Carbon Dioxide Activation at the Ni,Fe-Cluster of Anaerobic Carbon Monoxide Dehydrogenase

See allHide authors and affiliations

Science  30 Nov 2007:
Vol. 318, Issue 5855, pp. 1461-1464
DOI: 10.1126/science.1148481


Anaerobic CO dehydrogenases catalyze the reversible oxidation of CO to CO2 at a complex Ni-, Fe-, and S-containing metal center called cluster C. We report crystal structures of CO dehydrogenase II from Carboxydothermus hydrogenoformans in three different states. In a reduced state, exogenous CO2 supplied in solution is bound and reductively activated by cluster C. In the intermediate structure, CO2 acts as a bridging ligand between Ni and the asymmetrically coordinated Fe, where it completes the square-planar coordination of the Ni ion. It replaces a water/hydroxo ligand bound to the Fe ion in the other two states. The structures define the mechanism of CO oxidation and CO2 reduction at the Ni-Fe site of cluster C.

The biological redox transformations of CO2, N2, and H2 are essential processes in global biogeochemical cycles and are catalyzed by enzymes containing complex metal clusters based on iron and sulfur whose detailed function is still poorly understood (1, 2). Carbon monoxide dehydrogenases (CODHases) are the biological catalysts for the reversible oxidation of CO to CO2, with water as the source of oxygen: CO + H2O → CO2 +2e +2H+ (Eq. 1). Two principal types of CODHases have been described that differ in their cofactor composition, structure, and stability in the presence of dioxygen: Anaerobic bacteria and archaea use oxygen-sensitive Ni- and Fe-containing CODHases, whereas aerobic, carboxydotrophic bacteria use a Cu-, Mo-, and Fe-containing flavoenzyme (3). The Ni,Fe-CODHases are monofunctional or bifunctional enzymes associated with Ni,Fe-containing acetyl–coenzyme A (CoA) synthases (ACS) (4, 5)[for review, see (3, 6)].

CO oxidation and CO2 reduction at the active site, cluster C, of Ni,Fe-CODHases are proposed to require three different oxidation states differing by one electron (Cred1, Cint, and Cred2) (6). In this model, the Cred1 state of cluster C converts CO to CO2 and is formed at redox potentials below –200 mV (7). At pH = 7.0, the midpoint potential for the conversion of Cred1 to Cred2 is –530 mV (8, 9), similar to the normal potential of the CO2-CO couple of –558 mV (10). Consequently, Cred2 is generated by a two-electron reduction of the Cred1 state via Cint (11). The structure of cluster C was revealed by crystallographic analysis of CODHases isolated from Carboxydothermus hydrogenoformans (CODHIICh) and Rhodospirillum rubrum (CODHRr) at 1.6 and 2.8 Å resolution, respectively (12, 13). Cluster C in CODHIICh has been described as an asymmetric [NiFe4S5] center, which comprises an integral Ni ion coordinated by four sulfur ligands with square-planar geometry (12). An asymmetrically coordinated Fe ion (Fe1) is found close to the Ni ion. In CO-treated CODHRr, cluster C has a similar structure with a cubane [NiFe3S4] center linked to a mononuclear Fe site (13). Corresponding structures of cluster C were also identified in the crystal structures of bifunctional ACS/CODH isolated from Moorella thermoacetica (CODHMt) (4, 5). Mechanisms proposed for the reversible oxidation of CO to CO2 posit the activation of H2O and CO as well as the stabilization of a metal-bound COO(H) intermediate. However, the structure of these states remained elusive. We describe how H2O and CO2 are bound and activated and propose a structure-based model for CO oxidation and CO2 reduction at the Ni-Fe1 site of cluster C.

