Differential Reduction of CO₂ by Molybdenum and Vanadium Nitrogenases

Abstract

The Mo- and V-nitrogenases are two homologous enzymes with distinct structural and catalytic features. Previously, we demonstrated that the V-nitrogenase was nearly 700 times more active than its Mo-counterpart in reducing CO to hydrocarbons. Here, we report a similar discrepancy between the two nitrogenases in the reaction of CO₂ reduction, with the V-nitrogenase capable of reducing CO₂ to CO, CD₄, C₂D₄ and C₂D₆, and its Mo-counterpart only capable of reducing CO₂ to CO. Further, we show that V-nitrogenase may route the formation of CD₄ in part via CO₂-derived CO, but it does not catalyze the formation of C₂D₄ and C₂D₆ via this route. The exciting observation of C-C coupling by V-nitrogenase from CO₂ adds another interesting reaction to the catalytic repertoire of this unique enzyme system; whereas the differential activities of V- and Mo-nitrogenases in CO₂ reduction provide an important framework for systematic investigations of this reaction in the future.

Nitrogenases are a family of complex metalloenzymes that catalyze a key step in global nitrogen cycle: the reduction of atmospheric nitrogen (N₂) to a bio-accessible form, ammonia (NH₃).^[1–4] Apart from N₂, nitrogenases are also capable of reducing alternative substrates, such as acetylene (C₂H₂) and carbon monoxide (CO), thereby displaying a unique versatility in processing small, carbon-containing molecules.^[1,5] The molybdenum (Mo)- and vanadium (V)-nitrogenases are two homologous members of this enzyme family, sharing a good degree of homology in primary sequence and cluster composition.^[5,6]. Both enzymes are homologous binary systems, which consist of (i) a reductase component (nifH- or vnfH-encoded Fe protein), which contains one subunit-bridging [Fe₄S₄] cluster and one ATP-binding site per subunit; and (ii) a catalytic component (nifDK-encoded MoFe or vnfDGK-encoded VFe protein), which contains a P-cluster at the α/β-subunit interface and a cofactor (FeMoco or FeVco) within each α-subunit (Fig. 1A). Moreover, both enzymes use the same mode of action during catalysis, which involves the formation of a functional complex between the two component proteins,^[7,8], the ATP-dependent transfer of electrons from the [Fe₄S₄] cluster of the reductase component, via the P-cluster, to the cofactor of the catalytic component, and the eventual reduction of substrates at the cofactor site upon accumulation of a sufficient amount of electrons (Fig. 1A).

Figure 1.

Comparison between the Mo- and V-nitrogenases. Schematic presentations of the catalytic components (A) and structural models of the P-clusters (B) and cofactors (C) in Mo- (left) and V- (right) nitrogenases. Atoms are colored as follows: Fe, orange; S, yellow; Mo, cyan; V, magenta; O, red; C, gray.

Despite their homology in structure and function, the two nitrogenases are clearly distinct from each other with regard to their associated metalloclusters. The P-cluster of the Mo-nitrogenase assumes a ‘standard’, [Fe₈S₇] structure; whereas the P-cluster of the V-nitrogenase consists of a pair of [Fe₄S₄]-like clusters (Fig. 1B).^[5,7–9]. Likewise, despite a striking homology in structure, the cofactors of the Mo- and V-nitrogenases are distinguishable not only by heterometals, but also by electronic properties (Fig. 1C).^[10] The differences between the metal clusters in the Mo- and V-nitrogenases underline the differences in the catalytic behavior of these homologous enzymes. It has been documented that the V-nitrogenase is less efficient than its Mo-counterpart in terms of N₂ reduction; yet, this nitrogenase can reduce C₂H₂ to ethane (C₂H₆), a catalytic activity not observed in the case of Mo-nitrogenase.^[6,8]. Perhaps the biggest discrepancy between the catalytic properties of the two nitrogenases is their abilities to reduce CO to hydrocarbons, with the V-nitrogenase showing an overall activity that is nearly 700 times higher than its Mo-counterpart.^[11,12] This observation has prompted us to conduct a comparative study between the Mo- and V-nitrogenases to address the questions of (i) whether the two nitrogenases can also reduce CO₂ to hydrocarbons and (ii) if they have the same discrepancy in their activities to generate hydrocarbons from this substrate.

