View Online / Journal Homepage / Table of Contents for this issue
PERSPECTIVE
Dalton
Anne Volbeda and Juan C. Fontecilla-Camps*
Laboratoire de Cristallographie et de Cristallogenèse des Protéines, Institut de Biologie
Structurale J.P. Ebel (CEA-CNRS-UJF), 41 rue Jules Horowitz, 38027, Grenoble Cédex 1,
France. E-mail: [email protected]; Fax: 33-4-38-78-51-22; Tel: 33-4-38-78-59-20
www.rsc.org/dalton
Structural bases for the catalytic mechanism of Ni-containing carbon
monoxide dehydrogenases†
Received 14th June 2005, Accepted 22nd June 2005
First published as an Advance Article on the web 23rd September 2005
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
Significant progress has been made recently in our understanding of the structure/function relationships of the
catalytic C-cluster of carbon monoxide dehydrogenases.
Several structures of this enzyme have been reported, some
of them at very high resolution. One recurrent problem,
however, is the high degree of heterogeneity within each
structure, as well as between the different X-ray models.
Here, we have tried to relate the structural data with
the wealth of spectroscopic and biochemical information
gathered over many years. As a result, we propose a
catalytic cycle that is consistent with both observations and
stereochemistry. We also give alternatives to one of the most
difficult aspects of the cycle, namely, the location of the two
electrons in the most reduced state of the C-cluster.
1 Introduction
In anaerobic microorganisms the reduction of CO2 and the
oxidation of CO is mediated by the Ni-containing C-cluster of
the enzyme CO dehydrogenase (CODH) according to:
CO + OH− ⇔ CO2 + 2e− + H+
(1)
† Based on the presentation given at Dalton Discussion No. 8, 7–9th
September 2005, University of Nottingham, UK.
Ni-containing CODH’s can be divided into four classes.1 Class
I and II enzymes are found in methanogenic archaea such as
Methanosarcina thermophila, class III enzymes are normally
found in acetogens such as Moorella thermoacetica, whereas
class IV enzymes are found in anaerobic bacteria such as Rhodospirillum rubrum and Carboxydothermus hydrogenoformans.
Class I–III enzymes are bifunctional as they also catalyze either
the synthesis or the decarbonylation of acetyl-CoA, a universal
source of carbon and energy to the cell, according to:
CH3 –CoIII FeSP + CO + CoA-SH
⇔ CH3 CO-SCoA + CoI FeSP + H+
(2)
where CoFeSP is a heterodimeric corrinoid–cobalt iron–sulfur
cluster protein.
In methanogens, the different enzymes form a multimeric
abcde acetyl-CoA decarbonylase/synthase (ACDS) complex
where ae has CODH activity, b is the acetyl-CoA synthase (ACS)
and cd corresponds to CoFeSP.2 In acetogens, ACS activity is
located in the a subunit (homologous to b in ACDS) whereas the
CODH activity is found in the b subunit (homologous to a in
ACDS), both subunits forming an a2 b2 tetrameric structure.3
These microorganisms use the Wood–Ljungdahl pathway to
synthesize an acetyl group from two CO2 molecules.4 One of
these molecules is fully reduced to a corrinoid-bound methyl
group that is subsequently transferred to ACS whereas the
DOI: 10.1039/b508403b
Anne Volbeda obtained his Ph. D. degree in 1988 at the university of Groningen after working with Wim Hol on the crystallographic
analysis of spiny lobster haemocyanin, a large hexameric protein that uses binuclear copper centres to transport molecular oxygen.
During a post-doctoral fellowship with Dietrich Suck at EMBL Heidelberg he worked on the crystal structure of P1 nuclease, an enzyme
with a trinuclear zinc active site. In 1991 he obtained a permanent position at the laboratory of crystallography and crystallogenesis of
proteins (LCCP) in Grenoble, where he continued doing research on the structure and function of complex redox enzymes depending on
iron and nickel. His main themes of interest focus on electron, proton and gas transfer phenomena in proteins and on the functioning of
metal sites in biochemical energy transformations.
Anne Volbeda
This journal is
Juan C. Fontecilla-Camps
©
Juan C. Fontecilla-Camps received his undergraduate training at
the University of Concepcion, Chile, and his Ph.D. degree in protein
crystallography at the University of Alabama in Birmingham under
the supervision of Charles E. Bugg in 1980. Subsequently, he moved
to Marseilles, France, where he became group leader and then
head of a protein crystallography laboratory established by the
CNRS and the University of Aix-Marseilles. In 1991, he took a
job with the French Commissariat l’Energie Atomique (CEA) in
Grenoble. After having worked on the crystal structure of animal
toxins, in 1993 he turned his attention to the structural biology of
hydrogenases and other metalloenzymes. He and his group have
solved the structures of NiFe and Fe-only hydrogenases, pyruvate–
ferredoxin oxidoreductase and, very recently, carbon monoxide
dehydrogenase/acetyl-Coenzyme A synthase. In 2000, he was
awarded the medal of the European section of the International
Society of Biological Inorganic Chemistry for his work on hydrogenases. He is also “Chevalier dans l’Ordre des Palmes Académiques”.
The Royal Society of Chemistry 2005
Dalton Trans., 2005, 3443–3450
3443
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
View Online
Fig. 1 (a) Arrangement of metal clusters in CODH with Fe in red, S in yellow and Ni in green. (b) Coordination in the C-cluster of the Ni ion to
a cysteine and two l3 -S ligands and of the unique Fe ion to another l3 -S ligand (S1), a cysteine and a histidine (N in blue and C in white). Each
of the Fe ions in the [3Fe–4S] sub-site has in addition a cysteine ligand (not shown for simplicity). E1 and E2 are Ni coordination sites available to
non-protein ligands. (c) Three-dimensional fold of the CODHMt dimer and the N-terminal domains of ACSMt (grey) showing hydrophobic tunnels
(pink) and Xe sites (blue).
tunnel peters out close to a vacant coordination E1 site of the
C-cluster catalytic nickel. A second Ni–Fe bridging site (E2) is
also potentially available for binding of substrates, products and
inhibitors (Fig. 1b).
