7.8: Multisite Redox Enzymes (Part 4)

7. EPR, ENDOR, and ESEEM Studies

The FeMoco or M center has been identified spectroscopically within the FeMo protein;^239,248,273 it has a distinctive EPR signal with effective g values of 4.3, 3.7, and 2.01, and originates from an S = \(\frac{3}{2}\) state of the M center. The signal arises from transitions within the ±\(\frac{1}{2}\) ground-state Kramers doublet of the S = \(\frac{3}{2}\) system (D = +5.1 cm^-1, E/D = 0.04). The isolated cofactor (FeMoco) gives a similar EPR signal, but with a rather larger rhombicity (E/D = 0.12). Spectra from the C. pasteurianum nitrogenase and cofactor are shown in Figure 7.28, and comparative data are given in Table 7.7. The M-center EPR signal has proved useful in characterizing the nature of the site, especially when more sophisticated magnetic resonance techniques, such as ENDOR or ESEEM, are used.

Extensive ENDOR investigations^274,275,275a have been reported using protein samples enriched with the stable magnetic isotopes ²H, ³³S, ⁵⁷Fe, ⁹⁵Mo, and ⁹⁷Mo. The ⁵⁷Fe couplings have been investigated in the most detail. Individual hyperfine tensors of five coupled ⁵⁷Fe nuclei are discernible, and were evaluated by simulation of the polycrystalline ENDOR spectrum.²⁷⁵ The data from ³³S and ⁹⁵Mo were analyzed in less detail; ³³S gave a complex ENDOR spectrum, evidently with quite large hyperfine couplings, although no quantification was attempted because of the complexity of the spectrum.²⁷⁴ On the other hand, ⁹⁵Mo was shown to possess a small hyperfine coupling, indicating that the molybdenum possesses very little spin density (although the quantitative aspects of the conclusions of the ⁹⁵Mo ENDOR study have recently been shown to be in error²⁷⁶).

Although no nitrogen splittings were reported in any of the ENDOR studies, evidence for involvement of nitrogen as a cluster component has been forthcoming from ESEEM spectroscopy.^{277-279 14}N modulations are observed in the ESEEM of the M center. The observed ¹⁴N is not from the substrate (N₂), or from an intermediate or product of nitrogen fixation, because enzyme turnover using ¹⁵N as a substrate does not change the ESEEM spectrum. The isolated cofactor (FeMoco) does not show the modulation frequencies observed for the M center in the protein. These experiments suggest that the M-center ¹⁴N ESEEM arises from a nitrogen atom that is associated with the M center, and probably from an amino-acid side chain (most likely a histidine) ligated to the cluster.²⁷⁹ Recent evidence from site-directed mutagenesis of the Azotobacter vinelandii protein²⁸⁰ provides strong support for the presence of histidine ligation, and points specifically to His-195 of the \(\alpha\) subunit as the N ligand.

8. Mössbauer Studies

Extensive Mossbauer investigations of nitrogenase^271,281-283 and FeMoco^283a have been reported. Unlike EPR and EPR-based spectroscopies, which can be used to investigate only the EPR-active S = \(\frac{3}{2}\) oxidation state, all three available M-center oxidation states are accessible to Mössbauer spectroscopy. The fully reduced site was found to be diamagnetic with S = 0 (but see Reference 284), whereas the oxidized site was found to have S \(\geq\) 1. The zero-field spectrum of reduced C. pasteurianum nitrogenase is shown in Figure 7.29; the spectrum is comprised of four quadrupole doublets, one of which was concluded to originate from the M site.²⁸² Mössbauer spectra taken in the presence of applied magnetic fields were used to deduce the presence of four types of ⁵⁷Fe hyperfine coupling; these were called sites A1, A2, and A3, which have negative hyperfine couplings, and B sites, which have positive hyperfine couplings. The A sites were quantitated as a single Fe each; the B sites were estimated to contain three irons. These conclusions were largely confirmed and extended by later ENDOR investigations,²⁷⁴ although the B sites were resolved as two inequivalent, rather than three equivalent, sites. ENDOR is rather more sensitive to the nature of the hyperfine couplings than Mössbauer, although it cannot usually be used to count numbers of exactly equivalent sites. Thus the number of iron atoms in the M center is minimally five, although larger numbers cannot be excluded. Note also that some of the quantitative aspects of the earlier Mossbauer investigations have been criticized.²⁸⁵

Figure 7.29 - Mössbauer spectrum of C. pasteurianum nitrogenase FeMo protein,²⁸² indicating the various components (quadrupole doublets) and their assignments. The doublet labeled M is the cofactor signal; those labeled D, S, and Fe²⁺ are attributed to the P-clusters.

