Insights into properties and energetics of iron-sulfur proteins from simple clusters to nitrogenase.

Abstract

Some of the principal physical features of iron-sulfur clusters in proteins are analyzed, including metal-ligand covalency, spin polarization, spin coupling, valence delocalization, valence interchange and small reorganization energies, with emphasis on recent spectroscopic and theoretical work. The current state of structural, spectroscopic, and computational knowledge for the iron-sulfur clusters in the nitrogenase iron and iron-molybdenum proteins is examined by comparison and contrast to 'simpler' ironclusters. The differing interactions of the nitrogenase iron and iron-molybdenum clusters compared with those of other iron-sulfur clusters with the protein and solvent environment are also explored.

Full text

Page 1

259
Some of the principal physical features of iron–sulfur clusters
in proteins are analyzed, including metal–ligand covalency, spin
polarization, spin coupling, valence delocalization, valence
interchange and small reorganization energies, with emphasis
on recent spectroscopic and theoretical work. The current
state of structural, spectroscopic, and computational
knowledge for the iron–sulfur clusters in the nitrogenase iron
and iron–molybdenum proteins is examined by comparison and
contrast to ‘simpler’ iron–sulfur clusters. The differing
interactions of the nitrogenase iron and iron–molybdenum
clusters compared with those of other iron–sulfur clusters with
the protein and solvent environment are also explored.
Addresses
Department of Molecular Biology, TPC15, The Scripps Research
Institute, La Jolla, California 92037, USA
*e-mail: lou@scripps.edu
Current Opinion in Chemical Biology 2002, 6:259–273
1367-5931/02/$ — see front matter
© 2002 Elsevier Science Ltd. All rights reserved.
Abbreviations
AF
BS
DFT
ENDOR
Fd
HIPIP
HS
LUMO
PDR
RPBE
SCF
antiferromagnetic
broken symmetry
density functional theory
electron-nuclear double resonance
ferredoxin
high-potential iron–sulfur protein
high-spin
lowest unoccupied molecular orbital
phthalate dioxygenase reductase
revised Perdew–Burke–Enzerhoff
self-consistent-field
Introduction
Iron–sulfur proteins are among the oldest known biological
catalysts [1••]. Electron-transfer proteins (ferredoxins;
Fds) are associated with the earliest oxygen-evolving
photosynthesizers (cyanobacteria). Iron–sulfur clusters
embedded in protein frameworks catalyze a wide variety
of critical electron transfer and biosynthetic processes in
all living organisms. These processes include ‘simple’
electron transfer, proton-coupled electron transfer with
energy transduction, and biosynthetic electron transfer.
In this context, some of the most complex and energeti-
cally difficult processes are those of the multielectron
oxidoreductases, which carry out biosynthesis of small
molecules involving transfer of multiple electrons and
protons and breaking of strong multiple bonds, as found
in nitrogenase and sulfite and nitrite reductases. Iron–sulfur
proteins have important non-redox catalytic functions as
well as critical regulatory functions [1••,2].
High-spin (HS) iron sites with mainly tetrahedral
coordination to sulfur (either sulfide or thiolate) lead to
distinctive features of electronic structure that are
energetically and functionally significant. The iron sites
show large spin-polarization effects, strong sulfur-to-iron
covalency, and spin coupling via the bridging sulfur
ligands. Iron–sulfur complexes can display either local-
ized (trapped) or delocalized iron valence depending on
oxidation state, coordination geometry and other factors.
The active sites are anions with strong charged hydrogen
bonding with surrounding protein, and major interactions
with nearby solvent.
The combination of Heisenberg exchange coupling
between iron sites and valence delocalization within the
cluster can lead to complicated effects on spin alignments
and on the net total spin state of the cluster. Figure 1
shows typical spin alignments and patterns of valence
electron delocalization in simple iron–sulfur clusters.
Various spectroscopic methods are diagnostic of these
effects, and both first principles calculations by density
functional theory (DFT) and spin Hamiltonian methods
are very helpful in sorting out the contributing factors.
Theoretical and computational approaches
The conceptual framework for DFT calculations has been
summarized in recent reviews [3•,4,5]. Here, we emphasize
those features that are most important for spin-polarized
and spin-coupled systems found in iron–sulfur proteins.
Spin polarization and the inverted level scheme
Transition metal complexes with HS metal sites show a
substantial spin polarization splitting between majority
spin and minority spin levels. Figure 2 shows schematically
the ‘inverted energy level scheme’ for the molecular
orbitals for a tetrahedrally coordinated 1Fe complex
(M = Fe; L = SR1–). The greater number of spin-up com-
pared with spin-down (α versus β) electrons creates a large
difference in the self-consistent-field (SCF) potentials for
these electrons because the different α versus β electron
densities produce different exchange-correlation potentials.
A ferric (Fe3+) site is depicted with the filled majority spin
metal levels lying below the occupied thiolate ligand
levels. Above these lie the empty ligand-field type e and t2
minority-spin metal levels. On one-electron (1e–) reduc-
tion, one partner of the degenerate e level is filled to form
the ferrous (Fe2+) complex. This level pattern is not the
one generally depicted in textbooks where both the filled
and empty metal levels appear above all the filled ligand
levels (the normal level scheme). Further, with few exceptions,
spin polarization splitting is not often included in energy
Insights into properties and energetics of iron–sulfur proteins
from simple clusters to nitrogenase
Louis Noodleman*, Timothy Lovell, Tiqing Liu, Fahmi Himo and
Rhonda A Torres

Page 2

level diagrams, but it is important for most FeS complexes,
as verified by both optical and photoelectron spectro-
scopies. The consequences of large sulfur-to-metal
covalency are addressed later. We emphasize these points
because they are common both to 1Fe, 2Fe, and 4Fe
sulfur clusters (see Figure 1) as well as to more complicated
centers such as the MoFe7S9(FeMo cofactor) and Fe8S7
(P cluster) of the MoFe protein in nitrogenase.
Spin coupling and the broken symmetry method
Heisenberg interaction
For dinuclear and polynuclear FeS clusters, the spins on
the individual iron sites are spin-coupled [5]. These
bridged metal–metal interactions are a type of covalency
representing weak metal–metal bonding and are called
Heisenberg exchange coupling. The interaction energy
depends on the relative alignment of the site spins, and
Heisenberg coupling typically favors opposite (antiferro-
magnetic [AF]) alignment of neighboring site spins. In a
dinuclear system, according to quantum mechanics, the
two site spins with spin quantum numbers S1, S2can
couple to any total spin Stlying between |S1–S2| to |S1+S2|
in integer intervals. These states constitute the ‘proper
pure spin state multiplets’ of the system. Figure 3 shows a
ladder of spin states obeying a Heisenberg Hamiltonian
(for a diferric [oxidized] 2Fe cluster). The direct descrip-
tion of the individual pure spin states is difficult. Within
DFT, a better approach is to consider the SCF solution for
only two spin alignments, a HS state associated with the
maximum spin, S = 5, for parallel alignment of the two
S = 5/2 site spins, and a broken symmetry (BS) state, con-
structed by rotating one of the site spin vectors so that the
two site spins are oppositely aligned. (For example, the left
iron site can be spin-up and the right iron site spin down as
in Figure 1, or vice versa.) The BS state is not a pure spin
multiplet, but is instead a specific weighted average of
pure spin states, with closely related expressions for the
wave function |BS> and the energy EBS. Further, from the
energy difference, EHS– EBS, the Heisenberg coupling
parameter J can be calculated and then used to generate
the entire spin state ladder. In particular, the position
of the pure spin S = 0 ground state can be found as
indicated graphically. (The convention for the Heisenberg
Hamiltonian we use is H = JS
classical’ analog of the corresponding AF aligned state. For
poly-nuclear FeS complexes, often several different BS
states are considered, representing different spin align-
ments, and replacing a much larger number of pure spin
states of the spin-coupled system.
→
1S
→
2.) The BS state is a ‘semi-
Valence delocalization
Additional complexity arises in a mixed-valence dimer
with spin-coupled Fe2+–Fe3+ions, where each site is inter-
nally high-spin (S1= 5/2, S2= 2 or the opposite assignment
are the two possibilities) [5]. Then, for each of the states
of the Heisenberg spin ladder, there is an additional
260
Guest section: computational bioinorganic chemistry II
Figure 1
Clusters with multiple metal sites. 1Fe–, 2Fe–, and 4Fe–sulfur
clusters are depicted with their principal spin alignments (arrows) and
showing localization (2Fe) or delocalization (4Fe) for the sixth d electron.
The total spin states seen for each cluster type are listed. ox, oxidized;
red, reduced.
Fe
Cys-S
S-Cys
Cys-S
S-Cys
Fe
Cys-S
Cys-S
Fe
S-Cys
S-Cys
Fe
S
Fe
S
S
Fe
S
Fe
S-Cys
Cys-S
Cys-S
S-Cys
1Fe
e–
e–
e–
2Fe
4Fe
S = 5/2 (ox), 2 (red)
Fe3+ Fe2+
S = 0 (ox), 1/2 (red) (trapped)
9/2 (red) (delocalized)
S = 0 (ox), 1/2 (red), 3/2 (red), 7/2 (red) (trapped)
pairwise deloc.
Simple iron–sulfur clusters
Current Opinion in Chemical Biology
Figure 2
The inverted energy level scheme is shown for a ferric (Fe3+) 1Fe
complex with the filled majority spin iron (M) levels below the occupied
sulfur ligand levels. Above these, a single empty minority spin iron level
is shown. The ligand field splits this level into different patterns
depending on geometry of the site (see Figure 5).
α
β
L
MM
L
Current Opinion in Chemical Biology

