Belt-sulfur mobilization in nitrogenase biosynthesis and catalysis

Nitrogenase catalyzes the ambient conversion of N₂ and CO to valuable chemical commodities. Recent advances reveal the flexibility of the ‘belt’ region of the nitrogenase cofactor and illustrate the mechanistic importance of belt-sulfur mobilization for the biosynthesis and catalysis of this complex metalloenzyme.

Nitrogenase catalyzes some of the most extraordinary chemical transformations that are important for areas related to agronomy, energy, and environment [1]. Of particular note is the capability of nitrogenase to reduce N₂ to ammonia, and CO to hydrocarbons, respectively, in reactions mirroring the industrial Haber-Bosch and Fischer-Tropsch processes; however, contrary to their industrial counterparts, the reactions catalyzed by nitrogenase occur at ambient temperatures and pressures in aqueous solutions [2], highlighting the catalytic prowess of this unique metalloenzyme that could potentially be harnessed for the future development of homogeneous catalysts.

The ‘conventional’ Mo-nitrogenase functions as a binary system, with a reductase component (NifH) delivering electrons to its catalytic partner (NifDK) for substrate reduction (Figure 1A) [3,4]. Central to the reactivity of the Mo-nitrogenase is the active-site cofactor within its NifDK component, which is a complex metallocluster that has thus far evaded successful chemical synthesis. Designated the M-cluster (or FeMoco), this metallocofactor contains three μ₂ ‘belt’ sulfides and one μ₆ interstitial carbide that are bridged between a [MoFe₃S₃] and a [Fe₄S₃] subcubane (Figure 1B). The central cavity of the M-cluster had long been considered a void until electron density consistent with a light atom X (X=N,C,O) was observed and assigned as a μ₆-carbide (C⁴⁻) by spectroscopic and crystallographic methods [5,6]. Subsequent biochemical and spectroscopic studies traced the origin of carbide to S-adenosyl-L-methionine (SAM); further, they led to the mechanistic proposal wherein carbide insertion occurs concurrently with the synthesis of an 8Fe cofactor core via radical SAM chemistry [7]. This mechanism invokes transfer of the methyl group from one equivalent of SAM to a [Fe₄S₄] cluster pair (K-cluster) on NifB (a radical SAM enzyme), followed by abstraction of a hydrogen atom from the K-cluster-bound methyl group by a 5’-deoxyadenosyl radical that is derived from the homolytic cleavage of a second equivalent of SAM (Figure 1C). The resulting, cluster-bound carbon radical then initiates fusion of the two [Fe₄S₄] modules of the K-cluster into an 8Fe precursor, a structural homolog of the mature cofactor, while being further processing into a μ₆ carbide ion in the center of the 8Fe precursor (Figure 1C).

Figure 1. Biosynthesis of the nitrogenase cofactor.

(A) Crystal structure of the MgADP•AlF₄⁻−stabilized ‘transition-state’ complex of Azotobacter vinelandii Mo-nitrogenase containing NifH (γ2) and NifDK (α₂β₂) at a molar ratio of 2:1 (PDB entry 1N2C), and the electron pathway within the NifH/NifDK complex that permits electrons to flow from the [Fe₄S₄] cluster of NifH, through the P-cluster ([Fe₈S₇]), to the M-cluster ([(R-homocitrate)MoFe₇S₉C]) of NifDK to enable substrate reduction. The α-, β-, and γ-subunits of the complex are shown as transparent ribbons and colored red, blue, and grey, respectively. MgADP•AlF₄⁻ and metal centers are shown in ball-and-stick presentation, with the atoms colored as follows: Fe, orange; Mo, cyan; Mg, green; Al, dark gray; S, yellow; O, red; C, light gray; N, blue; F, light blue; P, dark orange. (B) Side and top views of the M-cluster (PDB entry 3U7Q) along with protein ligands and homocitrate (hc). The belt-sulfurs S2B, S3A, and S5A are indicated. (C) Biosynthesis of the M-cluster. A [Fe₄S₄] cluster pair (K-cluster) is transformed into an [Fe₈S₈C] cluster (L*-cluster) on NifB concomitant with insertion of an interstitial carbide via radical SAM chemistry, followed by insertion of a sulfite (SO₃²⁻)-derived ‘9^th sulfur’ in the ‘belt’ region of the cluster to yield an [Fe₈S₉C] cofactor core (L-cluster). The L-cluster is then delivered to NifEN and matured into an M-cluster upon insertion of Mo and homocitrate (hc) by NifH, followed by transfer of the M-cluster to its target location in NifDK. (D) Coordination and reduction of SO₃²⁻ at the belt-sulfur-free binding site of an under-coordinated L*-cluster. Shown are the reaction energies of two pathways calculated by DFT, which involve coordination of either two O atoms or one O atom and one S atom of SO₃²⁻ by a pair of Fe atoms across the ‘belt’ of the L*-cluster. The energies (kcal/mol) are indicated. The atoms in B-D are colored as described in A.

