Nitrogen Fixation and Hydrogen Evolution by Sterically Encumbered Mo-Nitrogenase

The substrate-reducing proteins of all nitrogenases (MoFe, VFe, and FeFe) are organized as α2ß2(γ2) multimers with two functional halves. While their dimeric organization could afford improved structural stability of nitrogenases in vivo, previous research has proposed both negative and positive cooperativity contributions with respect to enzymatic activity. Here, a 1.4 kDa peptide was covalently introduced in the proximity of the P cluster, corresponding to the Fe protein docking position. The Strep-tag carried by the added peptide simultaneously sterically inhibits electron delivery to the MoFe protein and allows the isolation of partially inhibited MoFe proteins (where the half-inhibited MoFe protein was targeted). We confirm that the partially functional MoFe protein retains its ability to reduce N2 to NH3, with no significant difference in selectivity over obligatory/parasitic H2 formation. Our experiment concludes that wild-type nitrogenase exhibits negative cooperativity during the steady state regarding H2 and NH3 formation (under Ar or N2), with one-half of the MoFe protein inhibiting turnover in the second half. This emphasizes the presence and importance of long-range (>95 Å) protein–protein communication in biological N2 fixation in Azotobacter vinelandii.


■ INTRODUCTION
The fixation of kinetically inert atmospheric dinitrogen (N 2 ) to ammonia (NH 3 ) is catalyzed in some specific microbes by a single family of enzymes known as nitrogenases, with turnover frequencies of around one N 2 fixed per second and a secondorder rate constant of ∼10 4 M −1 s −1 (k cat /K M ). 1−6 The Modependent nitrogenase consists of an N 2 -reducing molybdenum-iron (MoFe) protein and a corresponding reductase called the iron (Fe) protein ( Figure 1). The MoFe protein is a ∼230 kDa (αß) 2 heterotetramer (NifDK), where each αß half contains an electron-transferring [8Fe−7S] P cluster and a [7Fe−9S−C−Mo]/homocitrate FeMo cofactor (FeMoco). 2 The Fe protein is a homodimeric NifH 2 protein of ∼66 kDa containing a single [4Fe−4S] cluster and two MgATP binding sites. During turnover, each αß half of the MoFe protein accepts electrons from the ATP-hydrolyzing Fe protein, which are transferred via the P cluster to the FeMoco for N 2 fixation to NH 3 It is important to note that the above equation represents the optimized stoichiometry of one H 2 produced per N 2 fixed by the Lowe-Thorneley mechanism, with additional unproductive H 2 formation taking place under non-optimal turnover. 1,7−10 Each transient association of the Fe protein (Fe protein cycle) ultimately results in the transfer of 1e − to the FeMoco, where each Fe protein cycle consists of at least (i) MgATP-bound Fe:MoFe association, (ii) electron transfer (ET) from the P cluster to the FeMoco, (iii) ET from the Fe protein's [4Fe−4S] cluster to the P cluster, (iv) 2ATP hydrolysis, (v) the release of two inorganic phosphate (P i ), and (vi) MgADP-bound Fe:MoFe dissociation. The ratelimiting step of nitrogenase catalysis is thought to be the release of P i by the Fe protein, taking place with a rate constant of 25−27 s −1 . 11 This, in turn, implies that each Fe protein electron delivery cycle takes place with an overall rate constant of ∼13 s −1 .
Importantly, it has been shown that the two αß halves of the MoFe protein do not function independently during their repeated transient associations with the Fe protein (for electron delivery), suggesting that communication between the Fe proteins ∼100 Å apart takes place during turnover. 12 −14 In 2016, Danyal et al. reported that the Fe protein electron delivery cycles experience negative cooperativity in the presteady-state. 12 In other words, the Fe protein electron delivery cycle on one αß half suppresses ET in the other half. This was observed by quantitative measurements of Fe protein oxidation, ATP hydrolysis, and P i release. 12 In 2021, Truscott et al. reported positive cooperativity in the steady state for the reduction of acetylene, a non-physiological yet historically prominent substrate of nitrogenase. 13 In this approach, cooperativity was investigated by forming inactive Fe:MoFe complexes on one αß half of the MoFe protein (using AlF 4 − or nonstandard, tightly associating Fe proteins). Notably, cooperativity was not observed for 2H + reduction to H 2 , a physiologically relevant reaction that is catalyzed by nitrogenase in both the absence and presence of N 2 . In 2011, Eady and co-workers also observed that the MoFe protein containing only one FeMoco (half-populated) undergoes additional, non-electron-transferring interactions with the half-reactive MoFe protein, although the absence of the FeMoco was previously found to introduce a large change in the conformation of the MoFe protein. 15,16 More recently, cryogenic electron microscopic investigation into Fe−MoFe interactions during turnover identified a potential preference for the MoFe protein to associate to one Fe protein at a time, refocusing the spotlight on the MoFe protein's arrangement as a heterotetramer with two functional αß halves. 17 An important open question is therefore: how does cooperativity (negative, positive, or indeed none) impact N 2 fixation by nitrogenase during continued turnover? Harris et al. recently proposed that decreased selectivity toward N 2 fixation in the alternative vanadium-dependent and iron-only nitrogenases is due to decreased rate constants for the reductive elimination of H 2 (an activational step for N 2 fixation). 18 This reductive elimination requires the delivery of at least 4e − to the FeMoco for sufficient activation, which in turn requires ATPhydrolysis-coupled electron transfer from the Fe protein in one αß half of the MoFe protein. Stalled electron transfer to the FeMoco (by, for example, negative cooperativity induced by the second αß half of the MoFe protein) then provides time for the non-productive evolution of H 2 (and the loss of reducing equivalents) by the protonolysis of metal-hydrides on FeMoco. 10 This competition between reductive elimination and metal-hydride protonolysis explains the "optimal" stoichiometry of one H 2 evolved per N 2 fixed. 7 We hypothesized that the liberation of inhibited electron delivery to one αß half of the MoFe protein could therefore yield the stoichiometrically optimized production of one H 2 per N 2 fixed during continuous nitrogenase turnover, provided that cooperativity is not strictly necessary for N 2 fixation. However, in order to observe this, we deemed it of utmost importance to study the half-reactivity on a MoFe protein that (i) contained FeMoco in both αß halves (retaining its native conformation 16 ) and (ii) was not half-inhibited by a tightly associating Fe protein on one αß half, given that conformational changes transmitted between the Fe proteins bound to both αß halves are considered essential to cooperativity. 13,14 Here, we report on the reactivity of a MoFe protein (from Azotobacter vinelandii) wherein we sought to selectively inhibit Fe protein association on only one αß half by steric inhibition. To achieve this, we employed a MoFe protein mutant possessing a single solvent-exposed Cys residue in proximity to the P cluster (α-C45A/L158C, NifD = α); this mutant was previously employed to conjugate a Ru-based photosensitizer in the place of the Fe protein, enabling photo-excited electron transfer to the P cluster. 19 X-ray crystallography confirms the solvent accessibility (albeit somewhat geometrically hidden) of this cysteine residue. We subsequently employed the reactivity of this cysteine in a thiol-maleimide Michael addition to introduce a large synthetic Strep-tag-containing peptide (1.38 kDa) to facilitate both (i) Fe protein steric inhibition and (ii) the separation of inhibited MoFe proteins from the unmodified MoFe protein. This population of partially inhibited MoFe protein (lacking the uninhibited MoFe protein) was confirmed to be active for N 2 fixation, where a maximum velocity (V max ) of 66% was determined, consistent with negative cooperativity during N 2 fixation. Importantly, we observed that the selectivity (product/electron distribution) of this partially inhibited MoFe was practically unchanged between 0 and 1 atm N 2 , suggesting that cooperativity may not contribute toward nitrogenase's remarkable selectivity for N 2 . We conclude that negative cooperativity, globally, is employed by this nitrogenase for both H 2 production and N 2 fixation. As shown in Figure 1, residue α-L158 is located at the Fe:MoFe protein interface, and we therefore hypothesized that the functionalization of a Cys residue in this location with a steric inhibitor could prevent access of the Fe protein (and, therefore, nitrogenase catalysis). Indeed, a nitrogenase α-C45A/L158C MoFe mutant has been previously reported, yielding a single solvent-exposed Cys in proximity to the P cluster. 19 We first prepared this α-C45A/L158C MoFe with an N-terminal 8xHis tag on NifD (α subunit) for affinity purification, using a sacB-based markerless mutagenesis approach ( Figure S1, Supporting Information). 20−22 After verifying the introductions of the mutations by DNA sequencing (Supporting Information), we next crystallized the purified mutant MoFe protein and elucidated its structure by X-ray crystallography ( Figure 2). The X-ray crystal structure belonging to the P2 1 space group was solved by molecular replacement using the PDB 3MIN as a template. The structure was refined to a resolution of 3.03 Å and contained 2 MoFe proteins in the asymmetric unit (Table S2). The cell dimensions did not fit any of the previously solved structures (Table S3) and might be due to the introduced mutations that impacted the crystal packing ( Figure S2, Supporting Information). The α-C45A/L158C MoFe protein was found to overlay well with high-resolution structures previously reported for wild-type MoFe protein from A. vinelandii, suggesting a minimal impact of the introduced mutations on the overall conformation of the MoFe protein (Table S4 and Figure 2A). 23,24 Despite the rather low resolution, some distinct changes in the rotamers of amino acids around the introduced mutations were also observed ( Figure 2BC), once again being consistent with the mutations having been successfully introduced. Importantly, the P cluster and FeMoco of the α-C45A/L158C MoFe protein were both found to be present and intact (Supporting Information, Figures S3).

