8.22.2.1 FeProtein
8.22.2.1.1 Function and structure
The Fe protein (also callednifHafter the gene that encodes it) is the vehicle for a 4Fe/4S cubane- type cluster (Figure 2), in which the tetrahedral coordination at each iron atom is completed by a cysteine residue. The Fe protein is a homodimer, and the cluster lies at the interface between the two units, held by Cys97 and Cys132 from each subunit. The X-ray crystal structure of the Fe protein shows that the iron–sulfur cluster is exposed to solvent, a feature that presumably assists in acquisition of electrons from its natural electron donors (flavodoxin or ferredoxin).109,110
The Fe protein is the only known reductant that successfully induces the MoFe protein to catalyze N2reduction. The redox potential of the biologically important Fe4S42þ
/Fe4S41þ
couple depends on the presence or absence of nucleotide substrates: it is 0.30 V without nucleotides, 0.43 V in the presence of ATP, and 0.49 V in the presence of ADP.24This trend is consistent with a role where the Fe protein accepts an electron, and nucleotide binding and hydrolysis trigger electron donation. Because it is more effective than reductants with lower redox potentials, it is likely that binding between the Fe and MoFe proteins is accompanied by some conform- ational change that opens an efficient electron transfer pathway. The structural details have been elucidated using X-ray crystallography of nitrogenase inhibited with ADPAlF4
, which binds in the ATP site and stabilizes the docked Fe protein/MoFe protein complex.111 The structure shows that deep-seated conformational changes in the Fe protein upon binding112place the Fe4S4cluster about 14 A˚ from the P cluster (seeSection 8.22.2.2) of the MoFe protein. In addition, the bound complex has hydrogen bonding interactions between the peptide backbone of the MoFe protein and the sulfides of the cluster. The direct source of the shift of potential may derive from specific hydrogen bonding interactions, or from a change in the polarity of the environment of the cluster.
Redox potentials of iron–sulfur clusters in organic solution are typically much lower than those for the same site in water, suggesting that movement of the cluster from the surface to the interior of the protein upon nucleotide binding could make it a stronger reductant.113
Under standard in vitro conditions with dithionite as reductant, the Fe4S4 cluster is in the reduced Fe4S41þ state. Interestingly, the Fe4S41þ cluster can be reduced by titanium(III), chro- mium(II), or radiolytic reduction to an Fe4S40
(all-ferrous) form, an oxidation level that is
Cys-97(β)
Cys-132(β)
Cys-132(α)
Cys-97(α)
Figure2 Fe4S4cluster of the Fe protein. Coordinates from Schlessmanet al.110
unusual for Fe4S4 clusters.114,115 The structure of the Fe4S40
form has been determined using X-ray crystallography at moderate resolution (2.25 A˚).116The structure shows that the coordin- ation of the cluster does not change upon reduction. The Fe4S4cluster expands slightly, and the cysteine ligands attract seven hydrogen bond donors, presumably to dissipate the substantial negative charge on the cluster. Predicted metrical details also come from preliminary density functional calculations.121
Because nitrogenase catalysis consumes less ATP and shows no induction period with Ti(III) as reductant, it has been suggested that the Fe4S40
oxidation level could be physiologically rele- vant.117,118This proposal suggests that the Fe4S4cluster transfers pairs of electrons rather than undergoing single electron transfer events, consistent with the fact that all nitrogenase substrates are reduced by multiples of two electrons. However, the relevance of the all-ferrous statein vivois controversial: a study indicates that the redox potential of the Fe4S41þ/Fe4S40
couple in Av2 is 0.79 V, too low for reduction by any of the possible natural reductants, and not at 0.46 V as earlier reported.119 The lower value is more reasonable, based on the differences between Fe4S42þ/Fe4S41þand Fe4S41þ/Fe4S40
redox potentials in model complexes,120 and on the results of density functional calculations.121
Numerous synthetic Fe4S4cubanes with thiolate ligation like that of the Fe protein cluster are known in the Fe4S42þ
and Fe4S41þ
states (synthetic Fe4S4complexes are discussed inSection 8.22.2.2.4).
These have been studied mostly in nonaqueous environments, and have Fe4S42þ/Fe4S41þpotentials at0.7 V to1.3 V. In a few cases, it has been possible to reduce the clusters further to the Fe4S4 0
level in cyclic voltammetry experiments, but no all-ferrous Fe4S4 model compounds have been isolated.120,122,123It is possible that synthetic strategies using small peptides that incorporate protein-- like hydrogen bonding interactions may stabilize the substantial negative charge on a Fe4S4-(SR)44
unit and enable the isolation of this type of synthetic compound in the future.
