Introduction

Nitrogenases are a family of complex metalloenzymes that catalyze the reduction of nitrogen (N2) to ammonia (NH3) under ambient conditions [1–4]. Three homologous nitrogenases have been identified to date, which contain cofactors that are mainly distinguishable by the heterometals they contain at their respective cofactor sites: the “conventional” molybdenum-containing nitrogenase, the “alternative” vanadium-containing nitrogenase, and the “alternative” iron-only nitrogenase [2, 5]. Apart from these homologous forms of nitrogenases, several structural/functional homologs to nitrogenases have been identified in recent years, including proteins involved in nitrogenase assembly, bacteriochlorophyll biosynthesis and methanogenic process [6, 7]. Here, we provide an overview of the structures and functions of nitrogenases and homologs, which highlights the similarity and disparity of this uniquely diversified “super family” of enzymes.

The molybdenum (Mo) nitrogenase

The Mo nitrogenase, which has been isolated from Azotobacter vinelandii, Klebsiella pneumoniae and Clostridium pasteurianum [8–10], is undoubtedly the best characterized member of this enzyme family. It consists of two separately purifiable, oxygen-sensitive, metallosulfur proteins, which are designated the iron (Fe) protein and the molybdenum-iron (MoFe) protein, respectively [1]. The Fe protein belongs to a family of nucleotide-utilizing proteins [1, 8], and it serves as an ATP-dependent reductase in nitrogenase catalysis. Encoded by nifH, the Fe protein is a homodimer of M r ∼60 kDa. The primary sequence of the Fe protein has a GXGXXG consensus motif, which provides a binding site for MgATP in each subunit of this homodimeric protein. Moreover, the two subunits of this protein are bridged by a [Fe4S4] cluster (Fig. 1a, b) through four Cys residues, two from each subunit [11]. The MoFe protein is the catalytic partner of the Fe protein in nitrogenase catalysis. It is an α 2 β 2-tetramer of M r ∼220 kDa, and its α- and β-subunits are encoded by nifD and nifK, respectively [8, 9, 11]. The MoFe protein contains two unique metallosulfur clusters (Fig. 1a, b): the P-cluster ([Fe8S7]), which is bridged between each αβ-subunit pair by six Cys residues; and the FeMo cofactor, or the M-cluster ([MoFe7S9C-homocitrate]), which is located within each α-subunit and coordinated by a His residue and a Cys residue at the opposite ends of the cluster [3, 11].

Fig. 1
figure 1

a Surface presentation (transparent) of the structure of the MgADP·AlF4 −-stabilized Fe protein/MoFe protein complex. MgADP·AlF4 −, [Fe4S4] cluster, P-cluster and M-cluster are shown as space-filling models. The two subunits of the Fe protein (labeled as H) are colored gray and light wheat, respectively, and the α- and β-subunits of the MoFe protein are colored red (labeled as D) and light blue (labeled as K), respectively. b Components involved in electron transfer during catalysis. All clusters are shown as ball-and-stick models. Atoms are colored as follows: Fe, orange; S, yellow; Mo, cyan; O, red; C, gray; N, blue; Mg, dark green; Al, light green; F, light blue. PYMOL was used to create this figure (PDB IDs: 1N2C, 3U7Q)

Full size image

The X-ray crystal structures of both component proteins of the Mo nitrogenase have been resolved [12–17] and, while the structure of the MgATP-bound form of the Fe protein is unknown, the small angle X-ray scattering (SAXS) data have shown that the conformational change of the Fe protein upon MgATP binding/hydrolysis causes a contraction of the protein by 2.0 Å on average [18]. It has been proposed, therefore, that the conformational rearrangement of the Fe protein upon ATP binding/hydrolysis allows it to dock on the MoFe protein, which then facilitates the inter-protein electron transfer from the Fe protein to the MoFe protein and the reduction of substrates at the M-cluster of the MoFe protein. Consistent with this hypothesis, the x-ray crystallographic analysis of a MgADP·AlF4 −-stabilized complex between the Fe protein and the MoFe protein [15], which represents the “transition state” during MgATP hydrolysis, provides some compelling evidence that the electrons are transferred sequentially from the [Fe4S4] cluster of the Fe protein, via the P-cluster, to the M-cluster of the MoFe protein (Fig. 1a, b). Substrate reduction is believed to take place once the M-cluster accumulates a sufficient amount of electrons, followed by the release of the products, the dissociation of the complex between the Fe protein and the MoFe protein, and the initiation of a new round of substrate reduction.