Desthiobiotin-Maleimide Steric Inhibitor
As depicted in Figure 1, the α-L158C mutation is positioned at the Fe protein-binding interface on the MoFe protein. To interrogate the half-reactivity of nitrogenase's MoFe protein, we sought to modify only one-half of the α-C45A/L158C MoFe protein at this position with a steric inhibitor that would enable both (i) inhibition of the Fe protein on this half of the MoFe protein and (ii) affinity purification of this hybrid α-C45A/L158C MoFe protein.
Previously, iodoacetamide-Cys reactivity was employed to attach a Ru-based photosensitizer to this α-C45A/L158C MoFe protein. 19 In this work, we elected to utilize maleimide-Cys thiol-Michael addition chemistry to modify the α-158C residue due to its improved chemoselectivity over iodoacetamides. 25 Initially, a steric inhibitor was synthetized by coupling a poly(ethylene glycol) 3 -modified desthiobiotin moiety with N-aminoethylmaleimide via a peptide/amide bond formation ( Figure 3A; additional details are provided in the Supporting Information, Figures S4−S6). The maleimide moiety was incorporated for the site-selective modification of the α-C158 residue of the α-C45A/L158C MoFe protein, whereas the desthiobiotin moiety was included as a binding motif for avidin-based affinity purification post conjugation of the MoFe protein. It was hypothesized that the poly(ethyleneglycol) repeating units would both increase the size-in-space of the steric inhibitor (and therefore its potency) and increase its solubility during protein conjugation. Initially, this maleimidemodified desthiobiotin-containing inhibitor (referred to subsequently as "DTB") was incubated with both wild-type and α-C45A/L158C MoFe proteins, and western blotting with a Streptavidin-horseradish peroxidase (HRP) conjugate to confirm successful modification of the MoFe proteins (Figures 3B and S7−S8).
Sodium dithionite (DT) is commonly used as a reducing agent during the purification and handling of nitrogenases due to their deactivation by molecular oxygen (O 2 ). We hypothesized that, much like thiol-based reducing agents, DT could reduce the maleimide functional group of DTB and thus quash our Cys functionalization strategy. Therefore, DT-free MoFe protein samples were prepared (Supporting Information) and Cys-maleimide labeling was evaluated in the presence and absence of DT. 26 As qualitatively shown in Figure 3B, the presence of DT in the MoFe protein samples (additional DT was not included during the reaction) was observed to lower the overall labeling of the MoFe proteins with DTB (also Figure S8, Supporting Information). Importantly, this issue could not be completely resolved with the use of tris(2-carboxylethyl)phosphine (TCEP) as a replacement reducing agent, commonly used in thiolmaleimide Michael addition reactions. DT was therefore removed from all subsequent MoFe protein preparations prior to maleimide functionalization reactions, only being reintroduced to terminate the maleimide-Cys reaction and maintain reducing conditions after the incubation period. Importantly, the omission of DT from the purification procedure (removed during the first purification column) did not result in a statistically lowered specific activity of the α-C45A/L58C MoFe protein (t-test, P = 0.93).
We next evaluated the residual activities of the α-C45A/ L158C MoFe protein (DT-free) following incubation with varying molar equivalents of the DTB inhibitor, using the Fe protein as the electron donor for H + reduction under Ar and N 2 reduction ( Figure 3C; specific activities are provided in Figure S9). Although a marked decreased in activity was observed globally, the residual activities of the α-C45A/L158C MoFe proteins treated with only 0.5−1.5 molar equivalents of DTB were found to range from approximately 55−75%, indicating that DTB effectively inhibits the Fe protein and subsequent substrate reduction by the MoFe protein.
Increasing molar equivalents of DTB were found to further decrease the specific activity of the MoFe protein (<20% with 5 molar equivalents). As previously shown, 10 mM DT significantly impeded the Cys labeling of the α-C45A/L158C MoFe protein with the DTB inhibitor, confirming the necessity to remove DT from the α-C45A/L158C MoFe protein prior to the reaction. DTB steric inhibition was also performed on the WT MoFe protein (exploiting the α-C45 residue), which exhibited a less pronounced decrease in specific activity ( Figure S9, Supporting Information).
Having observed a decrease in both H 2 formation (under Ar and N 2 ) and NH 3 formation (under N 2 ), we next determined whether the steric inhibition on one αß half of the α-C45A/ L158C MoFe protein impacts the distribution of electrons between H + and N 2 reduction (i.e., does a cooperativity mechanism contribute to nitrogenase's selectivity toward N 2 fixation?). According to the modified Lowe-Thorneley model of nitrogenase's enzymatic mechanism, the reductive elimination of one H 2 enables the binding and subsequent reduction of each N 2 at the FeMoco, leading to the reaction stoichiometry in eq 1 ( Figure 4). 1,10 This has been  . Simplified Lowe-Thorneley scheme for nitrogenase. N 2 associates to the FeMoco at the E 4 state along with the reductive elimination of H 2 by a productive pathway. Non-productive H 2 release by the FeMoco is colored in gray. The release of 2NH 3 is placed after the E 8 state in this representation. Percentage electron distribution is calculated assuming that 2e − are required for 2H + reduction to H 2 and that 6e − are required for N 2 reduction to 2NH 3 .