9. X-ray Absorption Studies

One of the early triumphs of biological x-ray absorption spectroscopy was the deduction that the nitrogenase M center is an Mo-Fe-S cluster.²⁸⁶ (It is also worth noting that nitrogenase was the first enzyme to be studied by x-ray absorption spectroscopy.) Early work on lyophilized protein samples indicated the presence of two major contributions to the Mo K-edge EXAFS, which were attributed to Mo-S ligands, plus a more distant Mo-Fe contribution.²⁸⁶ Subsequently, these conclusions have been confirmed and extended, using samples in solution and with much more sensitive detection systems.

Most EXAFS studies to date have been on the molybdenum K-edge of the protein or of FeMoco, and indicate a very similar Mo environment in both (Table 7.7, Figure 7.30). A consensus of the best available analyses²⁸⁷ indicates that Mo is coordinated by three or four sulfur atoms at 2.4 Å, one to three oxygens or nitrogens at 2.2 Å, with approximately three nearby iron atoms at 2.7 Å. Of these, the EXAFS evidence for the oxygen/nitrogen contribution is weakest. However, comparison of Mo K-edge²⁸⁸ and Mo L-edge XANES²⁸⁹ spectra with model compounds indicates strong similarities with MoFe₃S₄ thiocubane model compounds possessing MoS₃O₃ coordination, and provides some support for the presence of O/N ligands.

Figure 7.30 - Mo K-edge EXAFS spectrum (left panel) and EXAFS Fourier transform (right panel) of Klebsiella pneumoniae nitrogenase MoFe protein. The solid line is the processed experimental spectrum and the dashed line a calculated one. ²⁸⁷

The iron EXAFS of FeMoco has been independently examined by two groups.^290,291 Both groups agree that the iron is coordinated largely to sulfur at about 2.2 Å, with more distant Fe-Fe interactions at about 2.6 Å. They differ, however, concerning the presence of short (1.8 Å) Fe-O interactions. Such interactions were apparently observed in the earlier study,²⁹⁰ but not in the later study.²⁹¹ One possible explanation for this discrepancy is that the short Fe-O interactions of the earlier study were due to extraneous iron coordinated to solvent, contaminating the FeMoco preparation.²⁹¹ A final resolution of this discord must, however, await the results of further experiments. Interestingly, a long Fe-Fe interaction at 3.7 Å was also observed in the later study.²⁹¹

Largely on the basis of the Mo K-edge EXAFS results and model studies discussed below, several proposals for the structure of the M center have been put forward. These are illustrated in Figure 7.31.

Figure 7.31 - Proposed models for FeMoco. (Compare with the recent model from the x-ray structure on page 444.)

The MoFe proteins from Clostridium pasteurianum²⁹² and from Azotobacter vinelandii²⁹³ have been crystallized. For the former protein, crystals of space group P2₁ are obtained, with two molecules per unit cell of dimensions 70 x 151 x 122 Å. There is good evidence for a molecular two-fold axis, which presumably relates equivalent sites in the two \(\alpha \beta\) dimers that make up the protein molecule.²⁹⁴ Preliminary refinement reveals that the two FeMoco units per protein are about 70 Å apart and the four P clusters are grouped in two pairs.

