al. 1994). SlyD contains a histidine-rich region near its carboxyl terminus and a domain with peptidyl-prolyl cis-trans isomerase activity. Deletion of the corresponding gene results in large reductions of cellular nickel content and hydrogenase activity (Leach et al. 2007; Pinske et al. 2015; Zhang et al. 2005). Of additional interest, SlyD of H. pylori interacts with the NiuD permease (Denic et al. 2021), hinting at a possible handoff of metal from a nickel uptake protein to a nickel delivery protein.

Upon nickel binding, a conformational change occurs in HyhL to bury the metallocenter and expose the histidine cleavage site to the endopeptidase HybD with a known structure (Kwon et al. 2016). Structures are also known for E. coli HybD and HycI, analogous proteases used for maturation of hydrogenase-2 and hydrogenase-3 (Fritsche et al. 1999; Yang et al. 2006). After HyhL cleavage and release of HypA, the small subunit of hydrogenase, HyhS, binds to form the mature, active [NiFe] hydrogenase.

4.7 Carbon Monoxide Dehydrogenase (CODH)

A critical reaction for many microorganisms is the reversible interconversion between carbon monoxide (CO) and carbon dioxide (Fig. 14.5g). In aerobic organisms, this reaction is carried out by using a copper-molybdopterin-[FeS] cluster-containing enzyme (King and Weber 2007). In contrast, anaerobes use an oxygen-labile nickel-dependent carbon monoxide dehydrogenase (CODH) of interest here (Can et al. 2014; Kung and Drennan 2017). CODHs are widely distributed in microorganisms, with one report indicating that over 6% of bacterial and archaeal genomes possess such genes that were resolved into six distinct clades (Techtmann et al. 2012). A more recent bioinformatics study identified 1942 non-redundant CODH genes in 1375 genomes covering 36 phyla and divided them into seven clades with 24 structural groups (Inoue et al. 2019). This section summarizes the properties of CODHs from carboxydotrophs that utilize the low potential electrons derived from CO (E°′ −524 mV) for energy metabolism, such as that resulting in hydrogen production, or for autotrophic growth. Described separately (Sect. 14.4.8) are CODHs present in larger protein assemblages containing acetyl-CoA synthase/decarbonylase (ACS) activities that are used for CO2 fixation and for aceticlastic growth of methanogens.

Particularly well-characterized sources of homodimeric CODHs are C. hydrogenoformans and Rhodospirillum rubrum, microbes in which CO oxidation is coupled to the production of hydrogen gas, along with D. vulgaris, which does not utilize CO as an energy source (Hadj-Saïd et al. 2015; Voordouw 2002). Notably, the C. hydrogenoformans genome encodes five CODHs (Wu et al. 2005), and each of the corresponding enzymes has been characterized [e.g., (Domnik et al. 2017)]. The R. rubrum enzyme was the first CODH purified and shown to possess a metallocenter containing nickel, iron, and inorganic sulfide (Bonam and Ludden 1987). The nickel-free forms of the R. rubrum and D. vulgaris proteins were obtained by growing cells in medium lacking this metal ion or by growing in the absence of the CooC accessory protein (see below), allowing for studies involving nickel-dependent activation (Bonam et al. 1988; Ensign et al. 1990; Hadj-Saïd et al. 2015). An extensive array of spectroscopic and mechanistic studies have been summarized for the R. rubrum CODH (Can et al. 2014; Kung and Drennan 2017) and will not be repeated here. Rather, I focus below on the structures of these proteins.

4.7.1 CODH Structure

The crystal structures of homodimeric CODHs from C. hydrogenoformans and R. rubrum were first reported in 2001 (Dobbek et al. 2001; Drennan et al. 2001) and shown to possess a [4Fe-4S] D-cluster that bridges the subunits, two additional [4Fe4S] B-clusters, and the deeply buried unique nickel-iron-sulfur C-clusters (Fig. 14.6g). The initial reports identified the R. rubrum CODH C-cluster as a [1Ni-3Fe-4S] cubane connected by one sulfide to a mononuclear Fe site, whereas the C. hydrogenoformans enzyme had an additional sulfide between the nickel and the non-core Fe (Dobbek et al. 2001; Drennan et al. 2001). Subsequent studies indicated the additional sulfide is not present in the catalytically active state of the enzyme (Kung and Drennan 2017). An essentially identical crystal structure was obtained for a nickel-depleted form of one CODH from Thermococcus sp. AM4 (two such enzymes are present in that microorganism) (Benvenuti et al. 2020). The CODH structure from D. vulgaris is of special interest. This enzyme exhibits redox-dependent changes in its C cluster such that the Ni and mononuclear Fe alter their positions (Wittenborn et al. 2018). In addition, this protein tolerates short-term exposure to small concentrations of oxygen because the exposed D-cluster has a more stable [2Fe-2S] cluster rather than a [4Fe-4S] cluster (Wittenborn et al. 2020). External electron transfer proteins deliver reducing equivalents to or remove electrons from the D-cluster that connects, via the B-clusters, to the active site C-clusters, which interconverts CO and CO2.