The overall reaction catalyzed by the Mo nitrogenase is usually depicted as follows: N2 + 8H+ + 16MgATP + 8e−→ 2NH3 + H2 + 16MgADP + 16P i [19]. Other than its physiological substrates [i.e., N2 and proton (H+)], the Mo nitrogenase is also capable of reducing a wide range of small doubly- and triply-bonded substrates, such as acetylene (C2H2), ethylene (C2H4), cyanide (CN−), nitrous oxide (N2O), nitrite (NO2 −), azide (N3 −) and hydrazine (N2H4) [1]; in particular, C2H2 is an efficient, alternative substrate that has been routinely used for the determination of the enzymatic activities of nitrogenases. Progress was made early on in understanding the interactions between the protein-bound M-cluster and substrates/inhibitors [e.g., C2H2, carbon disulfide (CS2) and carbon monoxide (CO)], as well as the interactions between the isolated M-cluster and substrates/inhibitors (e.g., CN− and CO) [20–26]. Recently, CO and carbon dioxide (CO2) have been either re-classified or defined as substrates for certain Mo nitrogenase variants, as well as substrates for the vanadium nitrogenase (see below), thereby broadening the substrate profile of nitrogenases [27–33]. Further, a combined genetic, biochemical and spectroscopic approach has led to the proposal of the first draft of nitrogenase mechanism, which involves reductive elimination of the M-cluster-bound hydrides (H−) as hydrogen (H2) that is obligate for the subsequent activation and reduction of N2 [4].

One interesting feature of the active site (i.e., the M-cluster) of the Mo nitrogenase is the presence of a μ6-coordinated light atom in its central cavity. Discovered more than a decade ago [16], this light atom was identified only recently by X-ray emission spectroscopy (XES) and combined high-resolution crystallography/electron nuclear double resonance (ENDOR)/electron spin echo envelope modulation (ESEEM) spectroscopy as a carbide (C4−) ion coordinated by six “equatorial” Fe atoms in the structure of the M-cluster [17, 34]. Biochemical characterization further established the methyl group of S-adenosylmethionine (SAM) as the source of the interstitial carbide and suggested a radical-dependent pathway for carbide insertion, which involves hydrogen abstraction of the SAM-derived methyl group by a 5′-deoxyadenosyl radical (5′-dA·) and further processing of the resultant radical carbon intermediate concomitant with radical-based coupling/rearrangement of FeS fragments till a complete Fe/S core of the M-cluster is formed with an interstitial carbide in place [35]. The interstitial carbide is not exchangeable upon turnover with N2, C2H2 and CO, which points to a structural function of this atom in stabilizing the active center of Mo nitrogenase [36]. Nevertheless, an indirect role of this atom in regulating the reactivity of the M-cluster cannot be excluded, particularly in light of a study describing the activation of N2 on iron metallaboratranes, which suggests that the interstitial atom may allow variation of the Fe–C bond distances and modulation of the metal-sulfur core geometry during catalysis [37]. Moreover, the recent observation of non-CO2-based methane (CH4) formation by Mo nitrogenase in the presence of CO2 suggests the interstitial carbide as one potential source for CH4 formation in this case, which may originate from the interaction between CO2 and the interstitial carbide and the subsequent release of this atom in the form of CH4 [27].

