Belt-Sulfur Mobilization as a Crucial Mechanistic Feature Shared between the Vanadium and Molybdenum Nitrogenases

SUMMARY

Nitrogenase catalyzes the reduction of N₂ to NH₃ at its active-site cofactor. Recent studies of the ‘conventional’ Mo-nitrogenase suggest a plausible involvement of all cofactor belt-S sites in catalysis. Here, we use analytical, enzymatic and spectroscopic methods to demonstrate the same dynamic belt-S mobilization by the ‘alternative’ V-nitrogenase during catalysis. Our results point to belt-S turnover as a common catalytic feature of the homologous Mo- and V-nitrogenases while identifying an activated, but N₂-free conformation of the V-nitrogenase that bears great potential for facilitating future mechanistic explorations of the intriguing nitrogenase enzyme.

Graphical Abstract

The eTOC Blurb

Nitrogenase catalyzes the reduction of N₂ to NH₃ at its active-site cofactor. Catalysis by the homologous V- and Mo-nitrogenases involves the same dynamic belt-S mobilization that occurs asymmetrically in the two cofactors, although V-nitrogenase differs from its Mo-counterpart in its inability to capture N₂ under sulfur- and reductant-depleted conditions. The shared feature between the two nitrogenases in belt-S turnover points to nitrogenase as a sulfite reductase, whereas their distinct abilities in N₂ capture could facilitate future mechanistic explorations of nitrogenase.

INTRODUCTION

Nitrogenases are complex metalloenzymes that catalyze biological nitrogen fixation, a key process in the global nitrogen cycle that converts the atmospheric dinitrogen (N₂) to the bioaccessible ammonia (NH₃) at ambient conditions. The molybdenum (Mo)- and vanadium (V)-dependent nitrogenases are two homologous members of this enzyme family (Figure 1).^1–10 Both nitrogenases are binary systems comprising a reductase component (collectively termed Fe protein) and a catalytic component (termed MoFe or VFe protein). Encoded by nifH or vnfH, the Fe protein of the Mo-nitrogenase (designated Av2) or V-nitrogenase (designated Av2^V) is a homodimer containing a subunit-bridging [Fe₄S₄] cluster and an ATP-binding site per subunit. Encoded by nifDK or vnfDGK, the MoFe protein (designated Av1) or VFe protein (designated Av1^V) has an α₂β₂-tetrameric core that houses a pair of homologous, high-nuclearity metalloclusters: a Fe₈ P-cluster (designated P- or P^V-cluster) at each α/β-subunit interface; and a MoFe₇- or VFe₇-cofactor (designated M- or V-cluster) within each α-subunit. Catalysis by both nitrogenases is believed to involve repeated association and dissociation between the two components that permits an inter-protein electron transport from the [Fe₄S₄] cluster (of Av2 or Av2^V), via the P- or P^V-cluster, to the M- or V-cluster (of Av1 or Av1^V), where substrate reduction occurs (Figure 1).^1–10 Given the high degree of sequence, structural and compositional homology between the Mo- and V-nitrogenases, it is not surprising that they display highly similar catalytic profiles, both of which can activate/reduce small molecules like N₂, H⁺, C₂H₂ and CO.^1–4,11,12 There are, however, clear distinctions between the two nitrogenases in terms of their catalytic behaviors; most notably, the Mo-nitrogenase is more efficient than its V-counterpart in reducing N₂ to NH₃, whereas the V-nitrogenase is more efficient than its Mo-counterpart in reducing CO to hydrocarbons.¹³ Explorations of the similarities and dissimilarities between the homologous Mo- and V-nitrogenases could shed important light on the intricate reaction mechanism of this unique metalloenzyme.

Figure 1. The homologous nitrogenases of Azotobacter vinelandii.

Crystal structures of (A) the Mo-nitrogenase, which consists of a γ₂-dimeric Fe protein (also termed NifH or Av2; PDB entry 1n2c) and an α₂β₂-tetrameric MoFe protein (also termed NifDK or Av1; PDB entries 1n2c and 3u7q); and (B) the V-nitrogenase, which consists of a γ₂-dimeric Fe protein (also termed VnfH or Av2^V; PDB entry 6q93) and an α₂β₂δ₂-hexameric VFe protein (also termed VnfDGK or Av1^V; PDB entry 5n6y). The two components form a functional complex during catalysis, mediating the ATP-dependent electron transfer from the [Fe₄S₄] cluster (of Av2 or Av2^V), via the P-cluster (of Av1) or P^V -cluster (of Av1^V), to the M-cluster (of Av1) or V-cluster (of Av1^V), where substrate reduction occurs. The active-site cofactors of the Mo- and V-nitrogenases (i.e., the M- and V-clusters), each containing an organic homocitrate (hc) entity and coordinated by a pair of His and Cys ligands, are also shown in top- and side-views, highlighting an overall structural conservation that features three catalytically relevant belt-S atoms (i.e., S3A, S2B and S5A) and strictly conserved (or highly similar) residues in their vicinities (i.e., two conserved Gly residues near S3A, a conserved His residue near S2B, and an Arg or Lys residue at S5A). Cluster composition: P, [Fe₈S₇]; M, [(R-homocitrate)MoFe₇S₉C]); P^V, [Fe₄S₄]-like pair or [Fe₈S₇]; V, [(R-homocitrate)VFe₇S₉C] or [(R-homocitrate)VFe₇S₈(CO₃²⁻)]. Note that the S3A site of the V-cluster (denoted as S3A*) is either occupied by S²⁻ or CO₃²⁻ (see SI, Note S1). ADP*, MgADP•AlF₄⁻. The two components of the Mo- or V-nitrogenase are shown as ribbons, with the two subunits of Av2 or Av2^V colored light yellow and light gray, the α- and β-subunits of Av1 or Av1^V colored blue and red, and the δ-subunit of Av1^V colored green. The clusters and nucleotides are shown in ball-and-stick presentation, with the atoms colored as follows: Fe, orange; S, yellow; Mo, cyan; C, light gray; Mg, green; O, red; Al, dark gray; F, light blue; P, dark orange.

