What Can be Learned From the Electrostatic Environments Within Nitrogenase Enzymes?

Abstract

Nitrogen fixation is a fundamental, and yet challenging, chemical transformation due to the intrinsic inertness of dinitrogen. Whereas industrial ammonia synthesis relies on the energy‐intensive Haber–Bosch process, nitrogenase enzymes achieve this transformation under ambient conditions—yet at the expense of a remarkably high ATP demand. Understanding their mode of operation could inspire the development of more efficient synthetic catalysts. In this study, we scrutinize the electrostatic environment surrounding nitrogenase's active site, the so‐called M‐cluster. Strikingly, we observe that all types of M‐clusters exhibit similar trends, with distinct patterns around the individual metal sites that have been proposed as potential N₂‐coordination sites. Specifically, a strong local electric field pointing away from the Fe2 site is identified, as well as a minor field pointing toward the Fe6 sites. Furthermore, a significant oriented long‐range field along the Fe2–Fe6 axis is computed across the entire family of nitrogenases. In the final part of the manuscript, we discuss how the observed electrostatic patterns may impact chemical reactivity, and how they can be connected to previously made mechanistic hypotheses. Overall, this study provides further evidence for the ubiquitousness of local electric fields in enzyme catalysis, even when substrates that seemingly have only limited electrostatic susceptibility are involved.

The catalytic M‐clusters of nitrogenase enzymes exhibit characteristic electrostatic patterns around the sites involved in N₂ fixation. Specifically, a strong local electric field pointing away from the Fe2 site is consistently identified, as well as a minor field pointing toward the Fe6 sites. Furthermore, a significant long‐range field along the Fe2–Fe6 axis is computed. These patterns may significantly impact the chemical reactivity of these clusters.

1. Introduction

Dinitrogen (N₂) is a remarkably inert molecule, owing to its strong triple bond and low reactivity toward proton and electron transfer. Breaking a single π bond requires approximately 110–120 kcal/mol, and even the gas‐phase addition of one H₂ molecule is endothermic by ∼48 kcal/mol (once the initial addition has taken place, however, further reduction to hydrazine becomes a downhill process).^{[ 1} , ^{2 ]} Additionally, N₂ has a large HOMO–LUMO gap (∼10 eV or 220 kcal/mol) and a relatively low proton affinity (∼110 kcal/mol), rendering its reduction particularly daunting.^{[ 2 ]}

Despite these well‐known challenges, nitrogen fixation is vital to life, providing the main source of organic nitrogen for plants and, by extension, nearly all life forms. Today, over 50% of the nitrogen entering the biosphere originates from industrial fixation—primarily via the Haber–Bosch (HB) process. While this process has dramatically improved food security and sustained exponential population growth, it remains highly energy‐intensive, consuming nearly 2% of global energy production.^{[ 3} , ⁴ , ^{5 ]} As such, there is growing interest in developing catalysts that can enable ammonia synthesis under milder conditions.

In a biomimetic spirit, biological nitrogenases, evolved through natural selection, offer a compelling blueprint for such efforts.^{[ 6} , ⁷ , ⁸ , ^{9 ]} These enzymes catalyze the six‐electron reduction of N₂ to ammonia under ambient conditions, driven by ATP hydrolysis. Despite their diversity, most nitrogenases share a similar metallocluster‐based active site—an MFe₇S₉C cofactor (M = Mo, V, or Fe), commonly referred to as FeMoco, FeVco, or FeFeco. Among the nitrogenases, FeMoco nitrogenase is the most widespread and efficient, making it the most widely studied class.^{[ 10} , ^{11 ]} In 2003, Seefeldt and coworkers provided strong experimental evidence for a preferential facet on the catalytic unit in FeMoco nitrogenases (Fe2‐S2B‐Fe6‐S5A, Figure 1) through a series of mutation studies on a representative example enzyme (PDB: 3U7Q).^{[ 12} , ¹³ , ¹⁴ , ^{15 ]}

Figure 1.

The FeMoco unit in nitrogenase (PDB: 3U7Q). Based on experimental evidence, Fe2 and Fe6 have been identified as the most probable coordination sites for N₂.^{[ 12} , ¹³ , ¹⁴ , ^{15 ]}

At the same time, the precise mechanism of N₂ reduction by nitrogenases remains elusive. State‐of‐the‐art quantum chemistry simulations often yield diverging conclusions, with details such as protonation states, N₂ binding geometry, and sequence of reduction steps remaining contentious (Figure 2A–C). These discrepancies reflect the inherent complexity of the system and the sensitivity of predictions to computational methodology.^{[ 2} , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ^{23 ]}

Figure 2.