JACS Au
pubs.acs.org/jacsau Article experimentally observed, with H 2 production remaining persistent under a high pressure of N 2 (50 atm). 7 The 6e − reduction of N 2 therefore requires a total of 8e − (the reductive elimination of H 2 requires 2e − ), with 75% of the electrons delivered to the MoFe protein ultimately being detected as NH 3 . However, in practice under typical laboratory conditions (i.e., 1 atm of N 2 ), the distribution of electrons toward N 2 fixation typically reaches an upper limit of 60%. 20,27−30 This can be explained by the non-productive release of H 2 from reduced FeMoco states by the protonolysis of metal-hydrides, which also explains nitrogenase's H 2 -production activity in the absence of N 2 ( Figure 4). 8 Here, we observed an electron distribution between 50 and 60% for N 2 , implying that at least 2 or 3 H 2 molecules are released for each N 2 molecule reduced (presumably including at least one H 2 reductive elimination step) ( Figure 3D). Interestingly, the observed distribution of electrons toward N 2 fixation remained around 50−60% upon the titration of increasing equivalents of the DTB inhibitor even though the overall specific activities were observed to decrease ( Figure 3D). This result provided an initial indication that both halves of the αß MoFe protein are not strictly required to function in order to fix N 2 in any given αß domain. However, it is necessary to control the homogeneity of the DTB-bound MoFe population to study the cooperativity of MoFe (i.e., to purify half-inhibited MoFe proteins from uninhibited proteins). Therefore, a purification protocol for the half-inhibited MoFe was established.
Although the functionalization of the α-C45A/L158C MoFe protein with the DTB inhibitor was detected by western blotting with a Streptavidin-HRP conjugate ( Figure 3B), our attempts to purify this functionalized protein with commercial StrepTactin-based affinity columns (i.e., "StrepTrap" by Cytiva) were unsuccessful. This was hypothesized to be due to poor affinity of this engineered StrepTactin protein for desthiobiotin over the conventional "StrepTag", and we therefore reoriented our strategy.