8.22.2.1.2 Spectroscopy
At the Fe4S41ỵoxidation level, the Fe4S4cluster is in a mixture ofSẳ1/2 andSẳ3/2 states.124–128 Electron paramagnetic resonance studies, magnetism, Mo¨ssbauer, and MCD studies have led to this conclusion, although the details of this spin crossover/admixture are not clear.18
As the Fe4S40
state is not stable in synthetic compounds, spectroscopists have taken advantage of the accessibility of Fe protein in the all-ferrous state to understand the spectroscopic char- acteristics of this novel oxidation level. The parallel-mode EPR spectrum shows a resonance at geffẳ16.4 from a ground stateSẳ4 spin multiplet, and the electronic absorption spectrum has a band at 490 nm that gives the all-ferrous protein a pink color.129,130 EXAFS shows iron–iron distances of 2.53 A˚ and 2.77 A˚, consistent with a tetragonally distorted tetrahedron with sub- stantial FeFe bonding interactions.131 However, the results of EPR and Mo¨ssbauer investiga- tions suggest that one of the four iron sites is different from the other three, inconsistent with any C2-symmetric structure.115 The X-ray crystal structure is not of high enough resolution to distinguish between these distortions.116
8.22.2.2 MoFeProtein: P Cluster 8.22.2.2.1 Function and structure
The P cluster is an Fe8S7cluster that lies at the interface of the- and -subunits of the MoFe protein. It is situated about 15 A˚ from the site at which the Fe protein binds and about the same distance from the FeMoco, suggesting a role in electron transfer. This idea was first supported by a tandem EPR/kinetics study, in which the (small) characteristic spectroscopic change observed corresponds to the rate of FeMoco reduction.132More direct evidence comes from the observa- tion in a mutant of a characteristic EPR signal that disappears during turnover and then returns.133 Therefore, the P cluster is generally viewed as a gateway for electron transfer into the catalytic FeMoco center. The P cluster is observed in several redox states: the most common are called P2þ(or POX), P1þ, and PN(fully reduced).
X-ray crystallography in the reduced PN form shows that it is a double cubane in which one sulfide is shared by both cubanes. Figure 3a shows the model from Av1 at 1.16 A˚ resolution.154 This structural feature is also present in the Kp1 structure.134 The bridging sulfur atom is six
coordinate, and the six bonds to this sulfur atom have equivalent lengths in the highest-resolution Kp1 and Av1 structures. The two cubane units are also bridged by the sulfur atoms from Cys88 () and Cys95 (), and the remaining coordination sites are filled by Cys residues from both subunits.
Upon transformation to POX (the POX/PN redox potential is roughly 0.3 V),21,135 there is a structural change where the two cubanes move apart, replacing two of the bonds to the central sulfide with O and N atoms from Ser188 () and the protein backbone of Cys88 (), respectively (Figure 3bshows the model of POXfrom Av1 at 2.03 A˚ resolution).136It is reasonable that these
‘‘hard’’ donors stabilize a more oxidized state of the redox center; however, the exact redox level of the iron atoms in the P cluster in the aforementioned states is not yet known. The presence of natural amide and alkoxide donors to an iron–sulfur cluster, as in POX, is unusual because iron–sulfur clusters are typically coordinated by cysteine thiolate ligands. In addition to their ‘‘hard’’
character, the high pKa values for the N and O donors admit the possibility that they may be protonated, i.e., a pH change may be mechanistically important in controlling the redox activity.
Indeed, the potential of the POX/P1þredox couple is pH dependent.135Mutation of Ser188 () to Gly does not alter the redox potential: this leaves the amide, a coordinated Cys, or some proton-mediated conformational change as potential pH-sensitive contributors to function.24 A crystal structure indicates that the coordinated amide nitrogen is not planar, suggesting that it might be protonated.134 The redox-coupled conformational change and the two-electron change from PNto POXhas led most researchers to view the P cluster as a diode and/or capacitor for electron flow into the FeMoco. This hypothesis explains the ability of the nitrogenase protein to build up sufficient reducing equivalents at one site (the FeMoco) to accomplish the difficult reduction of N2. However, more work is necessary to establish the ways in which the protein controls the rate and direction of electron transfer.24
8.22.2.2.2 Spectroscopy
Fe K-edge EXAFS of FeMoco-deficient MoFe protein gives a picture of the PNand POX states that is generally consistent with crystallography, although the presence of a large number of iron
Cys95 (β) Cys154 (α)
Cys62 (α)
Cys154 (α)
Cys62 (α)
Cys88 (α)
Ser188 (β) Cys153 (β)
Cys70 (β) Cys88 (α)
Cys95 (β)
Cys153 (β)
Cys70 (β)
(b) (a)
Figure3 P cluster of the MoFe protein in two different oxidation states: (a) PN; (b) POX. Carbon and nitrogen atoms are shown in white. Coordinates from Einsleet al.154and Peterset al.136
atoms in similar environments makes it difficult to distinguish specific distances.137It is interest- ing, though, that very short (2.42 A˚ and 2.57 A˚) FeFe distances consistently were obtained in the refinements of the POX and PN states, respectively.