4.7.2 CODH Maturation

CODH of R. rubrum is encoded by cooS that is immediately followed by cooC, cooT, and cooJ. Deletion analysis demonstrated that cooC is required for biosynthesis of the [1Ni-4Fe-4S] metallocenter in CODH unless high concentrations of nickel are provided in the medium, whereas deletions in the adjacent cooT and cooJ genes have less marked effects (Kerby et al. 1997). In C. hydrogenoformans, cooS-1 is located near cooC1, cooS-2 is adjacent to cooF, and two other cooC copies along with a cooT are found elsewhere in the genome (Merrouch et al. 2018). Both cooS copies in Thermococcus are located near cooC genes, a third cooC has no clear function, and no cooT or cooJ is reported in this genome (Benvenuti et al. 2020). In D. vulgaris, cooS is immediately adjacent to cooC, but no cooT or cooJ has been described (Hadj-Saïd et al. 2015). A bioinformatics analysis of microorganisms possessing CODH reveals the genes encoding CooT and CooJ often are missing and that some microbes also lack CooC; thus, these proteins are not strictly needed for CODH enzyme activation in all microbes (Merrouch et al. 2018). The following paragraphs briefly highlight these proteins.

CooC is an ATPase or GTPase with sequence similarities to the HypB and UreG proteins involved in activation of hydrogenases and ureases (Jeon et al. 2001). The structurally-characterized CooC from C. hydrogenoformans is a nickel-binding protein with the metal bound to four cysteine residues located at the interface of the subunits (Fig. 14.9a) (Jeoung et al. 2009, 2010). Deletion of cooC in D. vulgaris leads to the synthesis of a nickel-free CODH whose structure was determined (Wittenborn et al. 2019).

Fig. 14.9
figure 9

Structures of CODH maturation proteins. (a) Two views of CooC1 from Carboxydothermus hydrogenoformans (PDB ID 3kji) with bound ADP and zinc at the presumed nickel site. (b) CooT apoproteins from (left) R. rubrum and (right) C. hydrogenoformans (PDB ID 5n76 and 6fan) with their distinct nickel-binding sites. (c) Truncated CooJ from R. rubrum (PDB ID 6hk5) missing residues Asp69-Pro115 (the carboxyl terminal His-rich region) and highlighting the remaining nickel-binding site. Glu29″, not shown, from an adjacent dimer in the crystal coordinates as a bidentate ligand to the metal and completes its octahedral coordination sphere. The homodimeric proteins are shown in cartoon mode (green and cyan) with semi-transparent surfaces. Metals are shown as spheres and selected residues are depicted as sticks

CooT from R. rubrum and C. hydrogenoformans are nickel-binding proteins with known structures (Alfano et al. 2018; Timm et al. 2017). NMR, XAS, and modeling analyses of nickel-bound R. rubrum CooT suggested the presence of a unique square-planar nickel site at the subunit interface, with coordination by the side chains and amide groups of two Cys2 residues (note: the amino terminal Met1 residue was removed) (Fig. 14.9b, left) (Alfano et al. 2019c). C. hydrogenoformans CooT has a similar overall structure (Fig. 14.9b, right) despite being only 19% identical in sequence to that from R. rubrum; however, the presumed nickel site is modified and thought to involve coordination by the Cys2 and His55 side chains in each subunit (Alfano et al. 2018). Sequence comparisons show the two distinct nickel-binding motifs exist in other bacterial and archaeal CooT sequences (Alfano et al. 2019c).

