|Home | About | Journals | Submit | Contact Us | Français|
“Nitrogen fixation”—the reduction of dinitrogen (N2) to two ammonia (NH3) molecules—by the Mo-dependent nitrogenase is essential for all life. Despite four decades of research, a daunting number of unanswered questions about the mechanism of nitrogenase make it the ‘Everest of enzymes’. This Account describes our efforts to climb one “face” of this mountain by meeting two interdependent challenges central to determining the mechanism of biological N2 reduction. The first challenge is to determine the reaction pathway: the composition and structure of each of the substrate-derived moieties bound to the catalytic FeMocofactor (FeMo-co) of the molybdenum-iron (MoFe) protein of nitrogenase. To overcome this challenge, we need to discriminate between the two classes of potential reaction pathways: 1) a “distal” (D) pathway, in which H atoms add sequentially at a single N or 2) an “alternating” (A) pathway, in which H atoms add alternately to the two N atoms of N2. Secondly, we need to characterize the dynamics of conversion among intermediates within the accepted Lowe-Thorneley kinetic scheme for N2 reduction. That goal requires us to experimentally determine both the number of electrons/protons delivered to the MoFe protein and their “inventory”—a partition into those residing on each of the reaction components and released as H2 or NH3.
The principal obstacle to this “climb” has been the inability to generate N2 reduction intermediates for characterization. A combination of genetic, biochemical, and spectroscopic approaches recently overcame this obstacle. These experiments identified one of the four-iron Fe-S faces of the active-site FeMo-cofactor as the specific site of reactivity, indicated that the sidechain of residue α70V controls access to this face, and supported the involvement of the sidechain of residue α195H in proton delivery. We can now freeze-quench trap N2 reduction pathway intermediates and use ENDOR/ESEEM spectroscopies to characterize them.
However, even successful trapping of a N2 reduction intermediate occurs without synchronous electron delivery to the MoFe protein. As a result, the number of electrons and protons, n, delivered to MoFe during its formation is unknown. To determine n and the electron inventory, we initially employed ENDOR spectroscopy to analyze the substrate moiety bound to the FeMo-co and 57Fe within the cofactor. Difficulties in using that approach led us to devise a robust kinetic protocol for determining n of a trapped intermediate.
This Account describes strategies that we have formulated to bring this “face” of the nitrogenase mechanism into view and afford approaches to its climb. Although the summit remains distant, we look forward to continued progress in the ascent.
‘Nitrogen fixation’ – the reduction of dinitrogen (N2) to two ammonia (NH3) molecules – is essential to all life, as it is the exclusive pathway for the input of nitrogen from N2 into the global biogeochemical N-cycle. The majority of N2 fixation occurs biologically, by the action of the microbial enzyme, nitrogenase. There are four known types of nitrogenase, primarily distinguished by the metal composition of their active-site metalloclusters.1 The most important is the Mo-dependent enzyme2 on which this Account focuses, henceforth denoted, 'nitrogenase'. It comprises two components, denoted the MoFe and Fe proteins. The former contains the active-site, multimetallic iron-molybdenum cofactor (FeMo-co) cluster as well as an auxiliary 'P' cluster, presumed to mediate electron transfer from Fe protein to FeMo-co. The Fe protein contains a [4Fe-4S] cluster and delivers electrons one-at-a-time to the MoFe protein.
Nitrogenase has been subjected to four decades of intensive kinetic3,4 and spectroscopic studies,2 and structures have been determined for MoFe and Fe proteins and for multiple forms of their complex.5 Given the remarkable insights these studies provide into nitrogenase structure and function, one might expect that we would be well on the way to understanding the full mechanism of biological N2 fixation. However, as outlined by Howard and Rees6 and Schrock,7 a daunting number of mechanistic questions remain unanswered, in our view, making nitrogenase the 'Everest of enzymes’. This Account describes our efforts to climb one ‘face’ of this mountain by meeting two interdependent challenges that are central to determining the mechanism of biological N2 reduction. The first is to determine the reaction pathway for N2 reduction. The second challenge is to characterize the dynamics of conversion among intermediates by incorporating them into a ‘multidimensional’ formulation of the accepted Lowe-Thorneley (LT) kinetic scheme for N2 reduction.2–4 This task in turn requires an experimental determination for each intermediate of n, the number of electrons/protons delivered to MoFe protein, as well as the ‘electron/proton inventory route’ for catalysis, a partition of each intermediate’s n electrons/protons into those residing on FeMo-co, those residing on the substrate-derived moiety bound to it, and those released as H2 or NH3. These efforts seek some of the enzyme’s most closely held secrets. This Account both describes the barriers to ascent and summarizes recent progress8–10 in experiment and concept that are slowly advancing us towards the summit.
