7.9: Multisite Redox Enzymes (Part 5)

The Alternative Nitrogenases

1. Vanadium Nitrogenase

The "essentiality" of molybdenum for nitrogen fixation was first reported by Bortels in 1930.³⁰⁸ This finding led ultimately to the characterization of the molybdenum nitrogenases discussed in the preceding section. Bortels' work has been cited many times, and is often referred to without citation. Following this seminal work, many other Mo-containing enzymes were subsequently sought and found.^25,309 At present more than a dozen distinct Mo enzymes are known, and new ones are continually being discovered.

In addition to the classic 1930 paper, Bortels³¹⁰ reported in 1935 that vanadium stimulated nitrogen fixation. In contrast to the 1930 paper, the 1935 paper languished in obscurity. Then, starting in the 1970s, attempts were made to isolate a vanadium nitrogenase. In 1971, two groups reported isolating a vanadium-containing nitrogenase from A. vinelandii.^311,312 The interesting notion at this time was that V might substitute for Mo in nitrogenase, not that there was a separate system. The isolated enzyme was reported to be similar to the Mo enzyme, but had a lower activity and an altered substrate specificity. One of the groups carefully reinvestigated their preparation, and found small amounts of molybdenum, which were presumed to be sufficient to account for the low activity, although the altered selectivity was not addressed.³¹³ The vanadium was suggested to play a stabilizing role for [FeMo], allowing the small amount of active Mo-containing protein to be effectively isolated. Apparently the possibility was not considered that a truly alternative nitrogenase system existed, whose protein and metal centers both differed from that of the Mo nitrogenase.

The unique essentiality of molybdenum for nitrogenase fixation went unchallenged until 1980, when it was demonstrated³¹⁴ that an alternative nitrogenfixation system could be observed in A. vinelandii when this organism was starved for molybdenum.³¹⁵ Despite skepticism from the nitrogenase research community, it was eventually shown that even in a mutant from which the structural genes for the Mo nitrogenase proteins (nif H, D, and K) had been deleted, the alternative system was elicited upon Mo starvation. In 1986, two groups^316-321 isolated the alternative nitrogenase component proteins from different species of Azotobacter, and demonstrated unequivocally that one component contained vanadium and that neither component contained molybdenum.

One of the two components of the V-nitrogenase system is extremely similar to the Fe protein of nitrogenase. This similarity is evident in the isolated proteins from A. vinelandii³¹⁶ and in the genetic homology between nif H (the gene coding for the subunit of the Fe protein in the Mo-nitrogenase system) and nif H* (the corresponding gene in the V-based system). Both Fe proteins have an \(\alpha_{2}\) subunit structure, and contain a single Fe₄S₄ cluster that is EPR-active in its reduced state.

The FeV proteins from Azotobacter vinelandii and Azotobacter chroococcum each have an \(\alpha_{2} \beta_{2} \delta_{2}\) subunit structure.³²² Metal composition and spectroscopic comparisons between the FeMo and FeV proteins are shown in Table 7.9. Although there is the major difference involving the presence of V instead of Mo in the FeV protein and in the probable presence of the small \(\delta\) subunits (13 kDa), the two nitrogenase systems are otherwise quite similar.³²² In each, a system of two highly oxygen-sensitive proteins carries out an ATP-dependent N₂ reduction with concomitant H₂ evolution. The Fe proteins have the same subunit structure and cluster content, and are spectroscopically very similar. The V versions of the larger protein have somewhat lower molecular weights than their Mo analogues, and by MCD spectroscopy seem to contain P-like clusters.³¹⁸ The FeV site still may be an S = \(\frac{3}{2}\) center (by EPR, although its EPR differs significantly from that of the FeMo center).³²³ The V-S and V-Fe distances as measured by EXAFS^324,325 are similar to those in thiocubane VFe₃S₄ clusters and to Mo-S and Mo-Fe distances like those in [FeMo], which are in turn similar to those in MoFe₃S₄ thiocubanes. Likewise, XANES^324,325 indicates VS₃O₃ type coordination in [FeV] nitrogenase similar to the MoS₃O₃ coordination suggested by XANES for FeMoco. The "FeV cofactor" is extractable into NMF, and can reconstitute the nif B^-, FeMoco-deficient mutant of the Mo system.³²⁶ Despite the substitution of V for Mo, the proteins and their respective M-Fe-S sites do not differ drastically. However, the compositional changes do correlate with altered substrate reactivity.