A) Various protonation states for His195 adjacent to the FeMoco cluster have been proposed, resulting in marked changes in the preference between N₂‐coordination sites in QM/MM calculations reported by Bjornsson et al.^{[ 16 ]} B) Both a one‐ and two‐side structure have been characterized for the reduced FeMoco unit.^{[ 2 ]} C) In a recent study, Head–Gordon and coworkers proposed that upon (end‐on) coordination, N₂ forms a bridge between Fe2 and Fe6 (i.e., Fe2(m₂:h²–N₂)Fe6) prior to reduction. D) N₂‐activation by a Frustrated Lewis Pair. As recently suggested by Jain et al.,^{[ 37 ]} application of an external electric field aligned with the dipole moment of the transition state can significantly modulate the barrier associated with the breaking of the triple bond.

Instead of wading headfirst into the heated discussions surrounding the nitrogen fixation mechanism, we aim here to investigate the FeMoCo‐cluster from a different perspective. Specifically, we set out to scrutinize the electrostatic patterns exerted by the protein matrix surrounding the active site of nitrogenases.^{[ 24} , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ^{29 ]} We hypothesize that, just like in any other enzyme, the active site of nitrogenases is surrounded by an exquisitely tailored protein matrix that significantly affects catalysis, without direct interaction with the M‐cofactor or its substrates.^{[ 24 ]} Collectively, these (partial) charges exert strong local electric fields—experimental measurements, which corroborated preceding theoretical estimates (with reasonable accuracy), indicate that field strengths of several VÅ⁻¹ are not uncommon.^{[ 30} , ³¹ , ^{32 ]} Such field strengths can exert a marked influence on catalysis, whereby both kinetic and thermodynamic aspects can be affected, even for apolar substrates, as local electric fields tend to wake‐up dormant ionicity/charge separation in molecules. Many examples of electrostatic catalysis have been reported in recent years.^{[ 25} , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³³ , ³⁴ , ³⁵ , ^{36 ]}

Interestingly, simulations by Shaik and coworkers recently indicated that despite its limited polarizability in the gas‐phase, the energy barrier to N₂‐activation in Frustrated Lewis Pairs can be modulated by dozens of kcal/mol through the application of electric fields of a mere 0.2–0.3 VÅ⁻¹ (Figure 2D).^{[ 37 ]} Considering this field strength is rather modest, compared to what is often computed around the active site of enzymes, we surmised that nitrogenases may also take advantage of electrostatics, for example, by lowering the barrier to N₂ activation, and ultimately favoring its reduction.

In this study, we set out to quantify the electrostatic environment around nitrogenases, and we discuss the potential impact these fields may have on the reaction mechanism. Our hope is that our efforts will spark more in‐depth, and computationally demanding, mechanistic investigations that properly include the effect of the oriented local electric field (OLEF) on N₂‐coordination and activation.

2. Methods

To quantify the OLEFs exerted by the protein matrix surrounding the M‐cofactor of nitrogenase, a protocol inspired by Alexandrova et al.’s was developed. This protocol was previously demonstrated to enable analysis of electrostatic trends across entire enzyme families and to yield results that qualitatively reproduce more in‐depth/computationally expensive simulations.^{[ 38 ]} The approach started by extracting FeMoco/FeVco/FeFeco nitrogenase structures collected in the RCSB protein data bank.^{[ 39 ]} This initial dataset was filtered based on a series of structural criteria. First, only structures with a resolution ≤2.5 Å were retained. Next, we dismissed PDB entries if they a) were determined using CryoEM, b) contained selenium instead of sulfur, c) contained a bound CO‐ligand, d) were crystallized at a nonphysiological pH, e) contained xenon or f) were not fully resolved (e.g., due to disorder). Following this filtering, 26 FeMoco, 3 FeVco, and 2 FeFeco nitrogenases were retained.