CooJ from R. rubrum may assist in CODH maturation and is a potential nickel storage protein due to its His-rich carboxyl terminus (16 histidine residues in the last 34 amino acids) (Watt and Ludden 1998). The protein binds five nickel per dimer (four with a Kd ~ 1.6 μM and one with a Kd of 380 nM) (Alfano et al. 2019a). The structure of CooJ (truncated at the carboxyl region to remove the His-rich region) reveals nickel coordination by His22 and His26 from one subunit along with His18 and His22 of the second subunit, plus bidentate binding by Glu29 from another dimer (PDB ID 6hk5, Fig. 14.9c). The interaction with a neighboring dimer is consistent with the known ability of the protein to form higher-order oligomers in the presence of nickel (Alfano et al. 2019b). Direct evidence that the unique nickel-binding site is critical to CooJ function is lacking.

4.8 Acetyl-CoA Synthase/Decarbonylase (ACS)

In many microorganisms, CODH is a component of a larger protein complex known as acetyl-CoA synthase/decarbonylase (ACS) that includes a second nickel-containing metallocenter referred to as the A-cluster (Can et al. 2014; Kung and Drennan 2017). Versions of this enzyme are found in autotrophic anaerobes that use the Wood-Ljungdahl pathway for CO2 fixation, such as many acetogens, methanogens, and sulfate-reducing bacteria, and in aceticlastic methanogens that grow anaerobically on acetate using a version of the enzyme that operates in the reverse direction. In the Wood-Ljungdahl pathway, one CO2 undergoes three sequential two-electron reductions to generate a methyl group that is bound to tetrahydrofolate. The methyl group is then donated to a corrinoid-iron-sulfur protein (CFeSP), forming a methyl-Co bond. Meanwhile, the CODH component of ACS reduces one CO2 to CO. In a remarkable reaction, the A cluster of ACS condenses the CFeSP-bound methyl group, the CO, and CoA to generate acetyl-CoA (Fig. 14.5h), used to synthesize all cellular carbon compounds. By contrast, the A-cluster of ACS in aceticlastic methanogens splits acetyl-CoA into CoA, CO, and a methyl group that becomes bound to a CFeSP. Subsequently, the methyl group is transferred to tetrahydromethanopterin, then to 2-thioethanesulfonate (CoM), the product of which is reduced by methyl-S-coenzyme M reductase (MCR) to generate methane (see Sect. 14.4.10). The electrons required for this last reduction step are derived from the oxidation of CO to form CO2 by the CODH component of the protein.

4.8.1 Structure of ACS

Structures are available for the native ACS from Moorella thermoacetica (formerly Clostridium thermoaceticum) (Darnault et al. 2003; Doukov et al. 2002) and Clostridium autoethanogenum (Lemaire and Wagner 2020). In addition, both the CODH component of ACS from M. barkeri (Gong et al. 2008) and the CODH-free ACS component from C. hydrogenoformans (Svetlitchnyi et al. 2004) have been structurally defined. The native α2β2 proteins from M. thermoacetica and C. autoethanogenum (Fig. 14.6h) contain the central core, equivalent to the homodimeric CODH enzyme, with two active site [1Ni-4Fe-4S] C-clusters, two [4Fe-4S] B-clusters for internal electron transfer, and the single [4Fe-4S] D-cluster for electron exchange with an external electron carrier. Flanking the CODH component are subunits that contain the A-cluster as well as acetyl-CoA and CFeSP-binding sites. The highly dynamic nature of this subunit is shown by its two distinct conformations in crystallized M. thermoacetica and C. autoethanogenum enzymes and by the multiple structural arrangements found in C. thermoacetica ACS when examined by negative-stain electron microscopy (Cohen et al. 2020). The A-cluster is a novel [4Fe-4S]-Ni-Ni metallocenter that has a standard cubane cluster with one cysteine ligand bridging to a proximal nickel site (Nip). Two additional cysteines connect Nip and a distal nickel atom (Nid). The planar, 4-coordinate geometry of Nid is completed by two backbone amides. The tri-cysteine bound Nip is highly labile, with early structures depicting an enzyme lacking this metal, containing copper (Doukov et al. 2002), or with zinc (Darnault et al. 2003) at this position. The monomeric, CODH-free form of ACS from C. hydrogenoformans possesses the same [4Fe-4S]-Ni-Ni site (Svetlitchnyi et al. 2004). The structure of native ACS from M. thermoacetica revealed the existence of a 140 Å long molecular tunnel connecting the two C clusters in the CODH component and extending to Nip in each of the two A clusters. Direct evidence for this tunnel was obtained by showing that exogenous CO does not compete with CO produced at the C cluster for incorporation into acetyl-CoA at the A cluster and by demonstrating that addition of a CO-binding protein like myoglobin or hemoglobin has no effect on acetyl-CoA synthesis from CO2, methyl-CFeSP, and CoA (Maynard and Lindahl 1999; Seravalli and Ragsdale 2000). Notably, crystals treated with pressurized xenon gas confirms the presence of a channel by showing these electron dense atoms within the protein (Darnault et al. 2003; Doukov et al. 2008).