A major difference between the V and Mo enzymes lies in substrate specificity and product formation.³²¹ As is clearly shown in Table 7.9, the FeV nitrogenase has a much lower reactivity toward acetylene than does the Mo system. Furthermore, whereas the FeMo system exclusively produces ethylene from acetylene, the FeV system yields significant amounts of the four-electron reduction product, ethane.³²¹ The detection of ethane in the acetylene assay may prove a powerful technique for detecting the presence of the V nitrogenase in natural systems.³²² Moreover, this reactivity pattern is found in the nif B^- mutant reconstituted with FeVco, indicating that the pattern is characteristic of the cofactor and not the protein.³²⁶ The reactivity change upon going from Mo to V in otherwise similar protein systems clearly adds weight to the implication of the M-Fe-S center (M = V or Mo) in substrate reduction.

Table 7.9 - Comparison of alternative nitrogenase proteins^a

a) Av1 is the FeMo protein of Azotobacter vinelandii, Av1* is the FeV protein of A. vinelandii and Ac1* is the FeV protein of A. chroococcum. Data from References 317,377, and 319, respectively.

b) Atoms per molecule.

c) nmol product/min/mg of protein.

Property	Av1^47	Av1*^47	Ac1*^50
Molecular Weight	240,000	200,000	210,000
Molybdenum^b	2	< 0.05	< 0.06
Vanadium^b	—	0.7	2
Iron^b	30-32	9.3	23
Activity^c
H^+	2200	1400	1350
C2H2	2000	220	608
N2	520	330	350
EPR g values	4.3	5.31	5.6
	3.7	4.34	4.35
	2.01	2.04	3.77
		1.93	1.93

2. The All-iron Nitrogenase³²²

The first sign that there is yet another alternative nitrogenase again came from genetic studies. A mutant of A. vinelandii was constructed with deletions in both nif HDK and nif H*D*K*, i.e., the structural genes for the Mo and V nitrogenases, respectively. Despite lacking the ability to make the two known nitrogenases, the mutant strain nevertheless was able to fix nitrogen, albeit poorly. Moreover, this mutant strain's nitrogenase activity was clearly inhibited when either Mo or V was present in the culture medium. Preliminary studies indicate that the nitrogenase proteins produced by this organism are closely related to those previously isolated. A 4Fe-4S Fe-protein nif H^\(\dagger\) and a protein due to nif D^\(\dagger\) was produced. The latter appeared to contain no stoichiometric metal other than iron. Symmetry of nomenclature would suggest calling this the FeFe protein and its cofactor FeFeco. Interestingly, this nitrogenase seems to be the poorest of the set in reducing N₂ and makes ethane from ethylene. The finding of the all-iron nitrogenase, if fully confirmed, will add significantly to the comparative biochemistry of nitrogen fixation. Speculatively, one might suggest that the concomitant absence of V and Mo suggests that nitrogen fixation need not directly involve the noniron heterometal in the cofactor cluster. This result may explain the lack of direct implication of Mo in the nitrogen fixation mechanism, despite many years of intense effort by workers in the field. (The above discussion should be taken cum grana salis until the existence of the all-iron nitrogenase is confirmed.)