Before going any further into the structural analysis it is
necessary to introduce some nomenclature on the various redox
states of the C-cluster.1 A diamagnetic state Cox is obtained
at potentials below −200 mV. One-electron reduction of Cox
produces the Cred1 state that converts to Cred2 upon addition of
CO or other reductants. At pH 7 the midpoint potential of the
Cred1 to Cred2 conversion is −520 mV, close to the E 0 of −512
mV of the CO/CO2 redox couple of reaction 1. Both Cred states
exhibit EPR spectra typical for an S = 1/2 spin state. Another
undetected diamagnetic state called Cint is postulated to arise
from the one-electron reduction of Cred1 or from the one-electron
oxidation of Cred2 .
All the CODH atomic models reported so far have been
obtained from crystals grown under strictly anaerobic conditions and, in most cases, reducing agents have been added to
the crystallization solutions (Table 1). High temperature factors
and/or partial occupation for the C-cluster atoms in the refined
models indicate the presence of structural heterogeneity in all
the structures deposited with the Protein Data Bank. By far,
the best resolutions and refinement statistics have been reported
for the mono-functional CODHCh enzyme.7–8 In four of the five
CODHCh structures an inorganic sulfur atom (labeled S2) has
been modeled at the E2 site (Fig. 1b). This labile sulfur atom is
absent from the structures of the Rr and Mt enzymes. Most COtreated crystals show the presence of a ligand at the apical E1
coordination site of the Ni, located at the end of the hydrophobic
tunnel. This ligand was tentatively assigned as partially occupied
CO in CODHMt 11 and as an unknown species in CODHRr. 9
Recently, Dobbek et al. have reported that incubation of
CODHCh with CO leads to rapid inactivation, especially when
second CO2 is only partially reduced to CO by CODH according
to eqn. (1). Metal-based insertion/migration of the methyl and
CO groups at the ACS A-cluster produces an acetyl group
that binds CoA according to reaction 2. Class IV CODHs
are monofunctional and form a2 dimers. In these organisms
external CO can be used as a source of carbon and reducing
power.5 Aerobic bacteria such as Oligotropha carboxidovorans
utilize molybdopterin- and copper-containing CODHs,6 which
are unrelated to the nickel- and iron-containing CODHs found
in anaerobic bacteria and archaea.
2 The structure of the C-cluster
Crystal structures of nickel and iron-containing CODHs have
been reported for C. hydrogenoformans (Ch),7–8 R. rubrum
(Rr)9 and the bifunctional M. thermoacetica CODH/ACSMt. 10–11
These homodimeric CODHs contain a total of five FeS clusters
(Fig. 1a). The most exposed D-cluster that was first detected by
X-ray crystallography, is coordinated by two cysteine residues
from each subunit. At about 10 Å from this center, each
monomer binds a B-cluster that is also located at the subunit
interface. The more buried and catalytic C-clusters are located
at the interface of the three domains that compose each CODH
monomer7 and lie at about 11 Å from the closest B-cluster that
is located in the other subunit. Whereas the B- and D-clusters
are standard [4Fe–4S] cubanes, the structure of the C-cluster
is unprecedented: it contains a Ni and an atypical Fe ion that
is not a part of the FeS cluster and is thought to correspond
to the spectroscopically unusual Feu , also known as Ferrous
Component II or FCII;12 both ions are connected to a [3Fe–
4S] sub-site (Fig. 1b). In CODHMt an extensive hydrophobic
tunnel network links the two C-clusters to each other and to the
catalytic A-clusters of the ACS subunits.10 These tunnels have
been shown to bind xenon11 (Fig. 1c, see also next section). The
Table 1 Selected data on refined CODH crystal structures and C-clusters
a
3444
Source
Rr
Ch
PDB code
Reducing agents
Atmosphere
d max a /Å
Rb (%)
Rfree b (%)
Ni–S1/Å
Ni–Fe1/Å
E1 ligand
E2 ligand
1JQK
—
Ar, CO
2.8
25.0
28.1
2.6
2.6
Unk
C531Sc
1JJY
Na2 S2 O4
N2
1.63
16.1
20.9
3.8
2.8
—
S2
Mt
1SU6
DTT
CO (1 h)
1.64
15.9
19.9
3.7
2.8
—
S2
1SU7
DTT
N2
1.12
12.0
15.2
3.7
2.8
H2 O?
S2
1SU8
Na2 S2 O4
N2
1.10
12.3
15.2
3.8
2.8
H2 O?
S2
1SUF
DTT
CO
1.15
12.3
15.4
4.1
3.1
CO?
OH− ?
Maximal resolution of the X-ray diffraction data. b Crystallographic R- and Rfree -factors.47
Dalton Trans., 2005, 3443–3450
1MJG
DTT
N2
2.2
22.0
25.2
3.0
2.7
—
—
1OAO
Na2 S2 O4 + DTT
CO (10 )
1.9
14.7
17.9
4.4
3.5
CO?
—
—
Na2 S2 O4 + DTT
CO (10 )
2.5
14.0
19.6
4.1
2.8
CO?
C550Sc
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
View Online
no other reducing agents are present.8 These authors observed
a positive correlation between the presence of the inorganic
Ni–Fe bridging S2 atom and the enzymatic activity of the
corresponding crystalline enzyme; they reported that both the S2
ligand and the activity were lost after prolonged CO treatment.