The methanogen versions of ACS that operate in the direction of acetyl-CoA decomposition are best characterized for the proteins from M. barkeri and Methanosarcina thermophila (Grahame 1991; Terlesky et al. 1986). These corrinoid, nickel, iron, and sulfide-containing proteins each possess five types of subunits that aggregate into 2.4-MDa complexes with eight copies of each peptide. The structure of the intact complex is not known, but that for the α2ε2 component exhibiting CODH activity is determined (Gong et al. 2008). The α2 portion of this complex closely resembles the CODH structures described above, but it contains five metalloclusters; i.e., the typical [4Fe-4S] D-cluster bridging the subunits, two [4Fe-4S] B-clusters, and two [1Ni-4Fe-4S] C-clusters, along with two [4Fe-4S] E-clusters and two [4Fe-4S] F-clusters. The functions of the E- and F-clusters and that of the ε subunits, not present in the previously described CODH and ACS structures, remain unknown.

4.8.2 ACS Maturation

The remarkable complexity of the ACS metallocenters might lead one to suspect that multiple accessory proteins are required for their biosynthesis; however, that is not the case. Surprisingly, heterologous expression in E. coli of acsA and acsB (encoding the CODH component and the A cluster-containing protein) from M. thermoacetica provides a form of ACS that is indistinguishable from the native enzyme missing Nip; subsequent incubation with nickel yields active enzyme (Loke et al. 2000). Somewhat similarly, heterologous expression in E. coli of the gene encoding only the β subunit (i.e. the site of the A cluster) of M. thermophila ACS yields a nickel-free inactive protein that developes ACS activity after incubation with nickel (Gencic and Grahame 2003). These results suggest no accessory genes are needed for ACS maturation, but other studies provide evidence counter to this conclusion. Located in the same M. thermoacetica gene cluster as acsAB are cooC1 and acsF, homologues of R. rubrum cooC encoding a protein with a nickel insertion auxiliary role. M. thermoacetica CooC has not been studied, whereas AcsF was found to exhibit ATPase activity, but no involvement in ACS activation was identified (Loke and Lindahl 2003). By contrast, cooC2 in the ACS gene cluster of C. hydrogenoformans encodes an AcsF that clearly functions as an ACS maturation factor (Gregg et al. 2016). Two molecules of AcsF bind to inactive, monomeric ACS that lacks CODH (Svetlitchnyi et al. 2004), and the complex coordinates two atoms of nickel. Subsequent ATP hydrolysis is coupled to A cluster activation (Gregg et al. 2016). The authors concluded that AcsF installs the two nickel atoms of the A cluster.

4.9 2-Hydroxyacid Racemase/Epimerase (HARE) Containing NPN

The ability of microorganisms to interconvert d- and l-lactic acid was first described for Clostridium acetobutylicum and Clostridium beijerinckii (formerly C. butylicum) in 1936 (Tatum et al. 1936). This activity provides a means for some lactic acid bacteria to obtain d-lactate for incorporation into their cell walls (Chapot-Chartier and Kulakauskas 2014), for other microbes to obtain l-lactate for synthesis of odd-numbered fatty acids (Weimer and Moen 2013), and for lactate degraders to utilize both enantiomers. Extensive studies of lactate racemase were carried out in the 1960s (Desguin et al. 2017), but only since 2015 has a detailed understanding of this enzyme and identification of other nickel-pincer nucleotide (NPN)-containing racemases/epimerases (Fig. 14.5i) been accomplished.