Page 3

resonance delocalization splitting term, which is linearly
dependent on the total system spin S, ± B(S + 1/2)
(Figure 4). This represents the bonding or antibonding
Fe–Fe electron delocalization. The splitting is readily
calculated in the HS state, where it is (2Smax+ 1)B = 10B,
which again allows the splitting to be computed for each
total S state. The competition between the Heisenberg
term, favoring antiferromagnetism, and the resonance
delocalization term (also called ‘double exchange’ [DE] or
‘spin-dependent-electron-delocalization’ [SDD] is evident).
The important point is that this phenomenon occurs often
in polynuclear complexes, especially of 3Fe4S and 4Fe4S
type, where there are mixed-valence pairs of iron sites. It
occurs also, but more rarely, in mixed-valence 2Fe2S1+
systems [2,6•]. The most effective form of Fe2+–Fe3+
resonance delocalization arises from σ type Fe–Fe d–d
interaction (along the Fe–Fe axis) [7]. One partner of the
low-lying ligand field e orbital is occupied by the sixth
d electron, which then undergoes pairwise delocalization.
The energetic advantage of this pairwise delocalization is a
function of the approximately tetrahedral site geometry at
iron, allowing good Fe–Fe pair σ overlap, and is maximal
for iron sites with parallel spin. In more distorted systems
or with lower coordination at iron (trigonal, for example),
the spacing for the ligand field levels is different, the
corresponding orbital shapes are altered, and consequently
the pattern of electron delocalization is more complex.
Figure 5 shows the ligand field levels for tetrahedral and
trigonal coordination and distortions of these.
Electron trapping
Whereas resonance terms favor valence delocalization,
vibronic and solvent effects and also static asymmetries
combined with Heisenberg coupling generally favor electron
trapping with discrete iron valences, as shown by differ-
ences in Mössbauer, ENDOR (electron-nuclear double
resonance), and magnetic properties [8–12]. The stability
of a trapped valence-reduced cubane 4Fe4S1+with total
spin S = 7/2 is a rare event for 4Fe cubane clusters, occurring
only in a fraction of the 4Fe4Se1+clusters of selenium-
substituted Clostridium pasteurianum Fds. However, clusters
with S = 7/2 are also found as subunits of larger complexes,
as in the 8Fe7S P clusters and the Mo7Fe9S MoFe (Mcluster)
in the nitrogenase MoFe protein, as discussed below.
Role of ligands in metal–ligand covalency
The metal–ligand covalency is very high in these
complexes, and considerably larger for the Fe3+sites than
for the Fe2+sites. Further, bridging sulfide groups are
better donors than are terminal thiolates, and the sulfides
are more effective donors per Fe–S bond in 2Fe2S than in
4Fe4S complexes. In more reduced states, sulfur charges
become considerably more negative with increased
H-bonding to the backbone peptide and/or side chains.
The H-bonding also competes with sulfur-to-metal charge
transfer, leading to larger negative charges on bridging and
cysteine sulfur than for analogous synthetic systems, and
this also contributes to higher redox potentials in the
proteins than in synthetic systems. There are two
contributing effects here, one from the H-bonding potential
and the other from orbital polarization on sulfur and charge
transfer into the NH–S hydrogen bond. These effects can
be separated out by careful ligand K-edge spectroscopy,
which is sensitive to the S(3p) holes that become occupied
on S(K shell)→valence excitation [13•–15•].
Energetic and geometric consequences of
electronic interactions
In several papers, the electronic orbital, spin coupling and
valence delocalization contributions to iron–sulfur cluster
energetics and their effects in determining the spin ground
state have been assessed and the important effects of
solvation have been added [5,16]. Our group developed a
systematic assessment of how the balance of different terms
contributes to large variations in redox potentials comparing
1Fe, 2Fe2S, and 4Fe4S clusters (the latter for high-potential
and Fd-type redox couples) and this compares well with
experimental trends, although absolute redox potentials
remain difficult to predict with high accuracy. For redox
potentials, solvation effects counter a major part (but not all)
of the difference in electron–electron repulsion between
reduced and oxidized states. Spin barycenter-based analysis
clarifies the contribution of Heisenberg coupling to redox
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
261
Figure 3
The Heisenberg spin ladder for an oxidized 2Fe2S complex (2Fe3+,
with site spins S1= S2= 5/2 and total spin S = 0-5) is shown. These
and the ground S = 0 state are contrasted with the BS versus HS
(where S = 5) energies, right column. On the left side, the spin
barycenter (spin multiplicity weighted average of spin state energies) is
shown, which is the ‘effectively spin nonbonded’ state. By contrast, the
BS state contains most, but not all of the AF spin coupling in the spin
ground state (S = 0).
S = 5
S = 4
S = 3
S = 2
S = 1
S = 0
BS
S = 0
S = 0
Spin
barycenter
Uncoupled state
HeisenbergBroken symmetry
Current Opinion in Chemical Biology
9J
J
(25/2)J
(5/2)J
HS

Page 4

potentials, while resonance delocalization has a large effect in
4Fe4S systems, differentiating high-potential and Fd-type
couples along with electrostatic and solvation terms [17]. Our
group and others have examined vibronic influences on
cluster energetics, spin coupling parameters and valence
delocalization versus localization [18,19,20•,21].
Protein and solvent versus electronic effects
on redox potentials
For the redox potentials of proteins, the ‘reaction field
contribution’ (which has a large contribution from the
orientational polarization of aqueous solvent near the cluster
and also from protein polarization) makes a very large
contribution to the redox potential. However, in a comparison
of two 2Fe2S proteins homologous in the iron–sulfur binding
domain, we find that the difference in redox potentials
between these proteins is dominated by differences in
NH–S hydrogen bonding, particularly a flip of a single main
chain peptide bond orientation (comparing the two proteins
Anabaena Fd and phthalate dioxygenase reductase [PDR]).
Charged hydrogen bonding interactions between the
anionic cluster and the surrounding 8 to 10 main chain
(NH) and side chain groups are strong and increase in the
oxidized→reduced transition, having significant absolute
and differential effects on the redox potentials [22,23•,24•].
Reorganization energies and kinetics
As another consequence of large sulfur-to-iron covalency,
which is much greater for Fe3+sites than for Fe2+sites, on
1e–reduction (in a valence-trapped system) there is a
much smaller change in total electron density on iron than
is indicated by the formal valence difference. Systems with
valence delocalization, which are usually delocalized
between specific pairs of iron sites, have even smaller
electron density differences between redox states at iron.
The geometric changes are correspondingly smaller.
Both metal–ligand covalency and valence delocalization
lead to comparatively small reorganization energies. The
total reorganization energy is defined as the energy
required to move an electron from a reduced cluster (Ared)
to the corresponding oxidized cluster (Aox) when the
geometries of these clusters are not allowed to change from
the initial states and the surrounding protein and solvent
are not allowed to rearrange. It can be divided conceptually
into ‘inner sphere’ and ‘outer sphere’ parts. Inner sphere
reorganization energies include only the energies from
‘inner sphere’ geometric difference between oxidized and
reduced species, and excludes the cost of surrounding
solvent and protein reorganization, the ‘outer sphere’ term.
(The ‘inner sphere’ and ‘outer sphere’ terms are both positive
and additive.) The total reorganization energy (λ) can be
examined experimentally, whereas the ‘inner sphere’
reorganization energy has been the primary focus of recent
DFT computational studies. There is also a long history
of continuum models for the outer sphere λ and bond
distortion estimates for inner sphere λ. It is well known that
reorganization energies are one of the principal parameters
262
Guest section: computational bioinorganic chemistry II
Figure 4
Spin ladders for Heisenberg coupling and
with the additional resonance delocalization
energy (with resonance parameter B) are
shown for a mixed-valence reduced 2Fe2S
dimer (Fe2+–Fe3+). These are compared with
the BS picture showing the BS state and
HS delocalized bonding (g)and antibonding
(u) states with S = 9/2. Valence trapping is
neglected, so all states in the resonance B
column are totally delocalized.
S = 9/2
S = 7/2
S = 5/2
S = 3/2
S = 1/2
Current Opinion in Chemical Biology
10B
10J
2J
7J
9/2u
9/2g(HS)
9/2av
BS
S = 1/2
Heisenberg Resonance BBroken symmetry Uncoupled state
Spin
barycenter
2B
(3/2)J
S = 1/2

Page 5

determining electron transfer rates between different
centers in addition to the ∆G0for the reaction according to
the semiclassical Marcus theory [25]. Ryde and co-workers
[26•,27•] have quantitatively calculated inner sphere
reorganization energies with DFT in several iron–sulfur
complexes, comparing 1Fe, 2Fe, and 4Fe systems. They
found that the ‘inner sphere’ (first ligand shell) reorganiza-
tion energies for (Fe(SR)4) rubredoxin < (Fe4S43+,2+)
high-potential iron–sulfur proteins (HIPIPs) < (Fe4S42+,1+)
4Fe Fds < (Fe2S22+,1+) 2Fe Fds. Calculations in vacuum
give these energies as 0.41 eV (1Fe, Rd) < 0.45 eV
(4Fe HIPIP) < 0.64 (4Fe Fd) < 0.83 eV (2Fe Fd). It is
reasonable from the viewpoint of valence delocalization
(present in the 4Fe HIPIP and 4Fe Fd, but absent in most
2Fe Fds) that the 4Fe λ values would be lower than in the
2Fe Fd, but other aspects of these results are surprising,
for example, the low λ value for 1Fe rubredoxin. Ryde and
co-workers [26•] also found that their inner sphere λ
results also strongly depend on their treatment of the
environment, with the λ in the presence of either a fixed
or flexible protein environment being much lower than
found in their calculations of cluster in vacuum.
Bominaar and co-workers [28–30] have analyzed the
effects of valence delocalization on electron transfer kinetic
rates, considering both the expected effects of changes in
λ and changes in driving force ∆G0. Because there is a
low-lying ladder of spin states, which depends on both
Heisenberg exchange and resonance delocalization, these
also affect electron-transfer rates. Kummerle et al. [31•]
have evaluated the total reorganization energy experimental-
ly for electron transfer between the 2[Fe4S42+,1+] clusters
of Chromatium vinosum 8Fe Fd and its mutants. These
experiments are based on NMR line broadening as a
function of the driving force ∆G0(due to the redox potential
difference between clusters) and give a low total λ in the
range 0.2–0.5 eV. Babini et al. [32•]. examined electron
tunneling through ruthenium-bipyridine-histidine modified
HIPIP, and obtained a higher total λ 0.6–0.9 eV. These
results contrast with the calculated trend in ‘inner sphere’
λ found by Ryde et al., but the ‘inner sphere’ (vacuum) λ
can be altered by the protein surroundings, and experi-
mentally total λ is examined.
Patterns of localization, delocalization and
spin states
On ligand binding, the valence of various iron sites can
change and the pattern of spin alignment among these
sites is also altered. Valence interchange of the labile iron
site from Fe2.5+to Fe2+(where the exogenous ligand is
bound) with another cubane iron site (from Fe2+to Fe2.5+)
in Pyrococcus furiosis (on cyanide binding) and similarly in
reduced aconitase (on citrate binding) was monitored
by ENDOR, paramagnetic NMR, and by spin-coupling
analysis [33,34]. The use of spin projection coefficients
facilitates this analysis. Spin projection coefficients Kit(for
site i) give the projection of the iron-based site spins onto
the total system spin, Kit= < SiStotal> / <StotalStotal>, and
are essential tools for the analysis of hyperfine couplings
on the various sites. They depend on the spin-coupling
scheme and state. For observed hyperfine coupling
parameters Aiand intrinsic site couplings ai, the relation
Ai= Kiaiholds. aidepends also on metal–ligand covalency
through a reduction factor dB(i), so that aI= dB(i)i(aIis a
purely ionic intrinsic iron site coupling). DFT calculations
can be used to evaluate dB(i), or more phenomenological
approaches can be used based on ai, Kit. The Kitobey both
sum and chain rules, which further aids analysis [5,33–36].
Even without ligand binding, the spin coupling pattern
and associated Kitfor [Fe4S4]1+,3+clusters can be compli-
cated, and a comparison of experimental and calculated Kit
(the latter from spin-coupling models) is quite valuable.
Moriaud et al. [37•]. measured proton anisotropic and
isotropic hyperfine couplings. The through-space spin
dipolar coupling due to the effective spin densities on the
iron and sulfur sites produces the proton anisotropic
coupling. The protons serve as ‘probes’ of the ‘effective
local magnetic moments’ (integrated site spin densities on
iron and sulfur). With DFT covalency factors on the iron
and sulfur sites, experimental Kitwere determined and
compare well with some feasible spin-coupling models.
These methods can also be used to refine the proton
geometries with respect to the cluster core. Another
paramagnetic center can also serve as a probe of the effective
spin density distribution on an iron–sulfur cluster, which is
observed as a splitting or broadening of the EPR spectra.
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
263
Figure 5
Energy level splitting patterns for a single iron site in either a
mononuclear or polynuclear system showing the effects of
(a) tetrahedral or distorted tetrahedral iron coordination and
(b) trigonal or distorted trigonal coordination. A single minority spin
electron is placed in the lowest ligand field orbital in each case.
or
Trigonal
Distorted
Current Opinion in Chemical Biology
Tetrahedral
Distorted
a1
e
e
e
t2
(a)
(b)