From these observations Dobbek et al. concluded that C-clusters
that lack S2 represent non-functional states. However, Feng and
Lindahl13 have shown that sulfide reversibly inactivates both
CODHRr and CODHMt and that addition of Na2 S to the Cred1
state leads to changes in its EPR spectrum. The inhibitory effect
of sulfide can be reversed by reduction with CO and dithionite
because, as with many other anions, sulfide will not bind to the
Cred2 state. For instance, CN− that is known to interact with
FCII, as indicated by Mössbauer spectroscopy12 and ENDOR
results,14 can be dissociated from the C-cluster under the same
reducing conditions.15–16 Moreover, CODH is known to reduce
the CO2 analog COS to CO and SH− , in a reaction equivalent
to (1). Since the bridging E2 site is likely to bind OH− during
catalysis13 it should also bind the analog SH− . Taken together,
these observations suggest that the bridging S2 in CODHCh is
a signature of the sulfide-inhibited state reported by Feng and
Lindahl13 and that that center mimics the OH− bound C-cluster.
A comparison of the various refined C-cluster structures
shows that the [3Fe–4S] sub-site is relatively well ordered whereas
S1, Fe1 (the putative Feu or FCII) and especially the Ni are
either significantly agitated or only partially occupied (Table 1,
Fig. 2). In the crystal structure of CO-treated CODHRr the Ccluster has been modeled as a distorted [Ni–3Fe–4S] cubane.9
The Ni coordination environment includes a bridging cysteine
ligand to Fe1 and it may be described as distorted trigonal
bipyramidal, with an exogenous ligand that could be CO, bound
opposite to S1. Fe1 has a distorted tetrahedral coordination. In
all the CODHCh structures7–8 the C-cluster adopts a more open
conformation with Ni–S1 distances that are 1.1–1.5 Å longer
than in CODHRr . The Ni has either distorted square planar or
square pyramidal coordination whereas Fe1 retains the distorted
tetrahedral coordination. In three of the deposited structures,
PDB codes 1SU6, 1SU7 and 1SUF, an apical Ni ligand has
been modeled as water (although in the latter structure it could
be CO, given the experimental conditions). Alternative positions
for some Fe ions and cysteine side chains were observed in
the CODHCh structures refined at very high resolution. An
alternative position for Fe1 in the 1SUF model (not shown)
would convert the [3Fe–4S] sub-site into an approximately
regular [4Fe–4S] cubane when Ni is absent. It is known that
Ni-depleted C-clusters are not catalytically competent and,
until recently, they have been considered to have spectroscopic
characteristics similar to those of a S = 3/2 [4Fe–4S] cluster.12
However, as further discussed below, very recent MCD and
Raman data have challenged this idea.17
In the first reported CODH/ACSMt structure10 (PDB code
1MJG) both the C-cluster Fe1 and Ni ions are coordinated
by only three ligands, a rather uncommon situation for these
metal ions. This C-cluster structure has not been described in
the literature but the Ni–Fe distance of 2.7 Å is compatible
with the presence of a bridging hydride, which would not be
detectable at 2.2 Å resolution. A bridging hydride would bring
the number of Fe and Ni ligands to four, a much more plausible
proposition. The largest displacement of the Ni with respect to
S1 and Fe1 is the one found by us in the structure of another
Fig. 2 Crystallographic C-cluster models of CODH. (a) Rr, (b–f) Ch, (g–h) Mt, with PDB codes 1JQK, 1JJY, 1SU6, 1SU7, 1SU8, 1SUF, 1MJG
and 1OAO, respectively, (i) Mt, another crystal of the same form as (h), refined at 2.5 Å resolution. See also Table 1.
Dalton Trans., 2005, 3443–3450
3445
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
View Online
crystal form of CODH/ACSMt , treated with CO and refined
at 1.9 Å resolution.11 The interpretation of this structure is
complicated by significant heterogeneity at the C-cluster that
included a low-occupancy persulfide bond between S1 and a
minor second conformation of Cys316, an alternative position
for Fe1 and the presence of three different conformations for
Cys550. Crystallographically, the Ni and its putative CO ligand
were best modeled with 50% occupancy. This structure resembles
the 1SUF model of CODHCh discussed above. In the latter, three
water molecules were included as Ni and/or Fe1 ligands in the
deposited atomic coordinates, but the authors did not discuss
them, presumably because they consider this model functionally
irrelevant.8
We have recently obtained a more homogeneous C-cluster in
the structure of another CO-treated crystal of CODHMt , refined
at 2.5 Å resolution (unpublished). In this case, we modeled the
occupation of the Ni and its putative apical CO ligand at 80%
and Cys550 refined to a bridging position between Ni and Fe1,
resulting in approximately tetrahedral coordination for these
two metal ions (Fig. 2i). What is the mechanistic relevance,
if any, of the NiFe bridging cysteine? Although it has been
observed in another structure (Fig. 2a) it blocks the E2 site
that is likely to bind OH− during catalysis. It may just be that, in
the absence of an exogenous bridging ligand, Cys550 preserves
a stereochemically feasible coordination for Ni and Fe1.
3 The hydrophobic tunnel network
The first hint pointing out the existence of tunnels connecting the
C- and A-cluster came from the work by Grahame and DeMoll
who, in 1995, concluded that the physiological substrate of
ACDS from M. barkeri was CO2 and not CO.18 This conclusion
was based on the fact that CO-dependent reactions had a rate
of acetyl-CoA synthesis that was about one-half of the value
obtained using CO2 plus reduced ferredoxin. These authors
postulated that this was due to the artificial nature of CO as
a substrate. However, as we shall see below, CO is the natural
substrate of the A-cluster when supplied endogenously. Synthesis
or cleavage of acetyl-CoA depended on the amount of H2 present
and on a specific hydrogenase.
In 1999, Maynard and Lindahl showed that the rate of acetylCoA synthesis by CODH/ACSMt in the presence of reductants
such as Ti(III) citrate or methyl viologen, was 40 times faster
under CO/CO2 than under CO/Ar.19 Furthermore, the rate
did not depend on the presence of CO. Using hemoglobin
(Hb) as a CO scavenger, they also showed that this hemecontaining protein did not affect acetyl-CoA synthesis. To rule
out the possibility that Hb would reversibly bind CO under their
experimental conditions Maynard and Lindahl supplied Hb-CO
to the assay and showed that there was no acetyl-CoA synthesis.