4.9.1 Structure of Lactate Racemase and Discovery of the NPN Cofactor

The gene encoding lactate racemase (larA) was first identified in L. plantarum and shown to be located in the larABCDE gene cluster, adjacent to and transcribed opposite larRMNQO (Desguin et al. 2014). These operons are regulated by LarR, a lactic acid-dependent transcription factor (Desguin et al. 2015a). LarD is a lactate permease that allows for cellular entry/egress of both enantiomers (Bienert et al. 2013). The sequences of LarMN, LarQ, and LarO resemble those of ABC-type nickel transporters (Sect. 14.2.2.2), leading to studies demonstrating that lactate racemase is a nickel-dependent enzyme requiring LarB, LarC, and LarE for activation (Desguin et al. 2014). The structure of LarA (Fig. 14.6i) reveals the presence of a covalently tethered cofactor, pyridinium-3-thioamide-5-thiocarboxylic acid mononucleotide (Desguin et al. 2015b). Because this cofactor is a nickel-containing pincer nucleotide complex (Nevarez et al. 2020), it is abbreviated NPN. Notably, the nickel in this coenzyme exhibits square-planar geometry with coordination by the pyridinium C4 carbon, the two sulfur atoms of NPN, and a histidine side chain. This cofactor participates in catalysis by a proton-coupled hydride transfer mechanism as shown by kinetic isotope studies demonstrating cleavage/reformation of the substrate C2-H bond, trapping of the pyruvate intermediate by two different methods, substrate-dependent perturbation of the NPN spectrum consistent with its reduction, and structural studies (Rankin et al. 2018). The presence of the novel NPN cofactor in LarA provides an explanation for why LarB, LarC, and LarE are required for racemase activity; they synthesize the coenzyme.

4.9.2 NPN Cofactor Biosynthesis

Biosynthesis of the NPN cofactor (Fig. 14.10) initiates by LarB catalyzing dual activities with its substrate, nicotinic acid adenine dinucleotide (NaAD): carboxylation at C5 of the pyridinium ring and hydrolysis of the phosphoanhydride to produce pyridinium 3,5-dicarboxylic acid mononucleotide (P2CMN) while releasing AMP (Desguin et al. 2016). The structurally-characterized LarE then catalyzes a three-step sequence involving P2CMN carboxylate activation by adenylylation (using ATP and releasing pyrophosphate), formation of a covalent bond between the mononucleotide and a cysteine side chain (while releasing AMP), and sacrificial loss of the cysteine sulfur atom (leaving the protein with dehydroalanine at this position) to generate the thioacid product. These steps are repeated for the second substrate carboxyl group using another LarE subunit, thus generating pyridinium 3,5-dithiocarboxylic acid mononucleotide (P2TMN) (Desguin et al. 2016; Fellner et al. 2017). LarC inserts nickel into this precursor via a stoichiometric cyclometallase reaction in a manganese- and CTP-dependent process (releasing CMP and pyrophosphate) (Desguin et al. 2018). LarC binds multiple atoms of nickel (Desguin et al. 2014), presumably using a histidine-rich sequence in its amino terminal domain, and may function as a metallochaperone. Efforts to crystallize LarC led to protein cleavage—perhaps involving trace amounts of protease acting on a highly susceptible sequence—to release the carboxyl-terminal domain (starting at residue 272) from the rest of the 420 residue full-length protein. The crystal structure was obtained for a hexamer of the smaller fragment, which lacks the nickel-binding region and the P2CMN-binding site; however, this domain binds Mn2·CTP in a new nucleotide-binding configuration (Desguin et al. 2018).

Fig. 14.10
figure 10

Biosynthesis of the nickel-pincer nucleotide (NPN) cofactor. LarB carboxylates C5 of the pyridinium ring and hydrolyzes the phosphoanhydride of nicotinic acid adenine dinucleotide (NaAD) releasing AMP and pyridinium 3,5-dicarboxylic acid mononucleotide (P2CMN). The carboxylic acids of P2CMN are converted to thioacids in pyridinium 3,5-dithiocarboxylic acid mononucleotide (P2TMN) by duplicate sacrificial sulfur transfer reactions from Cys side chains of two copies of LarE, generating dehydroalanine residues. LarC installs nickel into P2TMN to generate NPN in a reaction that requires CTP hydrolysis to CMP and pyrophosphate

4.9.3 Other NPN-Binding Proteins

Approximately 10% of bacterial and archaeal genomes possess homologs to larA, larB, larE, and larC (Desguin et al. 2014), indicating a widespread ability to synthesize NPN for incorporation into lactate racemase-type enzymes. Significantly, many species possess more than one paralog of larA—up to eight copies in Clostridium asparagiforme and Thermosinus carboxydivorans—consistent with activities other than lactate racemase (Hausinger et al. 2018). Furthermore, gntE, a larA homolog from Thermotoga maritima, is proposed to encode an epimerase that interconverts d-mannonate and d-gluconate (Rodionova et al. 2012). Biochemical studies with purified proteins confirm this suspicion and identify other 2-hydroxyacid racemases including those for malate, phenyllactate, 2-hydroxyglutarate, and less specific short-chain aliphatic 2-hydroxyacids (Desguin et al. 2020). Of additional interest, many cyanobacteria and some other microorganisms possess homologs of larB, larE, and larC, but lack any homologs of larA, suggesting these microorganisms may synthesize NPN for incorporation into proteins with a distinct architecture and possibly a unique function.