With this set of structures at hand, a series of modifications was introduced to each of them, to facilitate a faithful determination of the electrostatic patterns around the respective active sites. In first instance, to enable proper detachment of the catalytic cluster from the protein matrix without inducing artificial distortions of the external electric field, the cysteine residue, coordinating the catalytic unit through a (deprotonated) thiol, was replaced by an alanine residue. Subsequently, Propka^{[ 40 ]} was used to assign physically relevant protonation states throughout the protein matrix. Next, the histidine residues adjacent to the catalytic unit, for which the protonation state depends on the parametrization of the cofactor, were corrected so that their states match those in respectively Head–Gordon's,^{[ 2 ]} as well as Bjornsson's^{[ 16 ]} and Siegbahn's,^{[ 17 ]} computational work. In the following step, all “nonresidue” substances (e.g., cofactors, waters, substrates, counterions, etc.) were removed so that only the actual protein matrix remained. The positions of the hydrogens on the residues were subsequently refined through a minimization with 250 cycles of steepest descent, followed by 750 cycles of conjugate gradients, while keeping the positions of all the heavy atoms fixed. Minimizations were run using the sander module of AMBER suite.^{[ 41 ]}

Next, the electrostatic environment, induced by the protein matrix, was extracted as point charges localized at the atomic position of the minimized structure, with partial charges assigned based on the ff14SB protein force field.^{[ 42 ]} To this end, electric field strengths were computed on spherical grids around the positions of the Fe atoms of the FeMoco/FeVco/FeFeco cluster with the help of the TITAN code,^{[ 43 ]} making use of Coulomb's law,

$${\mathit{F}}_{\mathrm{i}}\left(\mathit{r}\right)=\sum _i\frac{q_{\mathrm{i}}}{4\pi {\epsilon }_0}\frac{\mathit{r}-{\mathit{r}}_{\mathrm{i}}}{{\left|\mathit{r}-{\mathit{r}}_{\mathrm{i}}\right|}^3}=\sum _{\mathrm{i}}\frac{q_{\mathrm{i}}}{4\pi {\epsilon }_0}\frac{{\widehat{\mathbf{R}}}_{\mathrm{i}}}{{\left|\mathit{r}-{\mathit{r}}_{\mathrm{i}}\right|}^2}$$ (1)

in which ${\mathit{F}}_{\mathrm{i}}(\mathit{r})$ is the local electric field vector, ε ₀ is the permittivity of the vacuum, $q_{\mathrm{i}}$ is the atomic charge of atom $i$, $\mathit{r}-{\mathit{r}}_{\mathrm{i}}$ is the distance between the point charge and the point of evaluation and ${\widehat{\mathbf{R}}}_{\mathrm{i}}$ is the unit vector pointing from ${\mathit{r}}_{\mathrm{i}}$ to $\mathit{r}$.

We primarily focused on the OLEF components pointing toward to metallic centers from the catalytic unit's exterior, as these are the possible N₂ coordination sites (400 points distributed on spherical grids with radii 1.5, 2.0, 2.5, and 3.0 Å around the Fe2 and Fe6 sites were sampled, respectively). Throughout the discussion below, we consistently adhered to the so‐called “physics” convention for electric fields, that is a positive electric field is obtained when the direction vector runs from positive to negative charge, and vice versa (Figure 3A). Note that within this convention, a positive field stabilizes a molecular system with a negative dipole moment (µ), and vice versa (Figure 3B).

Figure 3.

A) The physics convention used for electric fields. Z is the direction axis pointing to the Fe site, F_Z is an oriented external electric field along this axis. B) According to the established convention, a positive F_Z stabilizes a compound with a negative dipole moment along this axis (m_z).

To validate the quality of our computed OLEFs, we also set up an alternative workflow where electrostatic environments were determined from the permanent multipoles and induced dipoles extracted from (polarizable) AMOEBA force field calculations,^{[ 44} , ^{45 ]} performed through OpenMM^{[ 46 ]} (see Section S1 of the Supporting Information for the equations used to quantify the electrostatic contributions of each outputted component). For the test cases considered (see Section S2 and S3 of the Supporting Information), we observe that despite some local distortions in the electrostatic environments — particularly on the evaluation sites close to the atoms of the protein matrix — the overall OLEF patterns are fully retained. The fact that both fixed‐charge and polarizable force fields reproduce the trends described below, provides confidence that our presented findings are robust.