The vanadium (V) nitrogenase

Although the role of V in enhancing nitrogen fixation by diazotrophic microorganisms was documented as early as that of Mo [38], the existence of an “alternative”, V-containing form of nitrogenase was only verified in the late 1980s, when this enzyme was purified from the tungsten (W)-tolerant mutants of A. chroococcum [39–41] and A. vinelandii [42–44] in which the structural genes of Mo nitrogenase were deleted. Like its Mo counterpart, the V nitrogenase is also a two-component enzyme system consisting of the Fe protein and the vanadium-iron (VFe) protein [2, 5]. Moreover, the two nitrogenases share a good degree of homology in the primary sequences and the cluster compositions of their component proteins. For example, the vnfH-encoded Fe protein (also designated VnfH) in the V nitrogenase, which is a homodimer of M r ∼60 kDa, has the same subunit composition and molecular mass as its nifH-encoded counterpart (also designated NifH) in the Mo nitrogenase. It shares 91 % sequence identity with its nifH-encoded counterpart [5], and it also has the same, conserved Cys ligands for the subunit-bridging [Fe4S4] cluster, as well as the GXGXXG consensus nucleotide-binding motif, in its primary sequence. Likewise, despite the presence of an extra vnfG-encoded δ-subunit, the vnfD- and vnfK-encoded α- and β-subunits of the VFe protein share ~33 and ~32 % sequence identity, respectively, with the nifD- and nifK-encoded α- and β-subunits of the MoFe protein [5], and the ligands for both the P- and M-clusters in the MoFe protein are also conserved in the sequence of VFe protein. The subunit composition of VFe protein, however, seems to vary between organisms, with the VFe protein purified from A. chroococcum being an α2β2δ2-hexamer of M r ∼240 kDa [2] and the VFe protein purified from A. vinelandii being an α2β2δ4-octamer of M r ~270 kDa [45].

Consistent with the presence of the same cluster ligands as those in the Mo nitrogenase, the V nitrogenase contains a set of metal centers that are highly homologous to those in its Mo counterpart; yet, these clusters also display structural/redox features that are clearly distinct from those in the Mo nitrogenase. For example, the nifH- and vnfH-encoded Fe proteins are believed to contain identical [Fe4S4] clusters; however, a recent XAS/EXAFS investigation revealed a less ferric nature of the Fe atoms in the cluster of the vnfH-encoded Fe protein than those in the cluster of the nifH-encoded Fe protein. Moreover, this study demonstrated that the [Fe4S4] cluster in the vnfH-encoded Fe protein was best modeled by two stacked [Fe2S2] rhomboids offset by 90°; whereas the cluster in the nifH-encoded Fe protein could also be modeled by two unequal [Fe2S2] rhomboids bent to a greater degree out-of-plane [46]. Likewise, the P-cluster of the VFe protein has long been regarded equivalent to its counterpart in the MoFe protein [47, 48]. However, a more recent electron paramagnetic resonance (EPR) investigation indicated that the P-cluster in the VFe protein existed in a more oxidized state than its counterpart in the MoFe protein in the resting state. Additionally, XAS/EXAFS analysis of the cofactor-deficient VFe protein [5, 49] supplied further proof for the presence of a different P-cluster in the VFe protein, suggesting that this P-cluster (designated the P*-cluster) may consist of paired [Fe4S4]-like clusters instead of the usual [Fe8S7] structure (Fig. 2)Footnote 1. Finally, the FeV cofactor, or the V-cluster, of the VFe protein has also been shown to be highly similar to the M-cluster of MoFe protein [42, 47, 50–53]. However, the EPR features of the two cofactors are clearly distinct from each other, suggesting a difference between the electronic properties of the two cofactors, as well as the interactions between the cofactors and their respective host proteins [45]. Consistent with the observed spectral differences, XAS/EXAFS analysis of the N-methylformamide (NMF)-extracted cofactors [54] revealed that despite a close resemblance between their metal–sulfur core structures, the isolated V-cluster was of a more octahedral symmetry, whereas the isolated M-cluster was of a tetrahedral/trigonal pyramidal structure (Fig. 2). Together, these observations point to a distinctive electronic structure of the V-cluster, which originates, at least in part, from the unique properties of V.