NH3 formation by nitrogenase proceeds through a series of intermediates in which N2 and its reduced forms bind to FeMo-co.2,3 However, during the first four decades of studying purified nitrogenase, no reduction intermediate was trapped for characterization, although intermediates that accumulate during CO inhibition were trapped early on.11,12 Instead, the 1970s and 1980s witnessed the formulation of a kinetic scheme for substrate reduction, with some insights into the nature of intermediates being provided by studies of pre-steady-state kinetics. These efforts culminated in the LT kinetic scheme,2–4 which has proven to satisfactorily describe the kinetics of nitrogenase catalysis in full.
Such efforts surprisingly showed that nitrogen fixation under optimized conditions exhibits the reaction stoichiometry,
Thus, the enzymatic reaction is both expensive, requiring the hydrolysis of two MgATP/reducing equivalent, and curious, in that eight reducing equivalents, not six, are needed to reduce each N2 to two NH3, with two equivalents being 'wasted' through the evolution of H2.2,3
The LT scheme, shown in part in Fig 1, is formulated in terms of states, denoted En, that are indexed by the number of electrons and protons, n, that have been delivered to the MoFe protein during their formation: n = 0 (resting) ↔ 8 (as required by eq 1); it is characterized by the rate constants for transformations among those states. Particular features of note include the following: during catalysis the Fe protein delivers one electron at a time to the MoFe protein as its 4Fe-4S cluster cycles between (1+) and (2+) oxidation states; protein complex formation and single-electron transfer from Fe protein to MoFe protein is driven by the binding and hydrolysis of two MgATP within the Fe protein; the release of the Fe protein after delivery of its electron is the rate-limiting step of catalysis.3 N2 reduction involves eight of these cycles to deliver the eight electrons necessary for substrate binding and reduction, eq 1.
There is little evidence for binding of any substrate to the resting state of the MoFe protein, and N2 does not appear to bind until the MoFe protein has been activated by the accumulation of three or four electrons and protons (E3 or E4).3 In the absence of other substrates, activation of MoFe protein leads to the reduction of protons to form H2, with the protein thereby cycling back to its resting state. In the presence of N2, this hydrogenase capability leads to a competition between H+ and N2 reduction. As a result, even the optimum stoichiometry of one H2 formed per N2 reduced (eq 1) represents only a lower limit.
The X-ray structure of MoFe protein was first determined sixteen years ago.5 It revealed the active-site FeMo-cofactor to be an unprecedented [Fe7S9Mo; homocitrate] cluster, Fig 2, with roughly trigonal symmetry along the terminal Fe-Mo axis. Subsequently, refinement showed the presence in FeMo-co of an interstitial first-row atom, X = C, N, or O, making its final composition, [Fe7S9MoX; homocitrate],13 and structures were obtained for multiple [Fe; MoFe] protein complexes.5
The structure of FeMo-co (Fig 2) was a major surprise, and its determination provides the foundation for any discussion of mechanism. But in some ways it deepened the mystery. Firstly, examination of the resting-state structure does not reveal the site of N2 binding/reduction! Mo is an obvious candidate for the active site, as it is incorporated in the only known inorganic metal complexes that catalytically reduce N2.14 However, Fe is an equally attractive candidate, given that it is the catalytic metal in the commercial Haber-Bosch process for NH3 formation and that there are V and Fe nitrogenases that reduce N2 but do not have Mo.15 Then, if Fe is involved, the structure does not indicate which one of the three 4Fe-4S 'faces' of the central [Fe6] iron prismane is active, much less identify the active Fe(s). In addition, there are such questions as: why homocitrate;16 and what is X?17
Thus, at the beginning of this millennium we knew the structure of nitrogenase in its resting state, but not where catalysis occurred within that structure. Kinetics measurements culminating in the LT scheme provided an excellent understanding of how nitrogenase converts among the En intermediates during N2 reduction, but we had absolutely no experimental evidence as to the identity and structure of any of these intermediates.