Interestingly, the radical SAM-based conversion of the K-cluster (2 x [Fe₄S₄]) into a carbide-containing 8Fe structure (designated L*-cluster, [Fe₈S₈C]) precedes the insertion of a ‘9^th sulfur’ in the belt region, which completes the stoichiometry of the cofactor core (designated L-cluster, [Fe₈S₉C]) [8]. The L*- and L-clusters show nearly identical EPR and XAS features that are characteristic of an 8Fe cofactor core; however, only the L-cluster can be transferred from NifB to the next biosynthetic scaffold, NifEN, where it is matured into an M-cluster and subsequently delivered to the catalytic component of Mo-nitrogenase, NifDK [8]. Conversion of the L*-cluster (8Fe,8S,C) to an L-cluster (8Fe,9S,C) occurs upon treatment with dithionite, a sulfur-containing reductant; yet, dithionite can be replaced by Eu(II)-EGTA (a sulfur-free reductant) when it is combined with sulfite (SO₃²⁻; a known decomposition product of dithionite) [8]. Given that SO₃²⁻ represents a major ‘hub’ of sulfur metabolism, this observation points to SO₃²⁻ as a potential in vivo source of the ‘9^th sulfur’. The incorporation of SO₃²⁻ into the L-cluster can be further traced with ³⁴SO₃²⁻ or SeO₃²⁻; moreover, DFT calculations suggest an energetically feasible route involving the ‘in situ’ reduction and insertion of SO₃²⁻ as a belt sulfide (S²⁻) (Figure 1D) [8].

The observation of the lability of the belt-sulfur—likely accommodated by the presence of an interstitial carbide that maintains the structural integrity of the M-cluster—has shifted the focus from the central cavity to the belt region of this cofactor as the potential site of catalysis. Indeed, crystallographic analyses of a CO-bound form of NifDK revealed displacement of the belt-sulfur S2B with a μ₂-CO ligand between Fe2 and Fe6 of the M-cluster [9], and subsequent crystallographic pulse-chase experiments demonstrated migration of a Se reporter through the whole belt region of the M-cluster upon turnover with C₂H₂ [10]. These results illustrate a highly dynamic nature of the belt region during catalysis while pointing to belt-sulfur displacement as a potential route to activated diiron species that is required for ligand binding. Consistent with these observations, a crystal structure of NifDK (designated NifDK*) was captured under physiological N₂-turnover conditions without externally supplied dithionite [11], which revealed an asymmetric displacement of S2B or S3A/S5A in the two cofactors of this protein, as well as elongation of either the Mo-O7 or Mo-O5 distance that changes the Mo-homocitrate ligation from bidentate to monodentate (Figure 2A, B). Subsequent biochemical, GC-MS, EPR and NMR experiments [11,12] demonstrated binding of N₂ in a catalytically competent state in NifDK*, leading to an asymmetric assignment of three dinitrogen species at the three belt-sulfur-displaced sites in the two cofactors of this protein (Figure 2A, B).

Figure 2. Catalysis at the nitrogenase cofactor.

(A, B) Side and top views of the asymmetric binding of dinitrogen species in the two M-clusters of NifDK* (PDB entry 6UG0), with one dinitrogen species displacing a belt-sulfur (S2B) and bound in a pseudo μ1,2-bridging mode between Fe2 and Fe6 in one M-cluster (A), and two dinitrogen species displacing two belt-sulfurs (S3A and S5A) and bound as asymmetric μ1,1-ligands bridged between Fe4 and Fe5 (at the S3A site) and between Fe7 and Fe3 (at the S5A site), respectively, in the other M-cluster (B). (C) Proposed mechanism of stepwise reduction of N₂ via cofactor rotation. This mechanism involves binding of N₂ at the S3A site via belt-sulfur displacement, followed by rotation of the cluster and reduction of N₂ to a diazene-level species (indicated by *) at the S2B site, prior to rotation of the cluster and further reduction of the diazene-level species to ammonia at the S5A site. The final reduction step at the S5A site signals binding of the ‘next’ N₂ (colored black) to the S3A site via belt-sulfur displacement, followed by release of NH₃ from the S5A site via belt-sulfur replacement, which signifies the next rounds of N₂ reduction via cluster rotation. It is plausible that displacement of the belt-sulfur at the S3A site (S²⁻→SO₃²⁻ at the S3A site) is coupled with binding and reduction of N₂, whereas replacement of the belt-sulfur at the S5A site (SO₃²⁻→S²⁻) is coupled with release of NH₃. (D) Coordination and reduction of SO₃²⁻ at the S5A site of the M-cluster. Shown are the reaction energies of two pathways calculated by DFT, which involve coordination of either two O atoms or one O atom and one S atom of SO₃²⁻ by the Fe3/Fe7 pair across the ‘belt’ of the M-cluster. The energies (kcal/mol) are indicated. The atoms in A-D are colored as described in Figure 1.