Efforts along this line have thus-far been focused on the ‘conventional’ Mo-nitrogenase, for which a wealth of knowledge has been acquired through biochemical, spectroscopic and structural investigations of both the wildtype and the variant forms of this enzyme.¹⁴ Of particular note, crystallographic analyses revealed binding of CO (an inhibitor and a substrate)^15,16 or N₂ (the physiological substrate)¹⁷ to the M-cluster of Av1 via displacement of one or more of the three μ₂-belt-sulfides (belt-S) that, along with a μ₆-interstitial carbide, bridge the [MoFe₃S₃] and [Fe₄S₃] subcubanes of the M-cluster (see Figure 1A). These unexpected findings point to a previously unsuspected, yet crucial role of belt-S mobilization in nitrogenase catalysis. Interestingly, binding of CO occurs symmetrically via displacement of one belt-S atom (S2B) in both M-clusters of Av1; whereas binding of N₂ occurs asymmetrically in the two M-clusters of an Av1 species (designated Av1*) obtained under N₂-turnover, but electron/sulfur-depleted (i.e., dithionite-free) conditions, with one belt-S atom (S2B) displaced by one N₂ moiety in one M-cluster and two belt-S atoms (S3A and S5A) displaced by two N₂ moieties in the other M-cluster (Figure S1A). In the latter case, there is also an alternate elongation of the two Mo-O bonds between Mo and its homocitrate ligand in the two M-clusters of Av1*, consistent with the dynamic nature of the Mo-homocitrate ‘end’ observed in the cryo-EM structure of a turnover complex of Mo-nitrogenase.⁶ Collectively, these structure-based assignments have led us to propose a mechanistic model that involves a stepwise reduction of N₂ at the three belt-S sites via rotation of the M-cluster (Figure S1B). Taking into consideration the absence (near S3A) or presence (near S2B and S5A) of proton donors at the belt-S locations, our model posits that a rotation of the M-cluster in the direction of S3A→S2B→S5A allows binding of N₂ at the S3A site via belt-S displacement, followed by a sequential hydrogenation of N₂ at the S2B and S5A sites, and release of NH₃ at the S5A site concomitant with belt-S replacement. Additionally, this model depicts an asynchronous rotation of the two M-clusters via an alternate docking of Av2 on the two αβ-dimers of Av1 (Figure S2), which drives the ATP-dependent rotation of the M-cluster in one αβ-dimer via elongation of one of the two Mo-O bonds while keeping the M-cluster in the other αβ-dimer in place to complete a certain reaction step. Such an asynchronous rotation of the two M-clusters would result in the same sequence of reaction steps—but one step apart from each other—in the two M-clusters of Av1.

Subsequent biochemical, spectroscopic and structural analyses not only validated the presence of N₂ species in Av1*, but also demonstrated the catalytic competence of Av1* to enable ATP-dependent, single turnover (i.e., without any externally supplied N₂) of its bound N₂ moieties into NH₃.¹⁸ More importantly, these studies pointed to the presence of multiple N₂ species in Av1*, which gave rise to EPR features with different temperature dependencies.¹⁸ The disappearance of these N₂-related EPR signals upon turnover of Av1* with ATP, along with the XAS/EXAFS observation of a dynamic in-and-out of belt-S/Se during catalysis, provided strong support for our proposed mechanistic model wherein the stepwise N₂ reduction is coupled with belt-S mobilization, with the events of substrate binding and product release facilitated by belt-S displacement and belt-S replacement, respectively (Figure S1B). In a recent study, we supplied further evidence for the presence of distinct N₂ species on Av1* by subjecting this protein to ATP-independent turnover with Eu^II-EGTA and SO₃^2-.¹⁹ The rationale for this strategy was that Av1* would be unable to load additional N₂ in the absence of ATP, thereby allowing us to focus solely on the existing N₂ moieties in Av1* and further distinguish these species based on their ability to undergo direct displacement with H₂ (i.e., if the N₂ species is not hydrogenated) or further reduction to NH₃ (i.e., if the N₂ species is hydrogenated). Excitingly, not only did we detect ATP-independent NH₃ production concomitant with N₂ release in the presence of H₂, but we also observed a ~1/3 reduction of NH₃ formation under H₂ compared to that under Ar.¹⁹ Both observations are significant: the former indicates a sufficient activation of the cofactor in Av1* such that it can undergo either the ‘forward reaction’ (i.e., reduction of N₂ to NH₃; see Figure S3) or the ‘reverse reaction’ (i.e., displacement of N₂ with D₂-derived D⁻ and incorporation of D⁻ into the product of C₂H₂- or H⁺-reduction; see Figure S3) without extra energy input from ATP hydrolysis; whereas the latter aligns well with our assignment of one non-hydrogenated N₂ moiety (at the S3A site) and two hydrogenated N₂ moieties (at the S2B and S5A sites) in Av1*. It is important to note the strict dependence of both the reduction and the displacement of N₂ on a suitable sulfur source alongside the electron source (supplied by dithionite or Eu^II-EGTA/SO₃²⁻), which once again highlights a critical role of belt-S mobilization in nitrogenase reactivity.

The recent developments in our mechanistic exploration of the ‘conventional’ Mo-nitrogenase have led to the question of whether our key observations, such as the dynamic turnover of belt-S in catalysis and the potential asynchrony of the reaction sequences in the two cofactors, are commonly featured in the catalytic mechanisms of the homologous members of this metalloenzyme family. The ‘alternative’ V-nitrogenase is an excellent candidate for this line of investigations. Despite the differences between the P/P^V-clusters and M/V-clusters in the Mo/V-nitrogenases (see Supplemental Note S1),^10,20,21 their active-site cofactors share the same architecture, with three belt-S atoms positioned near identical, or highly similar protein residues (see Figure 1). As such, a comparison of Av1^V prepared under N₂-fixing, but dithionite-free conditions (designated Av1^V*) with its Av1* counterpart prepared under the same conditions could provide strong validation for the results derived from the studies of Av1* while offering additional tools for probing the mechanistic details of nitrogenase.

Here, we report an analytical, biochemical and spectroscopic study of Av1^V*, the Av1* counterpart of the V-nitrogenase system. Our GC-MS, NMR and EPR analyses demonstrate an absence of N₂ from Av1^V* but illustrate a distinction of this species from the resting-state, belt-S replete Av1^V; whereas our activity, elemental and XAS/EXAFS analyses suggest a possible asymmetric displacement of three belt-S in Av1^V* that is analogous to that observed for Av1*, as well as the same, dynamic belt-S mobilization by Av1^V* as that observed for Av1* upon turnover, which potentially involves a cluster-assisted, in situ conversion of SO₃²⁻ to belt-S²⁻ as an important step in catalysis. Together, these observations point to belt-S turnover as a common catalytic feature of the homologous Mo- and V-nitrogenases, leading to our proposal that nitrogenase functions as an atypical sulfite reductase to enable belt-S mobilization during substrate turnover. Moreover, they suggest the potential utility of Av1^V* as an activated, but N₂-free intermediate in the future investigation of the mechanistic details of nitrogenase.