The LT scheme reveals that the major obstacle to generating an En intermediate in sufficient amounts for characterization, and thus perhpas the reason for such an unsatisfying state of affairs, is that the MoFe protein acquires electrons one-at-a-time from its partner Fe protein and this process involves the nucleotide-dependent association and dissociation of the two proteins.2 As a consequence, multiple electron additions to MoFe protein cannot be synchronized, for complex dissociation represents the rate-limiting step of substrate reduction,2,3 while electron accumulation is opposed by the competing loss of H2. EPR has been the primary tool in monitoring nitrogenase under turnover, and efforts to trap turnover intermediates for study indeed showed loss of the resting-state FeMo-co EPR signal under turnover, indicating that it has converted to intermediate En states. However, no signals from such states appeared in its place, indicating that the system distributes itself among multiple intermediates and/or that the preponderant states are EPR-silent. This barrier to study contrasts, for example, with the ready use of flash photolysis in photosynthesis to synchronously generate a particular desired intermediate for study.18
A way up this face was revealed9,10 in a series of studies that combined genetic, biochemical, and spectroscopic approaches that: (i) identified the specific site of reactivity as the 4Fe-4S face of FeMo-cofactor defined by Fe atoms 2, 3, 6, and 7 (numbering based on pdb 1M1N); (ii) demonstrated that the sidechain of residue α70V acts as a 'gatekeeper' that controls access to the FeMo-co active site; and (iii) supported the idea that proton delivery involves the sidechain of residue α195H.10,19,20 These structural features all are shown in Fig 2.
These conclusions stem from the observation that decreasing the size of the α70V residue by substitution with alanine allows the enzymatic reduction of molecules excluded from the wild type (WT) active site, such as propyne, propargyl alcohol, and 1-butyne, while increasing its size by substitution with isoleucine represses reaction of all WT substrates except H+.9,10 In both variants, the change leaves the reactivity of FeMo-co unaltered. These observations were accompanied by the discovery that reduction intermediates could be trapped in MoFe variants with mutations at α70V and/or α195H, and subsequently in WT enzyme itself. 10 This discovery open a trail to the summit.
Beginning in 2004, a series of reports described the first freeze-quench trapping and characterization of enzymatic intermediates,8,9 initially ones that form during the reduction of alkyne substrates,21,22 then one formed during the reduction of H+ under Ar,23 then four intermediates associated with N2 reduction.24–27 In all but one of these cases a key was the use of MoFe proteins that contain the amino acid substitutions, separately or in combination, that inhibit proton delivery to substrates (α-195Gln) and modulate substrate accessibility to the reduction site (α-70Ala). The trapping of an intermediate during N2 reduction by WT nitrogenase supports the view that these intermediates reflect normal nitrogenase function. They are being characterized biochemically, and as is now illustrated, their structures are being determined by the spectroscopic methods of choice, electron-nuclear double resonance (ENDOR) and electron spin-echo envelope modulation (ESEEM).28,29
The first use of ENDOR to study any nitrogenase intermediate employed 57Fe to prove that the EPR signals from the two CO-inhibited forms of the MoFe protein arise from the FeMo-cofactor, not the P cluster,30 used 13C ENDOR to demonstrate that one CO binds to the ‘lo-CO’ state formed under low CO pressure while two bind to ‘hi-CO’, and offered proposals for the CO binding modes.31
The first determination of the structure of a nitrogenase catalytic intermediate combined 13C and 1,2H Mims 35 GHz ENDOR techniques to reveal the structure of the substrate-derived moiety bound to FeMo-co in the intermediate formed during reduction of the alkynes propargyl alcohol (PA; HCCHCH2OH)21 and HCCH22 by MoFe variants. This study provides the 'template' for the ongoing applications of ENDOR/ESEEM spectroscopies to determine structure in such a complicated system. These techniques in general do not provide sufficient information with which to directly deduce the nature and binding geometry of the substratederived moiety, and quantum mechanical computations on FeMo-co are not advanced enough to unambiguously interpret the hyperfine/quadrupole couplings that are measured. Instead, the measurements supply constraints against which possible models can be tested and the correct model selected.