PEPTIDE PERMITS THE PURIFICATION OF PARTIALLY REACTIVE MOFE PROTEIN CONJUGATES
We next sought to replace the DTB inhibitor with an alternative steric inhibitor that would additionally enable affinity-based purification. Given the widespread success of "Strep-tag" peptides for affinity purification, we elected to employ an N-terminal-modified Strep-tag peptide for the purification of α-C45A/L158C MoFe proteins that had been s u c c e s s f u l l y m o d i fi e d ( s e q u e n c e : m a l e i m i d e -GGGWSHPQFEK, referred to herein as the "Strep" inhibitor) ( Figure 5AB).
The Strep inhibitor was reacted with α-C45A/L158C MoFe with a 0.5:1 molar equivalent of the Strep/α-C45A/L158C MoFe protein (four accessible Cys158 residues per Strepmaleimide) to favor the formation of the half-functionalized MoFe protein over the doubly inhibited MoFe protein (an additional discussion can be found in the Supporting Information, Figure S10). After 4 h of reaction, the maleimide was quenched by the addition of 1 mM DT, and the mixture was loaded onto a commercial pre-packed StrepTactin column ( Figure S11, Supporting Information). Unmodified (uninhibited) MoFe protein was collected in the flowthrough fraction, and a dark band was observed to bind to the top of column, consistent with the successful functionalization of the α-C45A/L158C MoFe protein. This functionalization reaction was performed in triplicate on the same sample of the α-C45A/L158C MoFe protein. Figure 5C highlights the purity of the uninhibited (flow-through) and Strep-inhibited MoFe proteins (StrepTactin-bound), where western blotting with a StrepTactin-HRP conjugate highlighted the presence of the Strep moiety on only the inhibited (modified) MoFe protein samples. A lower molecular weight impurity with high affinity to the StrepTactin conjugate was identified, although this was not expected to impact subsequent studies of the Strepinhibited MoFe proteins due to its comparatively low abundance on the SDS-PAGE gel (this was also not identified during subsequent proteomics analysis discussed below). Quantification of the total Strep-inhibited MoFe protein fraction revealed a functionalization efficiency of 14% ± 1; this reflects the inefficiency of the maleimide-Cys labeling reaction that may be in part due to the inward-facing geometry of the solvent-exposed α-L158C residue, which further reduces the probability of obtaining di-functionalized MoFe proteins ( Figure 2C).
LC-ESI-MS/MS was performed on a sample of the Strepinhibited α-C45A/L158C MoFe protein to confirm the presence of the maleimide-Strep modification on the intended α-C158 residue. LC-ESI-MS/MS analysis of the SDS-PAGEexcised sample (81% α-subunit coverage, 91% ß-subunit coverage) confirmed that the Strep-modification was principally performed on the peptide fragment (trypsin-digested) containing the α-C158 residue, although an additional P cluster-coordinating Cys residue is also present on this fragment (α-C154). A potential functionalization was also identified on the ß-subunit/NifK (a single partially solventexposed peptide fragment), although this low-score hit could not be fully assigned (further discussion in the Supporting Information). Thus, the WT MoFe protein incubated with an excess of the Strep-maleimide inhibitor per MoFe (5 molar equivalents, subsequently purified over a StrepTactin column) was also analyzed by LC-ESI-MS/MS (88% α-subunit coverage, 92% ß-subunit coverage). The Strep-inhibitor modification was not detected on the α-C154-containing fragment and is consistent with good selectivity of the Strep inhibitor to the surface-exposed α-C158 residue. Further, the potential modification on the ß-subunit was not identified, further supporting its identification on the α-C45A/L158C MoFe protein to be a false positive due to its low-quality spectra (further discussion in the Supporting Information). While our data are indicative of partial MoFe protein labeling, it is not possible to quantify the fraction of half-inhibited MoFe (the target) vs doubly inhibited MoFe with this approach.
Anoxic native-PAGE analysis was performed to confirm that the Strep-inhibited α-C45A/L158C MoFe protein retained its heterotetrameric organization ( Figure S12, Supporting Information). Interestingly, the unmodified MoFe protein appears as a major product at ∼240 kDa although a faint product with a slightly lower molecular weight was also identified systematically during repeated native-PAGE analysis of both the WT and α-C45A/L158C proteins. Due to the high purity of the MoFe proteins (by SDS-PAGE analysis), we hypothesized that this second product was due to a different conformation of the MoFe proteins that was realized predominantly during electrophoretic analysis. Different conformations of MoFe proteins have previously been observed during native-PAGE analysis. 31 Interestingly, Strep-inhibited MoFe protein samples (WT and α-C45A/L158C) qualitatively exhibited more of this second conformation than the proteins that did not bind to the StrepTactin solid phase during purification. Subsequent limited proteolytic analysis ( Figure S13, Supporting Information) indicated that uninhibited and Strep-inhibited α-C45A/L158C MoFe did not exhibit a significant difference in their conformational flexibilities, and we concluded that this additional conformational state is therefore indeed induced during electrophoretic analysis, which is further pronounced following the anchoring of this 1.4 kDa Strep-inhibitor peptide to the α-subunit of the MoFe protein (in either the WT α-C45 or α-C158 position).
These results were confirmed by gel filtration in which uninhibited and Strep-inhibited α-C45A/L158C proteins share very similar elution volumes ( Figure 5D). We also used this method to investigate whether a Fe 2 :MoFe transition-state complex could be formed between the Strep-inhibited α-C45A/L158C MoFe protein and the Fe protein as evidence for inhibition of the Fe protein association in proximity to the P cluster site on the MoFe protein ( Figure 5D). A non-ATPhydrolyzing L127Δ Fe protein mutant was employed, which was previously found to form a tight complex with the MoFe protein. 32 To confirm the possibility of the L127Δ Fe protein to associate to the α-C45A/L158C Strep-inhibited MoFe protein, both proteins were incubated and the complex was subsequently purified using the N-terminal His-tag on the αsubunit of the MoFe protein. SDS-PAGE analysis of the obtained sample confirmed the presence of protein bands corresponding to both the MoFe and Fe proteins ( Figure S14, Supporting Information). After incubation of the uninhibited α-C45A/L158C MoFe protein with 5 molar equivalents of the L127Δ Fe protein, virtually all of the MoFe protein was found to shift to a complex of increased molecular weight, consistent with the formation of a tight Fe 2 :MoFe complex ( Figure 5D). On the other hand, incubation of the Strep-inhibited α-C45A/ L158C MoFe protein with the L127Δ Fe protein resulted in a broadened MoFe protein peak; further, a comparatively increased quantity of the non-complexed L127Δ Fe protein was observed. We hypothesize that this broadening corresponds to a mixed population of (i) non-complexed Strepinhibited MoFe protein, (ii) Fe 1 :MoFe protein complex, and (iii) doubly inhibited MoFe protein, consistent with inhibition of Fe protein association to the MoFe protein in the presence of the Strep-inhibitor at the α-C158 position. Multi-Gaussian peak analysis of this gel-filtration profile was performed in an attempt to provide indicative quantification of these different fractions ( Figure S15). We calculated that this broad feature may comprise ∼68% of the half-inhibited MoFe:Fe 1 complex and ∼32% of either non-complexed half-inhibited MoFe or doubly inhibited MoFe (we cannot discriminate between the latter two). Recent cryogenic electron microscopic investigation into Fe:MoFe complex formation during turnover suggests that the MoFe protein preferentially docks with only one Fe protein at a time. 17 The formation of a Fe 2 :MoFe protein complex (uninhibited and Strep-inhibited) was also evaluated by native-PAGE analysis, also suggesting that the Strep-inhibited α-C45A/ L158C MoFe protein does not form a tight Fe 2 :MoFe complex ( Figure S16, Supporting Information). Uninhibited WT and α-C45A/L158C MoFe proteins were observed to form two different complexes of an apparent larger size. Keeping in mind the proposed alternative conformation of the Strep-inhibited α-C45A/L158C MoFe protein during native-PAGE analysis,