Note that in quite a number of the PDB entries considered during our analysis, the interstitial carbide^{[ 12} , ^{47 ]} is either not refined or incorrectly characterized as a nitride. Since we are exclusively interested in the electrostatic patterns induced by the protein matrix here, the correct assignment of the interstitial atom does not play a role, and hence we opted not to modify the extracted structures.

Quantum chemical calculations were performed using Density Functional Theory (DFT), as implemented in Gaussian16.^{[ 48 ]} Geometry optimizations and frequency calculations were carried out at the wB97X‐D/def2‐TZVP level of theory.^{[ 49} , ^{50 ]} Oriented External Electric Field (OEEF) effects were included through the “Field” keyword available in Gaussian16. Field strengths ranging from −2.05 to +2.05 V Å⁻¹ (−0.040 au, to +0.040 au: 1 au = 51.4 V Å⁻¹) were applied along the Z‐direction, which was aligned with the intrinsic dipole moment in the case of H₂S, and with the bond axis in the case of N₂ (as there is no intrinsic dipole moment in the latter molecule).

3. Results and Discussion

3.1. Analysis of the Electrostatic Environments Around Plausible N₂ Coordination Sites

First, we start by considering the electrostatic environments around the main candidate coordination sites for N₂ in the 3U7Q nitrogenase,^{[ 16 ],} that is the Fe2 and Fe6 sites on the FeMoco unit. In Figure 4A, the computed magnitudes of the OLEFs, that is F_z, pointing toward/away from these two sites are presented. From this quiver plot, it is apparent that the Fe2 site is surrounded by a strongly negative local electric field in all directions away from the core of the M‐cluster. The magnitudes of the computed OLEFs range from +0.1 VÅ⁻¹ to −1.6 VÅ⁻¹, depending on the precise direction and radius around the metal center considered.

Figure 4.

OLEF maps for the FeMoco unit of 3U7Q^{[ 12 ]} for different protonation states of His195 (cf. Figure 2A): A) the e monoprotonated form, B) the d monoprotonated form, and C) the doubly protonated form. D) The map obtained when the charges for His195 are discarded altogether. Note that arrows are scaled relative to the biggest one in the respective plot, only the color code is absolute.

A major contributor to the pronounced field at this site is the presence of the neighboring His195 residue, due to the presence of lone pairs on the N atoms in its imidazole ring pointing toward the FeMoco. As mentioned above (cf. Figure 2A), the protonation state of this residue has been the cause of significant controversy: both a (neutral) monoprotonated form, with the proton on either the e or d site of the imidazole ring, as well as a doubly protonated, cationic form (imidazolium), has been suggested for this residue at different stages of the catalytic mechanism.^{[ 2} , ^{15 ]} In Figure 4A, we assumed this residue to be in its e monoprotonated form, as this is the experimentally determined protonation state of the resting E₀ state, and in the mechanism proposed by Bjornsson and coworkers, it was also suggested as the most stable one in the E₄ state, that is preceeding N₂ coordination.^{[ 16 ]}

Remarkably, our calculations reveal that, in qualitative terms, the electrostatic environment around the FeMoco is fairly robust with respect to changes in the His195 protonation state. For the alternative d form, the F_Z is enhanced further by a significant margin, reaching values as high as −2.8 VÅ⁻¹ (Figure 4B). For the doubly protonated form, the direction of the OLEF is reversed in the region around the introduced protons on His195 (locally reaching values as high as +1.5 VÅ⁻¹), but it nevertheless retains a nonnegligible negative direction, with magnitudes up to −0.7 VÅ⁻¹, in all the other directions surrounding the Fe2 site (Figure 4C).

Interestingly, upon excluding the charges stemming from the adjacent His195 during the electric field calculation altogether, an almost uniform negative electric field is computed in all directions around the Fe2 site away from the FeMoco, with the OLEF reaching magnitudes up to −0.6 VÅ⁻¹ (Figure 4D). Furthermore, the attenuation of the field's magnitude with distance is remarkably small. This is reflected by the size of the quivers in the inner spherical ring around Fe2, which have nearly the same size as those in the outer ring. This highlights the fact that the field is not caused exclusively by a handful of charges in close proximity to the FeMoco, but that it is in fact a large collection of distant charges/dipoles that contribute to the overall OLEF (cf. Equation 1; if $r_{\mathrm{i}}$ is already big for most $i$ at the outer evaluation radius around the metal site, then the field strength is barely reduced by increasing the distance from the edge of the protein by 1.5 Å, to reach the inner evaluation radius).