Page 6

This is particularly useful in determining the geometric
position of the paramagnetic center, and has been effec-
tively developed by Bertrand and co-workers [38,39•].
Using the spectroscopy of the mixed valence delocalized
Fe2+–Fe3+dimer, [Fe2(OH)3(Me3TACN)2]2+(where TACN
is triazacyclononane) as a guide, the geometric depen-
dence of the B resonance term in mixed valence 2Fe2S
dimers was estimated, giving a large B value at the Fe–Fe
minimum of B = 965 cm–1at r(Fe–Fe) = 2.73 Å compared
with B = 1350 cm–1in the diiron-trihydroxyl species,
r(Fe–Fe) = 2.51 Å [7]. The large B value and the vibronic
coupling (‘dynamic trapping energy’) also influences the
‘observed’ effective Jeffvalue in the mixed-valence 2Fe2S,
causing a decrease in AF coupling and partial delocaliza-
tion compared with the ordinary Heisenberg coupling term
[19,40]. On the basis of DFT calculations, we showed that
there is significant geometric dependence of spin states
and Heisenberg J parameters in oxidized diferric 2Fe2S
dimers; by contrast, mixed-valence 2Fe2S dimers are
expected to have very similar geometries and also fairly
close energies because of the competition between
Heisenberg exchange coupling favoring trapped valence
and low spin (St= 1/2) and resonance valence delocaliza-
tion favoring delocalized high spin (St= 9/2) [4,20•]. This
provides a rationale for why Fe–Fe distances are similar for
mixed valence pairs in 2Fe2S complexes and in 4Fe4S
complexes, and also for the feasibility of Cys→Ser 2Fe2S
mutants displaying a physical mixture of St= 1/2 localized
and St= 9/2 delocalized states [6•].
To determine J couplings and resonance delocalization B
parameters, a valuable alternative to magnetic susceptibil-
ity methods is provided by solid state 13C NMR studies,
which have been applied to synthetic Fe4S42+complexes
[41•]. Improved methods for experimentally determining
J and B coupling parameters in a variety of systems will
prove important for understanding the physical basis of
spin couplings and spin alignment patterns.
Proton coupling to electron transfer in Rieske
iron–sulfur proteins
Rieske iron–sulfur proteins play a vital role in the linkage
between electron transfer and proton pumping through the
electron transport chain of mitochondria and in photosyn-
thesis (as parts of the bc1and b6f complexes containing one
Rieske iron–sulfur protein two b-type hemes and one
heme c1or f). The linkage between electron and proton
transfer is indicated by the pH dependence of the redox
potential now measured over a wide range in the Rieske
fragment from Thermus thermophilus [42•]. This and
other experimental and computational evidence ([43•];
M Ullmann, L Noodleman, DA Case, unpublished data)
points to an imidazolate-to-imidazole protonation event
associated with 1e–reduction at the Fe3+→Fe2+coordinated
to two histidines. The cluster has the coordination
(Cys)2FeS2Fe(His)2and is diferric in the oxidized form and
mixed valence Fe3+,Fe2+(trapped) in the reduced form.
Our group’s recent calculations on the structurally
characterized bovine Rieske fragment agrees well with the
independent measurements of Zu et al. [42•] on the
T. Thermophilus protein. In the oxidized form, one of these
two histidines loses a proton at about pH = 7.5 (bovine) or
7.9 (T. thermophilus). (The second proton is lost at higher
pH near 9.) For the reduced cluster, the pKas are much
higher, near 12. Our calculations indicate that for the
oxidized cluster, the apparent pKainvolves a statistical
mixture of deprotonations at either histidine imidazole.
This clear example of linkage between 1e–reduction and
protonation now needs to be compared with other systems
having either different or similar protonation sites.
Nitrogenase overview
Biological nitrogen fixation involves two different proteins
and three iron–sulfur cluster types in the nitrogenase
enzyme complex [44•,45]. The iron protein contains a
single 4Fe4S, which transfers electrons probably in 1e–
264
Guest section: computational bioinorganic chemistry II
Figure 6
For the docked iron protein to the MoFe protein of nitrogenase, the Fe
clusters probably participating in electron transport (4Fe cluster of the
Fe protein, 8Fe P cluster in the MoFe protein) to the catalytic FeMo
cofactor (M center) are shown along with protein residues along
the pathway.
MoFe
Fe Protein
Leu158
Pro155
2.3 Å
Cys154
P Cluster
Cys62
2.3 Å
Gly61
Gln191
Hca494
His442
Cys275
Current Opinion in Chemical Biology
MoFe
Fe P MoFe
Possible electron transfer path
3.5 Å
3.0 Å
2.3 Å
2.3 Å
2.8 Å
2.8 Å
AlF4– stabilized
nitrogenase

Page 7

steps to the MoFe protein, where multiple electron and
proton transfer steps are utilized for the reduction of
molecular nitrogen to ammonia plus hydrogen, specifically
N2+ 8e–+ 8H+→2NH3+ H2. Docking of the iron protein
to the MoFe protein is dependent on MgATP binding
(and perhaps also on ATP hydrolysis) to the iron protein.
In Figure 6, we show the arrangement of the three
iron–sulfur cluster types after protein–protein binding
based on the protein–protein structure found in the
ADP.AlF41–stabilized form in the X-ray structure of
Schindelin et al. [46]. This structure is very suggestive of
the likely electron transfer path from the 4Fe4S protein
cluster to the P cluster and then to the FeMo cofactor. ATP
binding causes both a conformational change in the iron
protein that is transmitted to the 4Fe4S cluster, and there
is also substantial evidence showing that the redox poten-
tials of the 4Fe4S cluster and the P cluster are shifted after
nucleotide binding and docking [47]. The combination of
ATP binding, docking, and electron transfer, as well as the
strong dependence of properties of the 4Fe4S cluster on
solvent show that this cluster has novel and functionally
critical structural and redox properties.
The MoFe protein contains two P clusters of composition
(8Fe7S) involved in electron and (probably) proton transfer
and two FeMo cofactor clusters (also called M centers,
Mo7Fe9S) where the first three or four electrons and
protons are accumulated (based on substantial kinetic and
spectroscopic evidence) [45] and subsequently nitrogen is
bound and reductive cleavage initiated. The MoFe protein
has an α2β2subunit composition with each M center
bound within an α subunit, whereas each P cluster is
bound between α and β subunits. Each P cluster and the
single closest M center acts as an independent electron
transfer and catalytic unit for nitrogen reduction. The
detailed electron transfer and catalytic transformation
steps are under extensive experimental and theoretical/
computational study.
The all-ferrous 4Fe4S cluster in the iron protein
of nitrogenase
The 4Fe4S cluster in the iron protein has a 1e–redox
potential for the [4Fe4S]2+,1+couple of –360 mV, and electron
transfer to the MoFe protein was thought to involve this
couple exclusively [48]. However, Watt and Reddy [49]
concluded from redox experiments that moderate
reducing conditions could generate the all-ferrous cluster
[4Fe4S]0; this is Fe4S4(SR)44–including the terminal cys-
teines. The question then arises as to whether the 4Fe4S
Cluster is a 1e–transfer protein using the [4Fe4S]2+,1+
couple, or performs 2e–transfers (either in 1e–or in 2e–
concerted steps). The 2e–option would potentially require
only half the number of dockings, and could change the
assessment of the role of ATP binding and hydrolysis as
well. Further, the P cluster has two successive redox
potentials that are nearly the same (–300 mV) and physio-
logically relevant, so that 2e–transfer by the iron protein
would be potentially efficient. This iron–sulfur cluster is
more solvent-exposed when compared with those in 4Fe
HIPIP and Fd proteins. One X-ray crystal structure probably
contains the 4Fe4S cluster in the all-ferrous state [50•].
However, the 2.25 Å resolution is not accurate enough to
provide cluster geometries that can clearly distinguish
different Fe–Fe distances; at this stage, the distances
derived from EXAFS spectroscopy are probably more
accurate [51]. The best fit to the EXAFS data gives a
4:2 short-to-long pattern of Fe–Fe distances, with Fe–Fe
distances near 2.52 Å and 2.77 Å, respectively. By compar-
ison, the X-ray structure gives a 2:4 short-to-long pattern
with Fe–Fe in the ranges 2.54–2.57 Å and 2.66–2.79 Å
[50•]. The iron protein has also been examined by
Mössbauer and EPR spectroscopies [52•]. These show a
total cluster spin Stotal= 4 composed of four HS Si= 2 sites
coupled in a 3:1 alignment of up:down spin vectors. From
the spin alignment pattern alone, one would expect a 3:3
pattern of short-to-long Fe–Fe distances reflecting three
ferromagnetic and three AF Fe–Fe interactions. The
observation of a less symmetric pattern by EXAFS and
X-ray structure probably reflects cysteine ligand side chain
orientation or longer-range protein and solvent effects.
We have very recently calculated the electronic structure
and optimized geometries of the all-ferrous 4Fe4S cluster
in the iron protein using non-local DFT methods
(VWN–Becke–Perdew potential), and including the
interaction with the protein and solvent environment
(beyond the cluster thiolate ligands) using electrostatic
Poisson–Boltzmann methods for the protein and reaction
fields (combined ADF/MEAD method) (RA Torres,
T Lovell, L Noodleman, DA Case, unpublished data).
Three different BS spin coupling alignments were compared
for the all-ferrous quantum cluster Fe4S4(SCH3)44–: first,
all Fe2+spin vectors parallel aligned (HS with Stotal= 8);
second, three parallel (spin-up) Fe2+spin vectors with one
oppositely aligned (spin-down) (BS1 with Stotal= 4); third,
two spin-up and two spin-down vectors (BS2 with
Stotal= 0). The gas phase energies for the geometry-
optimized clusters give the two BS states with Stotal= 0 or 4
clearly as the lowest energy states, but these are very close.
When the interaction with the protein and solvent
environment is included, the Stotal= 4 state is the most stable
by 0.4 eV. We then calculated the redox potential for the
Fe4S41+,0couple in the full protein and solvent environ-
ment. Our predicted redox potential is –0.799 V. This is
considerably more negative than the redox potential
reported by Watt and Reddy (–0.460 V) using methyl
viologen. However, they stated that this 1+/0 reduction
gave no change in the UV/vis spectrum, and no spectro-
scopic characterization of this state has appeared. Burgess
and co-workers [53•] have very recently reinvestigated this
problem. They were unable to reduce the iron protein
below the 1+ oxidation state using methyl viologen.
However, they did produce a ‘pink’ all-ferrous form of the
iron protein using lower potential reductants, and
structurally and spectroscopically characterized this state.
The redox potential required to attain the all-ferrous state
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
265

