1
Introduction
1.1
Research on Catalytic Reactions Involving
CO and CO2: Relationship to Energy, the Environment, Biogeochemistry,
Toxicology, Health, and Technology
The major environmental-
and energy-related problems facing our planet directly relate to carbon
dioxide and the carbon biogeochemical cycle, which includes the biological
fixation of CO2 into organic carbon and the oxidation of
fixed carbon back to CO2. Thus, the ultimate source of
all fossil fuels is CO2, which has been fixed into organic
carbon and deposited in the earth’s crust over the last 4 billion
years. Because modern life is so reliant on energy, particularly on
fossil fuels, there is intense competition for these nonrenewable
resources, thus creating a problem that has significant economic and
political impacts.
1
A related problem of
growing concern is the rising level of greenhouse gases, especially
CO2 and methane.
In nature, fixation of CO2 occurs on a huge scale, with photosynthetic CO2 fixation
occurring at a rate of 200 gigatons per year.
2
There are six known pathways by which CO2 is fixed,
3
with the Calvin cycle and photosynthesis providing
most of this fixed carbon. Under anaerobic conditions, the Wood–Ljungdahl
pathway is a predominant CO2 sink, and CO dehydrogenase
(CODH) and acetyl-CoA synthase (ACS), the subjects of this review,
are the key enzymes in this pathway.
4
The oxidation of organic carbon to CO2 slightly outpaces
CO2 fixation, leaving a balance in the atmosphere. In May
2013 at Mauna Loa Observatory, the atmospheric CO2 levels
reached 400 ppm—their highest value since records began—and
the levels are increasing at a rate exceeding 2 ppm per year.
5
Further increases are predicted to produce large
and uncontrollable impacts on the world climate, and evidence suggests
that these changes are underway.
5,6
Thus, it is important
to develop renewable nonfossil energy supplies that are CO2-neutral and easily stored,
distributed, and used. CODH/ACS and the
Wood–Ljungdahl pathway of CO and CO2 fixation could
play a role in this development.
Our ability to deal with these
environmental- and energy-related
problems will depend upon our understanding of the biology related
to the global carbon cycle, especially those processes that lead to
and limit CO2 fixation. One might imagine biotechnological
solutions to both the greenhouse gas and energy-limitation problems.
For example, supplying limiting nutrients, e.g., iron fertilization
in the Ironex experiments, can stimulate CO2 fixation in
the ocean.
7
Similarly, given the high efficiency
and rates of enzymatic CO2 activation and fixation, principles
borrowed from nature are being explored to design better CO2-reactive catalysts.
8
While CO2 is a relatively inert and nontoxic product
of the complete oxidation of carbon, CO is a reactive, toxic gas that
is produced naturally in some anaerobic bacteria by the two-electron
reduction of CO2 and in aerobic organisms by heme oxygenase-catalyzed
decomposition of porphyrins.
9
CO also is
generated anthropogenically by the incomplete combustion of organic
materials, predominantly by the oxidation of methane and other hydrocarbons.
In the United States, poisoning by CO is responsible for ∼1000
accidental deaths,
10
while more than 50 000
people per year seek medical attention for CO poisoning.
11
Faulty furnaces, inadequately ventilated heating
sources, and engine exhaust exposure are the main culprits of CO poisoning.
The mode of toxicity appears to be inhibition by binding tightly to
the metallocenters in heme proteins, such as hemoglobin, myoglobin,
and cytochrome oxidase.
12
CO emissions
lead to atmospheric levels of CO ranging from 0.05 ppm in rural areas
to as high as 350 ppm in some urban settings.
13
Though this level is below the toxicity threshold, the OSHA limit
for CO is 50 ppm continuous exposure for 8 h. Mild effects of CO poisoning
are observed in humans when CO levels remain as high as 200 ppm for
2–3 h and exposure to 1000 ppm for 1 h is fatal. Though it
may seem counterintuitive, given its reputation as “silent
killer” and environmental pollutant, CO, at low levels, is
cytoprotective and therapeutic applications for cardiovascular diseases,
inflammatory disorders, and organ transplantation are being explored.
14
This strategy follows the recognition that heme
oxygenase-1 is induced during tissue injury and oxidative stress.
15
Diverse microbes can grow on CO as their
sole source of carbon
and electron-equivalents.
16
This includes
anaerobes such as Moorella thermoacetica,
17
some purple sulfur bacteria akin to Rhodospirillum rubrum,
18
and Carboxydothermus hydrogenoformans,
19
as well as some aerobic carboxydobacteria
like Oligotropha carboxidovorans.
20
These are the organisms in which CO metabolism
has been most thoroughly studied. As indicated by its low half-cell
potential (−0.52 V, below), CO is a potent electron donor—approximately
1000-fold stronger than NADH—and life forms have probably utilized
that property of CO as an energy source ever since life emerged 4
billion years ago in the archaean eon. Approximately 1 century ago,
Haldane
21
and Leduc
22
suggested that the earliest organisms were likely to have
been anaerobic autotrophs, and it has been proposed that life emerged
within anaerobic hydrothermal vents by exploiting CO as a carbon and
energy source.
23
The early atmosphere,
which was formed by outgassing from the earth’s interior by
volcanoes and hydrothermal vents, is expected to have a similar composition
to that of modern volcanoes and vents, with little to no O2 and relatively high concentrations
of CO2, CO and CH4. Hydrothermal vents, which contain dissolved CO at about
100 nM concentrations,
24
still support
diverse populations of anaerobic CO oxidizers.
16,25
It has been suggested that a version of the Wood–Ljungdahl
pathway may have been the first metabolic sequence to emerge, with
early organisms metabolizing CO and CO2 using ancestral
forms of CODH/ACS.
3,26
Contemporary bacteria that use
this pathway, such as M. thermoacetica and C. hydrogenoformans, have been
proposed as models for these early chemolithotrophs.
19
Anthropogenic CO production amounts to about 2 billion
tons per
year,
27
while microbial CO metabolism is
partly responsible for maintaining the ambient CO below toxic levels
by removing an approximately equal amount of CO from the Earth’s
atmosphere.
28
As described in more detail
below, the microbial enzymes responsible for CO oxidation can operate
at rates as high as 40 000 (mol CO)(mol enzyme)−1 s–1 and catalytic efficiencies reaching
2 ×
109 M–1 s–1.
29
Anaerobic microbes can grow on CO by use of
CODH/ACS, the topic of this review, to initiate metabolism by the
Wood–Ljungdahl pathway, while aerobes use a Cu Mo-pyranopterin
CODH that is coupled to the Calvin–Benson pathway.
30
CO is also of great importance in the
chemical industry and its
reactivity is linked to formation of metal–CO bonds. For example,
the M–CO complex plays a key role in industrial organometallic
catalysis reactions, including the Monsanto process for acetate synthesis;
the industrial Reppe process leading to the carbonylation of alkenes,
alkynes, and conjugated dienes; Fischer–Tropsch reaction; hydroformylation;
homologation; the water–gas shift (WGS); and hydrogenation
using water as the hydrogen source.
8,31
Furthermore,
we feel it is likely that interdisciplinary research on the enzymology
of CO oxidation will lead to the development of novel catalysts that
follow principles used by the natural catalysts for carbonylation
(ACS) and reversible CO2 reduction (CODH).
1.2
CO and CO2 Chemistry
Carbon
dioxide is the final product of the compete oxidation of carbon. A
comprehensive review on CO2 activation and reduction is
available;
8
thus, we will summarize only
those aspects of CO and CO2 reactivity that are most relevant
for the present review on CODH/ACS. CO2 is very abundant
in the atmosphere and stored as various forms of carbonate, yet it
is relatively inert, which raises the stakes for researchers to describe
strategies to convert CO2 to useful products.
This
review focuses on a two-enzyme complex that couples two extremely
important reactions in biology and industry. CODH catalyzes CO2 reduction to CO and
ACS catalyzes C–C bond formation
using CODH-generated CO and a methyl group to generate the key metabolic
intermediate acetyl-CoA. This coupled reaction is a highly efficient
biochemical equivalent of coupling the WGS reaction to the Monsanto
process in a single reaction mixture. Here we will briefly review
the chemical principles related to the activation and reduction of
CO2 and to the use of CO in carbonylation and C–C
bond-forming reactions.
1.3
Introduction to CODH- and
ACS-Dependent Microbial
CO and CO2 Fixation
The use of CO, a toxic gas
to animals, as a metabolic building block is an interesting property
of certain classes of diverse organisms that can fix CO2 and are capable of converting
CO into CO2. CODH reversibly
oxidizes CO to CO2. This activity allows organisms to grow
on CO as a sole source of carbon and energy. The CO2 is
then fixed into cellular carbon by one of the six known reductive
CO2 fixation pathways.
3
A review
is available that covers the history of microbial CO oxidation and
our understanding of the catalytic mechanism of CODH and ACS up until
∼2003.
30
Aerobic CO metabolism
is performed by carboxydotrophic bacteria, which are aerobic microbes
that grow on CO as their sole source of carbon and energy,
16
fixing CO according to eq 1.
32
Aerobic CO oxidizing bacteria are
taxonomically diverse, including α-, β-, and γ-proteobacteria;
Firmicutes; and Actinobacteria, including pathogenic and nonpathogenic
mycobacteria.
16,33
These microbes transfer the electrons
derived from CODH-catalyzed CO oxidation to O2 through
an electron transport chain involving quinones.
34
The CO2 is assimilated into cell carbon through
the Calvin–Benson–Basham pathway.
16,35
The enzyme responsible for CO oxidation is called MoCu–CODH
because it contains a binuclear Mo–Cu center in which the Cu
is thiolate ligated to a molybdopterin center.
36
The CODH of Oligotropha carboxidovorans is the most thoroughly characterized MoCu–CODH
enzyme.
36,37
This three-subunit enzyme also contains two [2Fe–2S] clusters
and FAD, in common with other members of the xanthine oxidoreductase
family.
36,38
1
The Ni–CODH plays a similar role in anaerobic microbes that
the Mo–Cu enzyme plays in aerobic metabolism, allowing organisms
to grow autotrophically on CO by coupling CO oxidation to CO2 fixation. Purple sulfur
bacteria like Rhodospirillum
rubrum and Rubrivivax gelatinosus(18) couple CO oxidation to the Calvin–Benson–Basham
cycle, while methanogenic Archaea and sulfate reducing and acetogenic
bacteria use the Wood–Ljungdahl pathway.
30
The CODH from the latter organisms contains a tightly associated
ACS, which is purified either as an α2β2 complex containing a central core of two
CODH subunits that
are associated on either side by two ACS subunits
39
or as a larger complex containing other components of the
Wood–Ljungdahl pathway (e.g., the corrinoid iron–sulfur
protein and methyltransferase).
40
The association of CODH with ACS confers the ability to utilize
the Wood–Ljungdahl pathway to perform diverse reactions in
the carbon cycle (Figure 1). As shown in the
top left panel, CODH/ACS allows organisms to grow autotrophically
on CO and CO2. In this pathway, CODH catalyzes CO2 reduction into CO; then, ACS catalyzes
the condensation of in situ
generated CO with CoA and a methyl group bound to the cobalt center
in a B12-containing protein, to generate the key metabolite,
acetyl-CoA. This mode of autotrophic growth is used by a variety of
anaerobic microbes, including acetogenic bacteria and methanogenic
Archaea. The Wood–Ljungdahl pathway is found in a wide distribution
of phylogenetic classes, including Clostridia, Deltaproteobacteria,
Chloroflexi, and Spirochaetes, and is also found in two domains (Archaea
and Bacteria); however, it is found in only a few species within these
classes, suggesting that this pathway was distributed by horizontal
gene transfer of the core genes (CODH, ACS, MeTr, CFeSP).
41
The marker genes for the Wood–Ljungdahl
pathway are acsB (ACS) and the two subunits of the
CFeSP (acsC and acsD); as the only
genes that co-occur and are co-omitted among the sequenced bacterial
genomes,
41
these enzymes are undoubtedly
crucial for acetogenesis.
Figure 1
The Wood–Ljungdahl pathway of CO/CO2 fixation
and its involvement in acetogenesis and methyltrophy, as well as in
the oxidation of acetate to methane. The methanogenic CODH/ACS is
often called ACDS, acetyl-CoA synthase decarbonylase.
In acetogenic bacteria, this pathway generates
acetate (eq 2), conserving energy through electron
transfer-linked
phosphorylation. As shown in the bottom left panel (Figure 1), coupling methanogenesis
to this pathway (operating
in reverse) gives organisms the ability to convert acetate to methane.
Sulfate-reducing bacteria can utilize the eight electrons generated
during acetate oxidation (again using this pathway in reverse, bottom
right scheme) to reduce sulfate to H2S. The pathway also
allows organisms to grow on various methyl donors, such as methanol
and aromatic methyl ethers (top right panel). Furthermore, any oxidative
pathway that generates CO2 can potentially couple to the
Wood–Ljungdahl pathway. For example, heterotrophic growth on
sugars allows organisms to stoichiometrically convert glucose into
3 mol of acetate by capturing the reducing equivalents and the 2 mol
of CO2 generated by glycolytic oxidation of pyruvate to
acetyl-CoA and generating a third mole of acetyl-CoA.
2
42
2
CO Dehydrogenase
2.1
Redox Chemistry and Enzymology
Involving CO
and CO2
Because there is such a large reservoir
of CO2 and its potential for conversion into useful products,
there is much interest in the activation and reduction of CO2. The energetic requirements
for CO2 reduction (eqs 3–8) at pH 7 vs NHE depend
on the number of electrons in the redox half-reactions, as shown in
eq 3–8.