Fig. 2
figure 2

a Schematic presentation (transparent) of the hypothetical structural layout of the MgADP·AlF4 −-stabilized Fe protein/VFe protein complex. MgADP·AlF4 −, [Fe4S4] cluster, P*-cluster and V-cluster are shown as space-filling models. The two subunits of the Fe protein (labeled as H) are colored gray and light wheat, respectively, and the α-, β- and δ-subunits of the MoFe protein are colored red (labeled as D), light blue (labeled as K) and dark gray (labeled as G), respectively. b Components involved in electron transfer during catalysis. All clusters are shown as ball-and-stick models. The structures of the clusters are derived from the available XAS/EXAFS and XES data. Atoms are colored as follows: Fe, orange; S, yellow; V, dark gray; O, red; C, gray; N, blue; Mg, dark green; Al, light green; F, light blue. HCs hydrocarbons

Full size image

The striking homology between the primary sequences and the metal centers of V- and Mo-nitrogenases suggests that the V nitrogenase likely follows the same mode of action as its Mo counterpart during catalysis, forming a functional complex between its two component proteins to enable the ATP-dependent, inter-protein transfer of electrons from the [Fe4S4] center of the Fe protein, via the P*-cluster, to the V-cluster of the VFe protein for substrate reduction (Fig. 2). The overall homology in structure and mode-of-action renders a functional similarity between the V- and Mo-nitrogenases, which is reflected by the abilities of the two nitrogenases to catalyze the reduction of a similar set of substrates, such as N2, H+, C2H2 and CN− [5]. However, the unique structural features of the V nitrogenase, including those of its P*- and V-clusters, as well as those of the protein environments surrounding these clusters, underlie the distinct catalytic properties of this nitrogenase. For example, it has been documented that the VFe protein is less efficient than the MoFe protein in reducing N2, H+ and C2H2 and, in the case of N2 reduction, the VFe protein generates NH3 and H2 at a lower NH3/H2 ratio than its Mo counterpart.

But, perhaps the biggest difference between the catalytic properties of the V- and Mo-nitrogenases is their differential abilities to reduce CO and CO2. The V nitrogenase is capable of reducing CO to hydrocarbons of up to C4 in length [i.e., CH4, C2H4, ethane (C2H6), propylene (C3H6), propane (C3H8), butylene (C4H8) and butane (C4H10)] [29–31]. In comparison, except for CH4, the Mo nitrogenase can generate the same set of hydrocarbons from CO reduction; however, the total amount of hydrocarbons formed by the Mo nitrogenase is only 0.1 % of that by the V nitrogenase, and the product distribution and deuterium effect of the Mo nitrogenase-catalyzed reaction also clearly differ from those of the V nitrogenase-catalyzed reaction. Similarly, the V nitrogenase is able to reduce CO2 to C1 (i.e., CH4) and C2 (i.e., C2H4 and C2H6) hydrocarbons in the presence of D2O, but it does not form hydrocarbons from CO2 in the presence of H2O. In contrast, the Mo nitrogenase is only capable of generating CH4 in the presence of H2O, yet the CH4 generated in this case does not originate from CO2 [27]. It should be noted that certain Mo nitrogenase variants have been shown to exhibit altered CO- and CO2-reducing activities, although how substitution of one or two amino acid residues in the Mo nitrogenase could cause these changes remains unknown at the present time. What is known, however, is that synthetic V compounds are more active than synthetic Mo compounds in the reductive coupling of two CO moieties into functionalized acetylene ligands [55], which suggests that V (a group VB transition metal) is superior to Mo (a group VIB transition metal) in this particular type of reaction. Importantly, the ability of nitrogenase to form hydrocarbons from CO is analogous to the capacities of late transition metal catalysts in the industrial Fischer–Tropsch (FT) process [56]. However, contrary to the industrial process, the enzymatic reaction utilizes H+ instead of H2 (an expensive syngas) as the hydrogen source, and it occurs at room temperature and pressure, making it an appealing template for future development of energy-efficient approaches for CO-based hydrocarbon production [31].