JACS Au pubs.acs.org/jacsau
Article incubation with 5 molar equivalents of the L127Δ Fe protein yielded one major (and one significantly weaker) complex of apparent increased size, consistent with inhibition of the Fe protein access to primarily one-half of the MoFe protein.

Reactivity of the Strep-Inhibited MoFe Protein
After having successfully purified the Strep-inhibited α-C45A/ L158C MoFe protein, we next evaluated its remaining residual specific activity. Importantly, the mean specific activities for both H 2 (1 atm Ar) and NH 3 (1 atm N 2 ) production of the MoFe proteins that did not react with the Strep-inhibitor (the flow-through fractions) were found not to differ from the unreacted control protein (one-way ANOVA, P = 0.0833 and 0.3323), indicating that neither the reaction conditions nor the handling of the samples drastically impacted the enzymatic activity.
We then compared the specific activities of the uninhibited and Strep-inhibited α-C45A/L158C MoFe proteins for H + reduction under 1 atm Ar. As shown in Figure 6A, the specific activities of the Strep-inhibited MoFe proteins were found to be 79 ± 4% (mean of the three reactions) of the specific activities of their uninhibited counterparts. In addition, the specific activities of the three Strep-inhibited α-C45A/L158C MoFe protein samples were not found to significantly differ from one another (H 2 production under 1 atm Ar, one-way ANOVA, P = 0.8755). Our observation of >50% residual H + reduction activity per MoFe protein following Strep-inhibition (where at least one half of each MoFe protein has been inhibited by the Strep-inhibitor) suggests that the uninhibited α-C45A/L158C MoFe protein (more globally, nitrogenase) employs a negative cooperativity mechanism for H + reduction in the steady-state/continued turnover. This is consistent with the previous observation that the electron delivery cycle of the Fe protein also experiences negative cooperativity (cooperativity in terms of product formation in the steady state was not evaluated). 12 Another recent study observed that the MoFe protein does not exhibit cooperativity when producing H 2 under Ar. 13,33 As mentioned in the introduction, this approach inhibited Fe protein access to one αß half by introducing the L127Δ Fe protein to competitively form a non-dissociating complex (or by using a mismatched Fe protein from a different organism). Our gel filtration data ( Figure 5D) indicate that MoFe proteins half-inhibited by the L127Δ Fe protein may, in fact, not form non-dissociating complexes on only one αß half of the MoFe protein, leading to a population of Fe 2 :MoFe and free MoFe proteins (i.e., in the case of a 1:1 Fe L127Δ :MoFe protein ratio). The Strep-inhibition approach reported here is not anticipated to introduce a long-range conformational change at the second Fe protein-binding site. Further, we observed that the affinity of the Fe protein to the MoFe protein is relatively unchanged (H 2 production under 1 atm Ar) in the presence of the Strep inhibitor on the MoFe protein (Supporting Information, Figure S17).
The specific activity for NH 3 formation under 1 atm N 2 was investigated after Strep-inhibition of the MoFe protein, wherein a residual specific activity of 53 ± 4% (mean of the three reactions) was observed in comparison to that of the uninhibited MoFe protein ( Figure 6B, and Figure S18, Supporting Information). The specific activities of these three functionalization reactions were found to differ for NH 3 production only weakly under 1 atm N 2 (one-way ANOVA, P = 0.0467). As shown in Figure 6C, the % electron distribution toward N 2 fixation remains between 50 and 60% for all three Strep-inhibited MoFe proteins under 1 atm N 2 (one-way ANOVA, P = 0.1896), confirming that under these conditions nitrogenase's selectivity toward N 2 on both αß halves is not a result of cooperative behavior across the MoFe protein. The determination of whether a cooperativity mechanism is at play during N 2 fixation (in terms of specific activity) is more delicate and is treated in the following section.
A Strep-inhibition reaction was performed on a larger batch of α-C45A/L158C MoFe protein (∼18 mg of Strep-inhibited protein obtained) to evaluate Michaelis−Menten kinetic parameters and the distribution of electrons toward N 2 fixation under a range of N 2 partial pressures ( Figure 7A). Neither the Michaelis constant (K M ) nor maximum velocity (V max ) for the α-C45A/L158C MoFe protein was found to differ statistically from the wild-type MoFe protein (P = 0.2065 and P = 0.3873, Figure S19/Table S5, Supporting Information). Importantly, the V max for the Strep-inhibited α-C45A/L158C MoFe protein was found to be 66% of that of the uninhibited MoFe protein, while the affinity toward N 2 was unchanged (P = 0.2457, Table  1). The k cat normalized per FeMoco was observed to significantly increase following Strep-inhibition of the MoFe protein from 1.3 to 1.7 s −1 , P = 0.0102; the possible presence of a doubly inhibited MoFe protein in the sample suggests that the determined value of 1.7 s −1 per FeMoco could indeed be larger.