In the case of the PA intermediate,21 13C ENDOR showed that the PA backbone remains intact and is covalently bonded to FeMo-co, and that no two of the backbone 13C are equivalent. Thus, there are only three possible species that might be bound to FeMo-co: PA itself; singlyreduced PA; the alkene product (allyl alcohol), Eq 2.
1,2H ENDOR further disclosed that the terminal carbon (C3) of PA is bonded to a non-exchangeable H of the reactant and a solvent-derived H, and that the two are symmetry equivalent through a reflection in a g-tensor plane; in addition, there is another, more weakly-coupled, solvent-derived H, indicated in red. Upon testing these constraints against every relevant structure in the Cambridge Structural Database, we reached the remarkable conclusion that the intermediate contains a bio-organometallic complex – the alkene reduction product, allyl alcohol, bound 'side-on' to one Fe of FeMo-co (Fig 3) – and a similar conclusion was shown to apply to the corresponding intermediate trapped during C2H2 reduction.22 Subsequent experiments and DFT computations supported this conclusion.9
PA is not a substrate of WT nitrogenase, but becomes one when substrate access is expanded by reducing the size of the α-V70 side-chain through the α-V70A amino acid substitution. If the size of this side-chain is instead increased through the α-V70I substitution the result is a MoFe protein variant effective only in the reduction of 2H+ to H2. During turnover under Ar this variant accumulates a novel intermediate that was shown by 1,2H-ENDOR to have two chemically equivalent hydrides bound to metal ions of FeMo-co;23 the central role of this intermediate in N2 reduction is discussed below.
A molecular mechanism of N2 reduction must include three constituents: (i) the 'Reaction Pathway' – an enumeration of the states that arise during the conversion of reactants to products, (ii) the molecular structures of these intermediates; and (iii) a kinetic scheme, in this case the LT scheme, that incorporates information about the kinetics/dynamics of conversion among these intermediates. Considering constituent (i), we recently formalized the division of N2 reduction pathways into two classes. In a 'distal' (D) pathway, H-atoms add sequentially to a single N prior to N-N bond cleavage after the third hydrogenation, Fig 4. Catalytic NH3 formation at mononuclear Mo metal complexes follows such a pathway, as shown by the early work of Chatt and coworkers32 and recently by Schrock and coworkers.14 This contrasts with an 'alternating' (A) pathway in which H-atoms add alternately to the two N atoms of N2 before N-N bond cleavage at the fifth hydrogenation, Fig 4, often favored in computational studies, for example 33. Interestingly, current discussions of possible reduction mechanisms have tended to support one or the other class without explicitly noting that there are two classes, although this was well understood in early days.34
Until recently, the limited information about the nitrogenase pathway was predominanatly derived from kinetics studies, and these were interpreted as supporting the D version.3 As the substrate-derived species bound to FeMo-co along the alternative pathways of N2 reduction are different at many stages of N2 fixation, Fig 4, the enzyme’s pathway would be revealed if these species could be characterized by ENDOR/ESEEM. Thus it was of pivotal importance when in 2004 we first reported the successful trapping of intermediates associated with N2 reduction.25 To date, four such S = ½ states have been trapped:9,10,24–27 e(N2), an early stage in reduction obtained from wild type MoFe protein with N2 as substrate;25 m(NH=N-CH3), from α195Gln MoFe protein with CH3-N=NH as substrate;25,26 m(N2H2) prepared27 by freeze-quenching the MoFe protein α70Ala/α195Gln variant during steady-state turnover with in situ generated diazene; and l(N2H4), from α70Ala/α195Gln MoFe protein with H2N-NH2 as substrate.24,25