3
4
5
6
7
8
9
10
The one-electron reduction
of CO2 (eq 3) requires very negative
potentials,
due in part to the energy required for structural rearrangement of
linear CO2 to form the bent CO2 anion radical.
43
The high overpotential (cell potentials in excess
of 2.0 V) associated with formation of this radical anion intermediate
remains the major obstacle to rapid and efficient heterogeneous electrochemical
reduction of CO2.
8
On the other
hand, the two-electron reduction to either CO or to formate occurs
under much less demanding redox conditions (eqs 4,5). Both CO and formate formation
are pH-
and solvent-dependent,
44
being more favorable
at low pH. The optimal catalyst for CO2 reduction to either
CO or formate should avoid the highly energetically unfavorable formation
of an anion radical as a catalytic intermediate. This principle has
been demonstrated with an enzyme-based CO2 electroreduction
catalyst, which rapidly generates CO via eq 4 at −0.52 V, i.e., without any overpotential.
45
In biology, the two CO2 reduction reactions
to CO and
formate are catalyzed by CO and formate dehydrogenase, respectively,
which contain metal clusters to aid in CO2 activation and
electron transfer. These reactions are important in the global carbon
cycle and are keys to the activation of CO2 under anaerobic
conditions.
8,46
Similarly, the synthetic catalysts
that promote these reactions contain metals that bind CO2 and facilitate electron
transfer.
Homogeneous catalysts provide
one mechanism for the reduction of
CO2, by hydrogenation to formate, yet to increase its reductant
potential, high H2 pressures and/or bases are used to drive
the reaction.
8
The chemical interconversion
between CO and CO2 (eq 9) is an important
industrial reaction called the WGS reaction, which, in the reverse
direction, provides fuel-cell-grade H2 from steam reforming.
47
In this direction, the reaction is marginally
favorable with a ΔH
o of −41.2
kJ mol–1 and a ΔG
o
298 of −28.6 kJ mol–1
48
and is typically performed at temperatures greater
than 200 °C using d-metal catalysts on oxide supports.
49
Because of the industrial importance, a number
of laboratories in academia and industry are developing catalysts
that rapidly and efficiently produce H2 from CO and water;
for example, Ru3(CO)12 and a recent Au–CeO2 nanomaterial were described with a reactivity
of 0.01 (mol
H2) s–1 (mol metal carbonyl)−1 at 160 °C
50
and between 0.3 and
3.9 site–1 s–1 at 240 °C,
49b
respectively.
The WGS reaction is very
similar to the reaction catalyzed by the
enzyme CODH. In comparison, an enzyme (CODH)-based electrocatalyst
yields a value for CO oxidation of >3.2 s–1 at
30
°C.
51
In solution, the enzymatic oxidation
of CO by CODH I from C. hydrogenoformans (CODH
Ch
I) occurs with a turnover frequency
of ∼40 000 s–1;
29,45a
however, in the enzymatic reaction, electrons are transferred to
redox proteins (e.g., ferredoxin) that couple to other redox enzymes
like hydrogenase with proton reduction being a very slow side reaction.
52
By coadsorbing the C. hydrogenoformans CODH and Escherichia coli hydrogenase
to conducting graphite particles,
51
highly
efficient CO-dependent H2 production has been observed
with a turnover frequency at 30 °C comparable to that of conventional
high-temperature WGS catalysts (>2.5 s–1) (see
section 1.2).
51
This
biochemical
reaction performed on purified enzymes is similar to the mode by which
some anaerobic microbes grow. C. hydrogenoformans is an anaerobic organism that can
live on CO as sole carbon source,
evolving H2 as a byproduct.
29
A number of other microbes have been discovered that also adopt
this seemingly extreme life style.
25,30,53
Multielectron reduction of CO2 is
a very important reaction.
Given that this process is more thermodynamically favorable than the
two-electron reduction, it is somewhat surprising that such a reaction
has not been discovered in nature, which instead uses discrete two-electron
steps. For example, methanogenic archaea specialize in catalyzing
CO2 reduction to methane (eq 8),
which, when coupled to H2 oxidation, is thermodynamically
favorable and provides energy for cellular growth.
54
Similarly, acetogenic bacteria catalyze the reduction of
CO2 to acetic acid (eq 2), coupled
to the oxidation of H2 or other electron donors. In nature,
these eight-electron reduction reactions occur by discrete two-electron
steps through the formate (CO), formaldehyde, methanol, and methane
oxidation levels with the carbon from CO2 bound to and
transferred among organic or metallic cofactors during the process.
There are at least two reasons for the natural strategy of using
enzymes that catalyze discrete two-electron-transfer steps. One is
that the intermediates in the one-carbon metabolism branch off into
various directions to make important cellular metabolites. Another
is that the microbe is producing the final product (CH4 or CH3COOH) as a byproduct,
with energy being conserved
as ATP (through electron-transfer-linked phosphorylation) in the most
thermodynamically favorable reaction(s) in the sequence.
In
synthetic systems, multielectron CO2 reduction has
had limited success and the catalysts generally require large overpotentials,
are unstable, and exhibit low product selectivity and yields, with
the predominant industrial pathway for multielectron reduction being
through CO.
8
CO is readily available as
syngas (a mixture mainly of CO, CO2, and H2),
which is produced by steam reforming (or other gasification processes)
of reduced carbon-containing compounds like natural gas, coal, and
biomass; however, these processes require high temperatures and are
energy intensive. Thus, development of a highly efficient process
for converting CO2 to CO would have high impact on hydrocarbon
production from CO2.
Interestingly, there are no
known enzymatic catalysts for multielectron
CO reduction; however, nitrogenase, which functions in nature to catalyze
the eight-electron reduction of N2 and two protons to form
H2 and ammonia, providing fixed nitrogen into the global
nitrogen cycle,
55
has been modified by
mutagenesis to catalytically reduce CO directly, albeit very slowly.
55a,56
The related vanadium-based nitrogenase slowly reduces CO to form
a variety of short chain hydrocarbons, including ethylene, ethane,
propane, and propylene.
57
In the formation
of hydrocarbons from CO by nitrogenase, CO binds to Fe atom(s) on
one face of FeMo-cofactor.
58
A number
of chemical catalysts have been developed for multielectron
reduction of CO, though most require high temperatures and pressures
and produce mixtures of products.
8
For
example, Fischer–Tropsch conversion of CO to methanol and other
hydrocarbons using Cu/ZnO catalysts is a well-developed and efficient
process.
59
2.2
Characteristics
of Ni–CODHs
2.2.1
Enzymatic Activities
Ni–CODHs
can catalyze the reversible conversion of CO to CO2 with
specific activities as high as 15 756 U/mg (k
cat of ∼39 000 s–1) reported
at pH 8 and 70 °C for CODH I from C. hydrogenoformans (CODH
Ch
I) using conventional kinetic
assays.
29
Other two well-studied Ni–CODHs,
CODH from R. rubrum (CODH
Rr
) and CODH/ACS from M. thermoaceticum (CODH/ACS
Mt
), are reported to oxidize
CO at k
cat values of ∼10 000
and ∼3000 s–1, respectively.
60,60b
The high catalytic rates and their wide range among different CODHs
attract significant interest; however, various properties of these
enzymes have made it difficult to perform mechanistic investigations
and structural studies. Perhaps the most challenging issue is that
the Ni–CODH is extremely oxygen-sensitive; therefore, growth
of the organism and purification and manipulation of the enzyme require
the strict avoidance of contact with oxygen. This is most easily accomplished
by performing studies within an anaerobic chamber whenever possible
and by using Schlenck line techniques for any investigations outside
the chamber. For example, the glovebox within the authors’
laboratory maintains the oxygen level below 2 ppm. The rapid catalytic
turnover frequencies pose problems because most stopped-flow and freeze-quench
instruments have dead times in the 1 ms range, while under optimal
catalytic conditions, the half-time for all intermediate steps in
the reaction cycle must be greater than 0.2 ms (0.7/3000) for even
one of the least active enzymes (that from M. thermoacetica). Yet these issues have
been mostly overcome by performing rapid
kinetic experiments at low temperatures and/or at suboptimal pH values.
61
CO2 becomes a substrate for
CODHs at redox potentials below ca. −300 mV, and the turnover
frequency is in the range of 10 s–1, which is significantly
lower than the k
cat values for CO oxidation.
62,63
Electrochemical studies showed that CODH/ACS
Mt
catalyzes CO2 reduction very efficiently with almost
no overpotential.
64
Reduction of CO2 to CO plays a key role in the Wood–Ljungdahl pathway
65
(Figure 1) and could
allow fuel production if an efficient large-scale enzymatic electrocatalyst
could be achieved. Experiments with electrode-immobilized CODH are
described below in section 2.5.
Catalytic
reactions reported for Ni–CODHs are not limited
to CO/CO2 conversion. CODH
Rr
produces formate as a slow side reaction during CO2 reduction
in its nickel-containing and nickel-deficient forms.
66
CODH/ACS
Mt
can convert nitrous
oxide to dinitrogen in the presence of a low-potential electron donor.
67
CODH/ACS
Mt
has been
shown to catalyze the anaerobic reduction of 2,4,6-trinitrotoluene,
a dangerous pollutant.
68
Furthermore, CODH/ACS
Mt
can catalyze the oxidation of n-butyl isocyanide (n-BIC) to n-butyl
isocyanate (n-BICt).
69
In addition, the C531A and H265 V variants of recombinant CODH
Rr
catalyze H2 oxidation and hydroxylamine
reduction, respectively.
70
2.2.2
Structural and Spectroscopic Properties,
Metal Clusters, and Redox Chemistry
The X-ray structures
of five Ni–CODHs have been reported. These include structures
of three bacterial (M. thermoacetica, C. hydrogenoformans, and R. rubrum) and one
archaeal (Methanosarcina
barkeri) enzyme.
39,71
The bacterial
enzymes have sequence similarities between 46% (C.
hydrogenoformans and R. rubrum) and 63% (M. thermoacetica and R. rubrum) and structures
that are nearly identical
(RSMD of ∼0.95 Å according to PDB 1MJG and 1JQK). Crystal structures
clearly reveal the presence of five metal clusters per homodimeric
enzyme, two nickel–iron–sulfur clusters, called the
C-clusters, one Fe4S4 D-cluster; and two Fe4S4 B-clusters, as shown in Figure 2.
39b,71b,39a,71a
The structures also reveal why
all CODHs are dimeric—there is a single D-cluster that bridges
the two subunits; furthermore, the C-cluster of one subunit and the
B-cluster of the other are closer than those from the same subunit.
Thus, a functional dimer is required for rapid electron transfer.
The methanogenic CODH contains two more Fe4S4 clusters (E- and F-clusters) than the
bacterial enzymes. Since one
subunit is positioned over the D-cluster of this enzyme, E- and F-clusters
are proposed to be part of the electron transfer chain.
71i
This proposal is supported by the high sequence
similarity between the FeS domain bearing E- and F-clusters and M. barkeri pyruvate
ferredoxin oxidoreductase, electron
donor for ferredoxin, and the location of these clusters between the
surface and the B-cluster. Structures of the B-, C-, and D-clusters
are shown in Figure 2.
Magnetic circular
dichroism (MCD), resonance Raman (rR), and electronic absorption spectroscopic
studies on the nickel-deficient CODH
Rr
support the presence of two different types of [Fe4S4]2+/+ clusters, presumably consisting
of the bridging
D-cluster and the two B-clusters.
71b,71a,72
The midpoint potential of the B-clusters, between
−300 and −530 mV, is consistent with an electron transfer
role.
72
Interestingly, the D-cluster adopts
a diamagnetic 2+ state at potentials higher than −530 mV.
72
Although the D-cluster shows an unusually low
redox potential, its proximity to the surface and the B-cluster would
be consistent with an electron transfer role in the CODH mechanism,
though the role of this cluster has not been established.
Reversible
CO/CO2 conversion was shown to occur at the
C-cluster;
61,62a,73
thus, there is much interest in characterizing this metal center,
which is composed of an iron–sulfur cluster combined with a
nickel atom.
74−76
Four different oxidation states for the C-cluster
have been suggested: a catalytically inactive and EPR-silent Cox state; a one-electron
reduced Cred1 state, which
binds CO and has an electron paramagnetic resonance (EPR) spectroscopic
signal with g-values at 2.01, 1.81, and 1.65 (g
av = 1.82); a two-electron-reduced EPR-silent
Cint state;
77
and a three-electron-reduced
form, Cred2, which binds CO2 and has a distinct
EPR signature with g-values of 1.97, 1.87, and 1.75
(g
av = 1.86).
74,78
The electronic structure of these redox states is not clear yet;
however, the majority of unpaired electron spin density is localized
on Fe in both Cred1 and Cred2, which exhibit
large 57Fe and small 61Ni hyperfine values.
79
The g-values and midpoint redox
potentials for the metal clusters of CODHs from various organisms
are shown in Table 1.
Figure 2
(A) Structure of CODH
Rr
in cartoon
representation, (B) distances between the metal clusters, (C) structure
of the D-cluster, (D) structure of the B-cluster, and (E) structure
of the C-cluster. Atom colors: dark gray (iron), orange (sulfide),
red (oxygen), blue (nitrogen), white (carbon), dark green (nickel).
Generated using Pymol from PDB 1JQK.
The nickel, iron, and sulfide content; molecular structure;
and
redox properties of the C-cluster have been the subject of many spectroscopic
and structural studies (Figure 3).
74,75,79b,80−85
The X-ray diffraction structures and anomalous dispersion experiments
revealed that Ni in the C-cluster is a part of a slightly distorted
iron–sulfur cubane. Another iron atom in the C-cluster, but
outside the cubane, was assigned as ferrous component II (FCII) (also
called unique iron and the pendant Fe), according to a Mossbauer study.