The iron (Fe)-only nitrogenase

The Fe-only nitrogenase is the least understood of the three homologous nitrogenases. This enzyme has been isolated from A. vinelandii, Rhodospirillum rubrum and Rhodobacter capsulatus [2, 9, 10] and, like its Mo- and V-counterparts, it is a two-component system comprising the Fe protein and the iron-iron (FeFe) protein. The Fe protein of the Fe-only nitrogenase is encoded by anfH (also designated AnfH). An α 2-homodimer bridged by a [Fe4S4] cluster between the subunits, the anfH-encoded Fe protein shows a high degree of homology to both nifH- and vnfH-encoded Fe proteins, although the homology between the anfH-encoded Fe protein and nifH/vnfH-encoded Fe proteins is lower than that between the nifH- and vnfH-encoded Fe proteins. The FeFe protein of the Fe-only nitrogenase is the equivalent to the MoFe or VFe protein of the Mo- or V-nitrogenase. Like the VFe protein, the FeFe protein consists of α-, β- and δ-subunits, which are encoded by anfD, anfK and anfG, respectively. While progress has been made in improving the preparations of the FeFe protein from R. capsulatus, the successful purification of an intact form of this protein has remained a challenge and the most competent form of this protein reported so far has an α2β2δ2-hexameric subunit composition [2]. EPR, Mössbauer and EXAFS studies of the R. capsulatus FeFe protein have provided spectroscopic evidence that this protein contains metal centers homologous to the P/P*-cluster (designated the P’-cluster) and the M/V-cluster (designated the FeFe cofactor or the Fe-cluster), respectively [57, 58]. Given the overall homology between the Fe-only nitrogenase and its Mo and V counterpart, this nitrogenase likely utilizes the same mode of action during catalysis, which involves complex formation between the two component proteins and electron transfer from the [Fe4S4] cluster of the Fe protein, via the P′-cluster, to the Fe-cluster of the FeFe protein, where substrate reduction takes place. Interestingly, the Fe-cluster may very well resemble a so-called L-cluster in composition and structure, with the latter being an Fe/S precursor that can be matured into an M-cluster on NifEN (see below).

The nitrogenase assembly protein NifEN

NifEN is an indispensable assembly apparatus along the biosynthetic pathway of the M-cluster, receiving an Fe/S precursor from NifB and processing it further into an M-cluster before passing it on to the MoFe protein. Encoded by nifE and nifN, NifEN is an α 2 β 2-heterotetramer that shares a good degree of sequence homology with the MoFe protein [59, 60]. Sequence analysis reveals that NifEN contains cluster-binding sites that are homologous to those found in the MoFe protein: a “P-cluster” site at the interface of each αβ-subunit dimer, which houses a P-cluster analog; and an “M-cluster” site within each α-subunit, which houses the conversion of an Fe/S precursor to a mature M-cluster. Consistent with this observation, the “P-cluster” and the “M-cluster” in NifEN were identified by EPR, XAS/EXAFS and XES methods as a [Fe4S4]-type cluster [61] and an [Fe8S9C] cluster (designated the L-cluster) [62–64], respectively (Fig. 3a, b). Moreover, X-ray crystallographic analysis of an L-cluster-bound form of NifEN has further confirmed that the L-cluster is nearly indistinguishable in structure from the Fe/S core structure of the M-cluster [65, 66]; whereas biochemical experiments [67] have demonstrated that the NifEN-bound L-cluster can be converted to a fully matured M-cluster upon Fe protein-mediated insertion of Mo and homocitrate prior to its delivery to its target binding site in the MoFe protein (Fig. 3c).

Fig. 3
figure 3

a Surface presentation (transparent) of the hypothetical structural layout of the MgADP·AlF4 −-stabilized Fe protein/NifEN complex. MgADP·AlF4 −, [Fe4S4] clusters and L-cluster are shown as space-filling models. The two subunits of the Fe protein (labeled as H) are colored gray and light wheat, respectively, and the α- and β-subunits of the NifEN are colored red (labeled as E) and light blue (labeled as N), respectively. b Components involved in electron transfer during catalysis. All clusters are shown as ball-and-stick models. c Schematic presentation of the conversion of L- to M-cluster (c, top) on NifEN upon the replacement of a terminal Fe atom by Mo and homocitrate (HC). The Fe protein (c, bottom) serves as an ATP-dependent Mo/HC insertase in this process. Atoms are colored as follows: Fe, orange; S, yellow; O, red; C, gray; N, blue; Mg, dark green; Al, light green; F, light blue. PYMOL was used to create this figure (PDB IDs: 1N2C, 3PDI)