A combination of 15N and 1H ENDOR measurements (Fig 5) showed that three of the intermediates l(N2H4), m(N2H2), and m(NH=N-CH3) contain a substrate-derived [−NHx] moiety bound to the cofactor.24–27 In contrast, FeMo-co of e(N2) appears to bind a non-protonated N, which would indicate that e(N2) represents an earlier step in the catalytic pathway.25 Recent 14,15N and 1H ENDOR measurements showed that the substrate-derived moiety in turnover states trapped with diazene and hydrazine substrates have extremely similar characteristics.27 Hydrazine is well known to be a bona fide nitrogenase substrate35 and biochemical experiments showed that the same holds for diazene.27 Together, the above observations indicate that diazene enters the normal N2 reduction reaction pathway (Fig 4) and further suggest that during turnover and freeze-quench, diazene is reduced by the enzyme until it 'catches up' to the species trapped during turnover with hydrazine, with the two substrates thus generating a common intermediate that falls along the normal pathway of N2 reduction by nitrogenase. If this speculation proves to be correct, considerations of Fig 4 show that this interpretation favors an A mechanism.
Full characterization of these, possibly common, intermediates and of the others by ENDOR and by relaxation methods discussed below would definitively test this suggestion. However, the use of 1,2H and 14,15N ENDOR/ESEEM to determine the structures of the trapped nitrogenous reduction intermediates is a much more formidable task than was the determination of the structure of the alkyne reduction intermediates. Firstly, there are six stages of N2 reduction, not two (Eq 1). Secondly, when alternate A/D pathways are taken into account (Fig 4), there are numerous possible [NxHy] species to be considered at every stage of N2 reduction; this is illustrated by Chart 1 for later stages. Thirdly, one has less control over the placement of nuclear-spin probes. As the N2Hm (m = 0, 2, 4) substrates used to generate intermediates are symmetric, unlike PA, selective labeling of the N's thus is impossible—except with the unnatural substrate NH=N-CH3;26 all the N-H are solvent exchangeable, so selective deuteration as with PA is impossible. On the other hand, just a very few ENDOR-derived constraints can act as powerful means of narrowing the list of candidates. To characterize a [NyHx] cofactor-bound moiety one must answer the questions: (i) is the N-N bond broken (y =1 or 2), and if not, are the two N's equivalent; (ii) what is the reduction level of the bound fragment (what is x); (iii) what is the binding mode if y = 2? Some of these constraints will come directly from ENDOR results for the intermediates, as complemented and strengthened by comparisons36 with newly emerging biomimetic complexes,37,38 but some will come from a relaxation-kinetics protocol to be described below.
The nitrogenase reaction pathway is specified by determination of the composition and structure of the intermediates of N2 reduction, Fig 4. However, even when a N2 reduction intermediate has been trapped, this occurs without synchronous electron delivery to the MoFe protein. As a result, the number of electrons and protons, n, delivered to MoFe during its formation is unknown and the intermediate is untethered from the LT kinetic scheme for substrate reduction. Thus, n for the intermediate must be determined, but beyond this, one must determine how those electrons/protons are partitioned: are they on FeMo-co, P cluster, or substrate? One may loosely define an ‘N2 reduction mechanism’ as comprising complete answers to three conjoined questions for all intermediates: composition/structure; the value of n; and the partition of n. The next section discusses efforts to address the second and third of these.
The first effort to determine n for an intermediate focused on those generated during alkyne reduction. It introduced the concept of a MoFe protein 'electron inventory',22 seeking to determine the number (n) of electrons and protons that have been delivered to the MoFe protein to form an En state by partitioning n into distinct categories and determining the number in each: those that reside on the substrate-derived moiety bound to FeMo-co (s); those that remain on FeMo-co (m); the number of electrons/protons that have been 'released' (r) through liberation of H2 and reduced substrate (eg., NH3); further allowing for the possibility that P cluster has donated (p) electrons to FeMo-co. The index, n, thus can be written as the sum over categories,22
For the alkyne intermediates, r = p = 0, so n = m + s. The idea was to use ENDOR of nuclei associated with the substrate-derived moiety bound to FeMo-co to determine s and 57Fe ENDOR of FeMo-co to determine m.