The PN state is diamagnetic, as shown by its EPR silence and the lack of response of its Mo¨ssbauer signal to a magnetic field.138 This restrains the possible oxidation states to an even number of both iron(II) and iron(III) ions, but the exact oxidation level of the eight-metal cluster is not known. Mo¨ssbauer studies also suggest a primarily (if not exclusively) iron(II) formulation of PN.139Either an FeII8or an FeII6FeIII2formulation seems reasonable, because it is possible to oxidize the cluster to POX, which is thought to haveSẳ3 or 4 and be two electrons more oxidized than PN (redox potentialẳ 0.31 V). The integer-spin formulation of POX is supported by a characteristic signal in parallel-mode EPR spectra.140,141
An intermediate oxidation level can be identified in redox titrations below physiological pH:
this is P1ỵ, which has aSẳ1/2–5/2 mixed spin state.142This may arise from mixtures of isomers in which Ser188 () is bound or not, from states in which the ‘‘hole’’ is on different halves of the cluster, or in which different cysteinates are cleaved.134It has been possible to stabilize this state with a mutation Ser188Cys in which the electron-rich thiolate shifts the redox potentials to lower values.133The POX/P1þredox couple is pH dependent, as noted above. Another redox state, P3þ, has also been identified, and it also is apparently a mixed spin state withSẳ7/2 and 1/2.143In the R. capsulatus enzyme, though, neither P1þ nor P3þ is mixed.144 Clearly, additional work is necessary to understand the electronic structure of this cluster and its interplay with geometric structure and electron transfer.
8.22.2.2.3 Model complexes
It is now possible for synthetic chemists to create iron–sulfur clusters that mimic the unusual distribution of iron atoms in PN. The first stable double cubanes with thiolate bridges like those in the P cluster were created in 1997.145The challenge of making a6-sulfido bridge was conquered with the complex [(Cl4cat)6(Et3P)6Mo6Fe20S30]6, in which there are Fe6Mo2units.146A smaller cluster with similar structural analogies to PN is [(Cl4cat)4 (Et3P)4Mo4Fe12S20K3(DMF)]5 (10), which contains two [(catecholate)(phosphine)Mo]2Fe6S92
units bridged by two2-sulfide bridges and three potassium ions.147 Although the Fe6S9 [Mo(catecholate)(phosphine)]2 fragments have octahedral molybdenum atoms in place of the outer iron atoms, there is a clear analogy with PN, in which two bound cubanes are connected by -sulfide. Mo¨ssbauer and magnetization data support a primarily ferrous environment in (10). Finally, monomeric clusters (11) ([Tp2M2Fe6S9]n; MẳMo,nẳ3; MẳV,nẳ4) are stabilized by the tris(pyrazolyl)borate ligand, showing that the cluster topology found in PN can exist as a free entity in solution.148Complex (11b) has metrical parameters that are truly close to PN, with an r.m.s. deviation of only 0.33 A˚
for the M8S9core.
(10) (11a)(Mo, 3−), (11b)(V, 4−)
Cl Cl
Cl Cl
Cl Cl
Cl
Cl Cl
Cl Cl Cl Cl
Cl Cl Cl O
O
O O
O O
O
O O S
S
S S S
SH SH
S S S S S
S S S S S S
S S
S S
Fe Fe
Fe Fe Fe
Fe Fe
Fe Fe
Fe Fe
Fe Fe Fe
Fe Fe Fe
Fe P P
P P
N Mo
Mo Mo
K K K
H
S S
S S S
S S S S N N
N N N N
N N NN
N N M B B
M
H 5
The synthesis of (11) shows that the PN geometry can spontaneously assemble from a cubane synthon, hydrosulfide, and reducing agent. The Fe protein is essential in the natural synthesis of working MoFe protein (seeSection 8.22.5), suggesting that a similar reductive cluster fusion could
be relevant in the biosynthetic pathway. What challenges in model complexes for PNremain? The above clusters have not been synthesized with only iron. Apparently, the octahedral coordination at the terminal Mo sites is beneficial for steering the Fe3 units into appropriate fused clusters.