Consistent with an earlier report,^[13] the Mo-nitrogenase can reduce CO₂ to CO (Fig. 2A, triangles). Like its Mo-counterpart, the V-nitrogenase can also catalyze the reduction of CO₂ to CO (Fig. 2A, circles) in an ATP-dependent reaction (Fig. S1) that contains 20 mM dithionite at a pH of 8.5. The two nitrogenases display comparable efficiencies in H₂O-based reactions, forming approximately the same amount of CO from CO₂ over a time period of 180 min (Fig. 2A). Moreover, both nitrogenases exhibit roughly the same increase of activity in the formation of CO from CO₂ upon substitution of D₂O for H₂O, reaching a maximum increase of activity at 120 min (Fig. 2A). Apart from CO, CH₄—a further reduced C1 product—can be detected in reactions catalyzed by both Mo- and V-nitrogenases when CO₂ is supplied as a substrate (Fig. 2B). However, when H₂O is replaced by D₂O, the activity of CH₄ formation by V-nitrogenase increases from 0 to a maximum of 22.2 nmol/μmol protein (Fig. 2B, solid vs. open circles); whereas the activity of CH₄ formation by Mo-nitrogenase decreases from a maximum of 7.3 nmol/μmol protein to 0 (Fig. 2B, solid vs. open triangles). Such a disparate D₂O effect implies a difference in the routes to CH₄ formation taken by the two nitrogenases.

Figure 2.

Product formation by Mo- and V-nitrogenases in the presence of CO₂. Time-dependent formation of CO (A), CH₄ (B), C₂H₄ (C) and C₂H₆ (D) by Mo-nitrogenase in H₂O (blue open triangle, dashed line) or D₂O (blue solid triangle, solid line) and by V-nitrogenase in H₂O (red open circle, dashed line) or D₂O (red solid circle, solid line). Data are presented as mean ± SD (N=3) after background correction.

The difference between the V- and Mo-nitrogenases in CO₂ reduction is further illustrated by the difference in their abilities to form C-C bonds from CO₂. In the presence of H₂O, little or no C2 product can be detected in the reactions of CO₂ reduction by either Mo- or V-nitrogenase (Fig. 2C and D, open triangles and open circles). In the presence of D₂O, however, C₂D₄ (Fig. 2C, solid circles) and C₂D₆ (Fig. 2D, solid circles) can be detected as products of CO₂ reduction by V-nitrogenase; whereas these C2 products are hardly detectable in the same reaction catalyzed by Mo-nitrogenase (Fig. 2C and D, solid triangles). Thus, as is observed in the case of CH₄ formation, there is a clear increase in the activities of C₂D₄ and C₂D₆ formation by V-nitrogenase upon substitution of D₂O for H₂O; whereas these activities remain marginal in the reaction catalyzed by Mo-nitrogenase following such a substitution. Moreover, like the formation of CH₄, the formation of C2 products by V-nitrogenase is ATP-dependent, as no C₂D₄ and C₂D₆ can be detected in the absence of ATP (Fig. S1).

GC-MS analysis supplies further evidence for the difference between Mo- and V-nitrogenases in hydrocarbon formation from CO₂. When ¹²CO₂ is replaced by ¹³CO₂, ¹³CD₄ can be detected in the V-nitrogenase-catalyzed reaction in D₂O (Fig. 3B); yet, ¹³CH₄ is absent from the Mo-nitrogenase-catalyzed reaction in H₂O (Fig. 3A). This observation confirms CO₂ as a carbon source for the CD₄ generated by V-nitrogenase while suggesting a different carbon source for the same C1 product generated by Mo-nitrogenase. Aside from CD₄, CO₂ also gives rise to the C2 products in the reaction catalyzed by V-nitrogenase, as ¹³C₂D₄ (Fig. 3C) and ¹³C₂D₆ (Fig. 3D) can be detected in the presence of D₂O upon substitution of ¹³CO₂ for ¹²CO₂. Together, the GC-MS and activity data highlight the difference between the reactions of CO₂ reduction by V- and Mo-nitrogenases, showing the ability of V-nitrogenase to form C1 and C2 hydrocarbons along with CO and the inability of its Mo-counterpart to generate products other than CO under these experimental conditions. Given the previous observation that V-nitrogenase can reduce CO to hydrocarbons,^[11,12] the co-production of CO and hydrocarbons by this enzyme as products of CO₂ reduction raises a relevant question of whether it is the CO₂-derived CO that gives rise to the hydrocarbon products.