Page 8

is far more negative than found by Watt and Reddy,
consistent with our calculated values (–800 mV). Holm’s
group [54•] has studied analogous synthetic complexes for
the 1+/0 redox transition and also find quite negative
redox potentials.
For the geometry of the all-ferrous cluster, the DFT
calculated optimal geometry for BS1 Stotal= 4 shows a
short:intermediate:long ratio of 1:4:1, with distances
2.63:2.76 (mean):2.87 Å, with the shortest distance corre-
sponding to a ferromagnetically coupled Fe–Fe pair, the
longest to an antiferromagnetically coupled Fe–Fe pair,
and the intermediate distances to both F and AF pairs.
The calculated distances more closely resemble those from
the X-ray structure analysis than from the EXAFS data.
However, the calculated distances lie within the 95%
confidence interval of the EXAFS data, suggesting more
accurate structural data are needed. Isosurface plots of the
delocalized minority spin electrons (the sixth d electrons
on the three parallel spin Fe2+sites) show extensive iron-
to-iron d–d bonding interactions among these three iron
sites. These data show that the calculations give a good
account of redox properties, electronic structure in
complex environments, and structures that are reasonable
compared with available EXAFS and X-ray structural
data [48,50•,52•,55,56••].
P clusters of nitrogenase: X-ray structures,
spin coupling and redox properties
The all-ferrous (PN) form of the 8Fe7S P cluster (Figure 7)
is the probable electron donor to the FeMo cofactor center.
As shown in this Figure [35], in the 2e–oxidized (POX)
cluster form, the likely total cluster spin is Stotal= 3 or 4
(probably Stotal= 4) composed from cluster subunit spins of
S1= 1/2 and S2= 7/2. Formally, each subunit is like an
Fe4S41+cubane, but with each cubane sharing a sulfide
corner atom. We assigned the S2= 7/2 subunit to the more
distorted cubane as a consequence of the Ser-β-188 coordi-
nation and/or the deprotonated amide coordination from
Cys-α-88 to two of the iron sites in this cubane. In addition
to lying directly on the path from the 4Fe4S cluster of the
iron protein when docked to the MoFe protein as shown in
Figure 6, PNhas two successive redox potentials near
–300 mV, in the proper range for reduction by the iron
protein and electron transfer to the FeMo cofactor.
Potentially, it also could provide a proton transfer pathway.
The redox poise of the MoFe protein can be adjusted
to obtain clusters in the form M(OX)P(OX), whereas in the
‘resting’, more reduced environment, the spectroscopically
observed state is M(N)P(N)[57,58]. Lanzilotta et al. [58] have
found from the pH dependence of the redox potentials and
mutational studies that the relevant POX→PNredox steps
are POX+ 1e–+1H+→ P1+and P1++ 1e–→PN, with both
potentials near –300 mV. The mutation of Ser-β-188 to
glycine does not eliminate the pH dependence of the first
redox potential, but the spin state changes because the
integer spin EPR signal of the serine native form
disappears. Lanzilotta et al. infer that the protonation
occurs at the Cys-α-amide, not at Ser-β-188. However, the
X-ray structures of POXand PN(see Figure 7, based on
Peters et al. [57]) show clear structural changes at both
Ser-β-188 and Cys-α-88 as if two protons were involved.
This apparent contradiction is still unresolved. It may be
that the X-ray structure of M(N)P(N)or possibly M(OX)P(OX)
actually involves a mixture of oxidation states. We are
examining these different oxidation and protonation states
computationally to clarify the different possibilities.
Further structural, physical and spectroscopic studies of
this problem are clearly needed.
MoFe cluster of nitrogenase: oxidation state,
spin coupling pattern and electronic structure
Figure 8 shows schematically the structure of the ‘resting’
MNform of the FeMo cofactor cluster of nitrogenase. We
note the unusual three-coordinate iron sites of the central
6Fe prismane fragment. A more complete picture of the
full FeMo cofactor center is shown in Figure 6. From
another perspective, useful for spin coupling and bonding
analysis, there is a cubane type MoFe3fragment (with
spin Sa) that is coupled to a cubane type Fe4(spin Sb)
266
Guest section: computational bioinorganic chemistry II
Figure 7
The P clusters from the X-ray structures are shown for (a) the resting
PNand (b) the two-electron oxidized POXstates. The total spin (Stot) is
shown for PNand POX, and the proposed spins (S1, S2) of the subunit
4Fe cubanes in POXare shown. S2is for the more geometrically
distorted cubane where resonance delocalization is probably disrupted.
S
Fe
S
Fe
S
Fe
Fe
S
Fe
S
S
S
N
O
H
S
Fe
S
Fe
Fe
S
S
S
OH
S
S
S
S
S
Cys-β153
S
S
S
S
S
S
S
Fe
Fe
Fe
Fe
Fe
Fe
Fe
Fe
S
O
N
O
S
Cys-α88
POX
PN
Ser-β188
Cys-β70
Cys-β153
Cys-β95
Cys-α62
Cys-α154
Cys-α88
Ser-β188
Cys-β70
Cys-β95
Cys-α62
Cys-α154
Stot = 0
S2 = 7/2Stot = 3 or 4S1 = 1/2
(a)
(b)
Current Opinion in Chemical Biology

Page 9

fragment by three bridging µ2sulfide ligands. Compared
with typical cubane fragments, there is a missing sulfide
corner to each of these cubanes, producing the under-
coordinated central iron sites and the µ2sulfides replace
terminal sulfur (cysteine) coordination. One capping iron
has the more usual tetrahedral coordination to S(Cys) and
bridging sulfur, whereas the molybdenum is six-coordinate
to homocitrate, histidine and sulfide. Figure 8 (top) depicts
MNwith oxidation state Mo4+6Fe2+1Fe3+in accordance
with the spectroscopic properties and energies predicted
from our recent DFT and electrostatics calculations (using
a Becke–Perdew [BP86] exchange-correlation potential
and with the protein and solvent treated as dielectric
media with embedded charges by the Poisson equation)
[59••] (T Lovell, J Li, DA Case, L Noodleman, unpub-
lished data for the electrostatics and energetics in the
protein environment). These calculations included
geometry optimization of the large cluster, including the
full homocitrate ligand, the bonded histidine imidazole
side chain and cysteine side chain (represented by
SCH31–), as well as H-bonded Gln191 side chain and four
waters. Although the total cluster spin is known from
EPR and Mössbauer spectroscopy, Stot= 3/2, the cluster
oxidation state has been controversial because of different
analyses of Mössbauer [56••,60] and ENDOR spectro-
scopies [61], and the pattern of spin coupling has been
unresolved. By contrast, Mössbauer hyperfine parameters
for six iron sites are very consistent with those derived
from ENDOR, and the small Mössbauer hyperfine
coupling for the elusive seventh iron site has recently been
found [56••,61]. In the diagram, based on our calculations,
the thick spin vectors represent the majority spin electrons
(formally five d electrons per iron site), whereas the thin
spin vectors represent the sixth minority spin electron per
iron site for the 6Fe prismane (6Fe2+), the terminal iron
site being Fe3+. Of 10 basic spin alignment patterns, all
being consistent with Stot= 3/2, (10 different BS states
BS1–BS10), the one shown, called BS6, gave the best
assemblage of hyperfine, Mössbauer isomer shift, geometries,
spin densities and energetics for the resting MNstate, and
also rationalizes the Mössbauer isomer shift and hyperfine
data for one-electron reduced states (Figure 8; middle and
bottom). The spin coupling pattern in BS6 giving good
agreement with the iron hyperfine on all iron sites can be
summarized as MoFe3spin Sa= 2 antiferromagnetically
coupled to Fe4Sb= 7/2. There are, in fact, three energeti-
cally close-lying spin isomers, BS6-1, BS6-2 and BS6-3,
which differ only by cyclic permutation of the iron site
spins in the MoFe3fragment. There are several important
structural, energetic and spectroscopic issues that have
been analyzed [56••,59••,61,62••,63••,64].
Here, we briefly examine the question of the oxidation
state of the ‘resting state’ MNand then the character of the
one-electron reduced states MR, MI. The possible alterna-
tive oxidation states for MNare Mo4+6Fe2+1Fe3+(formally
43d total electrons on the metals, Mo4+(d2), Fe2+(d6),
Fe3+(d5)) and Mo4+4Fe2+3Fe3+(41d electrons) because
both can generate Stotal= 3/2. Yoo et al. [56••] proposed that
Mo4+4Fe2+3Fe3+is the likely state for MNbecause the
average isomer shift over all 7Fe sites is low, about
0.4 mm s–1, which lies well below that of the only known
Fe2+(SR)3three-coordinate model complex (0.57 mm s–1)
[65], and much closer to the expected weighted average of
4Fe2+3Fe3+. We have examined this model compound
interpolation argument using isomer shift calculations on
the Mo7Fe active site fit to the correlation for a series of
model complexes. We found that the interpolation from
model compounds to the actual Mo7Fe active site is not
reliable, and that a more direct isomer shift calculation for
this site favors Mo4+6Fe2+1Fe3+very significantly. Further,
the DFT calculations on the 4Fe2+3Fe3+active site shows
that this is highly oxidizing, and removes an electron from
the homocitrate ligand, yielding a homocitrate radical and
becoming effectively Mo4+5Fe2+2Fe3+. No such organic
radical is observed for the MNstate. By contrast, an organic
radical has been identified by EPR during turnover of
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
267
Figure 8
The core and flanking ligand atoms of the MoFe cluster of nitrogenase
are shown with the spin alignment proposed for state BS6-1 in
oxidation state MN. The spin alignment pattern is shown with thick
vectors representing the majority spins on the various sites and thin
vectors, the minority spin electrons (one electron per each Fe2+site as
shown). On 1e–addition of a spin-up electron, the catalytically active
reduced state MRis obtained (middle), while for 1e–addition, spin-
down, the radiolytically reduced state MIis obtained. The orbital shapes
for the MNLUMO filled for MRand MIare shown (middle, bottom).
Fe2+
Fe2+
S
Fe3+
Fe
Fe2+
S
S
S
Fe2+
S
S
Fe
S
S
S
Mo
O
O
N
S
S=3/2 MN
S=2, MR
Mainly Mo
O
O
N
S
S=1, MI
Delocal. Fe
2+
O
O
N
S
Current Opinion in Chemical Biology