A dynamic mechanism of N₂ reduction can be proposed to rationalize the asymmetric conformation of belt-sulfur displacement in NifDK* (Figure 2C). Such a mechanism involves rotation of the M-cluster in the direction of S3A→S2B→S5A, which allows binding of N₂ at the S3A site, followed by sequential reduction of N₂ at the S2B and S5A sites, and release of NH₃ at the S5A site; additionally, it invokes an asynchronous rotation of the two cofactors that renders the reactions in the two M-clusters a step apart from each other (Figure 2C). It is further hypothesized that rotation of the cofactor is facilitated by an alternating elongation/breaking of one of the two Mo-O bonds between Mo and homocitrate, and that the asynchrony in cofactor rotation is enabled by an alternating docking of NifH on the two αβ-dimers of NifDK. Interestingly, the proposal of an alternating binding of NifH to NifDK aligns well with the cryo-EM observation of a 1:1 turnover complex between NifH and NifDK [4]; whereas the hypothesis of a dynamic change in the Mo/homocitrate ligation is in line with the cryo-EM observation of unresolved electron densities at the Mo-O ‘end’ of the cofactor in the turnover complex [4].

Strikingly, while NifDK* was capable of substrate reduction when combined with NifH, ATP and dithionite, no product could be detected when dithionite was replaced with Eu(II)-EGTA, unless SO₃²⁻ was supplied alongside this sulfur-free reductant, illustrating a dual requirement of electron- and sulfur-sources for nitrogenase catalysis [12]. Subsequent EPR experiments revealed a restoration of the S=3/2 signal of the resting-state cofactor concomitant with the disappearance of the dinitrogen-associated features upon incubation of NifDK* with Eu(II)-EGTA and SO₃²⁻ [12]. Moreover, crystallographic analyses demonstrated a ‘return’ of belt-sulfurs of Eu(II)-EGTA/SO₃²⁻ treated NifDK* that was accompanied by dissociation of the bound dinitrogen species [12]. These results collectively point to a mechanism of product release via insertion of SO₃²⁻ as belt-S²⁻ during nitrogenase catalysis, and DFT calculations provide further support for an energetically feasible, ‘in situ’ incorporation of SO₃²⁻ as a belt-S²⁻ at the S5A site during catalysis (Figure 2D) [12] that mirrors the insertion of the ‘9^th sulfur’ during cofactor biosynthesis (Figure 1D). Perhaps even more intriguingly, the mobilization of belt-sulfurs is highly dynamic, as all belt-sulfur locations can be labeled with Se upon turnover in excess Eu(II)-EGTA and SeO₃²⁻ and, conversely, the Se labels can be ‘chased off’ upon turnover in excess Eu(II)-EGTA and SO₃²⁻ [12].

The dynamic ‘turnover’ of the belt-sulfurs during N₂ reduction points to a plausible, ‘in situ’ interconversion between SO₃²⁻ and S²⁻ in the cofactor belt region as an indispensable component that has yet to be taken into the mechanistic consideration of nitrogenase. Conceivably, the oxidative half (S²⁻→SO₃²⁻ at the S3A site) is coupled with the binding and reduction of N₂ to NH₃; whereas the reductive half of this interconversion (SO₃²⁻→S²⁻ at the S5A site) is coupled with the release of NH₃ (Figure 2C). Such a scenario is appealing given that the reactions of S²⁻→SO₃²⁻ and N₂→2NH₃ both involve six electrons, although the details of sulfur mobilization during nitrogenase catalysis and how it ties in with the classic Lowe-Thorneley model of N₂ reduction [1,2] remain unclear. Regardless, the crucial roles of belt-sulfur mobilization in both cofactor biosynthesis and nitrogenase catalysis have come to light with the recent studies, which provide a renewed context for recalibrating the mechanistic thinking of nitrogenase.