Single crystal EXAFS studies²⁹⁵ have provided important structural information on the molybdenum site. For different crystal orientations (relative to the polarized x-ray beam), the amplitude of the Mo-Fe EXAFS changes by a factor of 2.5, but the Mo-S EXAFS changes only slightly. Analysis of the anisotropy of the Mo-Fe EXAFS using the available crystallographic information²⁹⁴ is consistent with either a tetrahedral MoFe₃ geometry such as that found in thiocubanes (Figure 7.32) or a square-based pyramidal MoFe₄ arrangement of metals. This interpretation tends to rule out some of the structural proposals shown in Figure 7.33. The observed orientation-dependence of the iron amplitudes is too small for clusters containing a linear or planar arrangement of iron and molybdenum (e.g., Figure 7.33B,C), and too large for arrangements that involve regular disposition of iron about molybdenum. Moreover, the lack of anisotropy of the sulfur EXAFS (which was apparently not considered in the original interpretation²⁹⁵) argues against an MoS₃ (O/N)₃ model that has molybdenum coordinated by sulfur atoms that bridge only to Fe atoms disposed to one side of the molybdenum. Significant anisotropy for the Mo-S EXAFS (of opposite polarization, and smaller than that for Mo-Fe) would be expected for such an arrangement of sulfur atoms. However, the cubane model of Figure 7.33, which provides the best model of both geometric and electronic structure, remains viable if one of the nonbridging ligands to molybdenum is a sulfur atom (rather than oxygen or nitrogen) with a bond length similar to that of the bridging sulfides.

Figure 7.32 - Structures of thiocubanes that display Mo-S and Mo-Fe distances similar to FeMoco: (A) (Fe₃MoS₄)₂(SR)₉^3-; (B) (MoFe₃S₄)₂Fe(SR)₁₂^3-/4-; (C) MoFe₃S₄(SEt)₃(cat)CN^3-. (Data on A and B from References 328, 330a; data on C from References 331, 332.)

Figure 7.33 - FeMoS and FeWS structures of potential interest with respect to nitrogenase.^{331 ,332,332a-j}

10. Substrate Reactions

The two-component Mo-nitrogenase enzyme catalyzes the reduction of N₂ to 2NH₄⁺ as its physiological reaction. Concomitant with the reduction of N₂, H₂ evolution occurs, with electrons supplied by the same reductants that reduce N₂. The limiting stoichiometry appears to be

\[N_{2} + 10 H^{+} + 8 E^{-} \rightarrow 2 NH_{4}^{+} + H_{2} \tag{7.16}\]

If N₂ is omitted from the assay, all the electrons go to H₂ evolution. Indeed, to a first approximation the rate of electron flow through nitrogenase is independent of whether the enzyme is producing only H₂, producing both NH₄⁺ and H₂, or reducing most of the alternative substrates. As displayed in Table 7.8, many alternative substrates are known for this enzyme.^240,243,296 The most important of these from a practical perspective is acetylene, C₂H₂, which is reduced by the Mo nitrogenase exclusively to ethylene, C₂H₄. Acetylene can completely eliminate H₂ evolution by nitrogenase. Many of the substrates in Table 7.8 have a triple bond. Indeed, the only triple-bonded molecule not reduced by nitrogenase is CO, which nevertheless inhibits all substrate reactions, but not H₂ evolution (in the wild type). Triple-bonded molecules such as acetylene (H—C\(\equiv\)C—H) are useful probe molecules for related reactivity as discussed below for simple inorganic systems. All substrate reductions involve the transfer of two electrons or multiples thereof (i.e., 4,6,8 . . .). Multielectron substrate reductions may involve the stepwise execution by the enzyme of two-electron processes. Further, about as many protons as electrons are usually transferred to the substrate. One way of viewing the nitrogenase active site is that it can add the elementary particles (H⁺ and e^-) of H₂ to the substrate. This may have mechanistic implications.²⁹⁷

It is potentially fruitful to pursue the intimate connection between H₂ and the N₂ binding site in nitrogenase. It has been shown unequivocally^298,299 that one H₂ is evolved for each N₂ "fixed" even at 50 atm of N₂, a pressure of N₂ well above full saturation. Moreover, H₂ is a potent inhibitor of N₂ fixation, and under D₂, HD is formed, but only in the presence of N₂. These complex relationships between N₂ and H₂(D₂) have elicited a variety of interpretations.^255,300-302

Recently, it has been demonstrated that the FeMo protein alone acts as an uptake hydrogenase.³⁰³ Dihydrogen in the presence of [FeMo] causes the reduction of oxidizing dyes such as methylene blue or dichlorophenolindophenol in the absence of Fe protein. This is the only known catalytic reaction displayed by the FeMo protein alone. The hydrogen evolution and uptake by [FeMo] suggest that understanding hydrogen interaction with transition-metal/sulfur centers may be crucial to understanding the mechanism of nitrogenase action.