RESULTS AND DISCUSSION

GC-MS analysis of deuterium incorporation

GC–MS analysis was first used to examine whether Av1^V, like its Av1 counterpart, could incorporate D labels from D₂ into C₂H₄ or H₂, the product of C₂H₂- or H⁺-reduction, upon ATP-dependent turnover in the presence or absence of N₂. This experiment has been effectively used to illustrate a mandatory requirement of the coexistence of N₂ and D₂ for Av1 to catalyze the incorporation of D₂-derived D into C₂H₄ or H₂,^18,22–27 and to verify the binding of N₂ in an activated state in Av1* that permits this protein to carry out the same reaction in the sole presence of D₂.¹⁸ In both cases, the strict dependence of this reactivity on N₂, either supplied externally (in the case of Av1) or present in a bound state (in the case of Av1*), has been rationalized by a necessity to activate the cofactor upon N₂ binding, which then allows D₂ to displace N₂ and bind to the cofactor as D⁻, followed by incorporation of D into the product of C₂H₂- or H⁺-reduction (see Figure S3).

Consistent with earlier reports,^18,22–27 no C₂H₃D was generated by Av1 upon turnover under D₂/C₂H₂ (i.e., without N₂) in an assay containing Av2, ATP and dithionite, as indicated by the GC–MS fragmentation pattern (Figure 2A, ➀) and GC retention time (Figure 2B, ➀) of the un-deuterated product (i.e., C₂H₄) at m/z=28; yet, C₂H₃D was detected in the same assay upon turnover by Av1 under N₂/D₂/C₂H₂ (i.e., with N₂), as reflected by the GC–MS fragmentation pattern (Figure 2A, ➁) and GC retention time Figure 2B, ➁) of the deuterated product (i.e., C₂H₃D) at m/z=29. As observed for Av1, no C₂H₃D was generated by Av1^V upon turnover under D₂/C₂H₂ (i.e., without N₂) in an assay containing Av2^V, ATP and dithionite (Figure 2A,B, ➂). However, contrary to Av1, Av1^V could not generate C₂H₃D under N₂/D₂/C₂H₂ (i.e., with N₂; Figure 2A,B, ➃) in the same assay at the ambient pressure (total gas pressure: 1.1 atm); instead, it was only capable of forming C₂H₃D when the assay was conducted under ^OPN₂/D₂/C₂H₂ at over-pressurized (OP) conditions (total gas pressure: 2.5 atm) (Figure 2A,B, ➄).

Figure 2. GC-MS analyses of Av1 and Av1^V upon ATP-dependent turnover with dithionite.

Formation of C₂H₄ (A, B) and H₂ (C) upon turnover of Av1 (A-C, blue) or Av1^V (A-C, red) in the presence of Av2/ATP or Av2^V/ATP, was evaluated by GC-MS fragmentation patterns (A), GC elution profiles (B) and low-temperature GC analyses (C). Assays were run under D₂/C₂H₂ (➀,➂) and N₂/D₂/C₂H₂ (➁,➃,➄), either at atmospheric pressure (➀-➃) or under over-pressurized conditions (➄). The gas phases for ➀-➄ are indicated in the boxes on the left. See SI for details on the composition and total pressure of each gas phase, and a duplicated set of experiments (A) and (B) in Figure S4. Data in experiment C are expressed as mean ± s.d. (n=3). The GC-MS fragmentation pattern of the C₂H₄ standard is shown as grey bars (A). Formation of non-deuterated and deuterated products (i.e., C₂H₄ and C₂H₃D) with different masses in the same reaction is illustrated by overlaid GC traces collected on the same sample at different m/z ratios (B). The percentages of H₂ and HD formation are indicated (C), with the sum of both products set as 100%.

Incorporation of D₂-derived D was further probed by low-temperature GC analyses, which directly determined if Av1^V, like its Av1 counterpart, could produce HD upon ATP-dependent turnover in the presence or absence of N₂. In the case of Av1, no HD formation was detected upon turnover under D₂ alone in an assay containing Av2, ATP and dithionite (Figure 2C, ➀). In contrast, there was a substantial increase of the HD-to-H₂ ratio upon turnover of Av1 under N₂/D₂ in the same assay (Figure 2C, ➁). In the case of Av1^V, HD formation was not detected under D₂ alone upon turnover with Av2^V, ATP and dithionite (Figure 2C, ➂). Similarly, HD formation by Av1^V was hardly detectable when the same assay was conducted under N₂/D₂ at the ambient pressure (Figure 2C, ➃); yet, there was a substantial increase in the HD-to-H₂ ratio when Av1^V was subjected to turnover under ^OPN₂/D₂ at over-pressurized conditions (Figure 2C, ➄). Taken together, the results from the H⁺-reduction assays (Figure 2C) are in full agreement with those from the C₂H₂-reduction assays (Figure 2A,B), which collectively point to a lower affinity of the Av1^V-associated cofactor (V-cluster) for N₂ than its Av1-associated counterpart (M-cluster) that results in an inability of the V-cluster to interact with N₂ long/strong enough for the subsequent displacement of N₂ by D₂-derived D⁻ at the ambient pressure.

To further explore their differential capacities to interact with N₂, Av1 and Av1^V were isolated from the Azotobacter vinelandii strains expressing the Mo- and V-nitrogenases, respectively, under N₂-fixing conditions in the absence of dithionite, an artificial, dual electron- and sulfur-source. Such a procedure resulted in Av1* and Av1^V*, respectively, with the former previously established as an N₂-bound/activated species by structural, spectroscopic and biochemical studies.^17–19 Unlike the N₂-free Av1 (Figure 2A,B, ➀,➁), the N₂-bound Av1* could generate C₂H₃D under either D₂/C₂H₂ (i.e., without N₂) (Figure 3A,B, ➀) or N₂/D₂/C₂H₂ (i.e., with N₂) (Figure 3A,B, ➁) in an assay containing Av2, ATP and dithionite. Similarly, low-temperature GC analyses revealed a shift toward HD formation upon turnover of Av1* under both D₂/C₂H₂ (Figure 3C, ➀) and N₂/D₂/C₂H₂ (Figure 3C, ➁) in the presence of Av2, ATP and dithionite, although such a shift was much stronger in the latter case due to the turnover of additional, externally supplied N₂. When Av1^V* was subjected to turnover with Av2^V, ATP and dithionite, no C₂H₃D was detected under D₂/C₂H₂ (Figure 3A,B, ➂) or N₂/D₂/C₂H₂ (Figure 3A,B, ➃), nor was there a detectable level of HD formation under D₂ (Figure 3C, ➂) or N₂/D₂ at the ambient pressure (Figure 3C, ➃). At over-pressurized conditions, however, formation of C₂H₃D under ^OPN₂/D₂/C₂H₂ (Figure 3A,B, ➄), as well as a shift toward HD formation under ^OPN₂/D₂ (Figure 3C, ➄), was observed for Av1^V* in an assay containing Av2^V, ATP and dithionite, just like that observed for Av1^V in the corresponding assay containing Av2, ATP and dithionite (see Figure 2A–C, ➄).