To confirm this reasoning, we also computed the OLEFs when the charges belonging to any residue within a radius of 4 Å around the cofactor are discarded (see Section S3 of the Supporting Information). As expected, a consistently outward‐pointing field is indeed retained in this case. Note that the diffuse nature of the computed fields also means that local distortions of the region of the protein matrix in direct contact with the FeMoco cofactor during the catalytic cycle — which would be revealed by molecular dynamics calculations — are highly unlikely to completely disrupt the observed pattern.

Taken together, the observations made above underscore that the negative field direction at Fe2 is not solely caused by the nature/orientation of His195 (and/or the other residues adjacent to this Fe‐site in the first coordination shell). Instead, it is a persistent effect induced collectively by multiple residues in coordination shells further away from the FeMoco. Note also that when a cluster model is applied to the modelling of nitrogenase activity — a popular choice to reduce the computational cost and enable the application of more accurate electronic structure methods — this medium‐ to long‐range contribution to the OLEF is inevitably neglected, as even the most expansive clusters found in the literature do not include any residues beyond His195.^{[ 2} , ¹⁷ , ⁵¹ , ^{52 ]} To underscore the importance of this point, we computed how the electric field evolves as one removes residues in a gradually increasing radius around the FeMoco from the electric field calculation; almost 100 residues need to be removed for all the OLEFs to drop to values below ± 0.1 VÅ⁻¹ (when computing the electrostatics starting from the polarizable force‐field, the number of residues that need to be removed to this end increases even further, to 112, cf. Section S3 of the Supporting Information).

For the Fe6 site, the electrostatic environment is markedly different. Regardless of the protonation state of His195, the electric field strengths around this site tend to be small: except for a handful of evaluation points, the field strength never exceeds +0.4 VÅ⁻¹. In stark contrast to Fe2, the distance attenuation is relatively pronounced: as one approaches the Fe6 site, the size of the quivers decreases much faster than for the Fe2 site. Furthermore, in the regions on the outer evaluation spheres around Fe6 where the OLEF tend to be become significant, it uniformly points in the opposite direction than what is observed for Fe2, that is the field is directed toward the FeMoco. The relative radial isotropicity/consistency of the field surrounding Fe6 indicates that, here as well, the direction of the electric field direction is not caused by a single dominating residue but is the combined result of several residues surrounding the site.

Another important observation is that the combined effect of a (relatively weak) OLEF pointing toward Fe6, and a strong, long‐range OLEF pointing away from Fe2 induces a nonnegligible, and consistent, OLEF along the axis connecting both sites. Both for the e and d protonated form of His195, as well as when the charges of this residue are not considered, the field strengths along this axis vary from −0.2 to > −0.4 VÅ⁻¹ (for the doubly protonated form, the magnitude is reduced to a range between +0.1 and −0.2 VÅ⁻¹).

As noted in the Methodology Section, the trends described above are also recovered when electrostatic parameters derived from polarizable (AMOEBA) force field calculations are used to compute the electric fields (cf. Section S2 of the Supporting Information). Local distortions in the magnitude of the OLEFs are observed for some of the evaluation points directly adjacent to the atoms of the protein matrix, but most quivers are only slightly modulated;, for example, along the Fe2–Fe6 axis, the discrepancy in electric field magnitude between both descriptions amounts to a mere 10%.

Additionally, we also prepared an alternative version of the protocol in which crystal waters, as well as counterions are retained in the electric field quantification (see Section S4 of the Supporting Information). The exact same trends are recovered in this case as well, with modulation in the magnitude of the vectors limited to 2–3%. All these findings underscore the robustness of our findings discussed above.

It is important to note here that the computed patterns described above are not necessarily speculative in nature: they could in principle be validated experimentally, for example, through vibrational Stark effect measurements, cf. the previous work by Boxer and coworkers.^{[ 30} , ³¹ , ^{32 ]} This type of experimental validation would, however, require the installation of a vibrational probe, for example, a carbonyl or cyanide unit group on the respective metal sites of the FeMoco cofactor, which would bring with it a series of synthetic challenges. As such, direct experimental assessment of the local electric fields in FeMoco presumably remains, at present, out of reach. Nonetheless, the internal consistency of the computed trends across multiple protocols—and the growing body of theoretical and spectroscopic work connecting electric fields to enzymatic function^{[ 24} , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ^{34 ]}—provide compelling support for the reliability and relevance of the electrostatic features identified here.