Full size image

With a [Fe4S4] analog at the “P-cluster” site and an [Fe8S9C] homolog at the “M-cluster” site, NifEN has the equivalents (if simplified) for both clusters that are involved in the electron transfer within the MoFe protein (see Fig. 1). When combined with the Fe protein, NifEN is capable of reducing some of the substrates of Mo nitrogenase, such as C2H2 and N3 −, albeit at much lower efficiencies than the Fe protein/MoFe protein system [68]. Like the Mo nitrogenase, the Fe protein/NifEN system requires the Fe protein to function as an ATP-dependent reductase and deliver electrons to NifEN for substrate reduction, as neither C2H2 nor N3 − could be reduced by this enzymatic system if ATP was absent, if ATP was replaced by ADP or non hydrolyzable ATP analogs, or if the Fe protein was replaced by an Fe protein variant specifically defective in ATP hydrolysis [68]. Thus, formation of an Fe protein/NifEN complex that is homologous to the Fe protein/MoFe protein complex can be predicted during catalysis, which allows electrons to be transferred from the [Fe4S4] cluster of the Fe protein to the L-cluster of NifEN for substrate reduction (Fig. 3c).

There are, however, three interesting features of the Fe protein/NifEN system that set it apart from the Fe protein/MoFe protein system. One, the “P-cluster” analog in NifEN is a [Fe4S4]-type cluster, which is much simpler and, perhaps, far less efficient than the P-cluster ([Fe8S7]) in mediating the inter-protein electron transfer in this system. Two, the “M-cluster” homolog in NifEN is a Mo/homocitrate-free L-cluster ([Fe8S9C]), which is less complex and, possibly, less efficient than the M-cluster ([MoFe7S9C-homocitrate]) in catalyzing substrate reduction. Three, despite a similar location of the L- and M-clusters in their respective protein hosts, the L-cluster is nearly exposed at the surface of NifEN, whereas the M-cluster is buried within the MoFe protein. Taken together, these features contribute to the limited activity of the Fe protein/NifEN system in substrate reduction. However, such a “skeleton” system could be used to step-by-step add the “missing features” and address the functions of these features. Moreover, advantages could be taken of the slow turnover rates of the substrates by this enzymatic system, which may allow capture of substrates/intermediates for the subsequent mechanistic investigation.

The protochlorophyllide oxidoreductase (DPOR) and chlorophyllide oxidoreductase (COR)

The biosynthesis of chlorophylls and bacteriochlorophylls [6, 7] involves reduction of the C17–C18 double bond of protochlorophyllide (Pchlide) by dark operative protochlorophyllide oxidoreductase (DPOR) (Fig. 4a, b), which converts Pchlide to chlorophyllide (Chlide) (Fig. 4c). In addition, the biosynthesis of bacteriochlorophylls specifically requires another reduction of the C7–C8 double bond by chlorophyllide oxidoreductase (COR), which converts Chlide into bacteriochlorophyllide (Bchlide) (Fig. 4c). The enzymes involved in these reductive steps along the biosynthetic pathway of chlorophylls and bacteriochlorophylls (encoded by chl and bch, respectively) share certain homology with nitrogenase enzymes (encoded by nif, vnf and anf, respectively). Like nitrogenases, DPOR or COR consists of two separately-purifiable, metallosulfur proteins: a homodimeric reductase component (encoded by bchL/chlL or bchX/chlX, respectively) that shares up to 30 % sequence identity with the Fe protein (encoded by nifH); and a heterotetrameric catalytic component (encoded by bchNB/chlNB or bchYZ/chlYZ, respectively) that shares up to 15 % sequence identity with the MoFe protein (encoded by nifDK) [7]. Moreover, as is with the nitrogenase system, the two components of DPOR or COR form a functional complex during catalysis, which allows electron transfer from the reductase component to the catalytic component for the subsequent reduction of C17–C18 or C7–C8 double bond of the conjugated tetrapyrrole ring system [7].