Importantly, the quantity of H 2 produced under increasing pressures of N 2 is consistent with increased N 2 fixation by nitrogenase (electron allocation toward N 2 fixation), with uninhibited and Strep-inhibited MoFe proteins exhibiting similar trends. Figure 7B reports the percentage e − distribution between N 2 fixation and H 2 formation by uninhibited and Strep-inhibited α-C45A/L158C MoFe proteins with increasing N 2 pressures. A maximum electron distribution of approximately 65% toward N 2 fixation was observed under these conditions, with no clear difference in electron distribution between uninhibited or inhibited α-C45A/L158C MoFe observed over this range of N 2 pressures. In contrast to the k cat and k cat /K M parameters, this suggests that a cooperativity mechanism, or the arrangement of the MoFe protein as an α 2 ß 2 heterotetramer, does not contribute toward the selectivity of nitrogenase toward N 2 under these conditions.
In agreement with previous reports, the total electron flux (total electrons consumed determined by production quantification, where H 2 = 2e − and NH 3 = 3e − ) of uninhibited and Strep-inhibited α-C45A/L158C MoFe proteins was observed to decrease following the introduction of increasing N 2 partial pressure in the reaction vials ( Figure 8A, 100% electron flux represents activity under 1 atm Ar). 27−29 However, no clear difference in the decrease of electron flux with or without Strep-inhibition over 0−1 atm of N 2 was observed (electron flux decreases uniformly for uninhibited and strep-inhibited α-C45A/L158C MoFe protein), where Figure 8B compares the percentage remaining total electron flux over 0−1 atm N 2 after having Strep-inhibited the α-C45A/L158C MoFe protein (in comparison to the uninhibited α-C45A/L158C MoFe protein). Importantly, total electron flux remains <100% regardless of the N 2 partial pressure (mean = 75%), suggesting that the MoFe protein indeed follows a negative cooperativity mechanism with respect to electron delivery. This is consistent with the earlier finding that the Fe protein electron delivery cycle exhibits negative cooperativity in the pre-steady state. 12 We therefore conclude that negative cooperativity for electron delivery does in fact propagate itself in the formation of total products (including the production of NH 3 ), although this does not significantly impact the selectivity of N 2 fixation by this enzyme complex. Considering the proposal that negative cooperativity may arise from the MoFe protein only interacting with one Fe protein at a time during turnover alongside our observations of >50% activity for Strep-inhibited MoFe proteins, we hypothesize that negative cooperativity is introduced after the association of a second Fe protein to a Fe 1 :MoFe protein turnover complex. 17 Badalyan et al. employed voltammetry to determine the rate constant for electron consumption by nitrogenase to be 14 s −1 ; importantly, this value was observed to remain constant in the presence of N 2 , in fact suggesting that electron flux to nitrogenase remains constant. 30 In this study, 23% of electrons were unaccounted for (not detected as H 2 or NH 3 ). More recently, Lee et al. observed that nitrogenase proteins purified in the strict absence of DT were unable to undergo continued turnover, where it was proposed that sulfite (resulting from DT decomposition/oxidation) plays an additional sulfur-recharging role upon reduction at the FeMoco. 26 Such a reaction may explain the apparent decrease in total electron flux upon N 2 fixation by nitrogenase.

■ CONCLUSIONS
Long-range (>95 Å) communication between the Fe proteins interacting with the two αß halves of nitrogenase's MoFe protein is thought to be of mechanistic importance. 14 Previous studies have observed (i) negative cooperativity for electron delivery in the pre-steady-state, and (ii) no cooperativity for H + reduction by nitrogenase upon "locking" one αß MoFe half in an inactive Fe protein-bound state. 12,13 We sought to inhibit one αß half of the MoFe protein by an approach that was not expected to impact the Fe protein transient association behavior on the remaining uninhibited active site. More importantly, we sought to determine whether cooperativity is strictly necessary for N 2 fixation by nitrogenase. We conclude that negative cooperativity plays a role in both H + reduction to H 2 and overall electron delivery to the MoFe protein (and, thus, also for N 2 fixation) and that cooperativity across the MoFe protein is not strictly necessary for the fixation of N 2 to NH 3 . The k cat per FeMoco site for NH 3 production was observed to increase by 31% upon Strepinhibition (from 1.3 to >1.7 s −1 ), consistent with negative cooperativity in N 2 fixation by nitrogenase in terms of electron delivery (selectivity toward N 2 is not impacted). It remains difficult to precisely determine the magnitude of negative cooperativity-induced suppression of nitrogenase activity in a single αß MoFe protein half due to the possible presence of a doubly inhibited MoFe protein in our Strep-inhibited sample. A potential method to surmount this issue could be the coexpression of two copies of nif D (i.e., nif D and nif D*) followed by tandem affinity purification.
While an in vitro investigation into nitrogenase's cooperativity mechanism provides insight into its catalytic mechanism, it remains important to determine the importance and magnitude of negative cooperativity of nitrogenase in the context of in vivo turnover (in A. vinelandii), where (i) Fe/ MoFe protein ratios could be more dynamic or (ii) additional partners could play a role (such as maturases, activitymodulating proteins such as CowN, 35 O 2 -protection proteins such as the Shethna FeSII protein 36,37 or indeed an as-of-yet unidentified allosteric effector).