Since the pronounced electric field patterns described above are robust, and not the result of a couple of specific residues, but caused by a collection of them, one can speculate that they may not have arisen purely by accident, but may have emerged over the course of evolution. For comparison, the electrostatic environment around another pair of Fe sites, Fe4, and Fe5, is displayed in Figure 5A (plots comparing the OLEFs around all pairs of Fe sites, as well as the Mo site, can be found in Section S5 of the Supporting Information). Note that this pair has previously been considered as plausible alternative coordination sites for N₂ in the computational work by Siegbahn and coworkers.^{[ 17 ]} Our electric field calculations indicate that no unequivocal electrostatic patterns are observable around these sites; patches of field vectors of varying magnitude, pointing in both directions, and attenuating at different rates, can be found around both. Additionally, only faint field vectors, with magnitudes of the order of ±0.1 VÅ⁻¹, are observed along the Fe4–Fe5 axis.

Figure 5.

A) OLEF maps for the Fe4 and Fe5 sites of the FeMoco for 3U7Q¹² for the ɛ monoprotonated form of His195. OLEF maps for the Fe2 and Fe6 sites of the FeMoco for a selection of alternative nitrogenases extracted from the PDB database: B) 1M1N,^{[ 53 ]} C) 8E3U,^{[ 54 ]} D) 2AFH,^{[ 55 ]} E) 7MCI.^{[ 56 ]} OLEF maps for the Fe2 and Fe6 sites for F) the FeV nitrogenase 5N6Y.^{[ 57 ]} To clearly highlight the robustness of the electrostatic patterns, the OLEFs were consistently calculated without taking the influence of (the equivalent residue to) His195 into account.

Additional evidence in support of our evolutionary hypothesis concerning the emergence of the OLEFs around the Fe2 and Fe6 sites in FeMoco can be found in the persistence of the observed electrostatic patterns across the entire nitrogenase enzyme family. To highlight this point, we analyzed the electrostatic environments of the potential N₂‐coordination sites for all 26 FeMoco nitrogenases found in the PDB database which passed our filtering protocol (cf. Figure 5B–E for some representative examples; interactive plots in html format for each enzyme, and for each protonation state of His195, can be downloaded from https://figshare.com/articles/dataset/Output_files_nitrogenase_project/28370183). From these plots, it is clear that the same qualitative trends are observed in all of them. Interestingly, for some of the nitrogenases, we observe that the field strength along the Fe2–Fe6 axis, is even higher than for 3U7Q,^{[ 12 ]} reaching values on the order of −0.5 VÅ⁻¹.

Additionally, we also considered the electrostatic environments for the 3 FeVco and 2 FeFeco nitrogenases that were extracted from the PDB database, and which passed our filtering workflow. While these alternative nitrogenases are clearly distinct in their second coordination sphere, we observe similar trends in terms of the electrostatic environments as for the FeMoco nitrogenases, that is an outward pointing OLEF around Fe2, and an inward pointing field around Fe6 (Figure 5F). On average, the inward pointing OLEF on Fe6 tends to be slightly more pronounced than for the FeMoco analogs. It is however interesting that this more pronounced field around Fe6 coincides with a further strengthening of the long‐range electric field aligned with the Fe2–Fe6 axis, now reaching magnitudes as high as −0.7 VÅ⁻¹.

3.2. What are the Potential Mechanistic Repercussions of These Electrostatic Patterns?

As indicated before, controversies surround the mechanistic details of N₂ fixation in nitrogenases, and even slight changes in the computational methodology have been reported to lead to markedly different conclusions. Hence, it is not our goal here to provide a definite answer on how the computed electric field patterns impact the catalysis of the M‐cluster. Instead, we aim to obtain some qualitative insights into which plausible mechanistic steps are the most susceptible to electric field modulation, and to obtain a general sense about the potential impact of these computed fields on the catalytic reaction profile. We will start our analysis from previously considered mechanistic hypotheses, though we will also consider a more exotic hypothesis at the very end.