Figure 3. GC-MS analyses of Av1* and Av1^V* upon ATP-dependent turnover with dithionite.

Formation of C₂H₄ (A, B) and H₂ (C) upon turnover of Av1* (A-C, blue) or Av1^V* (A-C, red) in the presence of Av2/ATP or Av2^V/ATP, was evaluated by GC-MS fragmentation patterns (A), GC elution profiles (B) and low-temperature GC analyses (C). Assays were run under D₂/C₂H₂ (➀,➂) and N₂/D₂/C₂H₂ (➁,➃,➄), either at the atmospheric pressure (➀-➃) or under over-pressurized conditions (➄). The gas phases for ➀-➄ are indicated in the boxes on the left. See SI for details on the composition and total pressure of each gas phase, and a duplicated set of experiments A and B in Figure S5. Data in experiment C are expressed as mean ± s.d. (n=3). The GC-MS fragmentation pattern of the C₂H₄ standard is shown as grey bars (A). Formation of non-deuterated and deuterated products (i.e., C₂H₄ and C₂H₃D) with different masses in the same reaction is illustrated by overlaid GC traces collected on the same sample at different m/z ratios (B). The percentages of H₂ and HD formation are indicated (C), with the sum of the two products set as 100%.

NMR analysis of ammonia formation

The disparate behaviors of Av1* and Av1^V* in the D incorporation experiments strongly suggest that, contrary to the capture of N₂ on the Av1*-associated M-cluster via omission of dithionite at the ambient pressure, N₂ cannot be captured on the Av1^V*-associated V-cluster under the same conditions to enable the displacement of N₂ by D₂-derived D and the subsequent incorporation of D into the product of C₂H₂- or H⁺-reduction. In support of this argument, frequency-selective pulse ¹H NMR analysis demonstrated the appearance of the ¹⁵NH₄⁺ specific, 1:1 doublet signal at ~7.00 and ~7.15 ppm upon single turnover of the Av1* species prepared with ¹⁵N₂ in an assay containing Av2, ATP and dithionite (Figure 4, ➀). In contrast, no ¹⁵NH₄⁺ was detected upon single turnover of the Av1^V* species prepared the same way with ¹⁵N₂ in an assay containing Av2^V, ATP and dithionite (Figure 4, ➁), thereby verifying the absence of any bound, activated N₂ species from Av1^V*.

Figure 4. NMR analyses of NH4+ formation by Av1* and Av1^V*.

Shown are the frequency-selective pulse ¹H NMR spectra of Av1* (➀) and Av1^V* (➁) upon ATP-dependent turnover with dithionite under Ar. Note the presence of ¹⁵NH₄⁺-specific doublet signal in the spectrum of Av1* (➀), which is absent from the spectrum of Av1^V* (➁).

EPR analysis of sulfur dependent turnover

Despite the absence of N₂ from both Av1^V and Av1^V*, the two species were clearly distinct in their EPR properties. Most notably, compared to their respective counterparts in Av1^V (Figure 5, black trace), the P^V-cluster associated, S=1/2 signal in Av1^V* displayed a ~20% decrease in magnitude concomitant with a ~44% increase in the intensity of the V-cluster associated, S=3/2 signal (Figure 5, red trace). The attenuation of the S=1/2 signal could reflect an oxidation of the P^V-cluster of Av1^V*, a scenario that parallels the oxidation of the P-cluster to the P²⁺ (or P^OX) state in Av1*.¹⁷ Since the preparation of both Av1^V* and Av1* involves the omission of dithionite (a dual electron- and sulfur-source), oxidation of the P^V- and P-clusters likely results from a lack of continuous electron supplies to re-reduce these clusters following the delivery of electrons from these clusters to the V- and M-clusters, respectively. In the case of Av1*, such an interruption of the continuous flow of electrons from the P-cluster to the M-cluster, along with a shortage of sulfur supply to refill the belt-S sites of the M-cluster, ‘stalls’ the reaction and facilitates the subsequent capture of N₂ at the cofactor via belt-sulfur displacement.^17–19

Figure 5. EPR analyses of various Av1^V species.

Shown are the perpendicular-mode EPR spectra of Av1^V in the resting state (Av1^V, black), Av1^V subjected to ATP-dependent turnover with dithionite (Av1^V (TOD), gray), Av1^V* as prepared (red), and Av1^V* subjected to ATP-dependent turnover with dithionite (Av1^V*(TOD), brown). The spectra were recorded at 10 K using a microwave power of 5 mW. The g values are given for the V-cluster-specific S=3/2 signals and the P^V-cluster-specific S=1/2 signals. The S=3/2 signals are enlarged by 5-fold and shown as insets above the corresponding spectra. The relative intensities of the S=3/2 and S=1/2 signals are indicated, with the respective signal intensities of Av1^V set at 100%.

In the case of Av1^V*, however, the absence of N₂ binding renders the behavior of its V-cluster associated EPR feature different than that of its M-cluster associated counterpart in the N₂-bound Av1*. Specifically, the intensity of the S=3/2 signal in Av1* is reduced concomitant with the appearance of several previously unobserved N₂-related EPR features in the S=1/2 region;¹⁸ whereas the intensity of the S=3/2 signal in Av1^V* is increased without the concurrent appearance of additional features in any region of the EPR spectrum (Figure 5, black trace vs. red trace). Despite the opposite changes in the intensities of their cofactor-associated S=3/2 signals, the fact that both Av1* and Av1^V* display spectral changes of the cofactors relative to those of their respective resting-state counterparts points to a further ‘processing’, or a change of the redox properties of the cofactors in Av1* and Av1^V* upon turnover under N₂ in the absence of dithionite. Given the similarity between the cofactor species of the homologous V- and Mo-nitrogenases, it is plausible that binding of N₂ to the V- or M-cluster under turnover conditions occurs through a similar, belt-sulfur displaced mechanism that results in a change of the oxidation state of the cofactor. Consequently, Av1^V* and Av1* should adopt a similar belt-S displaced conformation following electron- and sulfur-depleted turnover under N₂; only in the case of Av1^V*, an immediate loss of N₂ upon its transient interaction with the V-cluster leaves ‘holes’ at the belt-sulfur locations, which cannot be refilled by sulfur (e.g., derived from dithionite), but are instead occupied/stabilized by other ligands (e.g., H₂O). In this scenario, turnover of Av1^V* in dithionite should restore the spectral feature of this protein species back to that of its belt-S replete Av1 counterpart.