Their conclusion was that, physiologically, it is CO from the
reduction of CO2 at the C-cluster that combines with CH3 at
the A-cluster to generate acetyl-CoA and that a tunnel connects
both active centers.
Seravalli and Ragsdale provided additional evidence for the
existence of such a channel.20 They studied the isotopic exchange
reaction between labeled 14 CO2 and the unlabeled carbonyl
group of acetyl-CoA according to reaction 1 and the reaction
below, catalyzed by the A-cluster:
(CH3 )12 CO–SCoA + 14 CO ⇒ (CH3 )14 CO–SCoA + 12 CO
(3)
They observed that 33% of the exchange reaction could still
take place in the presence of a high concentration of Hb, when
all the solution CO should have been complexed. From this they
concluded that under their conditions exchange was mediated
by CO2 -derived CO traveling through a tunnel connecting the
C- and A-clusters. Although high levels of external CO did not
inhibit incorporation of labeled carbon from CO2 to acetylCoA, the exchange represented only 60% of the expected value,
probably because some CO formed from unlabeled acetyl groups
3446
at the A-cluster, migrated back to the C-cluster and was oxidized
to CO2 . These authors concluded that CO generated at the Ccluster is preferred to CO from solution and that tunnels are
advantageous for animal hosts because the potentially toxic CO
remains confined to the enzyme interior.
The crystallographic studies of CODH/ACSMt have confirmed the existence of these extensive tunnels.10–11 However,
this tunneling system poses a major problem, namely, how does
CO2 gain access to the C-cluster. We postulated that CO2 could
diffuse from the A-cluster because this center can be accessible
to the medium when CoFeSP delivers the methyl group to the
Ni ion.21 However, Tan et al. have very recently shown that
(putatively) blocking the tunnel connecting the C- and A-clusters
by three mutations, A222L, A265M and A110C, does not
prevent significant CO oxidation activity (that involves diffusion
of CO2 to the external medium).22 Additional mutations at other
potential tunnels regions, transient and thus not detected by Xray crystallography, will be required before this issue is settled.
Determining which residues should be mutated may be helped
by molecular dynamics studies. It should also be noted that the
tunnels observed in Rr and Ch CODHs are different from those
found in CODH/ACSMt but in every case the tunnel points to the
same apical coordination site (E1) of the Ni ion of the C-cluster
(Fig. 3). At any rate, CO2 transit through the protein matrix is
unlikely as we have shown to be the case for H2 diffusion in NiFe
hydrogenases.23
Dalton Trans., 2005, 3443–3450
Fig. 3 Zoom at the part of the tunnel pointing to the Ni-ion of the
C-cluster, (a) in CODHMt refined at 2.5 Å resolution, (b) in CODHCh
model 1SU8.
Lindahl and co-workers have defined two modes of operation
for CODH/ACSMt : the “reductase” (i.e. CO2 reduction) and
the “synthase” (i.e. acetyl-CoA synthesis) modes.24 The fact
that Hb had no effect on the kinetics under “reductase” mode
indicated that there is no equilibrium between external and
internal CO. Under this mode, CO synthesis is less efficient.
Indeed, it has been shown that K cat /K m ratios of CO2 reduction
to CO differ by a factor of 100 depending on whether the bifunctional enzyme is producing acetyl-CoA or just reducing
CO2 . In the presence of an excess of CH3 -CoFeSP (which would
stabilize the open a-subunit and close the tunnel11 ) the amount
of external CO decreased, as measured by CO–Hb formation.
This effect has been assigned to coupling between the A and
C clusters.25 However, the crystal structure indicates that either
electronic or conformational changes induced by small effectors
such as CO2 are unlikely to connect the two active centers given
the long distance that separates them.10–11 Instead, we believe that
any correlated effects have to be attributed to the concentrations
of substrate and product traveling through the tunnels and
possibly having a crowding inhibitory effect depending on the
conformational changes that the a subunit undergoes that affects
opening and closing of the tunnel connecting the A and the C
clusters.
View Online
4 X-Ray absorption spectroscopy
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
26
From early EXAFS studies on CODHRr , Tan et al. suggested
that the Ni ion in the C-cluster was not part of a NiFe3 S4
cubane and that it had some N/O liganding amino acids. As
shown above, these conclusions have been challenged by the
structural results. More recent spectroscopic studies have been
carried out on the Ni-depleted CODHRr because it represents a
much more homogeneous system.17 One important conclusion
from this study using MCD is that the Ni-depleted C*-cluster
was more similar to the native C-cluster than to a classical S =
3/2 [4Fe–4S]+ , as previously reported.12 In fact, the signature of
the C*-cluster was found to closely resemble a [3Fe–xS] cluster.27
Therefore, removal of the Ni ion does not drastically distort the
geometry of the Fe–S site. This also applies to the Zn-substituted
C-cluster.17
Gu and co-workers reported no changes in the Ni absorption
edge upon reduction of CODHCh by either CO or Ti(III)
treatment and, consequently, concluded that nickel does not
get reduced under these conditions.28 However, both XANES
and EXAFS provided evidence for a redox-dependent structural
change in the NiFe4 Sx C-cluster. In the as-prepared CODHCh the
EXAFS could be simulated with 4 Ni–S bonds at 2.2 Å and no
light atom was required. After CO treatment, a new feature at
2.7 Å from the nickel ion appeared (Fe1?) and the Ni–S distances
expanded to 2.25 Å. This effect does not depend on CO as a
ligand to Ni but as an electron donor because it can also be
observed when the enzyme is reduced with Ti(III) citrate.28 So,
although nickel may not be the two-electron acceptor, reduction
significantly modifies the structure of the C-cluster. It should be
noted that similarly small changes in the absorption edge have
been observed in NiFe hydrogenases although it is now widely
accepted that nickel is found in at least two redox states: Ni(II)
and Ni(III).29 Charge delocalization over the thiolate and sulfide
ligands may explain this discrepancy.