Regardless of the exact sequence of elementary reaction steps, nitrogen fixation at the M‐cluster inevitably involves coordination of N₂ at some point throughout the nitrogen fixation mechanism. Simple DFT calculations suggest that it is highly unlikely that electric fields directly impact this initial binding event in a significant manner, since in isolation, N₂ has no intrinsic dipole moment, nor is it very susceptible to polarization: even applying an OEEF of −1 VÅ⁻¹ along the Fe─N₂ vector induces a dipole moment of only 0.75D, corresponding to <2 kcal/mol energetic stabilization (Figure 6A,B). A small dipole moment may in principle result from electron transfer to, or from, the M‐cluster upon coordination. However, due to the large HOMO–LUMO gap of isolated N₂, this is presumably a very minor effect: Head–Gordon and coworkers reported a transfer of a mere 0.03 e⁻ from (end‐on bonded) N₂ to the catalytic cluster in their advanced electronic structure calculations.^{[ 2 ]}

Figure 6.

Plots of A) the dipole moment and B) the change in the Gibbs Free energy, with respect to the field free situation, computed for N₂ and H₂S as a function of the magnitude of an OEEF aligned with the intrinsic dipole moment/bond axis, respectively.

The limited susceptibility of N₂ to long‐range electrostatics stands in contrast to the behavior of compounds with intrinsic dipole moments, even moderately polar ones. Let us consider H₂S, with an intrinsic dipole moment of 1.5D. Applying an OEEF of −1 VÅ⁻¹ aligned with its dipole vector induces energetic stabilization by over 8 kcal/mol. Of note is that for molecules with an intrinsic dipole, the direction of the field matters: applying a +1 VÅ⁻¹ OEEF to H₂S leads to an energetic destabilization of 3 kcal/mol.

Interestingly, the formation of a thiol end‐group — or an absorbed H₂S molecule — has been suggested, by some, as a critical preliminary step prior to N₂ coordination: rupture of the Fe2‐S2B‐Fe6 bridge in the M‐cluster results in the formation of a vacant Fe‐coordination site to which N₂ can bind more easily. The end result is then that the M‐cluster will carry two bonded/absorbed species: a thiol end‐group, or an absorbed H₂S moiety, on one Fe‐site, and N₂ absorbed on the other (cf. Figure 2C).^{[ 2 ]}

Based on the reasoning above, the protein matrix would then stabilize the configuration where N₂ ends up on the Fe2 site, and thiol/H₂S ends up on the Fe6 site, that is the situation obtained when the Fe6‐S2B bond breaks. Supporting evidence for this qualitative reasoning is provided by advanced QM/MM computations by Bjornsson et al. for 3U7Q.^{[ 16 ]} In this study, it was observed that when His195 is in its d monoprotonated form, the protein matrix induces a strong preference for N₂ coordination on Fe6; in the e monoprotonated form, exhibiting a reduced OLEF, this preference is significantly reduced. Notably, in Head–Gordon's work, which was based on a rather large cluster model, as well, a preference for cleavage of the Fe6‐S2B bond during the initial reduction was computed (note that in subsequent catalytic steps, they inadvertently display H₂S─Fe6 instead).^{[ 2 ]}

Next, we can turn to the potential role played by the persistent OLEF identified along the Fe2–Fe6 axis of all nitrogenases. A potential clue for why the protein matrices of nitrogenases are organized the way they are, can be found in Head–Gordon's recent study.^{[ 2 ]} After the initial N₂ coordination, their simulations indicate that the molecule changes orientation and takes on a bridging pose between Fe2 and Fe6 (cf. the right‐most structure in Figure 2C). It is in this position that the activation of N₂ is predicted to be maximized, and the first hydrogenation takes place.^{[ 2 ]} Note that this mechanistic hypothesis has striking similarities to the Frustrated Lewis Pair activation mechanism recently investigated by Shaik et al.,^{[ 37 ]} where an N₂ molecule binds in a bridging position between the Lewis acid and base, after which hydrogenation takes place (cf. Figure 3). In their simulations, an electric field of a mere 0.2–0.3 VÅ⁻¹ was sufficient to modulate the hydrogenation barrier by dozens of kcal/mol. Since our electric field calculations indicate that OLEFs with magnitudes ranging from −0.2 to −0.5 VÅ⁻¹ are present along the Fe2–Fe6 axis in the M‐cluster, it is reasonable to expect that the electrostatics could also markedly contribute to reduce this barrier.