85b
For the Cred1 state, a ferrous component
III (FCIII) was also described while other two irons were assigned
to be mixed valence Fe2+Fe3+.
74
Thus, according to this scenario, Cred1 would
consist of three ferrous and one ferric iron. The initial crystal
structure of C. hydrogenoformans CODH
II (CODH
Ch
II) included a bridging sulfido
ligand connecting nickel and the pendant iron, indicating the cluster
composition as [NiFe4S5],
71b
with the bridging sulfide proposed to serve an undetermined
catalytic role.
71c
However, crystal structures
for CODH
Rr
,
71a
CODH/ACS
Mt
,
39a,39b
and another CODH
Ch
II crystal structure
86
do not include the bridging sulfide. Furthermore,
sulfide appears to reversibly inhibit CODH
Rr
and CODH/ACS
Mt
.
87,88
Inhibition by sulfide and other ligands, which bind to different
oxidation states of the C-cluster, will be discussed in more detail
below in section 2.2.3. It is now accepted
by the community that there is no bridging sulfide between Ni and
the pendant Fe in the active form of the C-cluster. This Fe-bound
hydroxide is viewed as the nucleophile that attacks a Ni–CO
to generate a metal-bound carboxylate during the catalytic cycle.
71d,85c,94
It has been suggested that sulfide
acts as a reversible inhibitor by replacing the catalytically important
hydroxide.
87,88
Crystallographic studies of the
carboxylate-bound state,
71a
observation
of COS as a substrate,
95
and weak CO-dependent
hydrogen evolution activity of CODHs
96
support
this proposal. The CODH structure in its Cred1 state reported
by Jeoung and Dobbek also is interpreted to have a bridging hydroxide
between Ni and pendant Fe. ENDOR spectroscopy of Cred1 reveals
the proton from the metal-bound hydroxyl group while Cred2 appears to lack this spectral
feature.
85c
On the other hand, in the Cred2 state, a bridging hydride
was proposed upon computational calculations.
97,98
Structural changes upon catalytic activity will be discussed later.
Table 1
Spectroscopic and Electrochemical
Data for the Ni–CODHs from Different Sources
A-cluster
B-cluster
C-cluster
g-values (−CO)
E
0′
g-values
E
0′
g-values
E
0′
R. rubrum(74,89)
2.04,
1.94, 1.89
–418
2.03, 1.88, 1.71
–110
1.97, 1.87, 1.75
C. hydrogenoformans(88)
2.04, 1.93, 1.89
2.01, 1.89, 1.73
1.96, ?, 1.77
M. thermoaceticum(79b)
2.08, 2.07, 2.03
2.04, 1.94, 1.90
–440
2.01, 1.81, 1.65
–220
2.06, 2.05, 2.03
–530
1.97, 1.87, 1.75
–530
M. thermoaceticum with azide
90
2.34, 2.07, 2.03
2.34, 2.11, 2.04
M. barkeri(91)
2.05, 1.94, 1.90
–390
2.01,
1.91, 1.76
–35
?, ?, 1.73
M. soehngenii(92)
2.05, 1.93, 1.86
–410
2.01, 1.89, 1.73
–230
M. thermophila(93)
2.06, 2.05, 2.03
2.04, 1.93, 1.89
–444
2.02, 1.87, 1.72
–154
?, ?, 1.79
Figure 3
Structure of C-cluster including only one coordinating
residue,
cysteine, and the ligands from (A) CODH
Rr
(PDB 1JQK),
(B) CODH
Ch
II (PDB 1SU8), (C) CODH
Ch
II at 320 mV (PDB 3B53), (D) CODH
Ch
II at 600 mV (PDB 3B51), (E) cyanide-bound CODH
Ch
II at 320
mV (PDB 3I39), (F) CO2-bound CODH
Ch
II
at 600 mV (PDB 3B52), (G) cyanide-bound CODH/ACS
Mt
(PDB 3I04), (H) CODH/ACS
Mt
(PDB 3I01), (J) n-BICt-bound CODH/ACS
Mt
(PDB 2YIV), (K) CO-bound CODH
Mb
(PDB 3CF4). Atom colors: Dark gray (iron), orange (sulfide), red (oxygen),
blue (nitrogen), white (carbon), dark green (nickel).
Accessory proteins (CooC, CooT, and CooJ), whose
genes are part
of a CODH-containing gene cluster in R. rubrum, appear to be required for assembly
of the C-cluster.
99
Deletion of CooC, which has ATPase and GTPase
activity and a nucleotide-binding P-loop region, leads to a C-cluster
that contains the Fe–S but lacks Ni components of the cluster.
99,100
This Ni-deficient form of CODH
Rr
can
be activated in vitro by incubation of the reduced protein with NiCl2.
101
However, a similar role for
AcsF, the M. thermoacetica homologue
of CooC, could not be established.
102
On
the basis of homology with HypC, CooT may be involved in metal ion
discrimination.
99
CooJ has a histidine-rich
C-terminus and binds up to four nickel ions per monomer.
103
As shown in Figure 2, the C-cluster is deeply
buried inside the enzyme with the C-, B-, and D-clusters aligned as
an efficient redox wire with 10–11 Å intercluster distances
to allow rapid electron transfer.
71b
The
structures of CODH/ACS
Mt
, CODH
Rr
, and CODH
Ch
II are
very similar, with strict conservation of all amino acid residues
that ligate the metal clusters (Figure 2, Table 2). Other residues that are thought
to be important
in acid–base chemistry are also identified in Table 2.
Table 2
Key Residues in the
Primary and the
Secondary Coordinating Spheres of the Metal Centers in Different Ni–CODHs
organism (PDB ID)
A-cluster
B-cluster
C-cluster
D-cluster
His-tunnel
acid–base
Rr (1JQK)
C50
C300, C338
C41, C49
H95
K568
C53
C451, C481
C41′,
C49′
H98
H95
C58
C531, H265
H101
D223
C72
W575
Ch (3B51)
C48
C295, C333
C39,
C47
H93
K563
C51
C446, C476
C39′, C47′
H96
H93
C56
C526, H261
H99
D219
C70
H102
W570
Mt (1OAO)
C506, C509
C68
C317, C355
C59, C67
H113
K587
C518,
C528
C71
C470, C500
C59′,
C67′
H116
H113
C595, C597
C76
C550,
H283
H119
D241
G596
C90
H122
W594
2.2.3
Inhibition of CODH Enzymatic
Activity
Several molecules including nitrous oxide, sulfide,
azide, thiocyanate,
cyanate, cyanide, and n-BIC are known to inhibit
the catalytic activity of CODHs.
60b,67,69,71c,88,90,104
Here we will describe research on these inhibitors that has helped
to enlighten the CODH catalytic mechanism.
Electrochemical studies
combined with EPR spectroscopy showed that cyanate, an analogue of
CO2, binds the Cred2 state and inhibits CO2 reduction.
88
Most likely it binds
to the active site in a similar fashion as CO2 and could
be used in structural studies. Inhibition of CO oxidation is limited
to a very narrow potential range, with almost no inhibition occurring
at potentials more positive than −0.4 V.
88
Binding of cyanate is slow, requiring several seconds with
millimolar concentrations. On the other hand, isocyanides (e.g., n-BIC), which have
been previously used as CO analogues,
36,105
can act both as a substrate and an inhibitor of CODH/ACS
Mt
.
69,71h
Since CODH catalyzes the oxidation
of n-BIC to n-BICt much more slowly
(105-fold) than CO oxidation, n-BIC behaves
as a rapidly binding competitive inhibitor of CO oxidation with a K
i
value of 1.66 mM.
69
The crystal structure of CODH
Ch
II treated with n-BIC reveals the C-cluster
in an n-BICt-bound state containing a Ni–C
bond and a hydroxyl group attached to the pendant iron (Figure 3J).
71h
A hydrogen-bonding
network that likely plays a role in stabilizing the C-cluster-bound
CO2 includes the iron-bound hydroxyl, a free water molecule,
the oxygen of the n-BICt, and two residues, His93
and Lys563.
Cyanide, an analogue of CO, is a reversible inhibitor
of CODH.
71f,71g,82,94,104,106,107
Depending on the conditions,
cyanide can act as a
rapid reversible inhibitor or a slow binding inhibitor.
106a
When cyanide binds to the C-cluster in the
Cred1 state, it forms a complex with an EPR spectrum that
exhibits a g
av of 1.72 (g = 1.55, 1.78, 1.87).
104,106a,108
CN– does not interact with the Cred2 state nor does it inhibit reduction of CO2.
88
Several studies suggested the nickel as the
binding site for the cyanide,
104,106a,109
while, based on the results of ENDOR
85c
and Mossbauer
74
studies, the iron was
proposed as the binding site. Furthermore, it was proposed that cyanide
may bind to multiple sites.
94
Furthermore,
different binding modes, bent
71f
or linear,
71g
are suggested according to different crystal
structures (Figure 3E,G). In the bent binding
mode, there is still a water molecule bound to the pendant Fe, while
there is no pendant Fe-bound water in the linear cyanide binding mode.
A rearrangement is suggested to occur upon the rapid reversible binding
of cyanide to yield a more stable cyanide adduct represented by the
linear binding mode.
71g,94
ENDOR and Mossbauer results,
previously interpreted as an evidence for cyanide binding to the pendant
Fe, most likely represent a change on the water binding/leaving due
to the linear binding mode of cyanide.
Sulfide (S2-, HS–, or H2S) has been proposed to
act both as inhibitor
87,88
and as activator,
104,110
and its existence and role as
a bridging ligand between Ni and the pendant Fe in the C-cluster have
been controversial (as mentioned in the previous section). Sulfide
inhibits CO oxidation, but not CO2 reduction, as expected
given that there were no significant changes in the EPR spectrum upon
its addition to CODH in its Cred2 state.
87,88,104
Furthermore, Wang et al. showed that sulfide
binds the inactive Cox state of the C-cluster inhibiting
catalytic activity in the −50 and −250 mV potential
range.
88
2.3
Catalytic
Mechanism of CO Oxidation and CO2 Reduction
2.3.1
Metal-Based Catalysis of the Water–Gas
Shift Reaction
The proposed CO/CO2 conversion
mechanism discussed here is analogous to the water–gas shift
reaction described in Scheme 1.
In both
reaction mechanisms, CO and hydroxide ion are bound to two different
metal centers that should be positioned in a proper geometry during
the catalysis to allow the hydroxide to attack the M–CO intermediate,
resulting in the formation of M–COOH. Release of the CO2 from the metal complex is
coupled to a hydride shift, leaving
a metal hydride that undergoes protonation to generate H2.
Scheme 1
Mechanism of the Water–Gas Shift Reaction
Scheme 2
Proposed Catalytic Mechanism of Reversible
Carbon Monoxide Dehydrogenase
The most well-characterized
ferredoxin (Fd) from M. thermoacetica and many other organisms contains two [Fe4S4]
clusters and thus can accept two electrons. For a Fd containing
a single cluster, two Fd would be required.
2.3.2
Enzymatic Mechanism of CODH
Besides
the metal binding and positioning effects of the WGS catalysts, CODH
is able to increase the reaction rate by optimizing the ligand binding
geometry, controlling the acid–base reactions in and around
the active site, enhancing substrate and product transport, and using
the metal clusters as a wire to achieve a very fast electron transfer
to the corresponding electron acceptors.
71b,111
In the description below, all residue numbers refer to the CODH
Ch
II. Oxidation of CO in the C-cluster occurs
by a ping-pong reaction as shown in Scheme 2. In the first half reaction, the Cred1
state of the C-cluster
binds and undergoes reduction by CO and then transfers electrons from
the reduced C-cluster (Cred2) through the B- and D-clusters
in the enzyme. However, we should point out that this electron transfer
role for D-cluster has not been established. Furthermore, the D-cluster
is not reducible at potentials as low as −530 mV, indicating
that it may serve a structural, instead of an electron-transfer role.
72
In the second half-reaction, electrons are transferred
to the external redox partners, e.g., ferredoxin. The midpoint reduction
potential of the Cox/Cred1 redox couple is −200
mV, while it was reported as −530 mV for the Cred1/Cred2 redox couple. Cred1/Cred2
redox couple reduction potential matches well for the CO/CO2 redox potential.
Similar to the water–gas shift
reaction, the first catalytic step is the binding of CO and water
to the metal centers (Scheme 2). On the basis
of the results of ENDOR spectroscopic
85c
and X-ray crystallographic
71a,71b
studies, the catalytic
water (hydroxide) molecule binds to the pendant Fe site of the C-cluster
and also associates through H-bonding interactions with Lys563, His93,
and His263 (Figure 4). These residues are proposed
to participate in acid–base reactions, including formation
of active Fe(II)–hydroxide.
71a,71b
Site-directed
substitutions of Lys563 and His113 abolish enzymatic activity, confirming
the importance of these residues in catalysis.
112
A histidine tunnel composed of histidine residues located
on sequential turns of a helix starting near the C-cluster and ending
at the protein surface is proposed to facilitate transfer of protons
during the reaction (Figure 4).
71a,112
Steady-state kinetic studies conducted using NMR spectroscopy support
the presence of a rich proton reservoir inside the enzyme.
94
Figure 4
Structure of the C-cluster from CODH
Ch
II at 600 mV including only one coordinating residue: histidine
and the ligands proposed to be important in catalytic activities.
Atom colors: Dark gray (iron), orange (sulfide), red (oxygen), blue
(nitrogen), white (carbon), dark green (nickel). Unbound red spheres
represent the water molecules. Generated using Pymol from PDB 3B51.