Fig. 4
figure 4

a Surface presentation (transparent) of the structure of the MgADP·AlF4 −-stabilized ChlL/ChlNB complex. MgADP·AlF4 −, [Fe4S4] clusters and Pchlide are shown as space-filling models. The two subunits of ChlL (labeled as L) are colored gray and light wheat, respectively, and the α- and β-subunits of ChlNB are colored red (labeled as N) and light blue (labeled as B), respectively. b Components involved in electron transfer during catalysis. All clusters and Pchlide are shown as ball-and-stick models. Atoms are colored as follows: Fe, orange; S, yellow; O, red; C, gray; N, blue; Mg, dark green; Al, light green; F, light blue. PYMOL was used to create this figure (PDB ID: 2YNM). c DPOR catalyzes the formation of Chlide through ATP-dependent, stereospecific reduction of the C17-C18 double bond of Pchlide, while COR catalyzes the formation of Bchlide through ATP-dependent, stereospecific reduction of the C7–C8 double bond of Chlide

Full size image

Structural information has become available of the dimeric reductase component (designated the L2 protein) and the heterotetrameric catalytic component (designated the N2B2 protein) of DPOR from various organisms in the recent years [69–72]. Together with the emerging biochemical data [73–78], these structural data have led to the elucidation of the mechanistic details of Chlide formation by DPOR. Like the Fe protein of nitrogenase, the L2 protein belongs to a family of nucleotide-utilizing proteins, and it serves as an ATP-dependent reductase in Pchlide reduction. Encoded by bchL/chlL, the L2 protein is a homodimer of M r ∼60 kDa. Each of its subunits has a binding site for MgATP, and the two subunits are bridged by a [Fe4S4] cluster in between [69, 72]. The N2B2 protein is the catalytic partner of the L2 protein in Pchlide reduction. It is an α 2 β 2-tetramer of M r ∼210 kDa, and its α-and β-subunits are encoded by bchN/chlN and bchB/chlB, respectively [73–78]. The N2B2 protein contains, in each αβ-subunit pair, one [Fe4S4] cluster that is bridged between the α- and β-subunits, and one binding site for Pchlide that is mainly located within the α-subunit but also capped by the β-subunit [69–71]. Binding of Pchlide to the N2B2 protein is required for the formation of DPOR complex [76], and hydrolysis of ATP is believed to facilitate the association and dissociation between the L2 protein and the N2B2 protein [69, 76]. Consistent with this hypothesis, the X-ray crystal structure of a MgADP·AlF4 −-stabilized complex between the L2 protein and the N2B2 protein from Prochlorococcus marinus [69] suggests that electrons are transferred sequentially from the [Fe4S4] cluster of the L2 protein, via the [Fe4S4] cluster of the N2B2 protein, to Pchlide (Fig. 4a, b), which allows the reduction to occur at the C17-C18 position of its tetrapyrrole ring (Fig. 4c).

The reaction catalyzed by DPOR is depicted as follows: Pchlide + 2H+ + 4MgATP + 2e− → Chlide + 4MgADP + 4P i [69]. Interestingly, DPOR is capable of catalyzing the two-electron reduction of N3 − or N2H4 to NH3, which mirrors the ability of nitrogenase to catalyze the reduction of the same substrates [69]. However, despite sharing the two “simple” substrates with nitrogenase, DPOR is unable to reduce the more “complex” substrates of nitrogenase, such as N2 [1] and CO [29, 31], which require the transfer of more than two electrons through the enzymatic system. The narrow substrate spectrum of these two systems likely originates from the presence of a [Fe4S4] clusters instead of a high-nuclearity, [Fe8S7] cluster at its “P-cluster” site. Nevertheless, the striking similarities between the structures and functions of DPOR and nitrogenases point to a strong evolutionary tie between the two enzyme systems.

Compared to the well-studied DPOR system, information of the dimeric, reductase component (designated the X2 protein) and the heterotetrameric, catalytic component (designated the Y2Z2 protein) of the COR system has remained scarce. However, the homology between COR and DPOR suggests a close resemblance between the reactions catalyzed by these two enzyme systems [7, 79]. The bchYZ- and bchX-encoded subunits of COR share sequence identity of up to 22 and 35 %, respectively, with the bchNB- and bchL-encoded subunits of DPOR [7]. Sequence alignments between the bchX- and chlL/bchL-encoded subunits indicate that both the Cys ligands of the [Fe4S4] cluster and the ATP-binding sites are conserved in these proteins [80]. Moreover, as observed in the case of the L2 protein, EPR analysis revealed signals consistent with the presence of a [Fe4S4] cluster in the X2 protein [7]; whereas biochemical experiments demonstrated that the L2 protein could be replaced by X2 protein in the activity assay of DPOR when Pchlide was supplied as a substrate [75]. Similarly, sequence analysis of the Y2Z2 protein suggests the presence of Cys ligands for [Fe4S4] clusters at locations comparable to those in the N2B2 protein. Consistent with this observation, EPR experiments further revealed the presence of a redox-active [Fe4S4] cluster in the Y2Z2 protein [75]. Together, these results strongly suggest that COR employs a highly similar mechanism to DPOR for the reduction of the tetrapyrrole substrate.