To end this section, we would like to note that, building on the same FLP analogy invoked in the previous paragraph, an interesting alternative mechanistic hypothesis could also be put forward, which — as to the best of our knowledge — has not been considered in previous computational studies. More specifically, one could imagine that, if N₂ does end up coordination at the Fe2 site, then the subsequent hydrogenation reaction step could as well directly occur in this position: the nonprotonated imine nitrogen center of His195 could act as a Lewis base and Fe2 as a Lewis acid. The extremely strong computed OLEF — several times stronger than the maximal field strength considered in the work by Shaik and coworkers^{[ 37 ]} — around the Fe2 site could provide the necessary transition‐state stabilization for this FLP activation of dinitrogen (cf. Figure 2D). Since no in‐depth calculations have been performed on this alternative mechanism, this hypothesis is purely speculative for now. Nevertheless, we believe this to be an intriguing possibility, that would warrant further scrutiny. Overall, we believe that explicitly taking electrostatic environments imposed by the entire protein scaffold into consideration when constructing mechanistic hypotheses could be valuable, and provide a new way forward, for the investigation of complex enzymatic mechanisms.

4. Conclusion

In this study, we have investigated the role of electrostatics in the mechanism of nitrogenase‐catalyzed N₂ fixation, focusing on the local electric fields generated by the protein matrix surrounding the M‐cluster (i.e., FeMoco, FeVco, and FeFeco). Our analysis reveals that Fe2 and Fe6, the prime candidates for N₂ coordination and activation, experience markedly different electrostatic environments. The Fe2 site is surrounded by a strong and consistent outward‐pointing electric field, while the field around Fe6 tends to be weaker, and is oriented in the opposite direction. These patterns are recovered both with fixed‐charge and polarizable force‐fields, and they persist across all three M‐cluster types, suggesting an evolutionary adaptation rather than an incidental feature.

Connecting these electrostatic insights to recent computational studies, we highlight some possible explanations for why these electrostatic patterns emerged. First and foremost, we argue that the strong OLEF around Fe2 may not be there to directly stabilize N₂ coordination, but rather to promote the preliminary cleavage of the Fe2– m‐S₂B–Fe6 bridge, which frees a coordination site at Fe6 that subsequently enables N₂‐coordination. The emergence of a long‐range oriented electric field along the Fe2–Fe6 axis, in its turn, could then play a critical role in promoting the reactivity of the Fe2–(m₂:h²–N₂)–Fe6 binding mode, which corresponds to a key intermediate proposed by Head–Gordon and coworkers.^{[ 2 ]} We also propose a radically different alternative hypothesis, in which N₂ would actually bind on Fe2, after which hydrogenation could be facilitated by the His195, acting as a Lewis base, and the strong OLEF would contribute to stabilize the transition state toward N₂‐polarization and reduction.

It is important to underscore that advanced simulations will be needed to confirm the hypotheses outlined above, and this will be the subject of future work. Nevertheless, our exploratory study underscores the pervasiveness of (long‐range) electrostatics in enzymes, and that its role in the mechanism may sometimes be more complex and less intuitive than the traditional paradigm proscribes.

Note also that when a cluster model is applied to the modelling of nitrogenase activity, any medium‐ to long‐range contributions to the OLEF are by definition neglected, as even the most expansive clusters reported do not include any residues beyond His195. This suggests that while cluster models can be extremely valuable to characterize mechanistic hypotheses in a cost‐effective way, caution is required when the resulting barriers are interpreted quantitatively.

Overall, it is our hope that this study can contribute to a better understanding of the role played by electrostatic environments, which in turn could inform the design of synthetic nitrogen fixation catalysts that mimic nature's ability to control reaction pathways through exquisitely fine‐tuned electrostatic environments.

Supporting Information

Electric field quantification from AMOEBA force field output, OLEF patterns around Fe2 and Fe6 for the FeMoco unit of 3U7Q determined with polarizable force field parameters, OLEF patterns around the different metal sites of the FeMoco unit of 3U7Q, OLEF patterns around Fe2 and Fe6 for the FeMoco unit of 3U7Q when crystal waters and counterions are taken into account, OLEF patterns when removing residues in an increasing radius around the FeMoco unit of 3U7Q.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The Python code used for data analysis and visualization in this study is available on GitHub at https://github.com/chimie‐paristech‐CTM/electric_fields_nitrogenase. The dataset supporting the findings of this work has been deposited in Figshare and can be accessed at https://doi.org/10.6084/m9.figshare.28370183.v1.