Figure 3.

GC-MS analyses of hydrocarbon products formed by Mo- and V-nitrogenases. The products were generated by Mo-nitrogenase in H₂O (A) or by V-nitrogenase in D₂O (B–D) when ¹²CO₂ (1) or ¹³CO₂ (2) was supplied. The mass-to-charge (m/z) ratios at which the products were traced are indicated.

This question can be addressed by directly supplying CO to the V-nitrogenase in a concentration simulating the maximum concentration of CO achieved in the ‘equilibrated state’ of CO₂ reduction by this enzyme (see Fig. 2A) and monitoring the formation of C1 and C2 hydrocarbons in D₂O over a time period of 180 minutes. Interestingly, the CO-based formation of CD₄ by V-nitrogenase (Fig. 4A, dotted circles) displays an activity increase of 12.6 nmol/μmol protein between 0 and 30 minutes; whereas the CO₂-based formation of CH₄ exhibits a nearly identical activity increase of 11.8 nmol/μmol protein between 30 and 60 minutes after an initial ‘lag’ phase between 0 and 30 minutes (Fig. 4A, solid circles). This observation suggests the possibility for V-nitrogenase to route the formation of C1 hydrocarbon via CO, as the 30-minute delay could be correlated with a need for the enzyme to accumulate a sufficient amount of CO₂-derived CO to initiate further reduction of CO to CD₄. On the other hand, the time courses of CD₄ formation from CO (Fig. 4A, dotted circles) and CO₂ (Fig. 4A, solid circles) diverge beyond 60 minutes, with a gradual decrease of activity in the case of the former and a gradual increase of activity in the case of the latter. The difference between the two time courses (Fig. S2A) could represent the portion of CD₄ that is generated independently from CO₂-derived CO. Consistent with this hypothesis, there is a notable difference between the CO- and CO₂-based reactions in the percentage activity of CH₄ formation in H₂O relative to that in D₂O (Fig. S3A), with the CO-based reaction favoring the formation of CH₄ in H₂O over that in D₂O (42%) considerably more than the CO₂-based reaction (0%). It is possible, therefore, that the V-nitrogenase generates CH₄ both from CO₂-derived CO and from CO₂ and/or other CO₂-derived intermediate(s).

Figure 4.

Formation of hydrocarbon products by V-nitrogenase. Time-dependent formation of CD₄ (A), C₂D₄ (B) and C₂D₆ (C) from CO₂ (solid circle) or CO (dotted circle) by V-nitrogenase in D₂O. CO was added in a concentration of 110 pm in assays involving the direct formation of products from CO, which was equivalent to the maximum concentration of CO that could be generated from CO₂ reduction by V-nitrogenase (also see Fig. 2). Data are presented as mean ± SD (N=3) after background correction.

Contrary to CD₄, both C₂D₄ and C₂D₆ seem to be produced by V-nitrogenase via a CO-independent route, as no C2 products can be detected (Fig. 4B and C, dotted circles) upon direct addition of CO in the same amount produced by V-nitrogenase from CO₂ reduction in the ‘equilibrated state’ (see Fig. 2A). This observation suggests that, instead of CO, CO₂ and/or other CO₂-derived intermediates are responsible for the formation of C2 hydrocarbon products by V-nitrogenase. Indeed, as is observed in the case of CH₄ formation, there is a significant difference between the CO- and CO₂-based reactions in the percentage activity of C₂H₄ (Fig. S3B) or C₂H₆ (Fig. S3C) formation in H₂O relative to that in D₂O, with the CO-based reaction favoring the formation of C2 products in H₂O over that in D₂O (C₂H₄, 92%; C₂H₆, 65%) considerably more than the CO₂-based reaction (C₂H₄, 0.7%; C₂H₆, 0%). Such a disparate deuterium effect on the CO- and CO₂-based reactions further implies that V-nitrogenase routes the formation of C2 hydrocarbons via CO₂ or other CO₂-derived intermediate(s). The lack of contribution of CO to the formation of C2 hydrocarbons in this case could be explained by an insufficient CO concentration achieved by the reduction of CO₂, which does not allow the formation of C-C bonds. More excitingly, it defines the ability of V-nitrogenase to directly use CO₂ as a substrate for the initial C-C coupling and the subsequent carbon chain extension.