4.10 Methyl-S-Coenzyme M Reductase (MCR)

This chapter uses the name methyl-S-coenzyme M reductase (MCR) when referring to the enzyme of methanogenic archaea that forms methane as well as to the protein responsible for anaerobic oxidation of methane and short-chain alkanes in syntrophic methano(alkano)trophs. Excellent recent overviews of MCR including its proposed enzyme mechanism are available (Ragsdale et al. 2017; Thauer 2019).

4.10.1 Nickel-Containing MCR of Methanogenic and Methanotrophic Archaea

Methanogenic archaea reduce CO2 to methane by a pathway that includes the intermediate methyl-S-coenzyme M (methyl-S-CoM) (Taylor and Wolfe 1974). In addition, selected methanogen strains are capable of metabolizing CO, formate, methanol, methylamine, and acetic acid by pathways that also form this key chemical species (Thauer 1998). In all cases, methane release involves the action of MCR that reacts methyl-S-CoM with 7-mercaptoheptanoylthreonine phosphate (coenzyme B, HS-CoB) (Noll et al. 1986) to form the mixed disulfide CoM-S-S-CoB (Fig. 14.5j) (Bobik et al. 1987; Ellermann et al. 1988). The α2β2γ2 enzyme associated with MCR activity (Fig. 14.6j) was first isolated from Methanothermobacter thermoautotrophicus (formerly Methanobacterium thermoautotrophicum ΔH) and shown to contain nickel based on the incorporation of radioactive 63Ni (Ellefson and Wolfe 1981; Ellefson et al. 1982). Numerous subsequent biochemical and spectroscopic studies of MCR have used the enzyme from this species and from Methanothermobacter marburgensis (formerly Methanobacterium thermoautotrophicum strain Marburg). Some methanogens possess multiple forms of MCR that are regulated differently (Rospert et al. 1990) and have distinct catalytic properties (Bonacker et al. 1993). The nickel of MCR is associated with a yellow organometallic cofactor named coenzyme F430 that occurs as protein-associated and free forms, with subtle differences in their spectra (Hausinger et al. 1984). An extensive series of studies ((Pfaltz et al. 1985) and earlier citations) identified the structure of coenzyme F430 as a highly reduced and modified nickel-containing tetrapyrrole (see inset to Fig. 14.6j). Notably, derivatives of this cofactor exist in other methanogens and in anaerobic methanotrophs, as described below (Allen et al. 2014).

Microbial mats associated with anaerobic methane seeps in the Black Sea possess an abundant microorganism that contains an MCR-like protein with a nickel-containing cofactor resembling F430 (Krüger et al. 2003). This cofactor structure is the same as the methanogen-derived F430 except for the added presence of a methylthio group on the methylene carbon adjacent to the keto group in the carbocyclic ring (Mayr et al. 2008). The enzyme containing the modified cofactor is proposed to catalyze a reverse methanogenesis reaction (Shima and Thauer 2005), resulting in the anaerobic oxidation of methane. Indeed, M. marburgensis MCR also catalyzes this reverse reaction when subjected to appropriate experimental conditions (Scheller et al. 2010). Remarkably, MCR homologs in members of the Candidatus Syntrophoarchaeum, Candidatus Methanoliparia, Candidatus Ethanoperedens, and other genera of archaea are capable of specifically catalyzing the anaerobic oxidation of short chain alkanes, including butane and ethane (Chen et al. 2019; Hahn et al. 2020; Laso-Pérez et al. 2016, 2019; Wang et al. 2019, 2020).

4.10.2 Structure of Methyl-S-Coenzyme M Reductase

The first reported MCR structure was that from M. marburgensis (Fig. 14.6j) (Ermler et al. 1997). Additional MCR structures have been elucidated from other methanogens [e.g. (Grabarse et al. 2000)] and from the Black Sea mat methanotroph (Shima et al. 2012). The structures are quite similar; e.g., each is an α2β2γ2 enzyme with HS-CoB-binding channels that lead to the buried coenzyme F430. Of additional interest, multiple amino acid residues in these proteins are enzymatically derivatized, with at least some of the modifications affecting the activity of the enzyme (Lyu et al. 2020; Nayak et al. 2017, 2020; Selmer et al. 2000). Depending on the microbial source, highly specific changes include methylations (of histidine, arginine, glutamine, and cysteine), thioamidation (of glycine), desaturation (of aspartic acid), and hydroxylation (of tryptophan or methionine) [reviewed in (Chen et al. 2020)].