Also, the synthetic compounds have2-sulfide bridges instead of thiolates (cysteinate residues in the protein).
8.22.2.3 MoFe Protein: FeMo Cofactor 8.22.2.3.1 Function and structure
The iron–molybdenum cofactor of nitrogenase is one of the most fascinating exhibits in bioinor- ganic chemistry, because it is here that the enzyme somehow catalytically cleaves the strong triple bond of N2to give ammonia. As noted inSection 8.22.1.2, human ingenuity has not yet devised a catalytic method for this process that is comparably effective at room temperature and atmos- pheric pressure. Therefore, coordination chemists can gain valuable lessons in bond cleavage and catalysis from this metal cluster, its surroundings, and its metal-substituted analogues.
The iron–molybdenum cofactor (‘‘FeMoco’’ or ‘‘M cluster’’) is widely accepted to be the site of N2binding and reduction. Mutations near this site, or the absence of homocitrate (from nifV mutants) greatly affect the catalytic specificity and activity. The enzyme is inactive if the biosyn- thetic machinery for FeMoco synthesis is removed, but becomes active if ‘‘extracted FeMoco’’
(see Section 8.22.2.3.4) is added. The characteristic EPR signal for the FeMoco changes under turnover conditions, and, as shown below, some substrates have been shown to bind at the FeMoco.
The FeMoco has been observed in three redox states. When the enzyme is isolated in the presence of dithionite, the FeMoco is in the MN (‘‘native’’) form. Most crystallographic studies have taken place on enzyme in which the FeMoco is in the MNstate. MOX can be generated by one-electron oxidation of MN with dyes, and can be reactivated by reduction. The MOX/MN redox potential is dependent on the organism from which the nitrogenase is derived, lying in the range 0 mV to180 mV.149The X-ray crystal structure of MoFe protein with the FeMoco in the MOX state shows no major differences from MN.134,136 MR is the turnover state of the enzyme, corresponding to one-electron reduction of MN, and has not been crystallographically analyzed because it is only observed during catalytic turnover.
The basic structure of the FeMoco in the MNstate (Figure 4) is a MoFe7S9cluster, lying deep within the subunit, that has superficial similarity to the PN cluster. However, in the FeMoco there are fewer protein-derived ligands: a cysteine thiolate (Cys275) binds to the terminal iron site, and the central bridges are sulfides rather than thiolates. The molybdenum atom is octahedrally
His442
Homocitrate X
Cys275
Mo Fe
S
Figure4 Iron–molybdenum cofactor (FeMoco) of the MoFe protein. Carbon and nitrogen atoms are shown in white. Coordinates from Einsleet al.154
coordinated by three 3-sulfides, His442, and (R)-homocitric acid (2-hydroxy-1,2,4-butanetricar- boxylic acid). Homocitrate, the deprotonated form of the acid, coordinates through the alcohol oxygen (which may or may not be deprotonated) and in a monodentate fashion through the central carboxylate. One of the remaining carboxylates has a hydrogen bond to Gln191 (Figure 5b); the others are surrounded by water molecules. A theoretical study suggests that the histidine imidazole could be deprotonated at the uncoordinated"position.150
The FeMoco is surrounded by a number of very close amino acid residues. Figure 5a shows MN along with several residues that have metal-binding and/or hydrogen-bonding capabilities.
His195 is close enough for a direct hydrogen-bonding interaction with a nucleophilic 2-sulfide bridge. Two arginine residues also are quite close to the central region of the cluster, and may be involved in proton delivery to the FeMoco. The positive charges on these residues are also likely to affect the behavior of the FeMoco, through direct interaction with N2 or intermediates, or through electrostatic modification of the redox potential. There is a substantial ‘‘pool’’ of water molecules in the vicinity of the homocitrate ligand. Channels through which protons can approach and ammonia can leave have been located using theoretical calculations.151,152
8.22.2.3.2 Atoms in the belt region
The six iron atoms in the center of the cofactor (the ‘‘belt’’ or ‘‘waist’’ region) are unusual. Based on crystallography with moderate resolution, admirers of the FeMoco (the author included) long held the unlikely belief that the six iron atoms were three-coordinate, because the only visible ligands are the three bridging sulfides in a roughly trigonal planar arrangement. This rare coordination geometry in synthetic compounds is invariably enforced by very large ligands (seeSection 8.22.2.3.5), and therefore it is surprising that three-coordination would persist in the presence of potentially coordinating Arg and His residues. The 2.6 A˚ FeFe distances between adjacent iron atoms (Figure 6) could represent enough iron–iron bonding that the iron atoms are not electronically unsaturated.153Iron–iron bonds can also be seen as a storage site for electrons.