3. Model Systems

Three types of model systems for nitrogenase may be considered. First, there are transition-metal sulfide clusters that resemble the FeMoco or FeVco centers of the active proteins. Although there has been significant progress, there are not yet any definitive models (as there are for Fe₂S₂ and Fe₄S₄). A second approach uses the reactions of N₂ and related substrates or intermediates with metal centers in order to gain insights into the way in which transition-metal systems bind N₂ and activate it toward reduction. Here to date the most reactive systems bear little direct chemical resemblance to the nitrogenase active sites. Nevertheless, these systems carry out bona fide nitrogen fixation from which one may learn the various ways in which N₂ can be activated. Finally, there are other inorganic systems that display some of the structural and possibly some of the reactivity characteristics of the nitrogenase active sites without binding or reducing N₂ or precisely mimicking the active center. We may nevertheless be able to learn effectively about nitrogenase reactivity from these interesting chemical systems.

a. Transition-metal Sulfide Models for Nitrogenase Sites

Although there has been great activity in synthetic Fe-S cluster chemistry, there is to date no example of a spectroscopic model for the P-cluster sites in nitrogenase. If the P-clusters are indeed asymmetrically bound high-spin Fe₄S₄ clusters, then the recent work on high-spin versions of Fe₄S₄ clusters³²⁷ and site-selectively derivatized Fe₄S₄ centers¹⁴³ may hint that appropriate model systems are forthcoming.

b. Fe-M-S Cluster Models for FeMoco

Despite the importance of P-clusters, the modeling of the FeMoco center has properly received the most attention. The significant structural parameters that any model must duplicate are the Mo-S and Mo-Fe distances detennined by EXAFS. Spectroscopically, the S = \(\frac{3}{2}\) EPR signal provides a stringent feature that model systems should aspire to mimic.

Many FeMoS clusters have been prepared in the quest to duplicate the FeMoco center, but none of the chemically synthesized clusters can reactivate the (UW-45 or Nif B^-) cofactor-less mutants, perhaps because of their lack of homocitrate, which only recently has been discovered as a key component of FeMoco. Undoubtedly, new FeMoS clusters containing homocitrate will be prepared, and perhaps these will activate the mutant proteins, thereby revealing a close or full identity with FeMoco.

Despite the absence of homocitrate, some interesting model systems have been investigated. It is beyond the scope of this chapter to give a comprehensive account of FeMoS chemistry. We concentrate on the so-called "thiocubane" model systems. Heterothiocubane models were first synthesized using self-assembly approaches analogous to those used for the simpler Fe-S model systems. The reaction^328-330a

\[MoS_{4}^{2-} + Fe^{3-} + SR^{-} \rightarrow (MoFe_{3}S_{4})_{2}(SR)_{9}^{3-} \; and\; (MoFe_{3}S_{4})_{2}Fe(SR)_{12}^{3-,4-} \tag{7.17}\]

uses tetrathiomolybdate, MoS₄^2-,as the source of Mo, and leads to the double cubane structures shown in Figure 7.32A,B. The Fe₇Mo₂S₈ structure proved particularly interesting, since it was possible to complex the central ferric iron atom with substituted catecholate ligands^331,332 and eventually isolate a single thiocubane unit (Figure 7.32C). Significantly, the single unit has S = \(\frac{3}{2}\) and Mo-S and Mo-Fe distances that match precisely those found by EXAFS for the M center of nitrogenase. Single cubes with VMo₃S₄ cores have also been prepared.^160,160a Although the single thiocubanes display spectroscopic similarity and distance identity with FeMoco, they are not complete models. They are stoichiometrically Fe and S deficient, lack homocitrate, and most importantly, fail to activate the UW-45 and Nif B^- mutants.

Other interesting FeMoS (and FeWS) clusters with structurally distinct properties are shown in Figure 7.33. These include the "linear" (MoS₄)₂Fe^3- ion, the linear (WS₄)₂Fe[HCON(CH₃)₂]₂^2- ion, the linear Cl₂FeS₂MS₂Fe₂CI₂^2- (M = Mo, W), the "linear" (MoS₄)₂Fe₂S₂^4- ion, the trigonal (WS₄)₃Fe₃S₂^4-, the capped thioprismane Fe₆S₆X₆[M(CO)₃]₂^3- (X = CI, Br, I; M = Mo, W), and the organometallic clusters MoFe₆S₆(CO)₁₆^2-, MoFe₃S₆(CO)₆(PEt₃₎3, and MoFe₃S₆(CO)₆^2-. Structures suggested for FeMoco based on these and other chemically synthesized transition metal sulfides and on spectroscopic studies of the enzyme are shown in Figure 7.31.