5 Infrared studies of CO binding to the C-cluster
A detailed infrared (IR) study of CO binding to CODH/ACSMt
in the absence of external redox mediators was reported recently
by Chen and co-workers.30 They detected at least seven IR
bands between 1900 and 2100 cm−1 suggesting that many
different CO complexes are formed, consistent with the C-cluster
heterogeneity observed in the crystal structures of CO-treated
enzymes. The bands gradually disappeared with time, but with
different decay rates. The strongest band occurred at 1996 cm−1
and it was assigned to CO bound to the A-cluster (the NiFeC
state), in agreement with previous results.31 This was also the
last IR CO band to disappear. Using labeled 13 CO, an additional
band at 2278 cm−1 appeared indicating the formation of 13 CO2
that could be distinguished from the 2349 cm−1 band arising
from natural abundance 12 CO2 present as a contaminant in the
spectrometer. Consequently, all the added CO was gradually
transformed to CO2 , according to the following two reactions:
CODHox + CO + H2 O ⇔ CODHred + CO2 + 2H
(4)
CODHred + 2H+ ⇔ CODHox + H2
(5)
+
H2 formation in (5) was postulated because protons were the
only available electron acceptors in the experiment. This is also
in agreement with the small but significant hydrogen evolution
activity that has been reported for CODH/ACSMt. 32 Based on
the time dependence of the IR band decay the rates of (4) and
(5) were estimated at 30 and 0.018 mM−1 min−1 , respectively,
whereas the rate of CO2 formation was about 30 min−1 . The
expected enzymatic activity measured in the presence of redox
mediators, is more than three orders of magnitude higher but
this agrees with the relative rates of eqn. (5) and (4). Although
the formation of H2 suggests that there is at least a transient
state of the C-cluster with a bound hydride, hydrogen evolution
is clearly not part of the catalytic mechanism and the same may
be true for the involvement of many of the CO-bound states.
This will be further discussed below.
As no CO bands were detectable before CO treatment, there
are no bound intrinsic CO molecules, in contradiction with an
earlier proposal by Ludden and co-workers for CODHRr. 33 The
CO band at 2074 cm−1 decayed the fastest probably because it
corresponded to the initial binding of CO to oxidized enzyme
and this ligand was subsequently oxidized to CO2 . Bands below
2000 cm−1 could reflect CO binding to a reduced C-cluster, most
likely in the Cred2 state. Lower frequencies are to be expected
for CO bound to the reduced cluster because of the ligand pacceptor capability. Because, as mentioned above, the tunnel in
the crystal structures points at an empty apical coordination
site of Ni, CO should bind terminally to this ion. The observed
vibration frequency of 2074 cm−1 is compatible with CO ligation
to a diamagnetic Ni(II) and it is comparable to the value of
2056 cm−1 measured for exogenous CO binding to Ni(II) in
[NiFe] hydrogenase.34,35 Some of the lower frequency bands are
similar to those of an intrinsic CO ligand of Fe(II) in [NiFe]
hydrogenase, compatible with CO binding to Feu , but they could
also correspond to Ni(I)-bound CO, although no stable C-cluster
with this Ni valence state has been reported.
It is not clear how many of the states corresponding to the
observed IR bands play a role in catalysis. As discussed below,
the two bands that were reported at 1727 and 1741 cm−1 and were
assigned to metal bound carboxylate or carboxylic acid30 may
represent components of the catalytic cycle. These bands were
still present after all CO was converted to CO2 , as evidenced by
the disappearance of the 1996 cm−1 band, suggesting that CO2
may form a stable complex with the enzyme.
6 Effect of CO2 binding on the magnetic properties
of the C-cluster
The substrate/product CO2 binds to both Cred1 and Cred2 states.
However, the binding must be different in the two cases because
CO2 does not affect the gav = 1.86 signal (Cred1 ) but it does modify
gav = 1.82 (Cred2 ).16 One possible interpretation is that CO2 binds
terminally to complete the tetrahedral coordination of Ni in Cred2
(indeed, it has been shown that in the Cred2 state neither CO nor
its oxidation product bind to Fe36 ) and it binds non-catalytically
to Ni and Feu in Cred1 , when Ni is square planar. However, this is
more difficult to explain because the oxidative addition of CO2
to Ni would require two additional electrons. Binding of CO2 to
a center other than the C-cluster cannot be ruled out.
An effect of CO2 on raising the redox potential of the
Cred1 /Cred2 pair at the C-cluster has already been reported.37 An
equivalent process seems to operate at the A-cluster that cannot
be reduced by dithionite unless CO is present.37
7
Binding of cyanide to the C-cluster
As for other anions, CN− binds to the Cred1 (gav = 1.82) but not
to the Cred2 (gav = 1.86) species.15 Upon binding, CN− alters the
EPR signal yielding a gav = 1.72. Using ENDOR spectroscopy,
DeRose et al.36 have concluded that CN− and a Hx O species
bind to the same site of the C-cluster but only in the Cred1
state. An exchangeable proton found in Cred1 was absent from
Cred2 or the CN-inhibited enzyme. These authors concluded that
CN− inhibits CODH by displacing OH− bound to Feu . Bridging
of CN− between Ni and Feu was predicted from IR studies of
CODHMt 38 as a 2037 cm−1 band was assigned to a CN− bent
bridge connecting the Ni ion to a low-spin Fe(II). We believe
that slow release of CN− upon reduction (or CO binding) of
the C-cluster may be due to a square planar to tetrahedral
nickel coordination transition that effectively removes the Ni–Fe
bridging site by moving nickel away from Fe1 or to an excessive
negative charge at the reduced C-cluster, or both. An intriguing
fact is that the CN− -treated C-cluster cannot be reduced by
Dalton Trans., 2005, 3443–3450
3447
View Online
dithionite unless CO, CO2 or CS2 are also added. This argues
for a combined reductive and ligand binding event as a requisite
for CN− dissociation from the C-cluster. However, CO2 and CS2
are not direct competitors of CN− binding indicating that they
do not bind to the same site.16 This is consistent with the idea
that CO or CO2 and CS2 plus reductant bind terminally to E1
whereas CN− bridges Ni and Fe1 at the E2 site.
including the presence of a hydrophobic tunnel that points at
the apical Ni ligand site.45 A potential problem with the hydride
model is the absence of a strong ENDOR signal corresponding
to such a species in Cred2. 36 It is not clear, however, what would be
the electronic properties of a C-cluster with a bridging hydride.