The ability of certain variants of Mo-nitrogenase to reduce CO₂ to CH₄ was reported recently.^[14] To our surprise, contrary to what has been reported for these variants of Mo-nitrogenase, the wild-type Mo-nitrogenase cannot reduce CO₂ to CH₄; rather, it uses an unknown carbon source to generate CH₄ in the presence of CO₂ and H₂O. Considering the presence of an interstitial carbide^[15–18] and a homocitrate moiety in the FeMoco,^[15,17] it can be postulated that CO₂ or its derivative in H₂O somehow promotes the release of the central carbide ligand or the carbon-containing groups of homocitrate in the form of CH₄. Alternatively, the side chain groups of certain amino acids at the active site of Mo-nitrogenase may also serve as a carbon source for the production of CH₄ in the presence of CO₂. Remarkably, despite the unclear nature of the carbon source, the formation of CH₄ by Mo-nitrogenase is ATP-dependent and requires the presence of both component proteins; moreover, it only occurs in the presence of CO₂ and H₂O (Fig. S4). This observation points to a redox-dependent characteristic of this reaction, as the requirement for ATP and both components is specifically associated with the transfer of electrons through the enzyme system, which may permit the initial binding and processing of CO₂ or its derivative in H₂O and the subsequent interaction between CO₂ or CO₂-derived intermediate(s) and the carbon species that eventually gives rise to CH₄. Given the overall homology between the Mo- and V-nitrogenases, one would expect the V-nitrogenase to catalyze the same, unspecific formation of CH₄ as its Mo-counterpart from a different carbon source than CO₂. While this possibility cannot be ruled out, our current data (see Fig. 3A and B) clearly demonstrate that the CH₄ formed by V-nitrogenase is derived, at least in part, from CO₂. Further investigation of the origin of the different routes taken by the two nitrogenases to CH₄ formation could be informative, particularly with regard to the initial binding and processing of CO₂ by this enzyme system.

Based on the hydrocarbon products identified so far in the gas phase, the V-nitrogenase generates carbon-containing compounds at a slow rate from CO₂ reduction, forming 0.3 mol CO, 0.02 mol CH₄, 0.04 mol C₂H₄ and 0.002 mol C₂H₆ per mol protein. Nevertheless, the ability of V-nitrogenase to form hydrocarbons, particularly the C2 products, from CO₂, is a most exciting finding of the current study, because it adds another exciting reaction to the catalytic repertoire of this unique enzyme system. As was observed in the case of CO reduction,^[12] the V-nitrogenase is superior to its wild-type Mo-counterpart in generating hydrocarbons from CO₂. The disparate CO-reducing activities of V- and Mo-nitrogenases were compared with the differential capacities of synthetic V- and Mo-compounds to reductively couple two CO moieties into functionalized acetylene ligands;^[19] whereas an alteration of CO-reducing activities was reported for MoFe protein variants containing modified residues at the active site.^[20] By analogy, the disparate CO₂-reducing activities of the two nitrogenases could also stem from the structural/redox differences between FeVco and FeMoco, as well as the protein environments surrounding the two cofactors (see Fig. 1). Additionally, the different structural/redox properties of the P-clusters in the two nitrogenases could further contribute to the differences between their abilities to reduce CO₂ (see Fig. 1). In fact, the ability of nitrogenase to generate hydrocarbons from CO₂ was first described in the case of a cofactor-deficient variant of MoFe protein^[21] and attributed to its unique ‘P-cluster’ that contains a [Fe₄S₄]-like cluster pair instead of the ‘normal’ [Fe₈S₇] P-cluster.^[22] Interestingly, the P-cluster of V-nitrogenase also consists of a pair of [Fe₄S₄]-like clusters^[5,8,9] and could, in principle, serve as a site for CO₂ reduction on its own; only in the case of the holo V-nitrogenase, the presence of the cofactor ‘downstream’ of the P-cluster along the electron transfer pathway (see Fig. 1) may effectively ‘funnel’ the electrons toward the cofactor site and only allow a small, ‘leaky’ activity of CO₂ reduction at the P-cluster site. The possibility of two reaction sites (i.e., P-cluster and cofactor) and different reaction routes (i.e., via CO or other CO₂-derived intermediates) for CO₂ reduction makes it a challenging task to elucidate the mechanistic details of this reaction. Nevertheless, the work reported herein provides an important framework for systematic investigations of this unique reaction in the future, which will hopefully lead to development of nitrogenase-based strategies to recycle the greenhouse gas into the useful carbon fuel.