The inventory concept implicitly recognizes that the single index, n, is inadequate to specify the properties of an intermediate, and that additional indices are needed. For example, as in other formal valency schemes, the inventory depends on the nature of the binding of substrate-derived fragments to FeMo-co. Consider the En intermediate that contains the ethylene product of acetylene reduction bound to FeMo-co. If ethylene acts as a dative π-donor (π-C2H4; Fig 6), this binding does not alter the formal reduction level of the FeMo-co (m) or the alkene (s = 2), and the complex has an electron inventory, n = m + s = m + 2. However, one might instead imagine that C2H4 actually binds through C-Fe σ-bonds as the ferracyclopropane (σ-C2H4; Fig 6). Such a structure corresponds to oxidative addition of C2H4 to the FeMo-co, and in this case the alkene must be described formally as having accepted two additional electrons from FeMo-co, making a total of s = 4 electrons transferred to the initial C2H2 substrate. As a result, n = m + 4. Analysis of 57Fe ENDOR measurements of the acetylene-reduction intermediate, in combination with a variety of other considerations, led us to prefer n = 4 for this intermediate, compatible with the finding of Lowe et al. that C2H4 is released from the E3 and E4 states during C2H2 reduction by WT enzyme.40 Thus, depending upon the binding mode, the E4 MoFe protein would have accumulated four electrons, but FeMo-co of this intermediate could formally retain m = 2 electrons (π-C2H4; Fig 6), or none, m = 0 (σ-C2H4). This proposed n will be tested with a kinetic procedure described below.
Once one introduces the concepts of electron and proton inventories for intermediates, it becomes clear that there are not just two alternative pathways for the sequential addition of electrons/protons to substrate, D and A, (Fig 4), but each represents a multiplicity. For example, according to the LT scheme (Fig 1), N2 binding to the E4 state of nitrogenase occurs with loss of two electrons/protons as H2, leaving N2-bound FeMo-co reductively activated by the remaining m = 2 electrons. But these electrons can in principle be delivered to substrate at any subsequent stage in the reaction, and thus the A and D pathways in Fig 4 each really represents many possible 'electron inventory routes' of reduction. As one limiting route, transfer of the m = 2 electrons and protons to N2 could occur Promptly (P), to form bound diazene; in the opposite limit, these electrons might remain on FeMo-co until the final, Late (L), stages of N2 reduction; and of course there are intermediate cases. To describe the alternative intermediate states embodied in these multiple pathways/routes requires other 'quantum numbers' in addition to n. In this light, the LT scheme as typically drawn (eg., Fig 1) is best viewed as a 1D projection of a 'multi-dimensional' kinetic scheme.
One way to represent such a scheme is to extend the notation for an En intermediate formed during N2 reduction so as to include indices that specify the electron (e) inventory (eq 3): En[m, s, r, p]e. However, as indicated above, an electron inventory in general is as much a formal construct, analogous to formal atomic valencies in a molecule, as an experimental one. Instead, it seems more instructive to denote an En intermediate by its 'proton (H) inventory'. As each electron delivered to MoFe protein is accompanied by a proton, n = n(e−) = n(H+), we may denote the multiple possible intermediates at the En stage of reduction by specifying hm, the number of protons that remain bound to cofactor as protonated sulfide or hydride, hs, the number that are associated with the substrate-derived moiety bound to FeMo-co, and r, the number of protons/electrons that have been released as H2 or reduced substrate/fragment. The proton inventory analogue to eq3 then equates sum of these protons to n,
where we further define the total number of protons that have been transferred to substrate, regardless of their fate: ht = [hs+ r]. For clarity we will sometimes denote an intermediate as, En[hm, hs, r]H, but in fact, for a given n the alternate intermediates can be specified with only two ‘quantum number’: [hm, hs], [hm, r], or [ht, r]; the remaining index is fixed by Eq 4. For completeness, one should further incorporate into the notation the substrate-derived species (S) bound to FeMo-co, leading to a notation in which the alternate intermediates associated with a given value of n may be written in terms of the three inventory indices, En[hm, hs, r; S]H, or more economically in terms of two, as En[ht, r; S]H, etc. An advantage of this approach is that it is not merely formal: a proton on FeMo-co is a 'classical' particle; it really is bound either to an atom of substrate or one of FeMo-co.