If Cred2 does have a bridging hydride this species is not displaced
by other anions such as CN− that can only bind to Cred1. 15
8 The nature of the two-electron acceptor/donor at
the C-cluster
9 Towards a structural basis for CODH catalytic
mechanism
It has been suggested that Cred1 is two electrons more oxidized
than Cred2 16 and that both states are involved in catalysis
according to:
Three states have been assigned to tentatively catalytic relevant
intermediates: Cred1 , Cint and Cred2. 16 Cred1 is one-electron more
reduced than Cox , a species that is not involved in catalysis.
This reduction is thought to take place at the 3Fe4S sub-site.
Our approach to model a catalytic cycle will rely mostly on
the available structural data. In this respect, the first reported
structure of CODHCh is central to any postulate of active
intermediates.7 The fact that the E2 site is blocked by sulfide
in dithionite-treated enzyme may seem puzzling because several
lines of evidence mentioned above indicate that bridging anionic
species should bind only to the oxidized Cred1 form. It is
thus tempting to conclude that CODHCh was not significantly
reduced by the dithionite treatment and that the S2-containing
C-cluster represents a variant of the Cred1 state. Indeed, dithionite
alone cannot reduce the CN− -treated C-cluster (gav = 1.72),15
which can be considered a structural analog of the bridging
SH− -bound C-cluster in CODHCh . In other CO-reduced CODH
crystal structures, the Ni–Feu bridging site does not bind a
crystallographically visible ligand or it is occupied by a Nibound cysteine that moves from a terminal to a bridging position
(Cys531 in Rr and Cys 550 in Mt) (Fig. 2). In the former case, the
bridging site could be occupied by hydride, a proposition that
is consistent with the decay of CO-treated CODH discussed in
the previous section (eqn. (7)). As already mentioned, in the
latter case the shift of the cysteine residue to a bridging position
(Fig. 2a and 2i) may be required to complete the coordination
sphere of Fe1 in the absence of any exogenous bridging ligand.
Extensively CO-reduced C-clusters may represent nonphysiological species. For instance, in our 1.9 Å resolution
CODH/ACSMt structure, where the crystals were exposed to a
high-pressure CO atmosphere prior to immediate flash-cooling
for X-ray data collection, there is no visible Ni–Fe1 bridging
ligand. However, the Ni ion seems to bind CO.11 So, although
substrate/inhibitor binding to iron (E2) in Cred2 may not to be
feasible, the Ni ion could bind CO non-physiologically at E1 in
this state.
A plausible structure-based catalytic cycle is depicted in
Scheme 1. The different intermediates are numbered in the
direction of CO2 reduction. The structure of the C-cluster in
state (2) could correspond to either Ni–COO− or Ni–COOH
(only the latter is shown). These two states may give rise to
the IR bands at 1724 and 1741 cm−1 reported by Chen et al.,30
respectively.
The two-electron reservoir in the Cred2 state (1) is depicted
as a hydride based on the similar WGSR; alternatives are a
Ni(II)/Ni(0) couple or a CysSH/S− CysSS (persulfide) couple
formed between Cys316 and one of the labile sulfufides (not
shown). Oxidative addition of CO2 leads to either a Ni–COO−
or a Ni–COOH species (2). The C–OH bond is broken (3) and a
pentacoordinated Ni(II) carbonyl (4) is generated. As discussed
by Lu and Crabtree,41 Ni(II) carbonyls are rare, except when
the Ni ion is penta-coordinated. Subsequently, CO dissociates
from the C-cluster and the Cred1 state (5), and a bridging OH− ,
appears. One-electron reduction of Cred1 gives rise to Cint (6), an
intermediate that does not accumulate.46
The recipient of this electron remains ill-defined but possible
species are either a Ni(I) with the bridging OH− still bound, or
a Ni(III)–H− if the water molecule has already dissociated from
the C-cluster. An additional one-electron reduction regenerates
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
Cred1 n+ + CO + OH− = Cred2 (n − 2)+ + CO2 + H+
(6)
Where do the two electrons go upon reduction? Mössbauer
spectroscopy has shown that the unusual FCII species of the
C-cluster is present in both Cred1 and Cred2 so it is apparently
not redox-active.12 In addition, there are no significant changes
in either the UV/visible or the Mössbauer spectrum in going
from Cred1 to Cred2 , suggesting that the two electrons are not
located in the Fe-containing part of the C-cluster.12,39 We have
recently proposed that the proximal Ni of the A-cluster oscillates
between distorted tetrahedral Ni(0) and square planar Ni(II)
during acetyl-CoA synthesis based on the crystal structure
of ACS/CODHMt , that displayed open and closed a subunit
conformations.11 A similar mechanism could operate at the Ccluster. The ligand environment of Ni at this site shares attributes
with the proximal Ni of the A-cluster, both are three-coordinate
with all sulfur ligands and Ni coordination oscillates between
tetrahedral and square planar. However, there are two apparent
problems with the Ni(II)/Ni(0) assignment for the C-cluster:
1) the EXAFS experiment mentioned in the previous section
suggests that Ni is 2+ in both Cred1 and Cred2 (but the charge
could be delocalized, as suggested above) and 2) the requirement
for the Ni ligands to stabilize Ni(0).