CO binds to the Cred1 state of the C-cluster
with a
diffusion-controlled rate constant greater than 2 × 108 M–1 s–1 (a value that is 10-fold
faster than k
cat/K
m) according to rapid freeze quench EPR,
61
NMR, and steady-state kinetic studies.
94
However, the rate of reduction of the B-cluster (3000 s–1)
61
is only slightly higher
than the steady-state k
cat, indicating
that this step is partially rate-limiting in the CODH mechanism. On
the basis of NMR and steady-state kinetic studies, release of CO2 has also been proposed
to be partly rate-limiting.
94
Binding of CO to CODH/ACS is associated with
Fourier transform infrared (FTIR) bands at 1901, 1959, 1970, 2044,
and 2078 cm–1, assigned to the Ni–CO stretching
mode.
85a
The absence of any IR bands in
this region for the as-isolated CODH/ACS
Mt
suggests that the intrinsic Ni–CO ligand seen in hydrogenases
113
is not present in CODH. Extended X-ray absorption
fine structure (EXAFS) spectroscopy reveals the presence of Ni2+ in the as-isolated
Cred1 state of CODH
Ch
. Treatment of the enzyme with CO or Ti3+ changed the Ni K-edge shape slightly but
does not shift the edge
significantly. In both cases, the average Ni–S distance increases
to 2.25 Å, making the Ni site more tetrahedral. Similarly, significant
changes in the EXAFS analysis upon CO treatment suggest a structural
rearrangement in the C-cluster, but without any changes in the Ni
oxidation state. The only crystal structure that depicts a Ni–CO
complex in a CODH is that of the CODH (CODH
Mb
) portion of M. barkeri ACDS
(Figure 3K), which, like the other CODHs, shows
a water ligand bound to the pendant Fe.
71i
CO is bound to the Ni in a bent fashion, with an angle of 103°,
which could contribute to the high turnover numbers by destabilizing
the ground state of the Ni–CO intermediate. The crystal structure
of the complex between cyanide and CODH/ACS
Mt
reveals a similarly bent Ni–CN bond (Figure 3G),
71f
supporting a bent
Ni–CO bond with the substrate. It was proposed that a conserved
isoleucine residue very close to the bound-CO could sterically block
the linear binding of the CO.
71f
It should
be pointed out that an independent scrutiny of the crystallographic
data, including a recalculation of the electron density, did not find
evidence for the CO-ligand in the CODH
Mb
structure and for the CN ligand in CODH
Mt
structure.
114
In another structure of
the CN complex, in this case with CODH
Ch
, the Ni–CN is linear (Figure 3E).
A computational study indicated that Ile567 (Figure 4) plays a steric role and that
Lys563 and the histidine residues
are involved in acid–base chemistry during CO oxidation.
111
In the second step of the catalytic cycle,
the Fe-bound hydroxide
attacks the Ni–CO. FTIR studies support the formation of a
metal carboxylate.
85a
On the basis of the
crystal structure of a bicarbonate-soaked CODH
Ch
II crystal, the Ni and Fe subcomponents of the C-cluster are
bridged by a carboxylate, indicating that this could be a catalytic
intermediate formed by attack of the hydroxide to the Ni–CO
(Figure 3F).
86
Superimposition
of the C-clusters of CO-bound CODH
Mb
with
CO2-bound CODH
Ch
II suggests
a significant shift in the carbon atom’s position, which is
proposed to change the nickel coordination from tetrahedral to square
planar in the CO2-bound form.
The third step includes
the generation and release of CO2 and a proton, and the
reduction of the C-cluster from Cred1 to the Cred2 state, which thus should be two
electrons
more reduced than Cred1. While reduction of Cred1 to Cred2 upon reaction with CO is
very fast (>2 ×
108 M–1 s–1),
61,84
release of CO2 is proposed to be slow on the basis of
NMR and steady-state kinetic studies.
94
Note that in the WGS reaction (above), this step involves a hydride
migration, leaving the metal center in the same redox state. For several
reasons, including the similarity of the EPR signals of Cred1 and Cred2, it was proposed
that a metal hydride is also
formed during this part of the CODH reaction cycle.
114
A related proposal is that two-electron reduction of the
C-cluster generates a Ni0 state.
115
Because Ni(0) would be a diamagnetic species in a spin system with
most of the electron density in the Fe–S cluster, formation
of this low-valent Ni state would also be consistent with the minimal
EPR spectral differences between the Cred1 and Cred2 states.
In the fourth step, the C-cluster returns to its resting
Cred1 state upon transfer of two electrons to the B- and
D-clusters.
The distance between the metal clusters is approximately 11 Å
(Figure 2B), making it a good electron transfer
route.
71b,116
Rapid kinetic studies show that, at
high (>K
m
) CO concentrations,
internal electron transfer
(from the C-cluster to B- and D-clusters) can be rate-limiting during
the first half-reaction;
84
however, the
final step (the pong stage) of the mechanism appears to be rate-limiting
during steady-state turnover.
61,84
Step 5 involves electron
transfer to the final electron acceptor. CODH interfaces with many
electron carriers that support different specific activities,
29,117
including small redox proteins (ferredoxin, flavodoxin, rubredoxin);
cofactors [FAD and FMN, but not NAD(P)]; redox enzymes (couple directly
to CODH), like pyruvate:ferredoxin oxidoreductase (PFOR); hydrogenase;
and artificial electron acceptors, like bipyridyl (viologen) dyes
and methylene blue.
52,82
2.3.3
CO
and Water Channels
Given that
the CODH active site is buried deeply inside the protein and the catalysis
rates are very high, there must be highly efficient routes to achieve
optimal substrate and product flow. A very long hydrophobic channel
starting from the surface of the protein directing above the apical
coordination site of nickel in the C-cluster was proposed to be the
substrate channel, while another channel starting approximately at
the end of the proposed substrate channel and ending at the enzyme
surface near the B- and D-clusters was also proposed to be the water
channel.
71b
Although the recently published
crystal structures support the presence of the channels, experimental
support for these channels in monofunctional Ni–CODHs has been
lacking. A recent X-ray crystallographic study of the interaction
of CODH
Ch
II with the inhibitor and slow
substrate n-BIC revealed the presence of two different
channels: one similar to the substrate channel found in the CODH component
of CODH/ACS and another substrate channel unique for monofunctional
Ni–CODHs.
71h
This unique channel
is blocked by several residues in bifunctional Ni–CODHs, most
likely to avoid the escape of the substrates. Molecular dynamics and
density functional theory computations have provided evidence for
a dynamically formed gas channel in CODH/ACS for diffusion of CO2 from solvent to
the C-cluster.
118
Two cavities that are not apparent in the X-ray structures and are
transiently created by protein fluctuations are proposed to form this
channel.
2.4
Inorganic Modeling for
CODH
2.4.1
Structural Models for the C-Cluster
Spectroscopic studies had initially been interpreted to exclude the
possibility of Ni being within a cube.
119
Thus, the first publication of the crystal structure of CODH was
surprising to the bioinorganic chemistry community, because it revealed
the C-cluster to contain a NiFe3S4 cubane cluster
bridged to another iron.
71a,71b
This heterometallic
cluster has proven to be one of the most difficult metal centers to
model. Holm and co-workers successfully prepared the first [NiFe3S4] cubane model
complex 2 (Scheme 3) by reacting 1 with Ni(PPh3)4.
120
Changing the Ni ligand
resulted in the synthesis of many different complexes; for example,
with Ni(SEt)4, 3 is obtained. Manipulation
of the iron ligands by tailoring the starting linear ferric cluster 1 led to novel
NiFe3S4 clusters, e.g., 4.
121
Another modeling approach
began by preparing cuboidal Fe3S4 clusters,
122
5, and incorporating different
metal ions into this center, to generate a series of [MXFe3LS3] [where LS3 is 1,3,5-tris((4,6-dimethyl-3-mercaptophenyl)thio)-2,4,6-tris(p-tolythio)benzene(3−)]
complexes.
123
In these model complexes, the iron atoms are bound to LS3 ligands, making them structurally
analogous to the C-cluster.
Several [NiFe3S4] complexes, including a square
planar species, have been reported, e.g., 6.
124,124b
However, among these synthetic structural models, none has yet been
reported to be active in catalyzing the interconversion of CO and
CO2. Furthermore, no NiFe3S4 complex
bridged to a pendant Fe like that of the C-cluster has yet been reported.
Scheme 3
Schematic Views of Model Complexes Mimicking the C-Cluster
2.4.2
Functional
Models for CODH
As described
in a recent US Department of Energy (DOE) report,
125
“The major obstacle preventing efficient conversion
of carbon dioxide into energy-bearing products is the lack of catalysts...”;
thus, the development of effective catalysts for the activation, reduction,
and conversion of CO2, an abundant greenhouse gas, to fuels
and chemicals would have enormous economic and environmental impact.
As described in the introduction, CO2 reduction is difficult
because of both thermodynamic (the low redox potential required) and
kinetic (the chemical inertness of CO2) issues. The largest
barrier that the model complexes have to overcome is the very high
activation energy of the one-electron reduction of CO2 to
the radical anion (see the Introduction, the electrochemistry section below, and a
recent review
8
for details). Two detailed reviews covering catalytic
CODH models are available.
8,126
Thus, here we will
briefly describe important conclusions from the catalytic modeling
efforts and how they relate to the enzymology of CODH, as well as
suggest how principles uncovered from studies of the enzyme might
inform the next generation of CO2 reduction (or CO oxidation)
catalysts.
Initial efforts to accomplish CO2 reduction
included the synthesis of Co+ and Ni+ compounds
of cyclam and its variants.
127,128
These studies showed
the importance of the metal reduction potential, solvent effects,
and intermolecular and intramolecular hydrogen bonding on CO2 binding affinity and
kinetics.
129,130
In the enzyme,
these factors are optimized to promote proper H-bonding, salt bridge
and hydrophobic interactions among residues in the overall protein
structure, and appropriate geometries and distances for metals and
ligands at the active site, as well as in the secondary coordination
sphere.
Palladium phosphine complexes have been designed to
be highly active
molecular catalysts of CO2 reduction to CO.
131−133
In these complexes, Pd2+ is coordinated by three phosphorus
atoms, RP(CH2CH2PR′2)2, where R and R′ can be alkyl or aryl groups, and a
solvent molecule. According to the proposed catalytic cycle,
126
the initial step includes reduction of the
metal center from (2+) to (1+) oxidation state, as shown in Scheme 4. Then, in the
rate determining step (at low pH
values), CO2 binds to the Pd+ to form a metal
carboxylate at a rate that depends on the reduction potential of the
metal center, with rates increasing as the potential decreases.
134
Similar initial steps are observed in Fe, Co,
and Ni catalysts that require very negative potentials for one-electron
reduction; however, they exhibit different rate-determining steps.
135−137
The next step is the protonation of the metal carboxylate, which
promotes C–O bond cleavage and presumably is the origin of
the increase in rate of CO2 reduction as the acidity of
the reaction mixture increases.
138
Then,
solvent (a coordinating organic molecule, e.g., dimethylformamide)
dissociates from the metal center upon another 1e– reduction of the CO2H-bound complex,
leaving a vacant
site on the metal.
131
Protonation of this
complex forms LPd–COOH2 followed by C–O bond
cleavage and separation of CO and H2O on the metal center.
At low acid concentrations, the C–O bond-cleavage step becomes
rate-determining.
126
In the last step,
water and carbon monoxide are released from the complex and solvent
coordinates again to the metal. Dissociation of the M–CO bond
is very fast, since the CO affinity of Pd2+ is very low.
131,133
In order to increase the CO2 affinity of the Pd
catalysts
(and unwittingly generate an intermediate(s) like that observed in
CODH), the bimetallic compound 7 (Scheme 4) was prepared.
139
While one Pd
binds the carbon atom of CO2, the other acts as a general
base to bind the oxygen. This complex exhibits CO2 reduction
activity as high as 104 M–1 s–1; however, it becomes inactivated after several turnovers,
most likely
due to Pd–Pd bond formation. We surmise that formation of a
Ni–Fe bond would also be inhibitory to the enzymatic reaction
and that this is prevented in Ni–CODHs due to the different
reduction potentials of the metal centers. The general bimetallic
theme is not necessarily conducive to catalysis in that compound 8, which has a Ni–Fe
bimetallic model like CODH, has
no CO2 reduction activity.
140
As a result, there is still a need to prepare and explore metallic
catalysts to efficiently and economically reduce CO2.
Scheme 4
Schematic View of the Proposed Intermediates in CO2 Reduction
on Palladium Catalyst
2.5
Electrochemical and Environmental
Application
Efforts
Given an abundant source of CO2, an important
aim for technology would be to achieve rapid and efficient CO2 reduction to any of
its reduction products using energy provided
by electricity or solar sources. Electrochemical considerations are
important in each case; a reversible electrocatalyst operates close
to the reversible potential and is therefore by definition the most
efficient, and efficiency is important given the cost of electricity
and the need to exploit the visible region of the solar spectrum.
The first 2 equiv stage of CO2 reduction, namely, its conversion
to CO or formate, formally takes us into organic chemistry, but this
stage is the most demanding in terms of electrochemical potential.
There are numerous efforts to find suitable catalysts for CO2 reduction that are based
on first-row transition metals; so far,
the most successful electrocatalyst is Cu, although a sizable overpotential
is required to drive conversions to several products. Other catalysts
include polymeric Ru carbonyl complexes, compounds based on other
transition elements, and even pyridinium ions, but they all fall far
short of the performance observed in electrochemical studies of CODH.
Protein film electrochemistry (PFE) refers to a suite of electrochemical
techniques used to study an enzyme that is attached tightly to a suitable
electrode surface, usually by simple physical adsorption.