4.10.3 Biosynthesis of Coenzyme F430

The biosynthetic pathway for coenzyme F430 has long been known to begin with l-glutamic acid, with initial steps progressing through glutamyl-tRNA, glutamate 1-semialdehyde, 5-aminolevulinic acid, porphobilinogen, uroporphyrinogen III, and dihydrosirohydrochlorin [reviewed in Thauer and Bonacker (1994)]. In addition, early studies reported the purification and mass spectrometric identification of the intermediate 15,173-seco-F430-173-acid (Pfaltz et al. 1987). A critical breakthrough in further understanding the pathway derived from genomic analyses that revealed a cluster of five genes in methanogens and anaerobic methanotrophs with potentially relevant functions. The genes within the cfbAEDCB cluster of Methanosarcina acetivorans C2A or the alternately named cfbABCDE cluster of M. barkeri were demonstrated to encode enzymes catalyzing the reactions shown in Fig. 14.11 (Moore et al. 2017). Thus (using the nomenclature of the first report), CfbA incorporates nickel into sirohydrochlorin, CfbB converts two acetate side chains to acetamides, CfbC and CfbD together reduce the diamide-containing tetrapyrrole by six electrons and convert one amide to a lactam ring, and CfbE converts a propionate side chain into a cyclohexanone ring. In methanotrophs and some methanogens additional modifications occur by still unclear processes (Allen et al. 2014). It is uncertain how the resulting coenzyme species are incorporated into the corresponding MCR proteins.

Fig. 14.11
figure 11

Biosynthesis of coenzyme F430. CfbA inserts nickel into sirohydrochlorin with the loss of two protons. CfbB converts two carboxylic acid side chains into amides via ATP-dependent activation and nitrogen transfer from glutamine. CfbC and CfbD catalyze a six-electron reduction of the nickel-sirohydrochlorin a,c-diamide and generate a lactam from one side chain. CfbE forms the carbocyclic ring of the coenzyme in an ATP-dependent reaction

4.11 Other Nickel Enzymes

It is likely that other nickel-containing enzymes exist, and experimental evidence already points to some probable representatives. For example, Bacillus subtilis AraM is a glycerol-1-phosphate dehydrogenase that depends on nickel for greatest activity and co-purifies with nickel from cells grown on LB medium supplemented with 1 mM of this metal ion (Guldan et al. 2008). Two chromatographically-resolved proteins from cell-free extracts of Pyrococcus furiosus contain nickel, but no enzymatic activity has been demonstrated (Cvetkovic et al. 2010). PF0056, a possible sugar-binding protein, binds 0.47 Ni/subunit and is predicted to possess a cupin fold that coordinates the metal using three histidine and one glutamate residues (much like the quercetinase structure, Sect. 14.4.3). PF0086 is annotated as an alanyl-tRNA editing hydrolase and is predicted to bind its metal (0.86 Ni/subunit) using three histidine and one cysteine residue. None of these proteins has been extensively characterized and their structures have not been determined.

One must use caution when claiming an enzyme is nickel dependent. A nice illustration of this point is provided by peptide deformylase, the enzyme responsible for removal of the N-formyl group from the amino terminal methionine residue of newly synthesized proteins. Early studies had suggested the E. coli protein contains an active site zinc ion; however, substantial increases in enzyme activity occur when this enzyme is isolated from cells grown in the presence of nickel (Ragusa et al. 1998). The stable nickel-containing protein was extensively characterized, including determination of its structure (Becker et al. 1998). Nevertheless, compelling evidence was later generated that the native enzyme contains ferrous ion, which undergoes rapid inactivation by oxygen due to conversion to the non-functional ferric species (Rajagopalan and Pei 1998).

4.12 Nickel Storage Proteins

Several proteins (HypB, UreE, LarC, SlyD, CooJ, etc.) bind nickel for eventual incorporation into metalloenzymes, and these proteins could also reasonably store the metal within the cell. Unfortunately, experimental support for a nickel storage function is lacking in most cases. The exception might be B. japonicum HypB (Sect. 14.4.6.2), named nickelin on the basis of studies that were consistent with such a role (Olson and Maier 2000).