Page 10

acetylene in a nitrogenase MoFe protein mutant
(His195Gln) [62••,63••,64], which may be related to the
homocitrate radical we have found. This would again be
consistent with the Mo4+6Fe2+1Fe3+oxidation state for MN.
In the full protein-solvent environment, the three spin
isomers, BS6-1, BS6-2 and BS6-3 (lowest) are very close
in energy, within about 2 kcal mol–1. However, BS6-1
uniquely has lowest unoccupied molecular orbitals
(LUMOs) for the up and down spin electrons that best
match the expected character of the 1e–reduced state
as exhibited by Mössbauer isomer shifts and hyperfine
spectra. In Figure 8 (middle and bottom), we depict the
LUMOs filled on one-electron reduction: under catalytic
turnover, MN+1e–→ MRwith Stotal= 2, whereas under
(nonphysiological) radiolytic reduction, MN+1e–→ MIwith
Stotal= 1 as expected based on the Mössbauer spectra
[56••]. The catalytic state MRis generated by filling a
spin-up LUMO, aligned with the net system spin of
MN(Stotal= 3/2); this orbital has considerable molybdenum
character and less iron character delocalized over two of
the three iron sites (the two net spin-down sites) within
the Mo3Fe cubane. By contrast, the spin-down LUMO
filled to generate the radiolytic state MI(oppositely
aligned to the net spin of MN) is delocalized mainly across
four of the six central prismane iron sites, specifically
those Fe2+sites that are net spin-up (see the thick arrows
in Figure 8 [top panel]). These patterns occur because
the LUMO electron is always minority spin, and can
delocalize across iron sites with parallel spin alignment. In
Figure 9, the more typical pairwise valence delocalization
for a mixed-valence Fe2.5+–Fe2.5+pair in a reduced
[4Fe4S]1+cluster is shown, which contrasts with the delo-
calization pattern in the MoFe7S9 cluster (Figure 8, middle
and bottom). The MoFe cluster shows valence inter-
change on 1e–reduction of MN, with MRcorresponding to
Mo3+6Fe2+1Fe3+, whereas the energetically close MI
reduced form has valence composition Mo4+5Fe2+1Fe1+1Fe3+
with some multicenter valence delocalization in both
states. Based on the evolution of H2and the reduction of
N2to NH3, proton binding to the FeMo cofactor should
occur upon electron addition to the cluster. The main
questions involve where the first and subsequent protons
bind as electrons are transferred into the Mo7Fe cluster,
what are the energetics of this process, and how the pro-
ton binding at one redox step can affect subsequent redox
steps and eventually N2binding and transformation. The
LUMO orbitals shown in Figure 8, when filled, create
added electron density at selected metal sites, and are
suggestive of the lowest energy binding sites for the
first proton bound to the cluster in the MRand MIstates.
Quantitative DFT and protein electrostatics work
from our group on these important problems is ongoing
(T Lovell, J Li, DA Case, L Noodleman, unpublished data).
Several research groups have examined the electronic
structure of the FeMo cofactor and potential reactions
with molecular nitrogen (and CO in some cases)
[66–72,73••]. The calculations differ widely in the
complexity of the cluster models used, in the oxidation
state assumed for the cluster, and in the type of quantum
chemistry method used. They examined also the effects
of successive addition of electrons and protons to the
FeMo cofactor cluster ‘resting state’ MN. In the presence
of cluster bound N2, proton and electron transfer to nitro-
gen can occur with N–H bond formation and successive
breaking of the N2triple bond on the cluster surface or in
the cluster interior. The most important and recent
studies of this type employing DFT methods are by Rod
and Norskov [73••]. The earlier work by Siegbahn et al.
[72] with DFT methods and by Stavrev and Zerner
[69–71] using largely spin-restricted open-shell ZINDO
(Zerner’s intermediate neglect of differential overlap)
calculations are also notable.
268
Guest section: computational bioinorganic chemistry II
Figure 9
A [Fe4S4]1+reduced 4Fe cluster is shown with a valence delocalized
electron across the top Fe–Fe pair. The occupied delocalized minority
spin orbital composed of Fe(1)dx2–y2+Fe(2)dx2–y2is shown displaying
σ Fe–Fe bonding (z is the vertical axis). This simple σ delocalization
where each iron is four-coordinate with delocalization across the
top rhombic unit contrasts with the more complex delocalization
patterns in the MoFe cluster for states MRand MIassociated with the
three-coordinate central prismane iron sites and the molybdenum site.
S
Fe2.5+
Fe2.5+
S
Fe2+
Fe2+
S
S
Current Opinion in Chemical Biology

Page 11

Rod and Norskov [73••,74,75] performed three-dimensional
solid-state type (periodic boundary conditions) DFT
calculations on two model structures (I,II). Model I consists
of a periodically repeated head-to-tail chain structure of
composition MoFe6S9(neutral) with subunit spin Stotal= 0.
Model II consists of discrete molecules of composition
(SH)Fe7S9Mo(NH3)(OH)2with spin ground state Stotal= 3/2,
constructed as a 3D array of cluster supercells. This is also
neutral, so that the iron composition in Model II is
4Fe2+3Fe3+and 4Fe2+2Fe3+in Model I, assuming Mo4+for
both. They used an SCF spin-dependent Perdew–Wang
(PW91) potential, and subsequently either a non-SCF
revised Perdew–Burke–Enzerhoff (RPBE) potential for
structural Model I or an SCF RPBE potential for Model II.
Despite the difference in oxidation state between their
Model II and the Mo4+6Fe2+1Fe3+state we propose for
MN, they found a ground state with a spin alignment like
that in BS6 (Figure 8, top), similar to what we find for the
same oxidation state in our calculations. Models I and II
were compared for geometric and electronic structures and
for N2and CO binding energies. Both models give similar
weak N2-binding energies (about 2.5 kcal mol–1), and much
larger CO-binding energies (about 25 to 27 kcal mol–1) to
the same binding site with end-on binding to a single
prismane iron. This is suggestive of the inhibition of N2
binding and reaction in the presence of CO. Also, the
stabilization of CO binding by an external H-bonding
cationic group (a protonated base) and the resulting large
negative frequency shift of the CO stretch resembles the
large observed band shift on CO binding. The cluster
binding of protons and electrons and reactions with N2
were calculated for the simpler Model I. The first proton
binding in the 1e–reduced state is found to be most strongly
favored to µ2S (∆E = –2.5 kcal mol–1) followed by terminal
binding to an iron site (about 10 kcal mol–1less stable), and
with very poor binding to the µ3S (26 kcal mol–1) less
stable. This same trend persists with successive electron
and proton addition to the cluster; the first 3H+go to µ2S,
whereas the fourth H+goes to an iron terminal site as three
or four electrons are added to MN, followed by a reorganization
to form H2at iron, which may then desorb. Depending on
the electron transfer rate, H2may be released after two or
three hydrogen atoms are added in agreement with the
Thorneley–Lowe scheme. Very reasonable energies for
successive hydrogen atom transfers (coupled electron-
proton transfer) to N2are shown, with the first hydrogen
atom transfer to form bound N2H, the energy requiring
and therefore rate-limiting step as expected from the
energetics of free N2and hydrogen.
Siegbahn et al. [72] used simplified model clusters (assuming
ferromagnetic coupling) to study N2binding to reduced
iron–sulfide–thiolate clusters derived from fragments of the
FeMo cofactor center. First, iron dimer models were examined.
In their notation: (1a) (HS)(H2S)Fe(II)-S-Fe(II)(H2S)(HS) and
(1b) (HS)(H2S)Fe(II)-SH-Fe(I)(H2S)(HS). Structure (1b) is
derived from (1a) by addition of one H atom, which then
changes the formal Fe oxidation state of one site from
Fe(II)→Fe(I). Structure (1b) binds N2strongly in a bridging
mode while structure (1a) binds N2only very weakly end-on to
one Fe(II) site. They also examined hydrogen atom plus N2
addition to an all-ferrous [Fe8S9]2–(that is, replacing the
octahedral molybdenum site with a trigonal iron site, and
where all other iron sites are also three-coordinate) cluster with
no other ligands. This cluster opens a central cavity on binding
hydrogen and the combined hydrogen atom plus N2is
strongly bound, with N2side-on coordinated to 4Fe sites of the
central prismane.
The simplified clusters studied by Siegbahn et al. are more
highly reduced than those of Rod and Norskov. The energetic
results found by both groups are reasonable, but not at all
definitive. Both groups have emphasized neutral clusters
(Siegbahn considered also some anionic clusters, as above),
and both have focused nearly exclusively on electroneutral
reactions. There is little consideration of the protein/solvent
environment or the involvement of the terminal parts of
the cluster. In Rod and Norskov’s paper, there is, however,
a brief but valuable analysis of the case where 1e–is added
to the cluster with proton transfer from an external proton
donor cation, comparing proton transfer to N2with CO. By
contrast, our recent calculations and associated analysis
[59••] indicate that the active site cluster is probably
negatively charged in most relevant oxidation states (whether
or not the two terminal [noncoordinated] carboxylate
groups of the homocitrate are included in the charge
count). The negatively charged FeMo cofactor is closely
surrounded by several charged and polar side chains,
roughly with a cationic layer from arginine and lysine
inside an anionic layer from asparagine, glutamic acid and
homocitrate 2COO–and interacts as well with the protein
and solvent dielectric media. More globally, the peptide
main chain dipoles are also energetically important as they
orient to stabilize the alternating charged shells. This is an
unusual ligating structure for an iron–sulfur cluster where
typically main chain peptide groups H-bond to the cluster
via strong N–H–S hydrogen bonds [22,23•]. Apparently,
this unusual arrangement of charged and dipolar layers
organizes the active site for efficient electron and proton
transfer to the cluster. The principal questions include:
1. How (and to what extent) are electron transfer events
coupled to proton transfers, and how does this relate to N2
binding and reactions?
2. How are the redox potentials of the system and acidities
through the redox cycle related to simple hydrogen atom
addition energies?
3. To what extent is charge separation favored in various steps?
4. What are the preferred protonation sites, and do these
differ with the net cluster oxidation state?
5. How do these processes relate to N2binding and
subsequent reactions?
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
269