141
The electrode is rotated at various speeds
in an enclosed cell containing a small volume of buffered electrolyte
and connected to a gas supply that goes through the headspace and
equilibrates with solution. Reagents can also be injected into the
solution through a septum. Many redox enzymes have now been investigated
by PFE, revealing detailed information on their catalytic activity
in both oxidizing and reducing directions, as a direct function of
electrode potential (E). The primary observable is
the catalytic current (i), negative or positive for
net reduction or net oxidation, respectively, which is directly proportional
to the turnover rate at the particular electrode potential that is
applied by the instrument. Specifically, the current i
(E) observed at a particular potential
is related to net turnover frequency k
cat(E) at that particular potential by i
(E) = k
cat(E). nFAΓ, where n is the number of electrons involved (2 for CODH), F is the Faraday
constant, A is the electrode area,
and Γ is the electroactive coverage of enzyme. The latter is
usually <1 pmol cm–2, too low to observe any
signals due to electrons entering or leaving the enzyme when substrate
is not present to amplify the current. For a very active enzyme, the
catalytic current may be large and provide an excellent handle with
which to measure the extent and rates of reaction with inhibitors
as these are added or removed. Use of PFE has provided new insight
into why redox enzymes are so efficient, because not only does an
investigator measure rates, but also the energy (strictly speaking
the potential) that is required to achieve a particular rate. In principle,
PFE is a new way of studying enzyme kinetics, except that the enzyme
is probed using a potential in the same way as we would examine an
electronic device to obtain its iE characteristics.
Unlike potentiometry, which examines states of active sites poised
at redox equilibrium, PFE examines the steady-state flow of electrons
in a particular direction, and measurements may be made at any potential,
often well outside the boundaries imposed by redox mediators.
The basic technique is cyclic voltammetry, in which the electrode
potential is scanned linearly, back and forth, between two limits.
Cyclic voltammetry, long used to characterize the reduction potentials
and stabilities of small molecules in solution, has become a powerful
method for studying the catalytic electron-transport properties of
enzymes. Overlay of catalytic currents in each scan direction means
that the catalytic activity is constantly a simple function of electrode
potential; conversely, and assuming that the enzyme is stable on the
electrode, hysteresis means that a change in catalytic activity occurs
on a time scale that is slow compared to the scan rate. The information
obtained by cyclic voltammetry provides the broader picture of what
an enzyme can do over a wide potential range, paving the way for more
specialized investigations, including bulk solution spectroscopic
experiments designed to isolate states prevailing at a particular
potential. An important electrochemical technique at this stage is
controlled potential chronoamperometry, in which a reaction is initiated
by a step in potential or injection of a reagent and monitored as
a current–time plot; this technique is used to obtain rates
of interconversions between different states of the enzyme.
An early (2007) study of CODH
Ch
I by
PFE showed very easily how this enzyme operates under different mixtures
of CO2 and CO.
45a
The voltammograms
in Figure 5 reveal the intense electrocatalytic
activity of CODH
Ch
I adsorbed on a pyrolytic
graphite “edge” (PGE) electrode rotating at high speed
in an anaerobic sealed cell. Panels a and b reveal the separate reduction
and oxidation activities under 100% CO2 or 100% CO, while
panels c and d show combined reduction and oxidation
activities for a 1:1 CO2/CO gas mixture at two different
pH values. The cyclic voltammograms recorded in the presence of both
CO2 and CO immediately give us some idea of the catalytic
bias of the enzyme, as explained later: they show how the current
cuts cleanly through the potential axis at the values expected for
the equilibrium potential of the mixture. In other words, electrocatalysis
occurs close to the reversible limit with only a minuscule overpotential
required to shift the reaction from one direction to the other. Temperatures
are also shown and allow us immediate insight into activation energies
in each direction. Such a clear example of reversible electrocatalysis, otherwise
observed only in a few cases—notably
H2 on platinum—is emerging to be a distinctive feature
of enzymes such as hydrogenases, CODH, and several other enzymes.
Size, it appears, is no barrier to being the best electrocatalysts
so far investigated.
Figure 5
Protein film voltammograms showing CO2 reduction
and
CO oxidation activities of CODH
Ch
I adsorbed
on a PGE electrode under atmospheres of 100% CO2, 100%
CO, or 1:1 CO2/CO gas mixtures. Scan rate was 10 mV/s in
parts a, c, and d and 30 mV/s in part b. Electrode rotation 4000 rpm.
Reprinted with permission from ref (45a). Copyright 2007 American Chemical Society.
The K
M values for reaction in each
direction were investigated by monitoring the time course of current
decrease after injecting small aliquots of solution containing CO
or CO2 into the cell, under a continuous flow of inert
gas, while the rotating electrode is held at a fixed potential, i.e.,
−0.4 V for CO oxidation or −0.6 V for CO2 reduction. The principle of this method
is that the gas concentration
decreases exponentially but a drop in current is not observed until
the concentration of gas remaining approaches that of the respective K
M. At 25 °C, pH 6.0, the K
M
CO value (13 experiments) was estimated at
ca. 0.002 atm (ca. 2 μM, using the relevant Henry’s constant),
but only a lower limit (ca. 0.06 atm) could be determined for K
M
CO2
. Although 25 °C
is well below normal growth temperatures for C. hydrogenoformans, the low K
M
CO value reflects
very well the enzyme’s ability to scavenge low-level CO. The
very high K
M
CO2
value
means that in Figure 5, CO oxidation activity
(panel b) is saturated, whereas CO2 reduction (panel a)
is not, and at pH values below 6, extrapolated values of k
cat for CO2 reduction must be higher than for
CO oxidation; however, this comparison may be of academic rather than
physiological interest. What is important here is that CO can be scavenged
from dilute sources at potentials close to the reversible value of
the CO2/CO couple. The catalytic bias of an enzyme, in
this case the efficiency with which CO is oxidized relative to the
efficiency with which CO2 is reduced, is discussed further
later in this section.
Armstrong and Hirst have discussed the
factors that are important
for efficient electrocatalysis by enzymes.
142
The important considerations are (1) efficient long-range electron
transfer (in accordance with Marcus theory, reorganization energies
for electron-transfer sites are small); (2) ensuring that electrons
leave or enter the catalytic cycle at a potential close to that of
the redox reaction being catalyzed; (3) ideally, concerted proton–electron
transfers at the active site, avoiding charge separation; and (4)
the ability to provide all the electrons needed to convert reactants
to products in a single stage or stabilize intermediates sufficiently
to a free energy value level with that for the single multistep reaction.
These factors seem to be satisfied well for CODH: first, the Fe4S4 clusters (D- and
B-clusters) of the electron
relay are optimized for low reorganization energy with potential values
quite close to the CO2/CO couple (see later); second, the
active site has evolved to bind CO or CO2 (depending on
oxidation state) with little reorganization, undergo concerted proton–electron
transfer, and stabilize a bound intermediate. All these properties,
due to exquisite positioning of the supramolecular atomic framework
around each site, appear to have been refined through evolution. Hexter
and co-workers have formulated a basic model for the catalytic bias
of the turnover frequency of an enzyme attached to an electrode surface.
143
This model is simplified by restricting the
enzyme to have a single active form (i.e., neglecting resting states)
and deals only with the limiting k
cat that
should be obtained under substrate-saturated conditions and does not
deal with substrate binding affinity. Regarding the issue of catalytic
bias, the analysis based on this model asks the question “how
fast can enzyme catalysis run in one direction relative to the other,
when the substrate is saturating and the electrode potential in each
case is set so as to provide an appropriate thermodynamic driving
force?” The answer, according to the model, is that the catalytic
bias to operate preferentially in one direction or the other is related
to the difference between the equilibrium potential for the substrate
reaction being catalyzed and the reduction potential at which electrons
enter or leave the catalytic cycle, the latter being the potential-determining
step and associated with a component of the enzyme termed the “electrochemical
control center”. For CODH
Ch
I,
the electrons enter or leave the enzyme via the D-cluster, and the
fact that CODH
Ch
I is a good CO2 reducer owes much to the D-cluster having a very negative reduction
potential. Further insight into the catalytic bias is provided when
we consider, in addition, how tightly the different states of the
enzyme bind CO or CO2, and this aspect is discussed later.
Returning to the cyclic voltammetry of CODH
Ch
I, as the electrode potential is scanned to more oxidizing
potentials, the current trace reveals hysteresis that is due to slow
oxidative formation of an inactive state followed, upon the return
scan, by a relatively rapid reductive reactivation. The immediate,
simplistic interpretation of these results is that the oxidized inactive
state Cox is being formed at the electrode. From time to
time, some samples of enzyme show two reactivation processes, one
at a much lower potential than the other. The “sample history”
dependence of the observation of a second species reactivating at
a lower potential is discussed later.
The lower reductive current
obtained in the presence of CO, compared
to when it is absent, shows that conversion of CO2 by CODH
Ch
I is subject to strong product inhibition.
45b
With PFE, activity measurements can be made
at much more negative potentials than can easily be applied with chemical
electron donors. Under more reducing conditions, i.e., below −700
mV, CO becomes much less effective as an inhibitor, as established
by Lineweaver–Burk measurements of K
i as a function of potential, providing a clue that a more reduced
state of the active site is unable to bind CO. Further experiments,
shown in Figure 6A, showed that both oxidation
of CO and reduction of CO2 are strongly inhibited by CN–; however, below −600 mV,
the current due to
CO2 reduction increases strongly as CN– is released from the more reduced active site
that now prevails.
All these observations are explained in terms of CO and isoelectronic/isostructural
CN– targeting and stabilizing the state Cred1. The PFE technique clearly shows that
CN– ceases
to be an inhibitor when the active site is in the Cred2 state.
Figure 6
Voltammograms showing, for CODH
Ch
I,
(A) the potential dependence of inhibition of CO2 reduction
activity upon injection of cyanide (CO oxidation is completely inhibited),
pH 7.0, scan rate 1 mV s–1 and (B) inhibition of
CO2 reduction activity and shift in potential for CO oxidation
upon addition of cyanate, pH 7.0, scan rate 1 mV s–1. Adapted with permission from
ref (45b). Copyright 2013 American Chemical Society.
Reduction of CO2 is
inhibited by cyanate (NCO–), which is isoelectronic
and isostructural with CO2,
and PFE reveals an interesting effect on CO oxidation, in that a small
overpotential is required to achieve conversion (Figure 6B). The result is explained
by NCO– binding
preferentially to Cred2 and stabilizing this state, so
the catalytic current commences only after the potential favors Cred1 and causes release
of the inhibitor.
The inhibitors
CN– and NCO– are thus complementary:
each targets a different redox state of
the active site, as analogs of either CO or CO2. In terms
of kinetics, however, the inhibitors behave differently to the natural
substrates, as binding and release of CN– and NCO– from the active site are orders
of magnitude slower
than the turnover rates observed for CO and CO2. These
slow rates could stem from requirements for protonation and deprotonation
(HCN/CN–), differences in charge (NCO– is the conjugate base of a strong acid), and
ease of rehybridization
upon binding or release. The studies carried out originally with CODH
Ch
I have been repeated with CODH
Ch
II, showing that similar (but not identical behavior)
is observed with the crystallographically characterized isozyme.
144
For example, CO2 reduction by CODH
Ch
II is more strongly inhibited by CO than
CODH
Ch
I, a property that may have physiological
relevance. Studies of the potential dependence of inactivation and
reactivation rates for CN– show clearly that reactivation
(ligand OFF) becomes significantly faster under the more reducing
conditions that would favor Cred2, whereas inactivation
(ligand ON) rates do not depend so much on potential. Comparative
numerical data on binding affinities and rates are shown in Tables 3 and 4.
Table 3
K
M and K
i Constants for CODH
Ch
I and CODH
Ch
II at 25 °C,
pH 7.0, Unless Stated Otherwisea
–209 mV
–560 mV
–760 mV
K
M (CO)
K
M (CO2) (mM)
K
i (CO) (μM)
K
M (CO2) (mM)
K
i (CO) (μM)
CODH
Ch
Ib
2 ± 1
8.1 ± 2.1
46
7.1 ± 0.7
337
CODH
Ch
II
8.0 ± 1.6
5
6.0 ± 1.0
85
a
Data cited
from ref (144) unless
otherwise stated.
b
Data taken
from ref (45a), pH
6.0.
Sulfide is an unusual
inhibitor, as it has no effect until the
electrode potential is raised hundreds of millivolts above the reversible
CO2/CO potential–, an observation which
shows that sulfide (entering as HS– or H2S) does not directly target an active state
of CODH but promotes
oxidative inactivation at a higher potential.
45b
The fact that the reactivation potential is much lower
than that observed without sulfide shows immediately that a different
Cox state is being formed, one that probably has a sulfide
entity attached (in place of water or hydroxide). A similar reactivation
process has been observed in samples of CODH
Ch
I that have not been deliberately exposed to sulfide during
the experiment, suggesting that those samples already contained a
sulfide entity. Sulfide binding at the C-cluster and its role in activity
has been a controversial issue over many years of studying CODH; the
PFE results for CODH
Ch
I and II now show
that sulfide is associated with oxidized forms of the enzyme and may
be retained in the active site unless quite reducing conditions are
applied.