Table 7.8 - Table 7.8 Nitrogenase substrate reactions.^296,374-376

Two-electron Reductions

$$2e^{-} + 2H^{+} \rightarrow H_{2}

$$C_{2}H_{2} + 2e^{-} + 2 H^{+} \rightarrow C_{2}H_{4}$$

\[N_{3}^{-} + 2e^{-} + 3H^{+} \rightarrow NH_{3} + N_{2}\]

$$N_{2}O + 2e^{-} + 2H^{+} \rightarrow H_{2}O + N_{2}$$

Four-electron Reductions

$$HCN + 4e^{-} + 4H^{+} \rightarrow CH_{3}NH_{2}$$

$$RNC + 4e^{-} + 4H^{+} \rightarrow RNHCH_{3}$$

Six-electron Reductions

$$N_{2} + 6e^{-} + 6H^{+} \rightarrow 2NH_{3}$$

$$HCN + 6e^{-} + 6H^{+} \rightarrow CH_{4} + NH_{3}$$

$$HN_{3} + 6e^{-} + 6H^{+} \rightarrow NH_{3}+ N_{2}H_{4}$$

$$RNC + 6e^{-} + 6H^{+} \rightarrow RNH_{2} + CH_{4}$$

$$RCN + 6e^{-} + 6H^{+} \rightarrow RCH_{3} + NH_{3}$$

$$NCNH_{2} + 6e^{-} + 6H^{+} \rightarrow CH_{3}NH_{2} + 2NH_{3}$$

$$NO_{2}^{-} + 6e^{-} + 6H^{+} \rightarrow NH_{3}$$

Multielectron Reductions

$$RNC \rightarrow (C_{2}H_{6} , C_{3}H_{6} , C_{3}H_{8}) + RNH_{2}$$

$$NCNH_{2} + 8e^{-} + 8H^{+} \rightarrow CH_{4} + 2NH_{3}$$

11. The Role of ATP

ATP hydrolysis appears to be mandatory, and occurs during electron transfer from [Fe] to [FeMo]. Dissociation of [Fe] and [FeMo] following electron transfer is probably the rate-limiting step in the overall turnover of the enzyme.²⁵⁵ The fact that reductant and substrate levels do not affect turnover rates is consistent with this finding.

The role of ATP on a molecular level remains one of the great mysteries of the mechanism of nitrogen fixation. As discussed above, the overall thermodynamics of N₂ reduction to NH₃ by H₂ or by its redox surrogate flavodoxin or ferredoxin is favorable. The requirement for ATP hydrolysis must therefore arise from a kinetic necessity. This requirement is fundamentally different from the need for ATP in other biosynthetic or active transport processes, wherein the free energy of hydrolysis of ATP is needed to overcome a thermodynamic limitation.

What is the basis for the kinetic requirement of ATP hydrolysis in nitrogen fixation? To answer this question, we again look at the potential reduction products of the N₂ molecule. Of these, only N₂H₂ (diimide, three potential isomers), N₂H₄ (hydrazine and its mono and dications), and NH₃ (and its protonated form, NH₄⁺) are isolable products. (In the gas phase, other species such as N₂H, N₂H₃, or NH₂ also have a "stable" existence.) In the presence of H₂, only the formation of ammonia is thermodynamically favored (Figure 7.26). Clearly, the formation of the intermediate species in the free state cannot occur to any reasonable extent. However, this does not mean that nitrogenase must form NH₃ directly without the formation of intermediates. It is possible for these reactive intermediates to be significantly stabilized by binding to a metal-sulfur center or centers.

Detailed kinetic studies^255,304 have suggested a scheme in which intermediates with bound and probably reduced nitrogen are likely to be present. Rapid quenching experiments in acid solution lead to the detection of hydrazine during nitrogenase turnover.³⁰⁵ Likewise, studies of inhibition of N₂ fixation by H₂ and the formation of HD under D₂ have been interpreted in terms of a bound diimide intermediate.^306,307 Although a bound "dinitrogen hydride" is likely to be present, its detailed structure remains unknown.