Page 12

6. What redox potentials and surrounding proton sources
and associated cluster and protein pKas are needed to drive
nitrogen reduction?
7. What are the comparative roles of molybdenum, iron
and sulfide in these processes?
8. Why is such a complex active site cluster required for
the FeMo cofactor?
Experimental and theoretical/computational work is
proceeding in several groups to further understand the
molecular logic underlying the catalytic pathways of nitro-
genase, and to understand how these pathways with their
electronic structure and associated protein environment
compare and contrast with other iron–sulfur proteins.
Conclusions
The FeMo cofactors (M centers) and P clusters of the
nitrogenase MoFe protein are still recognizable members
of the broad family of iron–sulfur protein clusters. All of
these possess HS iron sites, large spin polarization on the
iron sites, strong metal–sulfur covalency, and complex
spin-coupling patterns. In the one-electron reduced states
MRand MI, the FeMo cofactor shows valence electron
delocalization like that in 4Fe4S systems, but with a more
complicated pattern. The distinctive electronic features
of the M centers are connected with multiple three-
coordinate iron sites and with the six-coordinate molybdenum
site. The protein surrounding the M centers is organized
with many charged side chains pointing inward toward the
negatively charged cluster in contrast to the dominant
main chain NH–S hydrogen bonding in simpler 2Fe2S and
4Fe4S clusters.
On the theoretical/computational side, our group has
focused so far on analysis of the oxidation state of the
resting MNstate of nitrogenase and on the electronic
structure of the one-electron reduced and oxidized states
with a large active site cluster model. Other groups have
examined reactions with molecular nitrogen and hydrogen,
but using simpler cluster models. There has been substantial
theoretical and spectroscopic progress, but many issues
remain open. The P clusters and the iron protein cluster
also possess distinctive and functionally significant features
not seen in simpler clusters. On the experimental side,
more work is progressing on isolating states that occur later
in the catalytic cycle and deriving insights from alternative
substrates. From the computational side, larger models
with improved physical accuracy are in progress with
more specific and detailed predictions of spectroscopic and
energetic properties and reactions, to establish how the
novel cluster and protein structure leads to effective catalysis.
Acknowledgements
We would like to thank our former group members J Li, J-M Mouesca,
JL Chen and M Ullmann for their contributions to our work on iron–sulfur
systems and our long-time collaborators DA Case and D Bashford for
ongoing contributions. We thank LC Seefeldt, JW Peters, P Siegbahn,
T Rod, B Burgess, E Munck, B Hoffman and B Hales for discussions on
nitrogenase, JA Fee and CD Stout for discussions on the Rieske protein,
and ML Ludwig and D Ballou for discussions on phthalate dioxygenase
reductase and Fds. We also remember the helpful earlier work and
comments from MC Zerner who will be missed. This work was supported
by an NIH grant (GM 39914) to DA Case and L Noodleman.
References and recommended reading
Papers of particular interest, published within the annual period of review,
have been highlighted as:
• of special interest
••of outstanding interest
1.
••
This paper provides an excellent overview of the field from physical properties
to biosynthesis and regulatory aspects. Both electron transfer and non-electron
transfer catalysis are also covered.
Beinert H: Iron–sulfur proteins: ancient structures, still full of
surprises. J Biol Inorg Chem 2000, 5:2-15.
2.Johnson MK: Iron–sulfur proteins: new roles for old clusters. Curr
Opin Chem Biol 1998, 2:173-181.
3.
•
Li J, Noodleman L, Case DA: Electronic structure calculations:
density functional methods with applications to transition metal
complexes. In Inorganic Electronic Structure and Spectroscopy,
Vol. 1, Methods. Edited by Solomon EI, Lever ABP. New York:
John Wiley and Sons; 1999:661-724.
We present the methodology and applications of DFT applied to spin-
polarized and spin-coupled transition metal complexes. The electronic and
spin properties of iron–sulfur complexes are compared with a wide range of
transition metal complexes with one (metal-aquo and MnSOD), and two or
four metal sites (manganese-oxo). Solvation methods with applications to
redox potentials are also examined.
4.Li J, Noodleman L: Electronic structure calculations: density
functional methods for spin polarization, charge transfer, and
solvent effects in transition metal complexes. In ACS Symposium
Series 692: Spectroscopic Methods in Bioinorganic Chemistry.
Edited by Solomon EI, Hodgson KO. Washington, DC: American
Chemical Society; 1998:179-197.
5.Noodleman L, Peng CY, Case DA, Mouesca J-M: Orbital Interactions,
electron delocalization, and spin coupling in iron–sulfur clusters.
Coor Chem Rev 1995, 144:199-244.
6.
•
Achim C, Bominaar EL, Meyer J, Peterson J, Munck E: Observation
and interpretation of temperature-dependent valence
delocalization in the [2Fe–2S]+cluster of a ferredoxin from
Clostridium pasteurianum. J Am Chem Soc 1999, 121:3704-3714.
Mössbauer studies and temperature-dependent kinetics and spectroscopic
analysis are used to evaluate electron trapping versus delocalization in
the S = 1/2 and S = 9/2 fractions of the Cys56→Ser mutant of
C. Pasteurianum 2Fe Fd. The observed spectra are interpreted by a distribution
in delocalization B parameters, or alternatively by distributions in Heisenberg J
or reorganization parameters.
7.Gamelin DR, Bominaar EL, Kirk ML, Wieghardt K, Solomon EI:
Excited-state contributions to ground-state properties of mixed-
valence dimers: spectral and electronic structural studies of
[Fe2(OH)3(tmtacn)2 2]2+related to the [Fe2S2]+active sites of plant
type ferredoxins. J Am Chem Soc 1996, 118:8085-8097.
8. Noodleman L: Exchange coupling and resonance delocalization in
reduced [Fe4S4]+and [Fe4Se4]+clusters. 1. Basic theory of spin-
state energies and EPR and hyperfine properties. Inorg Chem
1991, 30:246-256.
9. Noodleman L: Exchange coupling and resonance delocalization in
reduced [Fe4S4]+and [Fe4Se4]+clusters. 2. A generalized
nonlinear model for spin-state energies and EPR and hyperfine
properties. Inorg Chem 1991, 30:256-264.
10. Auric P, Gaillard J, Meyer J, Moulis JM: Analysis of the high-spin
states of the 2[4Fe-4Se]+ferredoxin from Clostridium
pasteurianum by Mössbauer spectroscopy. Biochem J 1987,
242:525-530.
11. Gaillard J, Moulis JM, Auric P, Meyer J: High-multiplicity spin states
of 2[4Fe–4Se]+clostridial ferredoxins. Biochem 1986,
25:464-468.
12. Carney MJ, Papaefthymiou GC, Spartalian K, Frankel RB, Holm RH:
ground state variability in [Fe4S4(SR)4]3-. In synthetic analogues
of the reduced clusters in ferredoxins and other iron-sulfur
proteins: cases of extreme sensitivity of electronic state and
270
Guest section: computational bioinorganic chemistry II