Table 4
Comparison of Half-Times for Inactivation
by Cyanide– (0.5 mM) and Reactivation for CODH
Ch
I and CODH
Ch
II
CODH
Ch
I
CODH
Ch
II
potential
mV vs SHE
ON t
inact(1/2), s–1
OFF t
react(1/2), s–1
ON t
inact(1/2), s–1
OFF t
react(1/2), s–1
+140
83 ± 15
130 ± 25
–460
95 ± 15
307 ± 75
–560
95 ± 15
still inhibited
161 ± 16
still inhibited
–660
73 ± 15
143 ± 10
–760
64 ± 10
19 ± 7
54 ± 3
≪limit of detection
The fact that the binding
abilities of different inhibitors and
substrates depend strongly on potential demonstrates a further aspect
of catalytic bias that was not implicit in the basic model, which
dealt only with substrate-saturated conditions. The potential used
to drive the reaction in one particular direction also controls the
redox state of the active site prevailing during the catalytic cycle
and, hence, its ability to bind a particular agent. The range of potentials
over which different substrates and inhibitors CO, CN–, NCO–, and HS– target CODH
Ch
I is shown in Figure 7. A scheme outlining these conclusions is shown in Scheme 5.
Similar results have been obtained for CODH
Ch
II, suggesting that these features may
be characteristic properties of the C-cluster.
Figure 7
Potential dependence
of binding of inhibitors to CODH
Ch
I.
Red refers to the potential region over which
the enzyme is inhibited, gray indicates no binding, and green indicates
that binding leads to turnover. The dashed arrows indicate reactions
that are slow compared to those indicated by full arrows. Reprinted
with permission from ref (45b). Copyright 2013 American Chemical Society.
Scheme 5
Summary of the Interceptions of the Catalytic Cycle
of CODH
Ch
I by Small Molecule Inhibitors,
As Deduced from
PFE Experiments
The potentials −250
and −50 mV are the values observed for reactivation of enzyme
with and without sulfide. The potential −520 mV is the standard
potential for the CO2/CO half-cell reaction at pH 7.0.
Reprinted with permission from ref (45b). Copyright 2013 American Chemical Society.
The catalytic bias of CODH
Ch
I and
CODH
Ch
II may now be articulated as follows:
First, if binding affinity is ignored, i.e., assuming conditions in
which CO or CO2 levels comfortably exceed their respective
Michaelis constants, both CODH
Ch
I and
CODH
Ch
II are excellent catalysts of CO2 reduction, a factor that seems to relate to the
favorable
negative potential at which we suggest that electrons would enter
the enzyme via the D-cluster. If electrons could enter only at a much
higher potential (in effect a de-energization), CO2 reduction
would not occur. Second, if substrate binding is not ignored, the
fact that CO2 binding is weak and occurs only at the strongly
reducing state Cred2 means that the physiological bias
should lie against CO2 reduction and in favor of CO oxidation;
indeed, these enzymes can easily scavenge trace CO. The tighter binding
of CO to CODH
Ch
II compared to CODH
Ch
I certainly suggests that CODH
Ch
II should be the better CO scavenger.
The
highly active and reversible nature of CODH has stimulated
some unusual electrochemical experiments without an electrode, experiments
that demonstrate interesting benchmarks for technology. In one set
of investigations, molecules of CODH
Ch
I were coattached with a [NiFe]-hydrogenase (Hyd-2 from E. coli) to the surface of
graphite particles (platelets
formed by grinding pyrolytic graphite with a coarse abrasive) to make
a catalyst for the WGS reaction.
51
Aspects
of this experiment are shown in Figure 8.
Figure 8
(A) Cartoon
representation of an enzymatic device for catalysis
of the water–gas shift reaction. Electrons released by CODH-catalyzed
CO oxidation are transferred through a graphite particle to a CO-tolerant
hydrogenase that reduces protons to H2. (B) Typical cyclic
voltammograms (from separate experiments) showing the reversibility
of electrocatalysis by CODH
Ch
I and a
hydrogenase (Hyd-2) from E. coli, measured
at pH 6.0, 30 °C, scan rate 10 mV s–1, electrode
rotation rate 2500 rpm. (C) H2 production and CO depletion
over the course of 55 h at pH 6.0, 30 °C, as quantified by GC
analysis. Fresh aliquots of CO were introduced at the times indicated.
Adapted with permission from ref (51). Copyright 2009 American Chemical Society.
Panel A shows a scheme representing
the flow of electrons between
the two enzymes across the conducting particle. Panel B shows the
voltammograms for CODH
Ch
I and Hyd-2 enlarged
to focus on the regions where the catalytic current for each respective
system (50/50 CO/CO2 and 50% H2/pH 6) intersect
the potential axis. The 0.11 V difference in the two potential values
gives the thermodynamic driving force available for the WGS reaction,
and the fact that this difference is displayed so sharply is due to
the fact that these enzymes are reversible electrocatalysts. The particles
were then suspended in aqueous solution under an atmosphere of CO,
and the gas composition was measured at different time intervals by
gas chromatography. The graphite particle conducts electrons produced
from the oxidation of CO by CODH to Hyd-2, which converts H+ to H2. Details of the
experiment are shown in panel C,
which shows how CO depletion corresponds to the simultaneous formation
of H2. After CO is exhausted, recharging the vessel with
more CO restarts H2 formation.
On the basis of the
amount of enzyme attached to the particles
and the ambient temperatures used, the suspension gives a higher rate
of H2 production than industrial catalysts: for example,
a homogeneous catalyst Ru3(CO)12 is reported
as having a WGS reactivity of 0.01 (mol H2) s–1 (mol catalyst)−1 at 160 °C, and Au–CeO2
nanomaterials, regarded as being highly efficient heterogeneous
catalysts, show turnover frequencies up to 3.9 site–1 s–1 at 240 °C. The data correspond
to an
average H2 production rate of 2.5 (mol H2) s–1 (mol adsorbed Hyd-2)−1 and a CO
depletion rate of 0.07 (mol CO) s–1 (mol adsorbed
CODH I) s–1. The rates per enzyme molecule are lower
limits because it is assumed that all the adsorbed enzyme is electrocatalytically
active. The empirical turnover frequency is based on the less active
component, i.e, Hyd-2, and therefore, the particles display an equivalent
per “site” WGS turnover frequency of at least 2.5 s–1 at 30 °C. Importantly, significant
rates of
WGS conversion by this system are even detectable at ice temperature.
Finally, referring back to Figure 7, we note
that requirement for even a modest (0.1 V) overpotential for onset
of CODH or hydrogenase activity would result in no WGS activity being
observed.
CODH
Ch
I and II also have
been studied
to assess the possibilities for artificial photosynthetic CO2 reduction. The aim has
been to use the enzyme, with its superb efficiency,
to establish what should be possible using semiconducting materials
to harvest light and generate excited electrons, analogous to photosystem
I, the fuel-forming complex of natural photosynthesis.
145
One requirement of the semiconductor is that
the conduction band into which the electrons are injected has a potential E
CB that is sufficiently negative to reduce CO2 to CO. The principle is represented
in panel A of Figure 9.
Figure 9
Photoelectrocatalysis of CO2 reduction to CO
catalyzed
by CODH attached to light-harvesting nanoparticles. (A) The concept:
red arrows correspond to injection of electron into the conduction
band (potential E
CB) by a photosensitizer
(RuP) attached to the nanoparticle; green arrows correspond to injection
of electron into the conduction band by band gap excitation (potential
difference E
G) from the valence band (potential E
VB). The hole in either dye or valence band
must be filled more rapidly than the electron can return (the electron–hole
recombination rate). (B) Production of CO by visible light using a
photosensitizer. Experiments carried out by irradiating a vial containing
a 5 mL suspension of various semiconducting nanoparticles with visible
light (λ > 420 nm). In each case, 5 mg of nanoparticles (20
mg in the case of ZnO) was modified with CODH
Ch
I (total 2.56 nmol) and RuP (total 56 nmol). The buffer in
each experiment was 0.20 M MES, pH 6, 20 °C. (C) Production of
CO by visible light using direct band gap excitation of various types
of cadmium sulfide attached to CODH
Ch
I.
QD = quantum dot, NR = nanorod; calcined CdS was heated at 450 °C
for 45 min. The buffer in each experiment was 0.35 M MES, pH 6, at
20 °C. Adapted from refs (146a) (copyright 2011 The Royal Society of Chemistry)
and (147) (copyright
2012 The Royal Society of Chemistry) with permission.
The first of experiments used CODH
Ch
I attached to various nanoparticles for which the
natural band gaps E
G exceed the energy
available from visible light;
consequently, the nanoparticles were modified by coattachment of the
photosensitizing complex “RuP” = [Ru2+(bpy)2(4,4′-(PO3H2)2-bpy)]2+ (λmax 455 nm),
analogous to technology
introduced by Michael Grätzel for dye-sensitized photovoltaic
cells.
ref146
The relevant conduction band
potentials E
CB (measured for bulk materials)
are as follows: TiO2 (anatase), −0.52 V (note E
CB = 3.1 eV, hence the need to use UV irradiation
when RuP is not coattached); TiO2 (rutile), −0.32
V; ZnO, ca. −0.5 V; SrTiO3, −0.72 V. For
comparison the standard reduction potential for the CO2/CO couple at pH 6.0 is −0.46
V.
The results depicted
in panel B show that CO production by dye-sensitized
visible light excitation depends greatly on the nature of the metal
oxide semiconducting nanoparticles. Anatase is clearly supreme: the
nanoparticles known as P25 are a composite of anatase with some rutile
phase, although rutile itself is inactive (as expected, since E
CB is too positive to drive CO2 reduction)
and SrTiO3 is possibly inactive because E
CB is so negative that electrons easily transfer back
to RuP. The best rate, obtained with P25 and calculated on the basis
of total CODH
Ch
I used, equates to a CO
production rate of approximately 0.15 s–1 per molecule
of CODH.
146
This rate is much slower than
that achieved for a hydrogenase at the same material (50 s–1), a fact that is still
not resolved. One important difference between
the conventional electrochemical and photoexcitation experiments is
that, in the latter, electrons may recombine before being used by
the catalyst. The tentative conclusion is that a good photoelectrocatalyst
should be one that traps all the electrons required
to carry out the reaction and restricts their return to the semiconductor
and inevitable recombination.
Using semiconducting materials
with a smaller band gap, it is possible
to use visible light with the need for dye sensitization. Experiments
similar to those with RuP-modified anatase, but using band gap excitation,
were carried out using different types of nanoparticle formed from
cadmium sulfide, CdS. As a rough guide, for bulk CdS, E
g = 2.3 eV (corresponding to λ = 540 nm) and E
CB = −0.87 V. Using CdS nanoparticles
(nanorods, NR) or CdS quantum dots (QD), slightly higher rates were
achieved, 0.25 s–1 compared to the results obtained
with anatase (panel C).
147
The CdS quantum
dots have a typical radius that is half that of CODH; thus, in principle,
up to 10 QDs may bind to one CODH molecule, reversing the size ratio
indicated in panel A. Thermal calcination of CdS nanoparticles, which
results in irregular clusters of larger particle size, resulted in
no activity when CODH was attached.
The success of these artificial
photosynthesis experiments gives
strong encouragement for pursuing research in this area and for the
role that enzymes play in providing a reversible catalyst in which
many different properties can be modified by genetic engineering and
tested quantitatively by electrochemical methods.
3
Acetyl-CoA Synthase
3.1
Chemistry and Biochemistry
of C–C Bond-Forming
Reactions Involving CO2 and CO
Developing an industrial
process that efficiently couples CO2 reduction to CO with
a carbonylation reaction would be an important advance in the chemical
industry because carbon–carbon formation by reactions with
CO is instrumental in many industrial processes.
148
CODH/ACS catalyze such a coupled process as an important
component of the biological carbon cycle.
46
If fuels could be made from CO2, these C–C bond-forming
reactions will be of even more importance in energy generation.
Industrial processes involving carbonylation chemistry include the
Monsanto process, hydroformylation, and the Reppe process. As has
been pointed out elsewhere,
149
the intermediate
steps in the Monsanto process for acetic acid formation from methanol
and CO are nearly identical to those in the catalytic mechanism of
ACS, as described in the Introduction. Both
the biological and homogeneous catalysts use organometallic mechanisms
that feature low-valent metal centers [e.g., Rh(I) vs Ni(I)] to react
with CO and form a metal–carbonyl bond (M–CO) or with
a methyl donor and generate a methyl–metal bond (M–CH3). The key carbon–carbon bond-forming
reactions involve
a migratory insertion of the metal-bound CO and methyl groups to generate
an acyl–metal intermediate that undergoes reductive elimination
by a coordinated iodide in the chemical reaction or by the thiolate
of CoA in ACS to generate acetyl-CoA.
150
Acetyl-CoA then serves as a source of energy and cell carbon.
30
M–alkyl and M–CO are also key
intermediates in the hydroformylation reaction, to convert alkenes
to aldehydes. Similar organometallic intermediates are formed in the
Pd-based Reppe process.
31b
3.2
Characteristics of CODH/ACS
3.2.1
Enzymatic
Activity
The gene encoding
ACS (acsB) is a marker for the Wood–Ljungdahl
pathway, and whenever it occurs in a microbial genome, it is within
a gene cluster containing other pathway genes.
41
ACS associates tightly in a complex with CODH and utilizes
the product of the CODH reaction (CO) as its substrate in a kinetically
coupled reaction linked to generation of acetyl-CoA via eq 11.
61,71b,151
The second substrate of ACS is a methyl group donated by a methylated
B12 protein, the corrinoid iron–sulfur protein (CFeSP).
The third substrate is CoA, which reacts with CO and the Co-bound
methyl group to make acetyl-CoA, a cellular carbon and energy source.
As shown in Figure 1, ACS can catalyze this
reaction reversibly. Thus, in aceticlastic methanogens, it catalyzes
the disassembly of acetyl-CoA, breaking both the C–C and C–S
bonds to form CoA, the methylated CFeSP, and CO.