Consistent with this argument, turnover of Av1^V* with Av2^V, ATP and dithionite resulted in a protein species (designated Av1^V*(TOD); TOD, turnover with dithionite) that closely resembled the species generated upon turnover of Av1^V under the same conditions (designated Av1^V(TOD)). Specifically, the P^V-cluster specific, S=1/2 signal of Av1^V*(TOD) (Figure 5, brown trace), like that of Av1^V(TOD) (Figure 5, gray trace), was comparable with its counterpart in Av1^V (Figure 5, black trace) in intensity. Likewise, the V-cluster specific, S=3/2 signal of Av1^V*(TOD) (Figure 5, brown trace), like that of Av1^V(TOD) (Figure 5, gray trace), showed characteristic, disproportional decreases in the intensities of its two components (at g=5.5 and g=4.4, 3.8, respectively; see Supplemetal Note S2) relative to the two corresponding components of the S=3/2 signal of Av1^V (Figure 5, black trace).²⁸ Such a restoration of the EPR features of Av1^V* back to those of Av1^V upon turnover with dithionite parallels the conversion of the belt-sulfur displaced Av1* back to the belt-sulfur replete Av1 upon turnover with dithionite.¹⁸ More importantly, it implies that the homologous V-and Mo-nitrogenases may utilize a similar belt-sulfur mobilization mechanism during catalysis despite the differential affinities of their respective cofactors for N₂.

Enzymatic, analytical and XAS/EXAFS analyses of belt-sulfur turnover

The crucial involvement of sulfur mobilization in catalysis by V-nitrogenase was further verified by replacing dithionite with a sulfur-free electron source (i.e., Eu^II-EGTA), alone or in combination with a specific sulfur source (i.e., S²⁻, SO₃²⁻ or SO₄²⁻), in the Av2^V/ATP-dependent activity assay of Av1^V*. As observed for Av1*,¹⁸ Av1^V* was unable to catalyze substrate reduction in the absence of a sulfur source or in the presence of S²⁻ or SO₄²⁻ (Figure 6); yet, it was capable of reducing H⁺, N₂ and C₂H₂ in the Av2^V/ATP-dependent reactions in the presence of SO₃²⁻ (Figure 6). The same catalytic dependence of Av1^V* and Av1* on SO₃^2- once again points to sulfur mobilization as a common mechanistic feature of the homologous V- and Mo-nitrogenases. Moreover, it suggests the possibility of using SeO₃²⁻ (an analog of SO₃²⁻) as an effective marker to trace the fate of belt-S upon turnover of Av1^V* as previously demonstrated for Av1*.¹⁸

Figure 6. Dependence of substrate turnover by Av1^V* on a suitable sulfur source.

Av1^V* was subjected to ATP-dependent turnover in an in vitro activity assay containing Av2^V, ATP and Eu^II-EGTA in the absence or presence of one of the following sulfur sources: S²⁻, SO₄²⁻ or SO₃²⁻. The specific activities of Av1^V* to produce H₂ (under Ar), H₂ and NH₃ (under N₂), and C₂H₄ and C₂H₆ (under C₂H₂) in these in vitro assays were determined. Data are expressed as mean ± s.d. (n=3). The symbol * denotes assays performed under reductant-free, non-turnover conditions.

The belt-S turnover experiment was first conducted by subjecting Av1^V* to ATP-dependent turnover with Eu^II/SeO₃²⁻ and comparing with Av1* that underwent the same treatment. As described earlier,¹⁸ re-purification of Av1* upon turnover with Av2, ATP, Eu^II-EGTA and excess SeO₃²⁻, followed by elemental determination of the cofactor extracted from the resulting protein species (designated Av1*(TOSe)^+ATP; TOSe, turnover with SeO₃²⁻) by ICP-MS analysis, revealed a Se/Mo ratio of 2.70±0.17 (Figure 7A, ➂). This ratio closely resembles the belt-S/Mo ratio of the belt-S replete, resting-state Av1 (with its two M-clusters containing two Mo and six belt-S) and aligns well with the crystallographically confirmed ‘return’ of belt-S upon turnover of Av1* with Av2, ATP, Eu^II-EGTA and SO₃²⁻.¹⁸ Similarly, ICP-MS analysis of the cofactor extracted from the re-purified Av1^V* protein species (designated Av1^V*(TOSe)^+ATP) upon turnover with Av2^V, ATP, Eu^II-EGTA and excess SeO₃²⁻ revealed a Se/V ratio of 2.33±0.15 (Figure 7A, ➆), a number comparable with that of the resting-state Av1^V (with its two V-clusters containing two V and four¹⁰ or six²¹ belt-S). This observation suggests that, like its Av1* counterpart, Av1^V* undergoes the same ‘refilling’ event of all belt-S locations with Se upon ATP-dependent turnover with Eu^II/SeO₃²⁻.

Figure 7. Tracing belt-S turnover by Av1* and Av1^V* with Se labels.upon turnover.

(A) ICP–MS determination of Se/Mo or Se/V ratios of M- or V-clusters extracted from various Av1* (blue) or Av1^V* (red) species upon turnover. Av1*(-TO,Se) (➀) or Av1^V*(-TO,Se) (➄) was prepared by incubating Av1* or Av1^V* with SeO₃²⁻ under reductant- and ATP-free, non-turnover conditions; Av1*(TOSe)^−ATP (➁) or Av1^V*(TOSe)^−ATP (➅) was prepared by subjecting Av1* or Av1^V* to ATP-independent turnover with Eu^II-EGTA and SeO₃²⁻; Av1*(TOSe)^+ATP (➂) or Av1^V*(TOSe)^+ATP (➆) was prepared by subjecting Av1* or Av1^V* to ATP-dependent turnover with Av2 or Av2^V, ATP, Eu^II-EGTA and SeO₃²⁻; Av1*(TOSe)^+ATP,S (➃) or Av1^V*(TOSe)^+ATP,S (➇) was prepared by subjecting Av1*(TOSe)^+ATP or Av1^V*(TOSe)^+ATP to ATP-dependent turnover with Av2 or Av2V, ATP, Eu^II-EGTA and SO₃²⁻. Data are expressed as mean ± s.d. (n= 3). (B-D) Se K-edge XAS analyses of Av1*(TOSe)^+ATP (blue) or Av1^V*(TOSe)^+ATP (red). Shown are the k³-weighted EXAFS data (dotted) and the best fits of data (solid) (B), the Fourier transforms (FT) of the EXAFS data (dotted) and the best fits of data (solid) (C), and spectra of the rising edges (blue and red) compared with the Na₂SeO₃ standard (green) (D). See Tables S1 and S2 for detailed fits of EXAFS data.