The most extensively-studied and most convincing examples of nickel storage proteins are Hpn (alternatively called Hpn-1) and Hpn-like (Hpnl or Hpn-2) proteins of Helicobacter species. First discovered in H. pylori, Hpn is a 60-amino acid long protein (7077 Da) that contains 28 His residues (47%) and tightly binds zinc or nickel (Gilbert et al. 1995). Hpn accounts for up to 2% of the H. pylori cellular protein, a feature that is compatible with a role in cellular nickel storage. Analogous proteins are present in H. mustelae, H. felis, and several other gastric species of this genus (Vinella et al. 2015). Biochemical and spectroscopic studies of the purified protein have demonstrated that H. pylori Hpn is multimeric (with the 20-mer being most prevalent), able to coordinate ~5 nickel ions per monomer (K ~ 7.1 μM), and also capable of binding zinc and copper (~5 Zn/monomer, 19.3 μM; ~8.5 Cu/monomer, ~2.2 μM) (Ge et al. 2006a, b). Deletion of the gene encoding Hpn leads to increased toxicity from nickel ions, consistent with a protective role against nickel stress; furthermore, gene deletion reduces the nickel content of cells whereas gene overexpression results in increased cellular nickel content, indicative of a nickel storage function (Ge et al. 2006a; Vinella et al. 2015). A large subset of Hpn-containing Helicobacter strains also produce Hpn-2 (Vinella et al. 2015). This 72-amino acid long protein is rich in His (nearly 25%), but has an even greater percentage of Gln residues (42%) (Seshadri et al. 2007). Gene deletion studies suggested that it too functions in metal detoxification (in this case against nickel, cobalt, and cadmium) and a role in nickel storage was suggested. The purified Hpn-2 protein is oligomeric, like Hpn, and binds ~2 nickel ions per monomer (K ~ 3.8 μM) (Zeng et al. 2008). The two nickel-binding sites are distinct, with the region involving His29 to His32 being critical for one binding site, whereas substitutions of other His residues do not affect nickel binding (Zeng et al. 2011). About one atom of copper, cobalt, or zinc also binds per Hpn-2 monomer, consistent with distinct environments for the two nickel sites. Chemical cross-linking studies show that Hpn and Hpn-2 interact with more than 100 proteins in Helicobacter cells, including UreA/UreB urease subunits, UreG GTPase, and HypB [NiFe] hydrogenase accessory protein (Saylor and Maier 2018). Overall, these studies reveal that Hpn-type proteins bind nickel, reduce toxicity to this metal, and facilitate activation of nickel enzymes, but a nickel storage role remains less compelling.

5 Conclusions and Perspectives

Microorganisms have developed a vast array of proteins that function to sense nickel concentrations, import the metal ion when cellular needs are deficient, and export the cation when internal levels are too high. The basis for cellular toxicity by nickel is partially understood, as are the mechanisms used by cells to defend against this harmful metal ion. A number of nickel-containing enzymes have been identified, with several of these enzymes dependent on accessory proteins for their biosynthesis. Nickel storage proteins exist in at least one bacterial genus. Further studies are likely to identify additional nickel-dependent enzymes and are certain to expand our understanding and provide more examples of proteins involved in sensing, transporting, and utilizing.

Note Added in Proof

During the interval since this chapter was submitted, several relevant publications have appeared; selected examples are mentioned here. With regard to nickel-sensing transcription factors, additional structural and functional insights have been reported for SrnR (Mazzei et al. 2021). Nickel importers and exporters in H. pylori were reviewed (Camporesi et al. 2021). The maturation components responsible for activation of nickel-dependent hydrogenase and urease in H. pylori were summarized (Tsang and Wong 2022). Dynamics studies were carried out to better characterize a “bucket brigade” mode of nickel movement within the UreF and UreD/H proteins during urease activation (Masetti et al. 2021). Related to CODH activation, 150 homologues of CooJ were identified and three were characterized for nickel binding and other properties (Darrouzet et al. 2021). The structure and mechanism were revealed for LarB, catalyzing the first step of NPN cofactor biosynthesis (Rankin et al. 2021). Newly reported features of LarC, catalyzing the CTP-dependent nickel insertion during NPN cofactor biosynthesis, include formation of a novel reaction intermediate (Turmo et al. 2022). The structure was revealed for an MCR that functions in anaerobic ethane oxidation and contains a modified F430 cofactor (Hahn et al. 2021).