An expression system for CODHIICh established in Escherichia coli enables a one-step purification of active enzyme. Crystals of recombinant CODHIICh diffract up to 1.40 Å resolution on a rotating anode x-ray generator (table S1). The overall structure of recombinant CODHIICh is identical to the structure of native CODHIICh (12). CODHIICh crystals were held at a defined redox potential of –600 mV for 3 hours with use of Ti(III) citrate. These crystals were either directly frozen in liquid nitrogen, generating the –600 mV state; oxidized via incubation with methylviologen (MVox) and dithiothreitol (DTT) and then incubated with DTT to give the –320 mV state; or incubated in the –600 mV solution with NaHCO3 as the CO2 source, generating the –600 mV+CO2 state. The –600 mV state (equivalent in its redox potential to the Cred2 state) and the –320 mV state (equivalent to the Cred1 state) display practically identical structures for cluster C (Fig. 1, A and C). In both structures, the Ni ion is coordinated by three sulfur ligands with distorted T-shaped coordination geometry (Fig. 1, A and C, and table S3). Fe1 is coordinated by His261 (H261), Cys295 (C295), a μ3-sulfido ligand, and a monoatomic ligand (Fig. 1, A and C, and figs. S2 and S4). A weakly occupied alternative position for Fe1 (Fe1B) is observed in both states (Fig. 1 and table S2). The monoatomic ligand is a distance of 2.7 Å from the Ni ion and occupies the position that would complete the square-planar coordination of the Ni. The electron density of the ligand can be modeled as a light atom (C, N, or O) with occupancies of about 60 to 70% or a sulfur atom with 30% occupancy. The observed Fe-ligand bond lengths of 1.93 to 1.95 Å are atypical for Fe-S bonds, whereas a H2O/OH ligand is consistent with the refined bond length, the relative occupancies of ligand and Fe1 (table S2), and spectroscopic investigations of the Cred1 state. The long distance between Ni and the ligand suggests a weak Ni-OHx interaction. A H2O/OH ligand has been detected bound to a high spin Fe2+ ion called ferrous component II (FCII) in the Cred1 state (14, 15), so the crystal structure is consistent with Fe1 in the –320 mV state being FCII.

Fig. 1.

The –600 mV (A), –600 mV+CO2 (B) and –320 mV (C) states of cluster C. 2FobsFcalc maps in blue are contoured at 1σ, and FobsFcalc maps in green are contoured at 4.5σ. For the calculation of the FobsFcalc map, the OHx ligand [(A) and (C)] and the CO2 ligand (B) have been removed from the model. An alternative position found for Fe1, termed Fe1B, is depicted in transparent light gray. The occupancies for the alternative position have been estimated to 10 to 30%. Selected distances are shown in Å. For more details on the geometry of the three states, see figs. S2 to S4. All pictures were prepared by using PyMol (23).

In the presence of appropriate reducing agents, Ni,Fe-CODHases can catalyze the reduction of CO2 (16). The structure of CODHIICh with CO2 (–600 mV+CO2 state) reveals a triatomic ligand bridging Ni and Fe1, which replaces the water/hydroxo ligand at Fe1 (Fig. 1B). Modeling the ligand as CO2 fully satisfies the observed electron density maps, whereas modeling with one or two atoms does not. CO2 bound to cluster C acts as a η1 OCO ligand at Ni2+ with a Ni-C distance of 1.96 Å and completes the square-planar coordination geometry typically found for Ni2+ ions. CO2 acts as a η1 OCO ligand at Fe1 with an Fe1-O1 distance of 2.05 Å, resulting in a μ22 binding mode of CO2 bridging the Ni-Fe1 site (Fig. 1B). Like the H2O/OH ligand in the –600 mV and –320 mV states, O1 of CO2 is in hydrogen-bonding distance to Lys563 (K563) (Fig. 1B). O2 is in hydrogen-bonding distance to His93. CO2 binding to cluster C causes only minor changes in the geometry of the cluster (Fig. 2). The change of the distorted T-shaped to the square-planar coordination at the Ni ion induces a small shift in the Ni position of about 0.2 Å and widens the Cys526Sγ-Ni-S3 angle (table S3).