Fig 7 presents a 3D proton inventory plot that that displays the four limiting alternate routes of N2 reduction that involve N2 binding to E4 with release of H2: D vs A pathways; each with either Prompt (P) or Late (L) transfer to substrate of the m = 2 electrons/protons that remained on MoFe after N2 binding. The plot omits the corresponding routes associated with N2 binding at E3 because their inclusion would make it unreadable. In this plot, the alternate proton inventories available for En are denoted, En[ht, r; S]H, with a particular inventory route represented by a series of points for which the abscissa is n, the ordinate is ht, and the z coordinate is r, the number of protons (and electrons) released as H2 and NH3 during catalysis.
The routes displayed begin with the accumulation of n = 4 electrons/proton to generate the ‘hydride’ intermediate activated for N2 binding, E4 [ht = 0, r = 0]H, and hence m = 4 from Eq 4. Next, N2 binding and the accompanying release of H2 (r = 2) generates E4 [0, 2]H, with m = 2. The P/L routes then diverge, each subsequently splitting into D/A branches. Release of the first NH3 (r = 5) occurs at a different state on each of the four branches; the four finally converge to release the second NH3 (r = 8). The volume enclosed by these limiting paths contains the points for states that would arise in a 'mixed' pathway.
This approach also helps in thinking about such enduring mysteries as: why is release of H2 an obligatory feature of N2 binding (eq 1); why must N2 bind to MoFe protein that has acquired n = 4 electrons/protons (Fig 4), written most descriptively with the three inventory indices as state E4[hm, hs, r; −]H = E4[4, 0, 0; −]H, when the accompanying loss of H2 yields an N2-bound state in which MoFe protein retains only two equivalents, E4[2, 0, 2; N2]H, and analogously for binding to n = 3? Consider the ‘overpotential’ requirement, that N2 bind to a state with n > 2. If it were to be shown that catalysis follows a P route in which, for example, E4[4, 0, 0; −]H binds N2 and loses H2 to form E4[2, 0, 2; N2]H, with prompt followup reduction of N2 to generate E4[0, 2, 2;N2H2]H, then one would interpret this overpotential as reflecting a need to reductively stabilize bound N2 On the other hand, if an L route is operative, for example with the e(N2) intermediate corresponding to E4[2, 0, 2; N2]H, then one might speculate that the required high level of reduction (n = 4 or 3) is accompanied by a structural rearrangement of the FeMo-cofactor that is required for N2 binding.
The use of an 'electron inventory' to determine n for intermediates other than those trapped during alkyne reduction was blocked by several difficulties, one being that the effects of slow 57Fe nuclear relaxation made it difficult to analyze 57Fe ENDOR spectra. This led us to devise a robust new kinetic protocol for determining the En state of a trapped intermediate.39 It is founded on the recognition that no matter how an intermediate state of the MoFe protein has been trapped, it has accumulated a specific number of electrons, n, and it is in a specific En state. Thus, it acts as a 'synchronously prepared' initial state. In particular, we hypothesized that n for an intermediate state early in the kinetic scheme (Fig 1) could be revealed by following its 'synchronous' relaxation back to the resting state (E0) through the loss of one or more equivalents of H2, and ‘simply’ counting the number of such Δn = 2 steps.