Another possibility for a redox-active species is the
Cys316/cluster labile sulfide couple that we detected as
forming a minor persulfide component of the C-cluster in
ACS/CODHMt. 11 Recent results from Lindahl’s laboratory indicate that mutating Cys316 into Ala completely abolished the
C-cluster activity. Unfortunately, this was due to the complete
loss of the cluster.40 This result shows that CODH can fold
without the C-cluster and that Cys316 is essential for cluster
integrity. It does not, however, tell us anything about a possible
mechanistic role for this residue. One argument against the
persulfide hypothesis is its absence from all the other C-cluster
structures determined so far. For instance, the initial structure
of CODHCh reported by Dobbek et al.,7 that we will argue below
was in a state analogous to Cred1 , lacks the persulfide.
Finally, the two-electron acceptor could be a proton. The IR
experiments described above30 indicate that CO-treated CODH
decays evolving molecular hydrogen32 according to eqn. (5),
which can also be formulated as:
CO + H2 O → CO2 + H2
(7)
This process is equivalent to the industrial Water-Gas Shift
Reaction (WGSR). Briefly, in the WGSR CO binds to a
metal catalyst, gets attacked by a deprotonated water molecule,
forms a M–COOH species that decarboxylates to generate a
metal hydride that gets protonated, and molecular hydrogen is
evolved.41 This reaction implicates a hydride species as the twoelectron reservoir. Given the anionic character of the Ni–Feu
C-cluster bridging ligands (OH− , SH− , CN− ) it would not be
unexpected for H− to bind in the same fashion in CODH when
in the Cred2 state. Indeed, OH− , SH− and H− have been postulated
as bridging ligands between Ni and Fe in corresponding redox
states of hydrogenases.42–44 Furthermore, the Ni coordination
spheres in hydrogenase and CODH are remarkably similar,
3448
Dalton Trans., 2005, 3443–3450
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
View Online
Scheme 1
the Cred2 (1) species. This cycle is similar to the one proposed
by Feng and Lindahl.13 In fact, as these authors indicate in
the acknowledgements paragraph, it emerged from discussions
between our two groups.
The same intermediates and possibly others would be involved
in the decay of CO IR bands observed by Chen et al.30 in the
absence of electron acceptors other than protons: starting with
CO-bound (4), CO2 would be generated through (3) and (2). The
reductive elimination of CO2 would then lead to the appearance
of (1). Finally, protonation of this species would evolve H2 and
the oxidized C-cluster could bind OH− and CO to regenerate
(4).
for the binding of other anions, such as the substrate OH− and
the inhibitor CN− at the same site.
Although the nature of the species accepting the two electrons
when going from Cred1 to Cred2 remains elusive, we tend to favor
a proton. Besides the obvious similarities with the WGSR and
the results by Chen et al.,30 several reduced structures show that
Fe1 is only coordinated by three protein ligands. Consequently,
it is tempting to propose that the fourth ligand, invisible in the
crystal structures, is a hydride.
Both the structure and the catalytic mechanism of the Ccluster is clearly one of the most complex systems in biology.
Although great progress has been made in understanding
the subtleties of the CODH reaction, further advancements
will require the study of better-defined, more homogenous Cclusters.
10 Conclusions
In spite of the observed structural heterogeneity of the various
C-clusters some key points can be made: 1) as in the A-cluster,11
the Ni ion displays two coordination sites that can be occupied by
exogenous ligands. This appears to be a requirement, as both CO
and OH− should bind simultaneously to the C-cluster. Previous
models that proposed that CO would terminally bind to Ni
and OH− would terminally bind to Fu (Fe1)36 can now be ruled
out given the proximity of Ni to Feu that indicates that both
ligands should bind Ni; 2) the crystal structure of the S2-bound
C-cluster of CODHCh is most likely an inhibited version of the
Cred1 state. This structure that shows the binding of an anionic
SH− species that bridges the Ni and Fe1 ion is a very good model
References
1 P. A. Lindahl, Biochemistry, 2002, 41, 2097–2105.
2 B. Bhaskar, E. DeMoll and D. A. Grahame, Biochemistry, 1998, 37,
14491–14499.
3 J. Xia, J. F. Sinclair, T. O. Baldwin and P. A. Lindahl, Biochemistry,
1996, 35, 1965–1971.
4 (a) L. G. Ljungdahl, Annu. Rev. Microbiol., 1986, 40, 415–450;
(b) H. G. Wood, FASEB J., 1991, 5, 156–163.
5 V. A. Svetlitchnyi, T. G. Sokolova, M. Gerhardt, N. A. Ringpfeil,
N. A. Kostrikina and G. A. Zavarzin, Syst. Appl. Microbiol., 1991,
14, 254–260.
6 H. Dobbek, L. Gremer, R. Kiefersauer, R. Huber and O. Meyer,
Proc. Natl. Acad. Sci. USA, 2002, 99, 15971–15976.
Dalton Trans., 2005, 3443–3450
3449
Downloaded by Universidade de Lisboa on 15 November 2012
Published on 23 September 2005 on http://pubs.rsc.org | doi:10.1039/B508403B
View Online
3450
7 H. Dobbek, V. Svetlitchnyi, L. Gremer, R. Huber and O. Meyer,
Science, 2001, 293, 1281–1285.
8 H. Dobbek, V. Svetlitchnyi, J. Liss and O. Meyer, J. Am. Chem. Soc.,
2004, 126, 5382–5387.
9 C. L. Drennan, J. Heo, M. D. Sintchak, F. Schreiter and P. W. Ludden,
Proc. Natl. Acad. Sci. USA, 2001, 98, 11973–11978.
10 T. I. Doukov, T. M. Iverson, J. Seravalli, S. W. Ragsdale and C. L.
Drennan, Science, 2002, 298, 567–572.