The Se-incorporated Av1*(TOSe)^+ATP and Av1^V*(TOSe)^+ATP samples were then subjected to ATP-dependent turnover with Eu^II/SO₃²⁻ to determine whether the belt-Se in these proteins could be ‘chased off’ by SO₃²⁻-derived belt-S. Indeed, re-isolation of Av1*(TOSe)^+ATP upon Av2/ATP-dependent turnover with Eu^II-EGTA and excess SO₃²⁻, followed by ICP-MS analysis of the cofactor extracted from this protein species (designated Av1*(TOSe)^+ATP,S), revealed a decreased Se/Mo ratio of 0.51±0.17 (Figure 7A, ➃). Likewise, ICP-MS analysis of the cofactor extracted from the re-isolated protein species (designated Av1^V*(TOSe)^+ATP,S) after subjecting Av1^V*(TOSe)^+ATP to Av2^V/ATP-dependent turnover with Eu^II-EGTA and excess SO₃²⁻ revealed a decreased Se/V ratio of 0.63±0.18 (Figure 7A, ➇). Such a reversible ‘knock-in’ of belt-Se upon ATP-dependent turnover with Eu^II/SeO₃²⁻ and ‘knock-off’ of belt-Se upon ATP-dependent turnover with Eu^II/SO₃²⁻ highlights the dynamic in-and-out of belt-S during nitrogenase catalysis, and the close resemblance of the Se/V ratios of the Av1^V*-derived species to the Se/Mo ratios of the corresponding Av1*-derived species during these reversible processes strongly points to belt-S turnover as a crucial catalytic event shared by the homologous V- and Mo-nitrogenases.

Interestingly, when Av1* was subjected to ATP-free/independent turnover in the sole presence of Eu^II-EGTA and excess SeO₃^2-, the Se/Mo ratio of the cofactor extracted from the re-purified Av1* species after such a treatment (designated Av1*(TOSe)^−ATP) was 1.46±0.18 (Figure 7A, ➁), approximating the belt-S/Mo ratio observed for the crystal structure of the N₂-bound Av1* species (with its two M-clusters containing two Mo and three belt-S). This observation could be explained by a ‘refill’ of only three activated, N₂-bound belt sites with SeO₃²⁻-derived Se, an event not accessible by the other three un-activated belt-S sites under Av2/ATP-independent conditions. Strikingly, turnover of Av1^V* in an ATP-independent assay containing only Eu^II-EGTA and excess SeO₃²⁻ resulted in a Se/V ratio of 1.29±0.16 (Figure 7A, ➅), consistent with a refill of approximately three activated, belt-S displaced sites of Av1^V* in a scenario analogous to that observed in the N₂-bound Av1* (see Figures 2–4).¹⁸ This result implies that, like its Av1* counterpart, Av1^V* contains three belt-S-displaced sites that have been sufficiently activated via transient interactions with N₂ to enable the ATP-independent incorporation of SeO₃²⁻ as belt-Se²⁻. Moreover, the comparable Se/V (~1.3) and Se/Mo (~1.5) ratios upon ATP-independent Se ‘knock-in’ suggest that Av1^V* could assume an asymmetric conformation analogous to that previously assigned to Av1* (Figure S1A),¹⁷ with one belt-S atom sites displaced at the S2B site in one cofactor and two belt-S atoms displaced at the S3A and S5A sites in the other cofactor.

The turnover-dependent incorporation of Se was further examined by Se K-edge analyses of Av1^V*(TOSe)^+ATP and Av1*(TOSe)^+ATP. K-space and Fourier transforms (FT) of the EXAFS data of Av1*(TOSe)^+ATP (Figure 7B,C, blue) and Av1^V*(TOSe)^+ATP (Figure 7B,C, red) are similar to each other, both of which contain a prominent feature at R+Δ ~2 Å in the FT. As reported previously,¹⁸ this major FT feature at R+Δ ~2 Å correlates to the characteristic Se–Fe distances at ~2.4 Å (Tables S1 and S2), which strongly suggests the incorporation of SeO₃²⁻-derived Se²⁻-type species into the cofactor of Av1*(TOSe)^+ATP or Av1^V*(TOSe)^+ATP. Additionally, there are minor contributions from Se–O scatterers, with Se–O distances observed at ~2.0 Å for both Av1*(TOSe)^+ATP and Av1^V*(TOSe)^+ATP (Tables S1 and S2). Including these Se–O scattering paths enhances the fit of the data, indicating the presence of SeO₃²⁻-derived SeO species in the cofactors of Av1*(TOSe) and Av1^V*(TOSe). The coexistence of fully reduced (Se²⁻) and partially reduced (SeO) Se species in both Av1*(TOSe)^+ATP and Av1^V*(TOSe)^+ATP suggests a more oxidized state of Se compared to its sulfur counterpart (18). Additionally, as illustrated by the differences in the normalized fluorescence around their rising edges, there are some differences in the responses of the two protein species towards Se incorporation, likely due to the differential redox properties of the cofactors within these proteins (Figure 7D, blue vs. red). Regardless, the similarity between the Se K-edge data of Av1*(TOSe)^+ATP and Av1^V*(TOSe)^+ATP, along with the similar behaviors of the Av1* and Av1^V* derived species in the Se ‘knock-in’ and ‘knock-off’ experiments (see Figure 7A), points to a common mechanism shared by the homologous Mo- and V-nitrogenases for belt-S mobilization during catalysis.

Proposal of belt-sulfur mobilization as a common mechanistic feature of nitrogenase

Such a mechanism has been proposed for the Mo-nitrogenase, wherein substrate binding and product release are coupled with belt-S displacement and replacement, respectively (Figure S1B). The energetic feasibility of a cluster-facilitated, in situ pathway that reduces SO₃²⁻ to a belt-S²⁻ (i.e., belt-S replacement), a reaction that could very well facilitate the release of the reaction product (i.e., NH₃), has received strong support from previous DFT calculations of the M-cluster in Av1*.¹⁸ Conversely, oxidation of belt-S²⁻ to SO₃²⁻ (i.e., belt-S displacement), an appealing scenario given the same amount of electrons (i.e., six electrons) involved in the oxidation of S²⁻ (to SO₃²⁻) as that involved in the reduction of N₂ (to NH₃), has also been hypothesized to enable the binding and activation of the substrate (i.e., N₂).¹⁸ While this mechanistic proposal requires experimental verification, the observation of a strict dependence of substrate turnover on sulfur mobilization, along with a similar pattern of sulfur mobilization by the homologous Mo- and V-nitrogenases, points to the dynamic turnover of belt-sulfurs as an integral part of the nitrogenase reaction. It is interesting to note that the ability of nitrogenase to catalyze the interconversion between sulfite and sulfide (Figure 8A) mirrors the reaction catalyzed by the classical sulfite reductase (Figure 8B). However, contrary to the typical members from this enzyme family, nitrogenase uses a suite of metallocenters (instead of an FeS cluster and siroheme, a tetrapyrrole prosthetic group) to facilitate electron transfer for the plausibly reversible, six-electron interconversion between sulfite and a cofactor-bound sulfide (instead of donating sulfide for other cellular functions). In this context, nitrogenase could be considered as an atypical sulfite reductase that utilizes its cofactor both as a reaction site and a substrate (or intermediate) to enable the coupling between sulfur metabolism and nitrogen metabolism, a feat with possible implications for the evolution of life on Earth.