152
A convenient assay for ACS is to measure that rate of exchange
of 14C from [1-14C] acetyl-CoA with 12CO. For ACS (ACS
Ch
) and CODH/ACS (CODH/ACS
Ch
) from C. hydrogenoformans, exchange rates were reported to be 2.4 or 5.9 μmol of
CO
per min per mg, respectively, at 70 °C and pH 6 in the presence
of 3 mM Ti(III) citrate.
110
The exchange
rate reported for CODH/ACS
Mt
is 0.16 μmol
of CO per min per mg at 55 °C and pH 6, without addition of any
external reducing agent.
153
ACS
Ch
and CODH/ACS
Ch
also
catalyze acetyl-CoA synthesis from CFeSP, methylcobalamin, CoA, and
CO with activities of 0.14 and 0.91 U/mg μmol of acetyl-CoA
production per min per mg.
110
11
3.2.2
Active Site Metal Cluster
and the Importance
of Nickel in ACS
The active site of ACS, so-called the A-cluster,
was the first NiFeS cluster reported,
154
although the specific role of nickel in ACS activity was established
later.
155,156
In the A-cluster, a Fe4S4 cluster is bridged to a nickel, called the proximal nickel
(Nip) because of its proximity to the cluster, and also
thiolate-bridged to the distal nickel (Nid), which is coordinated
by two cysteine and two backbone amides as shown in Figures 10 and 11. The Nid, stabilized
due to its square planar geometry and oxidation state
(2+), is adjacent to a cavity that can accommodate the substrate and
products. Nip is coordinated by three S atoms in an apparent
T-shaped environment. Another ligand, which completes a distorted
square planar coordination, has been assigned as an oxygen ligand
donated by water
110
or an acetyl
39a
group, though, in the latter case, the structure
was of an enzyme containing Cu at the Nip site. The Nip is labile (i.e., easily replaced
by other metals) and is
thought to be the sole metal that is directly involved in binding
the substrates. Two different crystal structures showed copper or
zinc located at the Nip site (Figure 11),
39a,39b
and early studies indicated
a positive correlation between the copper content and ACS activity;
thus, copper was suggested to be a component of the active cluster.
157
However, studies over a much wider range of
Ni contents demonstrated that activity was positively correlated with
Ni and negatively related to the Cu content;
110,158
furthermore, copper was not responsible for, and even inhibited,
the activity of the enzyme.
159
The active
methanogenic enzyme was shown to contain two Ni per active center.
Thus, it is clear that the active A-cluster contains two Ni and four
Fe atoms. In almost all of the studies utilizing recombinant ACS,
the enzyme is activated by nickel reconstitution. CO binding to the
A-cluster upon the reduction by dithionite results in an EPR-active
species called NiFeC species, due to its hypefine broadening by 61Ni, 57Fe, and 13CO,
and is used to
determine the nickel incorporation into the A-cluster.
Figure 10
Structure
of CODH/ACS
Mt
. (A) Overall
structure of CODH/ACS. Green units in the center are the two CODH
homodimers; the left unit is the ACS in open conformation, and the
right unit is the ACS in closed conformation. Closer views of the
A-cluster pocket in (B) open conformation and (C) closed conformation.
Atom colors: Brown (iron), orange (sulfide), red (oxygen), blue (nitrogen),
light green (carbon), dark green (nickel), white (unassigned). Generated
using Pymol from PDB 1OAO.
Figure 11
Structure of A-cluster from PDB (A and
B) 1OAO, (C) 1MJG, and (D) 2Z8Y. Generated using
Pymol.
3.3
Structure
of the CODH/ACS
3.3.1
Inner Channel in CODH/ACS
The gene
encoding ACS is generally contiguous with that encoding CODH. This
genetic linkage parallels tight enzymatic coupling of CODH and ACS.
Kinetic coupling has been established by several experiments, including
one in which CO2 was used as a substrate and the incorporation
of in situ-formed CO into the carbonyl group of acetyl-CoA was monitored.
Unlabeled CO in solution does not decrease the rate or extent of incorporation
of labeled 14CO2 into acetyl-CoA.
160
Although CO is a substrate for the CODH/ACS,
absence of CO in the solution did not affect acetyl-CoA synthase activity,
while CO2 had a major impact on the reactivity.
63
Similarly, addition of hemoglobin or myoglobin
to the assay mixture as a CO scavenger only marginally inhibited acetyl-CoA
synthesis.
63,160
These and other results
63,160,161
suggest that CO produced in
the CODH subunit from CO2 remains sequestered within the
enzyme without equilibrating with solution as it is transferred to
the ACS active site, and it was proposed that CO migrates through
an inner channel within the CODH/ACS complex from the CODH to that
ACS active site.
63,160
The crystal structure
of CODH/ACS
Mt
showed that the A- and C-clusters
are separated by 67 Å, which would seem to be too long to allow
kinetic coupling of the CODH- and ACS-catalyzed reactions (Figure 10A).
39
However, interior
surface calculations and diffraction experiments on Xe-treated crystals
disclosed the presence of a continuous 140 Å long hydrophobic
tunnel that connects the active sites of CODH and ACS, the C- and
A-clusters, respectively (Figure 12).
39,71e
Since the van de Waals radius of Xe (2.16 Å) and CO (∼2
Å) are similar, Xe can be considered as a good mimic for CO.
A total of 19 Xe atoms were located in this hydrophobic tunnel. Examination
of the residues within 5 Å of Xe atoms shows an insignificant
degree of sequence homology but supports a highly conserved pattern
of hydrophobic residues (except for the positions and orientations
of C468 and T593, which are located near the C-cluster). The tunnel
is composed of a series of interconnected hydrophobic pockets that
can be conceptualized as a pinball plunger where launching of each
ball (gas molecule) from the trough into the playfield releases another
ball into the launching lane. Of course, with CODH/ACS, multiple balls
are at play in the channel and each has only one target, the A-cluster.
In each of the ACS subunits, one Xe atom was found 3.5 Å from
Nip (Figure 11D). Further experimental
support for a CO-binding pocket near the A-cluster is the finding
that, when CO-bound ACS is subjected to photolysis, the energy barrier
for recombination of Nip with CO is only 1 kJ/mol.
162
Figure 12
Structure of CODH/ACS
Mt
crystallized
in the presence of high pressures of Xe (PDB 2Z8Y) (shown as the blue
spheres) to reveal the hydrophobic CO tunnel. Adapted with permission
from ref (71e). Copyright
2008 American Chemical Society.
When residues (A578, L215, A219, A110, A222, A265) that are
located
within the hydrophobic channel in CODH/ACS
Mt
were substituted, ACS activity with CO2 as substrate
was severely diminished.
163
These results
support the importance of the tunnel for CO migration to the A-cluster.
Furthermore, the variants exhibit little inhibition of acetyl-CoA
synthesis by CO, in contrast to the wild-type proteins, indicating
that the channel plays an important role in cooperative inhibition
of A-cluster activity by CO. It was suggested that there may be at
least two ways for CO to reach the A-cluster: through the channel
and from the solvent. A water channel close to the ββ
interface is proposed to be the second way for the CO,
163a
but this idea is not well established yet.
The role of the CO channel is most likely to prevent the loss of
energetically expensive CO in the solution and to efficiently direct
this gaseous substrate to its site of reactivity at the A-cluster.
3.3.2
Conformational Changes
As shown
in Figure 10, ACS consists of three main domains.
The first domain, which interacts with CODH, starts with helices and
continues with a Rossman fold. This domain contains a ferredoxin interaction
domain.
164
The second domain includes six
Arg residues near Trp418 (Figure 10A). These
residues are involved in CoA binding according to fluorescence-quenching
studies of Trp418 and inhibition studies of CoA binding upon modification
of Arg residues.
165
The final domain bears
the A-cluster. This domain undergoes structural rearrangements during
turnover (Figure 10B,C).
ACS binds three
substrates of vastly different sizes: CO (30 Da), CoA (770 Da coenzyme),
and methylated CFeSP (88 kDa dimeric protein). CODH/ACS
Mt
is crystallized in two different forms that are
thought to be related to the catalysis: closed
39a
and open
110
conformations. Another
structure depicts both conformations (one in each CODH/ACS dimer)
(Figure 10).
39b
In
its closed conformation, the channel is open, allowing CO to pass
through the tunnel to the A-cluster; however, there is no apparent
access to the methylated CFeSP. In the open configuration, one of
the domains (domain 3) of ACS rotates, which partially exposes the
A-cluster, enabling interaction with the CFeSP and closure of the
CO tunnel.
Although the catalytic importance and the main trigger
of this
conformational change are not yet well established experimentally,
there appear to be at least four discrete conformations. Throughout
all of these conformational changes, both CO and the A-cluster must
be protected from exposure to solvent, because CO does not equilibrate
with solvent during catalysis.
160,63
In one closed conformation,
poised for binding CO, the CO channel is open to allow the CO to reach
the A-cluster, which is buried and unable to access the CFeSP (Figure 10C). In an
open conformation, ready to bind the
methyl group, the A-cluster is rotated to interface with the CFeSP
and the CO channel is blocked to avoid CO release (Figure 10B). Another closed (solvent-excluded)
conformation
is required during formation of the acetyl–metal complex to
avoid hydrolysis of the acetyl–metal center. Then, the A-cluster
must be rotated into a more open conformation to allow CoA binding,
thiolytic cleavage of the acetyl group, and acetyl-CoA release. A
crystal structure of the truncated ACS
Mt
is proposed to represent the CoA binding conformation of the enzyme.
166
While there is concrete crystallographic proof
for the first two conformations, more work is needed to reconcile
the other two conformations.
Experiments performed on the methanogenic
acetyl-CoA decarbonylase/synthase
(ACDS) suggested that the N-terminal region of ACS is involved in
C–C bond cleavage.
167
On the basis
of kinetic and spectroscopic data for different ACS enzymes, it appears
that conformational changes directly impact stability of the Ni–acetyl
intermediate. Steric hindrance around the Nip due to conformational
changes of a proximal phenylalanine (F512) is proposed to facilitate
C–C bond cleavage and to affect interaction of CO with the
enzyme.
168
Thus, conformational changes
clearly affect ACS enzymatic activity, and studies are needed to better
understand these impacts on the catalytic mechanism.
3.4
Catalytic Mechanism of Acetyl-CoA Synthesis
The chemistry
of the ACS reaction is catalyzed by the A-cluster;
surprisingly, even though this center contains six redox-active metals,
substrate binding seems to be confined to a single metal center, Nip. Pulse-chase
studies indicate that the steady-state mechanism
involves random order binding of the methyl group and CO, followed
by ordered binding of CoA.
169
The two competing
mechanisms that have been proposed differ in the oxidation state of
the Nip. The “paramagnetic mechanism” proposes
a Nip(I) catalyst and Ni(I)–CO [or methyl–Ni(III)
and methyl–Ni(II) for the other branch of the random mechanism]
and acetyl–Ni(II) intermediates (Scheme 6),
170
while the “diamagnetic mechanism”
proposes a Ni(0) active catalyst with Ni(0)–CO and methyl–Ni(II)
[without the paramagnetic methyl–Ni(III)] intermediates.
115
However, both mechanisms include organometallic
methyl–Ni, acetyl–Ni, and thiolytic cleavage of the
acetyl–Ni species by CoA. Because of their similarity, we will
focus here on the paramagnetic mechanism and include relevant aspects
of the Ni(0)-based mechanism. As described elsewhere, the mechanism
of acetyl-CoA formation resembles the Monsanto process, where acetic
acid is produced by the reaction of methanol and CO on a rhodium complex
through organometallic complexes.
171,172
Scheme 6
Proposed
Paramagnetic Mechanism of Acetyl-CoA Synthesis Catalyzed
by the A-Cluster
The essential role of the tetrameric α2β2 CODH (as it was known until 1985) in acetyl-CoA
synthesis
was predicated on studies of the isotope exchange reaction between
CO and the carbonyl group of acetyl CoA.
151
In this reaction, the C–C and C–S bonds of acetyl-CoA
are cleaved to generate enzyme-bound methyl, carbonyl, and CoA groups,
allowing the central carbonyl group to exchange with free CO; finally,
the C–S and C–C bonds must be resynthesized. That CODH/ACS
alone (and it was subsequently shown that ACS alone is required for
this reaction
61
) is required clearly demonstrated
that this enzyme is responsible for the key step in the Wood–Ljungdahl
pathway: condensation of methyl, CO, and CoA to form acetyl-CoA.
Because CODH/ACS catalyzes an exchange reaction between CoA and acetyl-CoA
much faster than the CO/acetyl-CoA exchange reaction, CoA was proposed
to be the final substrate that reacts with the bound acetyl group
to form acetyl-CoA.
173
Since methylation
of CODH/ACS by the methylated CFeSP can occur without any CO or CoA
and also faster than the overall acetyl-CoA synthesis reaction, the
methyl group was suggested to be the first substrate to bind Nip.
174−177
Since the back-π-donation upon CO binding to the metal is
expected to decrease the electron density on the metal center, its
reactivity with methyl could be decreased if methyl is bound as the
second intermediate. However, CO can also bind the enzyme in the absence
of a methyl donor or CoA; furthermore, a pulse-chase study of acetyl-CoA
synthesis with CODH/ACS
Mt
and with ACS-only
clearly indicated that either CO or methyl can bind first during catalysis.
169
In this pulse-chase (or isotope dilution) study,
CODH/ACS
Mt
is incubated with equimolar
amounts of a labeled substrate (14CH3–CFeSP, 14CO, or 3′-dephospho-CoA) and then mixed
with a solution
containing either (1) the other two substrates at high concentrations
or (2) all three substrates at high concentrations. Incorporation
of the label into product is measured. If the mechanism is strictly
ordered with labeled substrate being first to bind, addition of that
unlabeled substrate in excess will not lead to dilution of the isotope.