This approach, of course, requires that no additional electrons transfer from Fe protein to MoFe protein during the relaxation, which raises the question: How can one 'turn off' intercomponent electron transfer without perturbing the MoFe protein in a sample that retains both the reducing equivalents and ATP required for this electron transfer? The answer rests in the fact that each ET from the Fe protein to the MoFe protein absolutely requires dissociation/association of the Fe-MoFe protein complex.2,3 We recognized that this electron transfer would be abolished if Fe-MoFe association/dissociation were prevented by keeping the sample frozen, while we knew from our studies of heme monooxygenases41 that reactions within the MoFe protein can proceed in the frozen state. In the resulting relaxation protocol, a 77 K freeze-trapped intermediate relaxes in the solid (typically at 253K) without additional intercomponent electron transfer, while reaction progress is monitored periodically by cooling the sample to 2K for EPR measurements of the EPR-active species, at which temperature no reactions occur.
With this protocol we showed that the hydride intermediate (A) relaxes to the resting state (C) in two steps, A B C. Both steps show appreciable solvent kinetic isotope effects, KIE ~ 3–4 (85% D2O) (Fig 8), assigned to the generation of H2 during relaxation. According to the LT scheme, Fig 1, this intermediate must then be the catalytically central E4 state that has been activated for N2 binding and reduction by the accumulation of four electrons. Its relaxation by loss of H2 generates the state B = E2, and this relaxes with loss of H2 to the resting state, C = E0.
We propose that the combination of this relaxation protocol with ENDOR/ESEEM studies of structure offers a real prospect of connecting trapped En intermediates to mechanism. Beyond that, it will help to characterize individual intermediates and enable us to distinguish among the multiplicity of reaction pathways that are revealed when considering alternate 'inventory routes', as in Fig 7. For example, if the same intermediate is trapped during turnover with diazene and hydrazine, then the two should show the same relaxation behavior.
This Account has discussed progress and prospects for determination of an ‘N2 reduction mechanism’ defined as comprising: the reaction pathway detailing the composition and structure of intermediates; the substrate reduction ‘route’ that links those intermediates to kinetics through the determination of n and the electron inventory for each. As noted above, the Perspectives of Schrock7 and of Howard and Rees6 provide a context for this work and discuss challenges not addressed in this Account -- the other unclimbed ‘faces of the nitrogenase mountain’. Three issues noted there are of particular relevance to the efforts described here. (i) Do the bound intermediates 'migrate' over FeMo-co during catalysis and is Mo directly involved at some stage?7,9 (ii) What is X (Fig 1)?17 (iii) How does the P cluster participate?42 The first two of these can be addressed by ENDOR studies of trapped intermediates, the third by relaxation measurements. But such challenges will require much ‘climbing’ by us and others before they can be discussed in some future Account.
At the start of this milennium the ‘face’ of nitrogenase discussed here was shrouded in clouds. This discussion has argued that strategies have been formulated that bring this face into view and afford approaches to its climb. This does not mean that this summit will soon be reached, but rather that vigorous climbing will be rewarded.
BMH thanks Prof. R. H. Holm for insights into the mysteries of ‘mechanism’. This work has been supported by the National Institutes of Health (HL13531 [BMH], GM59087 [LS and DRD] and the National Science Foundation (MCB 0723330 [BMH]) and MCB 071770 [DRD]).
Brian M. Hoffman was an undergraduate at the University of Chicago, received his Ph.D. from Caltech, and spent a postdoctoral year at MIT. From there he went to Northwestern University, where he is the Charles E. and Emma H. Morrison Professor in the Departments of Chemistry and of Biochemistry Molecular Biology & Cell Biology.
Lance C. Seefeldt received a B.S. degree from the University of Redlands and a Ph.D. from the University of California at Riverside. He was a postdoctoral fellow in the Center for Metalloenzyme Studies at the University of Georgia and is now Professor of Chemistry and Biochemistry at Utah State University.
Dennis R. Dean received a B.A. from Wabash College and a Ph.D. from Purdue University. Currently he is the Stroobants Professor of Biotechnology at Virginia Tech, where he also serves as the Director of the Fralin Life Science Institute and Associate Director of the VT-Carilion Medical Research Institute.