Figure 8. Proposal of nitrogenase as an atypical sulfite reductase.

Comparison of (A) nitrogenase with (B) sulfite reductase in sulfur mobilization. In its proposed role as a sulfite reductase, nitrogenase parallels the classical members of this enzyme family by enabling the reversible, six-electron redox conversion between sulfite (SO₃²⁻) and sulfide (S²⁻), although nitrogenase differs from the typical sulfite reductases in using a series of metalloclusters (see A) instead of a [Fe₄S₄] cluster and siroheme (see B) for electron transfer. More importantly, contrary to sulfite reductases that play a key role in sulfur assimilation (B), nitrogenase could plausibly couple the oxidation of a belt-S²⁻ to SO₃²⁻ with the six-electron reduction of N₂ to NH₃ and, conversely, enable the reduction of SO₃²⁻ to a belt-S²⁻ concomitant with the release of NH₃ (A). The dashed green arrow (in A) indicates supplementation of SO₃²⁻ to the sulfur that is frequently “lost” during the cycling between SO₃²⁻ and S²⁻, which forms the basis of the belt-S turnover experiments in Figure 7A wherein SeO₃²⁻ or SO₃²⁻ is supplemented. Fdx, ferredoxin; GSH, glutathione.

Despite sharing a highly similar, dynamic belt-S mobilization that possibly occurs asynchronously in their two cofactors during catalysis (as indicated by the asymmetric conformations of Av1^V* and Av1*; see Figures 9A and S1A), the disparate behaviors of V- and M-nitrogenases in D incorporation (Figures 9B and S3)—an observation made possible by the differential affinities of N₂ for the cofactors in these homologous nitrogenases—seems to indicate a clear distinction between the ‘forward reaction’ and the ‘reverse reaction that goes well beyond a simple, reversible exchange between N₂ and H₂ at the cofactor. In particular, Av1^V is capable of reducing N₂ to NH₃ concomitant with H₂ evolution at the ambient pressure (forward reaction), yet it does not allow for the incorporation of D₂-derived D labels into the products of C₂H₂- or H⁺-reduction under the same reaction conditions (reverse reaction) (see Figure 3A–C, ➃). This observation casts doubt on whether the reverse reaction, which likely involves binding of H₂/D₂-derived H⁻/D⁻ to the cofactor for its subsequent incorporation into C₂H₄ or H₂, can be used to infer an analogous formation of H⁻/D⁻ that precedes N₂-binding in the forward reaction.²⁷ Indeed, the fact that D can only be incorporated into C₂H₄ or H₂ at over-pressurized conditions in the reverse reaction (see Figure 3A–C, ➄), where N₂ is forced to interact longer or more strongly with the cofactor to enable binding of D₂-derived D⁻ via displacement of N₂, could be interpreted by a pre-requisite of N₂-binding for, or concurrent occurrence of N₂-binding with, the formation of metal hydride(s)—be it originated from H₂ (in the reverse reaction) or H⁺ (in the forward reaction). This plausible explanation, along with the observation of a shift towards H₂ evolution in the reaction of N₂-reduction catalyzed by the V-nitrogenase (showing an NH₃:H₂ ratio of 0.9:1) relative to that by its Mo-counterpart (showing an NH₃:H₂ ratio of 2:1), brings into question whether N₂ binding can only occur via displacement of the cofactor-bound hydride(s) to facilitate an obligate evolution of H₂.²⁹ Regardless, the formation of metal hydride(s) as a part of the mechanism of N₂ reduction by nitrogenase—an appealing scenario that has been suggested for the reduction and coupling of CO (an isoelectronic analog of N₂) by the same enzyme³⁰—is worthy of further exploration. The activated, but N₂-free conformation captured in Av1^V* could provide an effective tool for studies along this line, addressing questions such as whether hydride(s) can be captured at the activated, belt-displaced sites in the absence of N₂ and, if so, what is the fate of these metal hydride(s) along the reaction pathway of nitrogenase. While challenging, a systematic study that targets the correlation between the formation of the two reaction products (NH₃, H₂) and how these events tie in with the recently discovered catalytic element of belt-S mobilization could contribute to a better mechanistic understanding of this unique metalloenzyme.

Figure 9. Proposed conformation and potential utility of Av1^V*.

(A) Proposed conformation of Av1^V*, which has its P^V- and V-clusters present in different oxidation states compared to those in the resting-state Av1^V (see Fig. 5), as well as the same asymmetric belt-S displaced conformations as Av1* with respect to its two cofactors (see Fig. 7). Like their counterparts in Av1*, the belt-S displaced sites in Av1^V* are activated and can undergo ATP-independent belt-S turnover (see Fig. 7). However, unlike Av1* that has N₂ species bound at the belt-S displaced sites (see Fig. S1), Av1^V* is free of any bound N₂ species (see Figs 2–4). The metalloclusters are shown in stick presentation, with the atoms colored as follows: Fe, orange; S, yellow; Mo, cyan. (B) Potential utility of Av1^V* in the mechanistic investigation of nitrogenase. Unlike its N₂-bound Av1* counterpart, Av1^V* is activated, but free of any N₂ species. Moreover, contrary to Av1*, while Av1^V* could enable the ‘forward reaction’ at the ambient pressure, it is only able to support the ‘reverse reaction’ under over-pressurized conditions. Given the ability of Av1^V* to “uncouple” the various catalytic events, this activated, yet N₂-free species could potentially be used to individually capture, and collectively address the correlation of three key events during catalysis: (1) H⁻ formation; (2) N₂ binding; and (3) belt-S turnover.