11 C. Darnault, A. Volbeda, E. J. Kim, P. Legrand, X. Vernede, P. A.
Lindahl and J. C. Fontecilla-Camps, Nat. Struct. Biol., 2003, 10,
271–279.
12 Z. Hu, N. J. Spangler, M. E. Anderson, J. Q. Xia, P. W. Ludden, P. A.
Lindahl and E. Münck, J. Am. Chem. Soc., 1996, 118, 830–845.
13 J. Feng and P. A. Lindahl, J. Am. Chem. Soc., 2004, 126, 9094–9100.
14 M. E. Anderson, V. J. DeRose, B. M. Hoffman and P. A. Lindahl,
J. Am. Chem. Soc., 1993, 115, 12204–12205.
15 M. E. Anderson and P. A. Lindahl, Biochemistry, 1994, 33, 8702–
8711.
16 M. E. Anderson and P. A. Lindahl, Biochemistry, 1996, 35, 8371–
8380.
17 J. L. Craft, P. W. Ludden and T. C. Brunold, Biochemistry, 2002, 41,
1681–1688.
18 D. A. Grahame and E. DeMoll, Biochemistry, 1995, 34, 4617–4624.
19 E. L. Maynard and P. A. Lindahl, J. Am. Chem. Soc., 1999, 121,
9221–9222.
20 J. Seravalli and S. W. Ragsdale, Biochemistry, 2000, 39, 1274–1277.
21 A. Volbeda and J. C. Fontecilla-Camps, J. Biol. Inorg. Chem., 2004,
9, 525–532.
22 X. Tan, H. K. Loke, S. Fitch and P. A. Lindahl, J. Am. Chem. Soc.,
2005, 127, 5833–5839.
23 Y. Montet, P. Amara, A. Volbeda, X. Vernede, E. C. Hatchikian,
M. J. Field, M. Frey and J. C. Fontecilla-Camps, Nat. Struct. Biol.,
1997, 4, 523–526.
24 E. L. Maynard and P. A. Lindahl, Biochemistry, 2001, 40, 13262–
13267.
25 E. L. Maynard, C. Sewell and P. A. Lindahl, J. Am. Chem. Soc., 2001,
123, 4697–4703.
26 G. O. Tan, S. A. Ensign, S. Ciurli, M. J. Scott, B. Hedman,
R. H. Holm, P. W. Ludden, Z. R. Korszun, P. J. Stephens and
K. O. Hodgson, Proc. Natl. Acad. Sci. USA, 1992, 89, 4427–
4431.
Dalton Trans., 2005, 3443–3450
27 R. C. Conover, A. T. Kowal, W. Fu, J.-B. Park, S. Aono, M. W. W.
Adams and M. K. Johnson, J. Biol. Chem., 1990, 265, 8533–8541.
28 W. Gu, J. Seravalli, S. W. Ragsdale and S. P. Cramer, Biochemistry,
2004, 43, 9029–9035.
29 A. Volbeda and J. C. Fontecilla-Camps, Dalton Trans., 2003, 21,
4030–4038.
30 J. Chen, S. Huang, J. Seravalli, H. Gutzman, Jr., D. J. Swartz, S. W.
Ragsdale and K. A. Bagley, Biochemistry, 2003, 42, 14822–14830.
31 M. Kumar and S. W. Ragsdale, J. Am. Chem. Soc., 1992, 114, 8713–
8715.
32 S. Menon and S. W. Ragsdale, Biochemistry, 1996, 35, 15814–15821.
33 J. Heo, C. R. Staples, C. M. Halbieb and P. W. Ludden, Biochemistry,
2000, 39, 7956–7963.
34 K. A. Bagley, C. J. Van Garderen, M. Chen, E. C. Duin, S. P. Albracht
and W. H. Woodruff, Biochemistry, 1994, 33, 9229–9236.
35 H. Ogata, Y. Mizoguchi, N. Mizuno, K. Miki, S. Adachi, N. Yasuoka,
T. Yagi, O. Yamauchi, S. Hirota and Y. Higuchi, J. Am. Chem. Soc.,
2002, 124, 11628–11635.
36 V. J. DeRose, J. Telser, M. E. Anderson, P. A. Lindahl and B. M.
Hoffman, J. Am. Chem. Soc., 1998, 120, 8767–8776.
37 W. K. Russell and P. A. Lindahl, Biochemistry, 1998, 37, 10016–
10026.
38 D. Qiu, M. Kumar, S. W. Ragsdale abd and T. G. Spiro, J. Am. Chem.
Soc., 1996, 118, 10429–10435.
39 J. Seravalli, M. Kumar, W.-P. Lu and S. W. Ragsdale, Biochemistry,
1997, 36, 11241–11251.
40 E. J. Kim, J. Feng, M. R. Bramlett and P. A. Lindahl, Biochemistry,
2004, 43, 5728–5734.
41 Z. L. Lu and R. H. Crabtree, J. Am. Chem. Soc., 1995, 117, 3994–
3998.
42 A. Volbeda, L. Martin, C. Cavazza, M. Matho, B. W. Faber, W.
Roseboom, S. P. Albracht, E. Garcin, M. Rousset and J. C. FontecillaCamps, J. Biol. Inorg. Chem., 2005, 10, 239–249.
43 Y. Higuchi, T. Yagi and N. Yasuoka, Structure, 1997, 5, 1671–1680.
44 M. Brecht, M. van Gastel, T. Buhrke, B. Friedrich and W. Lubitz,
J. Am. Chem. Soc., 2003, 125, 13075–13083.
45 A. Volbeda and J.C. Fontecilla Camps, Coord. Chem. Rev., 2005,
online 12 April.
46 P. A. Lindahl, E. Münck and S. W. Ragsdale, J. Biol. Chem., 1990,
265, 3873–3879.
47 A. T. Brünger, F., Nature, 1992, 355, 472–474.