Page 13

structure to extrinsic factors. J Am Chem Soc 1988,
110:6084-6095.
13. Anxolabehere-Mallart E, Glaser T, Frank P, Aliverti A, Zanetti G,
•
Hedman B, Hodgson KO, Solomon EI: Sulfur K-edge X-ray
absorption spectroscopy of 2Fe–2S ferredoxin: covalency of the
oxidized and reduced 2Fe forms and comparison to model
complexes. J Am Chem Soc 2001, 123:5444-5452.
The authors show that Fe(III)–sulfide bonds are more covalent than
Fe(II)–sulfide, and that the Fe(III)–sulfide polarity is increased, evidently by
NH–S hydrogen bonding in Fd compared with analogous synthetic
complexes based on a careful examination and calibration of sulfur K-edge
XAS. The effect of altered polarity on redox potentials is estimated.
14. Glaser T, Hedman B, Hodgson KO, Solomon EI: Ligand K-edge
•
X-ray absorption spectroscopy: a direct probe of metal–ligand
covalency. Acc Chem Res 2000, 33:859-868.
This reviews applications of ligand K-edge spectroscopy to blue copper, CuA
(dinuclear delocalized), and to dinuclear and polynuclear iron–sulfur
complexes. The poorer electron donor character of µ3sulfide in the 4Fe
compared with that of µ2sulfide in the 2Fe complexes per Fe–S bond is one
major conclusion, and is proposed to give weaker J coupling in 4Fe than in 2Fe
systems. This is one mechanism favoring delocalized valence in 4Fe complexes;
however, delocalized valence is also favored by spin frustration arguments.
15. Glaser T, Bertini I, Moura JJG, Hedman B, Hodgson KO, Solomon EI:
•
Protein effects on the electronic structure of the [Fe4S4]2+cluster
in ferredoxin and HIPIP. J Am Chem Soc 2001, 123:4859-4860.
Sulfur K-edge spectroscopy is used to analyze the differences in metal–sulfide
and metal–thiolate covalency in oxidized HIPIPs and Fds compared with in
synthetic 4Fe systems. Fe–S covalency decreases as the number of NH–S
hydrogen bonds increase in the proteins, and is less in the proteins than in
synthetic complexes.
16. Noodleman L, Case DA: Density-functional theory of spin
polarization and spin coupling in iron–sulfur clusters. Adv Inorg
Chem 1992, 38:423-470.
17.Mouesca J-M, Chen JL, Noodleman L, Bashford D, Case DA: Density
functional/Poisson–Boltzmann calculations of redox potentials
for iron–sulfur clusters. J Am Chem Soc 1994, 116:11898-11914.
18. Noodleman L, Li J, Zhao XG, Richardson WH: Density functional
studies of spin coupled transition metal dimer and tetramer
complexes. In Density Functional Methods in Chemistry and
Materials Science. Edited by Springborg M. New York: John Wiley
and Sons; 1997:149-188.
19. Noodleman L, Case DA, Mouesca J-M, Lamotte B: Valence electron
delocalization in polynuclear iron–sulfur clusters. J Biol Inorg
Chem 1996, 1:177-182.
20. Case DA, Noodleman L, Li J: Modern computational approaches to
•
modeling polynuclear transition metal complexes. In Metal–Ligand
Interactions in Chemistry, Physics, and Biology. Edited by Russo N,
Salahub DR. Dordrecht, Netherlands: Kluwer; 2000:19-47.
This work reviews Heisenberg spin coupling and valence electron delocalization,
showing how these are connected to vibronic coupling and static asymmetries.
The geometric dependence of different spin state curves and their minima is
demonstrated for a diferric [Fe2S2(SR)4]2–(R = methyl) model.
21. Bominaar EL, Borshch SA, Girerd J-J: Double-exchange and
vibronic coupling in mixed-valence systems. Electronic structure
of [Fe4S4]3+clusters in high-potential iron protein and related
models. J Am Chem Soc 1994, 116:5362-5372.
22. Li J, Nelson MR, Peng CY, Bashford D, Noodleman L: Incorporating
protein environments in density functional theory: a self-
consistent-reaction-field calculation of redox potentials of
[2Fe2S] clusters in ferredoxin and phthalate dioxygenase
reductase. J Phys Chem A 1998, 102:6311-6324.
23. Ludwig ML, Ballou DP, Noodleman L: Phthalate dioxygenase
•
reductase. In Handbook of Metalloproteins. Edited by Wieghardt K,
Huber R, Poulos T, Messerschmidt A. Chichester, UK: John Wiley and
Sons; 2001:652-667.
The biochemistry of PDR and related enzymes is reviewed. The structure of
PDR is compared with the latest accurate structure of Anabaena Fd with
emphasis on hydrogen bonding. Redox potential calculations using a
combined DFT-electrostatics methodology establishes that the redox shift
between PDR and Anabaena Fd is largely due to the relative flip of one main
chain NH–S hydrogen bond. In general, NH–S hydrogen bonds to first shell
ligands are strong and stronger in the reduced compared with the oxidized
form. The reaction field term, reflecting particularly solvent close to the
2Fe2S cluster makes a very large contribution to the redox potential in both
proteins, but is nearly the same for Anabaena Fd and PDR.
24. Morales R, Chron M-H, Hudry-Clergeon G, Petillot Y, Norager S,
•
Medina M, Frey M: Refined X-ray structures of the oxidized, at
1.3 Å, and reduced, at 1.17 Å, [2Fe–2S] ferredoxin from the
cyanobacterium Anabaena PCC7119 show redox-linked
conformational changes. Biochemistry 1999, 38:15764-15773.
The very-high resolution structure of Anabaena Fd is given in both oxidation
states. The two redox forms show a major conformational shift with the
CONH of the peptide bond linking Cys46 and Ser47 flipping from CO ‘out’
form, reduced, to CO ‘in’, oxidized.
25. Marcus R: Electron transfer reactions in chemistry: theory and
experiment (Nobel lecture). Angew Chem Int Ed Engl 1993,
32:1111-1121.
26. Sigfridsson E, Olsson MHM, Ryde U: Inner-sphere reorganization
•
energy of iron-sulfur clusters studied with theoretical methods.
Inorg Chem 2001, 40:2509-2519.
Although there are considerable quantitative uncertainties, these DFT
(B3LYP) calculations of inner sphere reorganization energies raise several
important issues, and provide a good starting point for future studies.
27.
•
Sigfridsson E, Olsson MHM, Ryde U: A comparison of the inner
sphere reorganization energies of cytochromes, iron–sulfur
clusters, and blue copper proteins. J Phys Chem B 2001,
105:5546-5552.
This comparison using DFT (B3LYP) leads to the conclusion that the inner
sphere reorganization energy in cytochromes with uncharged axial ligands is
much less than for either the dinuclear CuAsite of cytochrome c oxidase
and nitrous oxide reductase or for iron–sulfur clusters in proteins. Further
experimental examination of these questions would prove valuable.
28. Bominaar EL, Achim C, Borshch SA, Girerd JJ, Munck E: Analysis of
exchange interaction and electron delocalization as
intramolecular determinants of intermolecular electron transfer
kinetics. Inorg Chem 1997, 36:3689-3701.
29. Achim C, Bominaar EL, Munck E: Aspects of valence delocalization
relevant to the kinetic properties of electron-transfer chains. J Biol
Inorg Chem 1998, 3:126-134.
30. Bominaar EL, Achim C, Borshch SA: Theory for electron transfer
from a mixed-valence dimer with paramagnetic sites to a
mononuclear acceptor. J Chem Phys 1999, 110:11411-11422.
31. Kummerle R, Gaillard J, Kyritis P, Moulis JM: Intramolecular electron
•
transfer in [4Fe–4S] proteins: estimates of the reorganization
energy and electronic coupling in Chromatium vinosum
ferredoxin. J Biol Inorg Chem 2001, 6:446-451.
NMR line broadening measurements of electron transfer rates between the
2[4Fe4S] Fd clusters in native C. vinosum and mutants allows construction
of a ln(kET) versus driving force curve and estimation of the total reorganiza-
tion energy, a rarity for Fds.
32. Babini E, Bertini I, Borsari A, Capozzi F, Luchinat C, Zhang X,
•
Moura GLC, Kurnikov IV, Beratan DN, Ponce A, Di Bilio AJ,
Winkler JR, Gray HB: Bond-mediated electron tunneling in
ruthenium-modified high-potential iron–sulfur protein. J Am Chem
Soc 2000, 122:4532-4533.
Introduction of an exogeneous ruthenium complex covalently linked to the
HIPIP Fe4S4cluster allows measurements of electron transfer rates to
the cluster and estimates of the HIPIP total reorganization energy.
33. Telser J, Huang H, Lee H-I, Adams MWW, Hoffman BM: Site
valencies and spin coupling in the 3Fe and 4Fe (S = 1/2) clusters
of Pyrococcus furiosus ferredoxin by 57Fe ENDOR. J Am Chem
Soc 1998, 120:861-870.
34. Calzolai L, Gorst CM, Bren KL, Zhou Z-H, Adams MWW, La Mar GN:
Solution NMR structure of the electronic structure and magnetic
properties of cluster ligation mutants from the hyperthermophilic
archeon Pyrococcus furiosus. J Am Chem Soc 1997,
119:9341-9350.
35. Mouesca J-M, Noodleman L, Case DA: Analysis of the 57Fe
hyperfine coupling constants and spin states in nitrogenase
P-clusters. Inorg Chem 1994, 33:4819-4830.
36. Mouesca J-M, Noodleman L, Case DA, Lamotte B: Spin densities
and spin coupling in iron–sulfur cluster: a new analysis of
hyperfine coupling constants. Inorg Chem 1995, 34:4347-4359.
37.
•
Moriaud F, Gambarelli S, Lamotte B, Mouesca J-M: Detailed proton
Q-band ENDOR study of the electron spin population distribution in
the reduced [4Fe–4S]1+state. J Phys Chem B 2001, 105:9631-9642.
Proton Q-band ENDOR in combination with DFT derived Fe–S covalencies
parameters yield effective spin projection factors that can be compared with
those from spin-coupling models. With sufficient proton ENDOR hyperfine
data, the proton positions with respect to the 4Fe4S structure can be refined.
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
271