Mechanistic considerations of belt-S mobilization

Until the discovery of ligand/substrate-binding via belt-S displacement a few years ago,^15–17 the mobilization of belt-S had not been recognized as a crucial, integral feature of the nitrogenase mechanism. Consequently, aligning our proposed nitrogenase mechanism model (designated Hu-Ribbe or HR model)^17–19 with previously proposed models^2,29 presents a serious challenge. However, recent DFT calculations by Per Siegbahn and coworker have, for the first time, considered belt-S loss as a mechanistic component of the Mo-nitrogenase,^31–33 leading to a proposed mechanism (designated Per Siegbahn or PS model) in which sulfide loss occurs following four pre-activation reduction steps A₀–A₄ (Figure S6A, upper).³² Notably, the (pre)activation process of the PS model (designated PS activation process) results in the loss of the belt-S (bridged by Fe4/Fe5) at the S3A site of the cofactor, followed by an end-on binding of N₂ to Fe4.^31,32 Similarly, the activation process of our model (designated HR activation process), which presumably occurs through the loss of the belt-S²⁻ at the S3A site as SO₃²⁻ (Figure S6A, lower), would allow for an entry of N₂ at the same S3A site, with N₂ displacing S3A and forming an asymmetric μ_1,1 bridge between Fe4 (the primary or proximal Fe center) and Fe7 (the secondary or distal Fe center).^17–19 Thus, both activation processes share the common feature of belt-S removal at S3A, a previously not considered element for the binding and activation of N₂. In the PS model, the binding of N₂ to Fe4 at S3A represents the ‘A₄ state’³³ that occurs prior to the hydrogenation of N₂ (Figure S6B). In agreement with this proposal, our model calls for a rotation of the activated, yet non-hydrogenated N₂ from the S3A site to the S2B site, where the initial hydrogenation event occurs (Figure S6B).^17–19 Together, these events would give rise to the ‘E₄ state’ of the classic Thorneley-Lowe (TL) model (Figure S6B), at which stage the binding and initial hydrogenation of N₂ take place.² A similar (pre)activation process has also been proposed for the V-nitrogenase by Siegbahn and coworker.³⁴ Strikingly, the V-nitrogenase requires an additional activation step than its Mo-counterpart, consistent with our experimental observation of an increased difficulty to retain N₂ on Av1^V* than Av1* under the same, electron/sulfur-depleted conditions.

There are many questions that must be addressed prior to merging the various mechanistic models of nitrogenase, such as how belt-S exchange is coupled with substrate binding and product release, or if and how hydride formation is related to the turnover of N₂. Additionally, how cofactor rotation—as suggested by the catalysis-dependent migration of a S2B-specific Se label (generated with selenocyanate, an inhibitor and a weak substrate)³⁵ and the dynamic, turnover-driven in-and-out of the Se label (generated with an alternate use of the recently identified substrates, selenite and sulfite)^18,19—enables substrate reduction needs further exploration, along with the cooperativity between the two functional halves of the catalytic component that facilitates the asynchronous rotation of the two cofactors. Nevertheless, the recent experimental and theoretical studies^{17–19,31–33} effectively demonstrate that the existing models are not mutually exclusive; instead, they offer great opportunities to explore and incorporate previously unrecognized elements as the missing pieces for completing the mechanistic puzzle of nitrogenase. A (re)calibration of mechanistic thinking and existing data is required to solve this important riddle in the field of metalloprotein biochemistry and bioinorganic chemistry.

In conclusion, we demonstrate that the V-nitrogenase (Av1^V*) is unable to retain N₂ at the ambient pressure like its Mo-nitrogenase counterpart (Av1*) under N₂-fixing, but electron- and sulfur-depleted conditions; however, like Av1*, Av1^V* potentially contains three activated belt sites in its two cofactors that can be refilled with Se labels upon ATP-independent turnover with SeO₃²⁻. Moreover, both Av1^V* and Av1* can undergo Se incorporation at all belt-sites upon ATP-dependent turnover with SeO₃²⁻, possibly via a cluster-assisted, in situ reduction of SeO₃²⁻ to belt-Se²⁻ as predicted previously by DFT calculations.¹⁸ These observations collectively point to the same mechanism employed by the V- and Mo-nitrogenases for catalysis, which involves a plausible participation of the three belt-S sites via an asynchronous belt-S mobilization in the two cofactors of these homologous nitrogenases. Such a parallelism in the reaction mechanisms of the V- and Mo-nitrogenases not only provides strong validation for our prior interpretation of the biochemical, structural and spectroscopic data derived from the Mo-nitrogenase system, but also establishes this homologous group of nitrogenases as a previously unidentified type of sulfite reductases that couple the turnover of belt-S with that of N₂. Taking advantage of the activated, yet N₂-free reaction intermediate (i.e., Av1^V*) identified herein, the future studies will focus on addressing if and how metal hydride(s) can be accumulated on this intermediate and studied without the interference of N₂. Efforts along this line, coupled with further explorations of belt-S mobilization during catalysis, could prove instrumental in elucidating the correlation between three key events of the nitrogenase reaction (i.e., NH₃ formation, H₂ evolution, and belt-S turnover), thereby shedding crucial lights on the catalytic mechanism of nitrogenase.

METHODS

Detailed experimental procedures are provided in the Supplemental Information.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Prof. Yilin Hu (yilinh@uci.edu).

Materials availability

The materials generated in this study are available upon reasonable request to the Lead Contact.

Data and code availability

All data supporting this study are available in the manuscript and supplemental information.

THE BIGGER PICTURE.

Nitrogenase is a versatile metalloenzyme that is capable of activating a wide range of small molecules, such as N₂ and CO, at ambient conditions. The significance of this enzyme for areas related to agronomy, energy and environment makes it a key target of investigations that aim at elucidating the mechanistic details of this complex metalloenzyme. Previous studies have demonstrated that the belt region of the nitrogenase cofactor, which consists of three subcluster-bridging μ₂-sulfides (designated belt-S), is highly labile during catalysis. In this work, we report the common mechanistic feature of the V- and Mo-nitrogenases involving belt-S turnover that occurs dynamically and asymmetrically in the two cofactors of their respective catalytic components during catalysis. Our results point to a potential role of nitrogenase as an atypical sulfite reductase, thereby establishing a possible connection between the sulfur and nitrogen metabolism through this ancient metalloenzyme in the context of evolution on Earth. Moreover, our success in capturing an activated, belt-S-displaced yet N₂-free intermediate of V-nitrogenase offers a unique opportunity to uncouple the key events of N₂ reduction (N₂ binding, H⁻ formation, and belt-S turnover) and tackle their correlation during catalysis, which could greatly facilitate the elucidation of the intricate mechanism of nitrogenase and, in the long run, enable the development of nitrogenase-inspired catalysts for the cost-efficient and environmentally-friendly production of valuable chemical commodities.

Highlights.

• Belt-S mobilization is a common mechanistic feature of V- and Mo-nitrogenases

• The in-and-out of belt-S is likely coupled to substrate binding and product release

• Distinctions of belt-S displaced V- and Mo-nitrogenases shed light on mechanism

• Nitrogenase as a sulfite reductase that links evolution of N and S cycles on Earth