Culturing of A. vinelandii
Extensive procedures are reported in the Supporting Information. Briefly, MoFe and Fe nitrogenase proteins were isolated from various A. vinelandii strains and were cultivated on a modified Burke's medium. The nif operon was derepressed after overnight growth by centrifugation and resuspension into fresh Burke's medium lacking fixed N (NH 4 + ). Cells were harvested by centrifugation after ∼3.5 h and stored at −80°C until use.

Fe Protein Purification
All Fe protein purifications were performed in a COY anaerobic chamber (<5% H 2 />95% N 2 , Michigan, USA). The wild-type Fe protein was purified from a strain derived from A. vinelandii DJ that was modified to introduce an 8x His-tag to the N-terminus of NifD (A. vinelandii RS1). 20 The L127Δ Fe protein was purified from the A. vinelandii strain DJ1065 (provided by Dennis Dean, Virginia Tech). Cells from a 12 L culture were thawed and resuspended in anoxic lysis buffer (50 mM Tris/HCl, pH 8.0) containing 5 mM DT and 37% v/v glycerol. All Fe protein purification buffers contained 2 mM DT, except for the lysis buffer mentioned above. After incubation for 20 min, the cells were collected by centrifugation and resuspended in glycerol-free lysis buffer to induce cell lysis by osmotic shock. The lysate was incubated on ice for a further 15 min and clarified by centrifugation at 30,000 x g (4°C, 1 h). A post-lysis buffer (2 M NaCl, 234 mM Tris/HCl, 2 mM DT, pH 8.0) was added to the cellfree supernatant to achieve a final NaCl concentration of 0.3 M. The cell-free supernatant was next passed over a HisTrap HP column (5 1.3 ± 0.1 >1.7 ± 0.1 a k cat /K M (s −1 atm −1 ) 6.5 ± 1.7 11.0 ± 2.5 k cat /K M (x10 4 s −1 M −1 ) 1.0 ± 0.3 1.7 ± 0.4 a k cat is defined here as turnover frequency per active αß half (nmol NH 3 FeMoco −1 s −1 ). The possible presence of a doubly inhibited MoFe protein in this sample could result in a larger k cat value per FeMoco (αß half). Partial pressures of N 2 were from atm to M −1 using a Henry's law conversion factor of 6.4 × 10 −4 . 34 mL column volume, Cytiva) to remove the His-tagged MoFe protein (below). The column flow-through (containing the Fe protein) was next diluted with a NaCl-free buffer (50 mM Tris/HCl, pH 8.0, 2 mM DT) to obtain a final NaCl concentration of 0.1 M. The Fe protein was first purified over a HiPrep Q-Sepharose HP 16/10 column (20 mL column volume, Cytiva) and eluted over a linear NaCl gradient of 0.2−0.65 M NaCl. Following concentration to <10 mL using a Merck-Millipore stirred concentration cell (30 kDa molecular weight cut-off membrane), the Fe protein was next purified by size-exclusion chromatography over a HiPrep 26/60 Sephacryl S-200 HR column (320 mL column volume, Cytiva) using a Tris running buffer (50 mM Tris/HCl, pH 8.0, 0.5 M NaCl). The eluted Fe protein was concentrated further to >20 mg/mL and flash-frozen in liquid nitrogen until use.

MoFe Protein Purification
All MoFe protein purifications were performed in a COY anaerobic chamber (<5% H 2 />95% N 2 , Michigan, USA). The wild-type MoFe protein carrying an 8x His-tag on the N-terminus of NifD was purified from the A. vinelandii strain RS1. The α-C45A/L158C MoFe protein was purified from a derivative of A. vinelandii RS1 (A. vinelandii strain "M1", Supporting Information), which carried the same 8x His-tag on the N-terminus of NifD. The cell-free supernatant (prepared as above) was first passed over a HisTrap HP column (5 mL column volume, Cytiva). The running buffers for this step (50 mM Tris/HCl, pH 8.0, ±0.3 M imidazole) were incubated overnight in the anaerobic chamber and did not contain DT (all subsequent buffers did not contain DT). MoFe proteins bound to the His-resin were first washed with >3 column volumes of DT-free buffer (0 mM imidazole) to remove DT introduced during cell lysis. After a 20 mM imidazole washing step, MoFe proteins were eluted using 0.3 M imidazole. The eluted MoFe proteins were next diluted with Tris buffer (50 mM Tris/HCl, pH 8.0) to reach a final NaCl concentration of 0.1 M prior to being loaded onto a pre-equilibrated Q-Sepharose FF 16/10 column (20 mL column volume, flow-rate = 20 mL/min). DT-free MoFe proteins were eluted over a linear gradient of 0.2−0.65 M NaCl and concentrated to >20 mg/mL using a 100 kDa stirred concentrator cell (fed with ultra-high-purity N 2 5.0) prior to being flash-frozen in liquid nitrogen until use.