On the other hand, if it is actually the second substrate in the sequence,
it must dissociate to allow the true first substrate to bind before
it can form a productive complex. Dissociation leads to isotope dilution,
as detected in the product. This method is valuable because one can
determine how ordered (or how random) the reaction is. Nearly complete
dilution of dephospho-CoA in the pool of excess CoA is observed. However,
there is no measurable isotope dilution when ACS is treated with 14CH3–CFeSP or 14CO.
Thus, the
first substrate can be either the methyl group or CO group, and the
third substrate is CoA. However, it is important to note that CO but
not CO2 was used as the source of the carbonyl group of
acetyl-CoA; thus, possible regulatory effects of the tunnel and the
possible effects of the coupled reaction on the mechanism were not
addressed in this study.
For illustration, we show the mechanistic
scheme as an ordered
reaction with CO as the first binding substrate (Scheme 6). Before substrate binding,
a reductive activation by Ti(III)
citrate or another low-potential electron donor is required. The oxidized
state of the A-cluster, which has a configuration of [Fe4S4]2+Nip
2+Nid
2+, cannot accept a methyl from the CFeSP
176,178
or bind CO.
177
This Nip(I)
intermediate was trapped by photolysis of the Ni(I)–CO species
and its EPR spectrum was recorded, exhibiting g-values
of 2.56, 2.10, and 2.01.
162
Then, in a
kinetically coupled reaction, Nip(I) binds CO as the first
step in the mechanism. For this to occur, the tunnel must be open
to allow migration of the CO that is produced from CO2 in
the C-cluster. Two reduced states have been observed by Mossbauer
spectroscopy: [Fe4S4]+ [Nip]+ and [Fe4S4]2+ [Nip]+.
179
The Nip–CO species is proposed to form the well-characterized NiFeC
species.
30
DFT calculations combined with
EPR,
154
ENDOR,
180
Mossbauer,
85b,181
IR,
182
and X-ray experiments
183,184
indicate that NiFeC
species consists of a [Fe4S4]2+ cluster
bridged to a dinuclear Ni center, Nip
+–CO,
and Nid
2+.
185
According to the EPR spectral properties,
the unpaired electron density is delocalized over the Nip, the [Fe4S4] cluster, and
the terminal carbonyl
group.
154,186
Various experiments indicate the catalytic
competence of the NiFeC
species. It forms at the same rate and decays 6-fold faster than the
steady-state rate of acetyl-CoA synthesis.
177
The rate of the formation of NiFeC species monitored by EPR equals
the rate of the Ni–CO bond formation probed by IR, indicating
that Ni–CO is the only metal–carbonyl species formed
upon the reaction of ACS and CO.
182b
Controlled
potential enzymology studies revealed the need for only a single electron
transfer with a midpoint potential of −511 mV
187
to activate the A-cluster, a value that is very similar
to that reported for the formation of NiFeC species from acetyl-CoA
(−541 mV).
188
Ferredoxin-II (Fd-II),
which enhances the isotopic exchange rate,
154
is shown to activate the A-cluster most likely by forming this Ni+ species.
187
In the diamagnetic
mechanism, formation of a Ni(0) intermediate
is proposed
115
and is supported by the
ability of a model Ni(0)–phosphine complex to accept a methyl
group from a Co3+–CH3 complex;
189
however, a Ni(0) state on ACS has never been
observed or reported. Furthermore, two-electron reduction of Ni(0)
to Ni2+ would be extremely difficult, since even the reduction
potential for Ni2+–CO/Ni+–CO is
already very negative, below −550 mV. The presence of a Ni(0)
in a highly electropositive environment formed by Nid
2+ and [Fe4S4]2+ seems unlikely
because electron transfer to the cluster or the Nid would
be favored.
The second step of the mechanism is binding of the
methyl group
to the A-cluster. In this step, the protein is most likely in its
open conformation, where the A-cluster is accessible to the large
CFeSP. Rapid kinetic studies utilizing a chiral methyl donor suggested
the transfer of a methyl cation through an SN2 mechanism
where Nip attacks to the Co3+–CH3 on the CFeSP to leave behind a Co1+ and a methylated
Nip. The methyl, like CO, appears to bind to the Nip.
155,158,190
Although a radical methyl transfer is suggested according to the
model studies,
191,192
this is not feasible in biology,
since the reduction potential of Co3+–CH3/Co2+–CH3 is below −1 V, which
would be too low for physiological electron transfer.
193
Rapid kinetic studies indicate that both
methylated
177,190
and acetylated ACS
194
species are EPR-silent.
This represents a challenge for the paramagnetic mechanism, since
the SN2 addition of methyl cation to the Nip
+ should result in a Nip
3+. However,
since the Nip
3+ state is predicted to be highly
oxidizing and unstable, it should readily be reduced to the Ni2+ state. Since acetyl-CoA
synthesis does not require net electron
transfer from the environment,
151
this
reduction could be achieved by an internal electron transfer, as shown
in Scheme 6. Fd-II is shown to donate an electron
to the proposed Nip
3+ intermediate and to accept
an electron during the cleavage of the Nip–acetyl
intermediate, most likely by interfacing with an internal electron
shuttle.
187
However, this internal electron
shuttle has not yet been identified. Such an internal electron transfer
is not necessary for the diamagnetic mechanism, since Ni(0) is converted
to Ni2+–CH3. However, as mentioned above,
the diamagnetic mechanism has its own challenges.
The next step
involves a methyl migration (carbonyl insertion)
to form an acetyl–metal complex. A crystal structure of ACS
Mt
is proposed to represent the CoA binding
conformation of the enzyme.
166
Addition
of CoA is followed by the thiolytic cleavage of the acetyl-CoA product
and also the internal transfer of electrons.
3.5
Structural
and Functional Models of ACS
Modeling efforts for the A-cluster
of ACS up until ∼2005
have been reviewed.
195
Thus, we will only
briefly cover the Ni complexes reported (Scheme 7).
Scheme 7
Schematic Views of Model Complexes of A-Cluster
Since the [Fe4S4] complex
and the distal
Ni are thought to modulate the electronic and redox properties of
the active site but not to bind any ligands, most model complexes
have focused on imitating the Nip or the bimetallic Nip–Nid environment, omitting
the [Fe4S4] complex. Initially, compound 10 was prepared
by the reaction of 9 with Ni(cyclooctadiene)2 and CO as a very stable complex in anaerobic
solution that undergoes
immediate degradation upon air exposure.
196
The IR spectrum of 10 exhibits νCO bands at 1948 and 1866 cm–1. Crystallographic
and NMR spectroscopic characterization of the compound indicates the
presence of a Ni(0)Ni(2+) couple. The bimetallic Ni complexes, 12 and 14, have also
been reported.
197
Compound 12 was synthesized from the reaction of 11 and (R2PCH2CH2PR2)NiCl2 (R = Et,
Ph). Reaction of 11 with nickel chloride also
yielded a trinuclear nickel complex upon the dimerization of two units
of 11 around a nickel atom. The Ac–CycGlyCys–CONH2 is used as precursor for the synthesis
of compound 11. Synthesis of compound 12 is a significant
improvement, since it includes two sulfides and two phosphines to
mimic the environment of Nip. That Ni can be reduced to
form a Ni(1+)Ni(2+) complex. While the oxidized Ni(II) state cannot
bind CO, the reduced state can and be reduced further to the Ni(0)Ni(2+)
state. Compound 14, synthesized from 13,
contains a coordinating ring pattern and donor set for Nid that is almost identical
to that of the A-cluster. However, no ACS
activity or ligand-binding properties were observed for this interesting
compound. Furthermore, Harrop reported the synthesis and characterization
of new complexes 15–18.
198−200
Treatment of compound 15 with Cu(2,9-dimethyl-1,10-phenanthroline)Cl
resulted in a dinuclear Cu(I)–Ni(II) complex, which does not
bind CO and does not include a reducible nickel center.
198
Neither compound 15 nor compound 16 can be reduced easily or can bind CO. Compounds
15 and 16 were utilized as precursors to prepare 17 and 18, respectively. Reduction
of 17 with dithionite yields a five-coordinate Ni(I) complex in trigonal
bipyramidal geometry with an axial EPR signature of g = 2.226, 2.125. The Ni(I) state
of 17 binds CO to form
a complex with a rhombic EPR spectra (g = 2.223,
2.218, 2.019), which is typical for six-coordinate Ni(I)–CO
complexes
201
and with a Ni(I)–CO
band at 2044 cm–1. Compound 18 can
be reduced with dithionite or sodium borohydride to form a Ni(I) complex,
based on its EPR spectrum. As expected, 18 binds CO in
the Ni(I) state, exhibiting a strong Ni(I)–CO band at 1997
cm–1, a value that is very close to what is observed
in A-cluster (1996 cm–1).
149
These studies show the stability and inertness of Nid
2+ and reducibility and ligand affinity of the Nip atom. A Ni(II)–Ni(I) compound,
19, was
recently shown to accept methyl from methylcobaloxime and form thioester
upon CO exposure.
202
This result indicates
that a Ni(II)Ni(I) can afford the chemisty of the acetyl-CoA synthesis
reaction in a proper coordination and electronic environment.
Reactivity of a Nip(0) analogue, Ni(triphos)(PPh3) (compound 20), with a methyl–CFeSP
analogue, 21, yields compound 22.
203
While compound 20 was methylated by 21 in approximately 1 h, no methylation or acylation
was observed for
compound 23, even 24 h after of reaction. Furthermore,
reaction on compounds 24 and 25 with 21 and CO leads to acetylation of the S-ligand
of the methylated
nickel and dissociation of the thioester.
204,205
The viability of the Ni(II)/Ni(0) couple in Ni–acyl formation
is further supported by another binuclear nickel compound, 26 (Dmp is 2,6-dimesitylphenyl),
which forms the acetyl thioester upon
reaction with CO.
206
The methyl group in
compound 26 was donated either by compound 21 or MeI. These studies support the plausibility
of the methyl ligand
binding to the metal before CO binds. Similarly, a Ni(0)–CO
complex, compound 27 was prepared and shown to accept
methyl and to exhibit Ni–acyl bond formation.
207
As summarized above, inorganic model studies suggest
that Ni(II)
centers mimicking Nid are not reducible or catalytically
active. Ni(0) and Ni(I) complexes can bind CO as well as mimic ACS
activity. There are also examples of both Ni(I) and Ni(0) complexes
that bind methyl followed by CO and vice versa. Further studies are
necessary to clarify these mechanistic issues. Inclusion of the Fe4S4 cluster in the
inorganic models would provide
important information about the role of this redox-active center in
the ACS reaction and perhaps would afford new catalysts to afford
acetyl-CoA synthesis without enzymes.
4
Conclusions
and Future Directions
We have described studies on two remarkable
metalloenzymes that
have defined novel biochemical mechanisms involving organometallic
chemistry to catalyze their reactions. CODH catalyzes CO2 reduction, a reaction that
has important potential impact on the
generation of energy-rich compounds and on the environment due to
its involvement in the global carbon cycle. This is a catalyst that
has optimized its kinetics and thermodynamics, operating at high rates
and without an overpotential. These characteristics warrant further
studies of CODH aimed at understanding the principles that guide these
two enviable properties. Past studies outlined here have uncovered
novel metal clusters to bind, activate, and transform substrates (CO
and CO2) and macromolecular channels that enhance flux
of precious substrates between catalytic sites. CODH also is a wonderful
system to explore how chemical bond forming and breaking interfaces
with redox chemistry. Future research will define the kinetic and
structural properties and electronic states of the yet-to-capture
intermediates in CO oxidation/CO2 reduction and reveal
where the electrons reside during the two-electron redox interconversion.
Future studies on this enzyme will be greatly enriched with the development
of a well-defined and reproducible way to generate variants of CODH.
This enzyme, especially coupled to ACS and other enzymes of the Wood–Ljungdahl
pathway, offers great potential for biotechnology through the conversion
of simple abundant compounds into needed chemicals and fuels. To realize
this promise, host organisms must be developed or reconfigured to
foster an anaerobic environment that includes all of the metallochaperones
and accessory factors required to support the high activity observed
in the native organisms. These factors and their roles need to be
characterized.
In order to tap the potential of CODH, ACS must
be tamed. Above
we have described the highly unusual metal center at the heart of
this enzyme and provided information, gleaned by a mixture of biochemical,
biological, and biophysical methods, on the modular way that this
center forms organometallic (M–CO, M–CH3,
M–acetyl, and M–S) bonds en route to generation of the
compound at the center of our metabolic charts, acetyl-CoA. Though
we have defined the novel modular approach to synthesis of this key
metabolic building block, we do not yet understand the internal redox
chemistry that drives C–C and C–S bond formation to
generate acetyl-CoA. It is important to capture and characterize the
yet undefined intermediates in the ACS catalytic cycle. We have learned
to express ACS and reconstitute it in vitro to near full activity;
however, the same challenges remain in developing a genetic system
that produces a highly active enzyme.
It will be extremely important
to understand how the activities
of CODH and ACS are coordinated in the complex and to increase our
understanding of the dynamics and mechanics of the tunnel that carries
CO from the C-cluster to the A-cluster. With both CODH and ACS, it
is important to understand the movement of domains and how these proteins
interact with other components of the Wood–Ljungdahl pathway,
especially the CFeSP.Future high-impact papers will emerge that provide
an understanding of the structures of complexes between CODH/ACS and
the CFeSP, achievable by X-ray diffraction methods as well as other
methods that can define conformational, ligation, and electronic states
and measure distances among the redox centers in these various states.
Finally, ACS has been found in the multidrug-resistant human pathogen Clostridium
difficile, and better understanding of
these enzymes could foster the discovery of new therapeutic solutions
against C. difficile infections.
208