Page 14

38. More C, Camensuli P, Dole F, Guigliarelli B, Asso M, Fournel A,
Bertrand P: A new approach for the structural study of
metalloproteins: the quantitative analysis of intercenter magnetic
interactions. J Biol Inorg Chem 1996, 1:162-168.
39. Guigliarelli B, Bertrand P: Application of EPR spectroscopy to the
•
structural and functional study of iron–sulfur proteins. Adv Inorg
Chem 1999, 47:421-490.
Both electronic structure aspects and geometric implications of EPR
spectroscopy are developed. The position of a paramagnetic center with
respect to a dinuclear or tetranuclear iron–sulfur cluster can be determined.
40. Blondin G, Girerd JJ: Interplay of electron exchange and electron
transfer in metal polynuclear complexes in proteins or chemical
models. Chem Rev 1990, 90:1359-1376.
41. Crozet M, Chaussade M, Bardet M, Emsley L, Lamotte B,
•
Mouesca J-M: Carbon-13 solid-state NMR studies on synthetic
model compounds of [4Fe–4S] clusters in the 2+ state. J Phys
Chem A 2000, 104:9990-10000.
Carbon-13 solid-state NMR provides a valuable alternative to magnetic
susceptibility studies in this system, and analysis allows J coupling and
delocalization B parameters to be determined, although cluster distortion will
introduce additional J parameters and alter B as well. These methods should
be extendable to a wider range of systems and spin states. The data are
richer than with magnetic susceptibility, and could yield additional spin-
coupling information.
42. Zu Y, Fee JA, Hirst J: Complete thermodynamic characterization of
•
coupled protonation and electronation at the bc1-type Rieske
center of Thermus thermophilus. J Am Chem Soc 2001,
123:9906-9907.
These pH-dependent redox potential measurements over a wide range
between pH 3 and 14 show how electron transfer and proton transfer are
linked in the Rieske protein.
43. Onufriev A, Case DA, Ullmann GM: A novel view of pH titration in
•
biomolecules. Biochemistry 2001, 40:3413-3419.
The protonation of different titratable sites in proteins can be coupled. The
apparent pKas may actually involve combinations of sites. The statistical
mechanical connection between the microscopic protonation states and the
apparent pKas seen experimentally is shown.
44. Rees DC, Howard JB: Nitrogenase: standing at the crossroads.
•
Curr Opin Chem Biol 2000, 4:559-566.
The current state of knowledge of nitrogenase is presented combining struc-
tural, biochemical, biophysical (redox and kinetics) and spectroscopic insights.
45. Burgess BK, Lowe DJ: Mechanism of molybdenum nitrogenase.
Chem Rev 1996, 96:2983-3011.
46. Schindelin H, Kisker C, Schlessman JL, Howard JB, Rees DC:
Structure of ADP.AlF4–-stabilized nitrogenase complex and
its implications for signal transduction. Nature 1997,
387:370-376.
47. Lanzilotta WN, Seefeldt LC: Changes in the midpoint potentials of
the nitrogenase metal centers as a result of iron
protein–molybdenum–iron protein complex formation.
Biochemistry 1997, 36:12976-12983.
48. Angove HC, Yoo SJ, Munck E, Burgess BK: An all-ferrous state of
the Fe protein of nitrogenase. Interaction with nucleotides and
electron transfer to the MoFe protein. J Biol Chem 1998,
273:26330-26337.
49. Watt GD, Reddy KRN: Formation of an all-ferrous Fe4S4cluster in
the iron protein component of Azotobacter vinelandii nitrogenase.
J Inorg Biochem 1994, 53:281-294.
50. Strop P, Takahara PM, Chiu H-J, Angove HC, Burgess BK, Rees DC:
•
Crystal structure of the all-ferrous [4Fe–4S]0form of the
nitrogenase iron protein from Azotobacter vinelandii. Biochemistry
2001, 40:651-656.
The first X-ray structure of the all-ferrous nitrogenase iron protein at medium
resolution (2.25 Å) is presented. A higher resolution structure is very
desirable for detailed comparison with EXAFS, which indicates some
very short Fe–Fe distances.
51. Musgrave KB, Angove HC, Burgess BK, Hedman B, Hodgson KO:
All-ferrous titanium (III) citrate reduced Fe protein of nitrogenase:
an XAS study of electronic and metrical structure. J Am Chem Soc
1998, 120:5325-5326.
52. Yoo SJ, Angove HC, Burgess BK, Hendrich MP, Munck E:
•
Mössbauer and integer-spin EPR studies and spin-coupling
analysis of the [4Fe-4S]0cluster of the Fe protein from
Azotobacter vinelandii nitrogenase. J Am Chem Soc 1999,
121:2534-2545.
Mössbauer spectroscopy and integer spin EPR establish that the total spin
is S = 4 for the all-ferrous state of the iron protein.
53. Guo M, Sulc F, Farmer P, Burgess BK: Evaluation of the 4Fe–4S
•
1+/0 reduction potential of the nitrogenase iron protein. J Inorg
Biochem 2001, 86:243.
A very brief report from the International Conference on Bioinorganic
Chemistry that the redox potential of the 4Fe–4S 1+/0 is much more
negative than that found in the work of Watt and Reddy.
54. Zhou CY, Raebiger JW, Segal BM, Holm RH: The influence of net
•
charge on the redox potentials of Fe4S4cubane type clusters in
aprotic solvents. Inorg Chim Acta 2000, 300-302:892-902.
For synthetic cubane clusters, the redox potentials for the 4Fe–4S 1+/0
couple are always quite negative, consistent with the recent nitrogenase iron
protein values seen by Guo and Burgess.
55. Angove HC, Yoo SJ, Burgess BK, Munck E: Mössbauer and EPR
evidence for an all-ferrous Fe4S4cluster with S = 4 in the Fe
protein of nitrogenase. J Am Chem Soc 1997, 119:8730-8731.
56. Yoo SJ, Angove HC, Papaefthymiou V, Burgess BK, Munck E:
••
Mössbauer study of the MoFe protein of nitrogenase from
Azotobacter vinelandii using selective 57Fe enrichment of the
M-centers. J Am Chem Soc 2000, 122:4926-4936.
The Mössbauer isomer shifts, quadrupole splittings, and hyperfine tensors
are determined for all seven of the iron sites in state MNstrongly comple-
menting earlier ENDOR studies of True et al. The hyperfine splittings of the
catalytically reduced state MRand the radiolytically reduced state MIwere
determined and analyzed.
57.Peters JW, Stowell MHB, Soltis SM, Finnegan MG, Johnson MK,
Rees DC: Redox-dependent structural changes in the nitrogenase
P-cluster. Biochemistry 1997, 36:1181-1187.
58. Lanzilotta WN, Christianson J, Dean DR, Seefeldt LC: Evidence for
coupled electron and proton transfer in the [8Fe–7S] cluster of
nitrogenase. Biochemistry 1998, 37:11376-11384.
59. Lovell T, Li J, Liu T, Case DA, Noodleman L: FeMo cofactor of
••
nitrogenase: a density functional study of states MN, MOX,MR,
and MI. J Am Chem Soc 2001, 123:12392-12410.
Spin polarized BS DFT methods were used to calculate optimized geometries,
spin-coupling patterns and energies, Mössbauer isomer shifts and hyperfine
parameters, and information about the one-electron reduced and one-
electron oxidized states starting from MN. Analysis of the isomer shifts
distinguishes between the two oxidation state assignments that have been
proposed for MNfrom Mössbauer and ENDOR spectroscopies.
60. True AE, Nelson MJ, Venters RA, Orme-Johnson WH, Hoffman BM:
57Fe hyperfine coupling tensors of the FeMo cluster in Azobacter
vinelandii MoFe protein: determination by polycrystalline ENDOR
spectroscopy. J Am Chem Soc 1988, 110:1935-1943.
61. Lee H-I, Hales BJ, Hoffman BM: Metal-ion valencies of the FeMo
cofactor in CO-inhibited and resting state nitrogenase by 57Fe
Q-band ENDOR. J Am Chem Soc 1997, 119:11395-11400.
62. Sorlie M, Christianson J, Dean DR, Hales BJ: Detection of a new
••
radical and FeMo-cofactor EPR signal during acetylene reduction
by the α α-H195Q mutant of nitrogenase. J Am Chem Soc 1999,
121:9457-9458.
The new radical observed by EPR during acetylene reduction in the
His195Gln mutant is either an amino acid or a homicitrate radical species
produced during enzymatic turnover.
63. Lee H-I, Sorlie M, Christianson J, Song R, Dean DR, Hales BR,
••
Hoffman BM: characterization of an intermediate in the reduction
of acetylene by nitrogenase α α-Gln195 MoFe protein by Q-band
EPR and 13C,1H ENDOR. J Am Chem Soc 2000, 122:5582-5587.
X-Band and Q-band EPR show 13C hyperfine, indicating a C2H2species
bound to the FeMo cofactor. This species is proposed to bridge 2Fe sites of
a 4Fe4S ‘face’, and to stabilize the observed S = 1/2 spin state of the cluster.
64. Sorlie M, Christianson J, Lemon BJ, Peters JW, Dean DR, Hales BJ:
Mechanistic features and structure of the nitrogenase α α-Gln195
MoFe protein. Biochemistry 2001, 40:1540-1549.
65. MacDonnell FM, Ruhlandt-Senge K, Ellison JJ, Holm RH, Power PP:
Sterically encumbered iron (II) thiolate complexes: synthesis and
structure of trigonal planar [Fe(SR)3]- (R=2,4,6-t-Bu3C6H2) and
Mössbauer spectra of two- and three-coordinate complexes. Inorg
Chem 1995, 34:1815-1822.
66. Deng H, Hoffmann R: How N2might be activated by the FeMo-cofactor
in nitrogenase. Angew Chem Int Ed Engl 1993, 32:1062-1064.
272
Guest section: computational bioinorganic chemistry II

Page 15

67.Dance I: Theoretical investigations of the mechanism of biological
nitrogen fixation at the FeMo cluster site. J Biol Inorg Chem 1996,
1:581-586.
68. Dance I: Calculated details of a mechanism for conversion of N2
to NH3at the FeMo cluster of nitrogenase. Chem Commun
1997:165-166.
69. Stavrev KK, Zerner MC: A theoretical model for the active site of
nitrogenase. Chem Eur J 1996, 2:83-87.
70.Stavrev KK, Zerner MC: On the reduced and oxidized forms of the
FeMo-cofactor of Azotobacter vinelandii nitrogenase. Theor Chem
Acc 1997, 96:141-145.
71. Stavrev KK, Zerner MC: Studies on the hydrogenation steps of the
nitrogen molecule at the Azotobacter vinelandii nitrogenase site.
Int J Quant Chem 1998, 70:1159-1168.
72.Siegbahn PEM, Westerberg J, Svensson M, Crabtree RH: Nitrogen
fixation by nitrogenases: a quantum chemical study. J Phys
Chem B 1998, 102:1615-1623.
73.
••
Rod and Norskov use high-quality DFT calculations to study N2binding and
reaction with atomic hydrogen. Reaction energies are also calculated using
molecular hydrogen as a reference reactant. Despite a simplified model, a
wealth of interesting results are given for the proposed reaction cycle.
Rod TH, Norskov JK: Modeling the nitrogenase FeMo cofactor.
J Am Chem Soc 2000, 122:12751-12763.
Questions about the cluster oxidation state and the coupling between
electron and proton transfer, as well as the role of the surrounding protein
and solvent environment remain unresolved.
74. Rod TH, Logadottir A, Norskov JK: Ammonia synthesis at low
temperatures. J Chem Phys 2000, 112:5343-5347.
75. Rod TH, Hammer B, Norskov JK: Nitrogen absorption and
hydrogenation on a MoFe6S9complex. Phys Rev Lett 1999,
82:4054-4057.
Now in press
The work referred to in the text as (M Ullmann, L Noodleman, DA Case,
unpublished data) is now in press:
76.Ullmann GM, Noodleman L, Case DA: Density functional calculation
of the pKavalues and redox potentials in the bovine Rieske
iron–sulfur proteins. J Biol Inorg Chem 2002, in press.
Also, (T Lovell, J Li, DA Case, L Noodleman, unpublished data) is now in press:
77.Lovell T, Li J, Case DA, Noodleman L: FeMo cofactor of nitrogenase:
energetics and local interactions in the protein environment. J Biol
Inorg Chem 2002, in press.
78. Lovell T, Li J, Case DA, Noodleman L: Binding modes for the first
coupled electron and proton addition to the FeMo cofactor of
nitrogenase. J Am Chem Soc 2002, in press.
Insights into properties and energetics of iron–sulfur proteins from simple clusters to nitrogenase Noodleman et al.
273