MoFe Protein Functionalization with the DTB Inhibitor
The DT-free MoFe protein was treated with the DTB inhibitor (Supporting Information) freshly prepared in 0.1 M MOPS/NaOH buffer (pH 7.0) for 4 h at room temperature within the COY anoxic chamber. After 4 h, unreacted maleimide was quenched and MoFe proteins were reduced by the addition of DT to a final concentration of 2 mM.

MoFe Protein Functionalization with the Strep-Tag Inhibitor and Conjugate Purification
A lyophilized synthetic Strep-tag peptide containing an N-terminal maleimide functionality (sequence = GGGWSHPQFEK) was obtained from GenScript (USA) and used as received. This 1.4 kDa peptide functioned as both (i) a steric inhibitor of Fe protein association to the MoFe protein and (ii) a Strep-tag for purification of the MoFe protein conjugates. Strep-inhibited α-C45A/L158C MoFe protein was obtained following the incubation of DT-free α-C45A/ L158C MoFe protein with 0.5 molar equivalents of the Strep-tag inhibitor (0.5 inhibitor per MoFe protein, resulting in a 1:4 Strep-tag to surface Cys ratio, freshly dissolved in 0.1 M MOPS/NaOH pH 7.0 to a final concentration of 1−3 mg/mL) for 4 h within the anoxic glovebox. The reaction was quenched (and the MoFe protein reduced) by the addition of DT to a final concentration of 2 mM. The functionalized/inhibited MoFe protein was next purified over a StrepTrap XT column (5 mL column volume, Cytiva) preequilibrated with MOPS buffer (0.1 M MOPS/NaOH, 0.2 M NaCl, 2 mM DT, pH 7.0). Unreacted MoFe proteins were collected from the column flow-through, and Strep-inhibited MoFe proteins were collected following the application of 50 mM biotin to the column. MoFe proteins were concentrated as above and flash-frozen in liquid nitrogen until use.

Nitrogenase Activity Assays
Briefly, nitrogenase activity assays were performed in 13 mL septumsealed glass vials containing 1 mL of an ATP-regenerating MOPS buffer (100 mM MOPS/NaOH, pH 7.0, 5 mM ATP, 30 mM phosphocreatine, 1.3 mg of bovine serum albumin, 0.2 mg of creatine phosphokinase from rabbit muscle, and 10 mM DT). All activity assays employed DT as the electron donor. All reactions were assembled within an Ar-filled anoxic glovebox (Jacomex, France). MoFe proteins (0.1 mg) and Fe proteins (0.48 mg) were included with a 16.6:1 Fe to MoFe protein ratio, sealed, and vented to atmospheric pressure. The reaction vials were then heated to 30°C in a shaking water bath and the reaction initiated by the injection of MgCl 2 to a final concentration of 10 mM using a gas-tight syringe. After 8 min, the reactions were terminated by the addition of 300 μL of 400 mM EDTA. H 2 was quantified using a gas chromatographthermal conductivity detector (molecular sieve 5 Å column, Ar carrier gas, SRI Instruments model 8610C). NH 3 was quantified by the orthophthalaldehyde method (corrected to controls and assays performed under 1 atm Ar) using NH 4 Cl as the standard. 20,38 Crystallization of the α-C45A/L158C MoFe Protein The purified enzyme in Tris/HCl buffer (50 mM Tris/HCl, 300 mM NaCl, 2 mM DT, pH 8.0) was crystallized anaerobically at 8 mg/ mL −1 (under a 100% N 2 atmosphere) with an OryxNano (Douglas Instrument, UK). The initial screening was performed at 20°C using the sitting drop method on 96-Well MRC 2-drop crystallization plates in polystyrene (SWISSCI) containing 90 μL of crystallization solution in the reservoir. The protein sample (0.5 μL) was mixed with 0.5 μL of reservoir solution. Crystals were transferred and stored anaerobically in a Coy tent (N 2 /H 2 , 97:3). Thin brown plate crystals appeared after a few weeks. The reservoir solution contained 25% w/v polyethylene glycol 3,350, 100 mM BIS-TRIS at pH 5.5, and 200 mM Lithium sulfate.

X-ray Crystallography Data Collection and Refinement
Crystal handling was done inside a Coy tent under an anaerobic atmosphere (N 2 /H 2 , 97:3). Crystals were soaked in the crystallization solution supplemented with 15% v/v glycerol as a cryo-protectant before being frozen in liquid nitrogen. Crystals were tested and collected at 100 K at the Swiss Light Source, X06DA−PXIII. Due to the high anisotropy, data were processed and scaled with autoPROC. 39 The relative resolution limits along the unit cell axis are a = 4.34 Å, b = 4.07 Å, and c = 3.03 Å. The molecular replacement was done with PHASER from the PHENIX package. 40 The model was then manually built with COOT and further refined with PHENIX without hydrogens. 41 The last cycles of refinement were performed with Buster and the final one with PHENIX. The model was validated by the MolProbity server (used on the 15th of August 2022). 42,43 ■ ASSOCIATED CONTENT
Protein production and purification methodology; activity assays; analytical gel filtration methodology; limited proteolysis methodology; additional X-ray crystallography data; synthesis of DTB inhibitor; sacBbased markerless mutagenesis; additional SDS-PAGE and western blotting; activity assays for the DTBinhibited wild-type MoFe protein; binding and elution of Strep-inhibited MoFe on StrepTrap XT (Streptactin) columns; native-PAGE; Fe protein titration of the MoFe protein; specific activities for NH 3