The nitrogen-fixation-like (Nfl) proteins

Genomic analysis indicates the presence of nifH-like and nifD/nifK-like genes among all methanogenic microbes [81]. These genes clearly differ from the other sets of genes that encode proteins for nitrogen fixation in methanogenic organisms. Phylogenetic analysis has led to the conclusion that the nif-like sequences nflH and nflD (nfl stands for nitrogen fixation like) lie basal in the tree between the nitrogenase and the DPOR/COR clades [81], and another study has suggested that the two nfl-encoded proteins form an enzymatic system that represents a “simplified”, ancestral member of this family of homologous enzymes [82]. Sequence alignments between the nflH-encoded protein (also designated NflH), the Fe protein and the L2 protein reveal the presence of a conserved ATP-binding site in the three proteins; moreover, they reveal the presence of two conserved cysteine residues from each subunit that could coordinate an analogous [Fe4S4] cluster at the interfaces of these dimeric proteins (Fig. 5a). The nflD-encoded protein (designated NflD), on the other hand, is somewhat different from the nifD- and chlN/bchN-encoded proteins based on sequence analysis, as none of the ligands for the M-cluster can be found in the sequence of the nflD-encoded protein. Interestingly, some cysteine residues that ligate the P-cluster in the MoFe protein are conserved in the nflD-encoded sequence, which would be sufficient for the ligation of a [Fe4S4] cluster between the two subunits of a dimeric form of the nflD-encode protein (Fig. 5a).

Fig. 5
figure 5

a Schematic presentation (transparent) of the hypothetical structural layout of the MgADP·AlF4 −-stabilized NflH/NflD complex. MgADP·AlF4 −, [Fe4S4] clusters and cofactor F430 are shown as space-filling models. The two subunits of NflH (labeled as H) are colored gray and light wheat, respectively, and the two subunits of NflD (labeled as D) are colored red and light blue, respectively. b Components involved in the proposed electron transfer during catalysis. All clusters and cofactor F430 are shown as ball-and-stick models. Atoms are colored as follows: Fe, orange; S, yellow; V, dark gray; O, red; C, gray; N, blue; Mg, dark green; Al, light green; F, light blue

Full size image

Initial information regarding the functions of the nfl-encoded proteins has been obtained through studies of Methanocaldococcus jannaschii, a methanogenic microbe that does not perform nitrogen fixation or photosynthesis [7]. These studies have shown that the nflH- and nflD-encoded proteins interact with each other and that both proteins are expressed constitutively and independently of the availability of nitrogen. Combined with what is known about the homologous DPOR/COR systems, these observations have led to the speculation that the enzyme system comprising nflH- and nflD-encoded proteins may be involved in the biosynthesis of cofactor F430, the active site of methyl-coenzyme M reductase that catalyzes the final step of methanogenesis. Given the homology between the nfl- and bch/chl systems, it appears possible that the nflD- and nflH-encoded proteins could work together and catalyze one or more ring reduction steps that are required for the biosynthesis of cofactor F430 [83] and that the catalytic mechanism of this system likely mirrors that of the DPOR or COR system, which involves inter-protein electron transfer from the reductase component (the niflH-encoded protein) to the catalytic component (the nflD-encoded protein) and the subsequent substrate reduction at its binding site in the nflD-encoded protein (Fig. 5b). Clearly, further investigation is required to test this hypothesis and elucidate the functions of the nfl-encoded proteins. Perhaps more importantly, these studies may help establish an interesting evolutionary link between proteins involved in methanogenesis (Nfl), bacterial photosynthesis (Bch/Chl) and nitrogen fixation (Nif), which are the three key metabolic processes central to the emergence of life on Earth.