Our view of the FeMoco changed drastically in 2002. Crystallographic analysis to 1.16 A˚
resolution ofA. vinelandiiMoFe protein shows that there is a light atom X (C, N, or O) between the six iron atoms, whose presence had been masked by systematic termination errors that are inherent in Fourier analysis of diffraction data.154 This discovery is consistent with numerous data that are more difficult to explain with the ‘‘three-coordinate’’ model. First, each of the six
Arg359
Arg96
Arg359
Arg96
His442 Mo
Cys275
His195 His195
Gln191 Gln191
FeMoco à2-S
à2-S à2-S
à3-S à3-S
à3-S
(a) (b)
Figure5 FeMoco with nearby residues that could potentially coordinate and/or form hydrogen bonds with intermediates. Carbon and nitrogen atoms are shown in white: (a) front view; (b) side view. Coordinates
from Einsleet al.154
belt iron atoms slightly deviates from its S3-coordination plane toward theinsideof the cluster, to make the bond to X. Second, the distance from X to each belt iron atom is 2.0 A˚, a reasonable bond distance for an FeN, FeC, or FeO bond. (This distance is too short for X to be sulfur, and the thermal parameters of the iron atoms are symmetric, indicating full occupancy of X.) Third, the unusual lack of coordination from the protein surroundings is explained because the iron atoms are not actually three-coordinate in the isolated form. Finally, the presence of a structural stabilizer like X explains why the FeMoco can be chemically extracted from the protein without extensive changes in the geometric and electronic structure (see Section 8.22.2.3.4).
Therefore, the belt iron atoms in the FeMoco have a four-coordinate trigonal pyramidal geometry.
Figure 6andTable 1show the current model of the FeMoco. The electron density for X in the X-ray crystal structure is most consistent with a nitrogen atom,154but C and O models are within experimental error. Theoretical calculations reproduce the FeX bond lengths most accurately with XẳN.155X has been postulated to be a nitride derived from splitting of N2, and therefore the resting state of the catalyst would represent an FeMoco that has completed part of the nitrogen reduction.154 This is consistent with the observation that N2 uptake only occurs after addition of three or four electrons to the resting state (seeSection 8.22.3.1.1). However, spectros- copic investigations show that added 15N2is not incorporated into the cluster during turnover, ruling out the identification of X as a nitride that originates from the catalytic reduction of N2.156 Therefore, the available data suggest that X fulfills a structural role, although much remains to be done to delineate its identification and function in N2reduction.
Synthetic compounds show precedents for C, N, and O atoms that bridge six transition metals.
Hexanuclear nitrides are known for most transition metals in groups 8 and 9, and there are a few in which all six atoms are iron.157,158Most of these clusters have an octahedral arrangement of metal atoms, and Fe(6-N) distances are 1.84–1.89 A˚. Trigonal prismatic arrangements of metals around a nitride are less common, but one example is [Co6N(CO)15], in which the 6- nitrido ligand was characterized by 15N NMR spectroscopy.159 The many 6-oxo complexes typically have MO distances greater than 2.0 A˚. Synthetic iron–carbide clusters are known having carbide ligands with varying coordination numbers up to 6, but it is unclear how carbide could be formed in nitrogenase.160,161 Therefore, there are precedents for C, N, and O atoms fulfilling the high-coordinate bridging role of X, but N seems most likely at this early stage. It is notable that the metal atoms in the synthetic clusters invariably have strong-field ligands (pri- marily carbonyl) and are low spin, and therefore may not be directly comparable with the low coordination number and weak-field sulfide ligands found in the FeMoco. A review has recently appeared.350
SC FeB
SC FeD
FeD SC
FeD SG
(Cys)SA Mo
SG FeF FeF
SG FeF
O SE
SE
SE
O
N(His)
O R R'
X
Figure6
Table1 Average bond lengths (standard deviation) of the FeMoco from the 1.16 A˚ structure of Av1.154 Distances in A˚. Labels inFigure 6.
FeBSA 2.27(1) FeBFeD 2.67(1)
FeBSC 2.28(1) FeDFeD 2.65(2)
FeDSC 2.27(2) FeDFeF 2.59(2)
FeDSE 2.23(1) FeFFeF 2.62(2)
FeD,FX 2.00(3) FeFMo 2.70(3)
FeFSE 2.22(1) MoO(carbox) 2.20(1)
FeFSG 2.23(1) MoO(alkox) 2.18(1)
MoSE 2.34(1) MoN(His) 2.30(1)