Fig. 2.

Superposition of the –600 mV (blue) and –600 mV+CO2 (element colors) states.

The ability of CO2 to coordinate transition metal complexes is well documented (17). Coordination of CO2 at the carbon atom results in a net electron transfer from the metal into the antibonding lowest unoccupied molecular orbital of CO2. This activation of CO2 increases the negative partial charges at the oxygen atoms, which are stabilized by binding to electron-deficient centers like transition metals or by forming hydrogen bonds (17). In the cluster C-CO2 complex, Ni acts as the Lewis base, and Fe1 is the Lewis acid that together with K563 stabilizes the negative partial charge on O1. The deviation from linearity along the O-C-O axis [O-C-O ∼133° (table S3)] is in agreement with the activation of CO2 by binding to cluster C.

In previous structural characterizations of native preparations of CODHIICh, we identified a μ-sulfido ligand (S2) bridging Ni and Fe1 (12, 18). The enzyme used for crystallization as well as dissolved crystals had high specific activities of ∼14.000 units mg–1, and on the basis of a positive correlation between the presence of the S2 ligand and enzyme activity it was postulated that S2 is essential for the catalytic CO oxidation (18, 19). However, the necessity of the S2 ligand for catalysis was debated when sulfide was shown to reversibly inactivate CODHRr and CODHMt, leading to short lag phases (20), and no S2 ligands were identified in the crystal structures of CODHRr and CODHMt (5, 13). Here, we describe the structures of [NiFe4S4(OHx)/(CO2)] clusters without S2 ligand in crystals with high specific CO oxidation activities (11.000 to 13.500 units mg–1), showing that the presence of S2 is not necessary for catalysis. Furthermore, the S2 ligand occupies the binding site of two substrates of Ni,Fe-CODHases, water, and CO2. The H2O/OH ligand identified requires the same coordination site at Fe1 as the bridging S2 ligand (fig. S5), and CO2 binds to the two open coordination sites of Ni and Fe1. Thus, we suggest that the S2 ligand between Ni and Fe1 is absent in catalytically competent enzyme species and can be reductively or chemically replaced, activating the enzyme.

The three presented structures offer direct insight into the reaction mechanism (Fig. 3). The [NiFe4S4OHx] cluster determined in the –320 mV state is the functional state that activates CO and contains the H2O/OH ligand. The Ni2+ ion is positioned at the end of the substrate channel, and its three sulfido ligands act as π donors to the metal, enabling CO to bind to the Ni2+ ion (21). The Ni2+ ion has two open coordination sites, allowing either an apical binding of CO to form a distorted tetrahedral geometry or CO binding equatorially to complete the square-planar coordination geometry. Binding of CO in the apical position has been proposed for CO-treated crystals of CODHRr and CODHMt (5, 13). Modeling of CO in the apical position places the CO-carbon atom more than 3.5 Å apart from the H2O/OH ligand and makes further rearrangements necessary for the reaction to proceed. In contrast, CO binding to complete a distorted square-planar coordination of the Ni2+ ion results in a OC-OHx distance of less than 2 Å (Fig. 3, step II). The binding of CO to a weakly backdonating metal like Ni2+ results in an electrophilic carbon atom and facilitates its reaction with the Fe1-bound H2O/OH ligand to a metal-carboxylate species as observed in the –600 mV+CO2 state (Figs. 1B and 3, step III). Product release may be assisted by the reversible ligand exchange of CO2 against H2O at Fe1 and is accompanied by a two-electron reduction of cluster C, generating the Cred2 state. The H2O/OH ligand can be replenished by a neighboring network of solvent molecules (fig. S6). A comparison of the [NiFe4S4OHx]and [NiFe4S4(CO2)] states reveals the positions of Ni and Fe1 to be largely unaffected by the presence or absence of the CO2 ligand (Fig. 2). The [Fe3S4] site of cluster C provides a solid metal-sulfur frame in which Ni and Fe1 are held in place and serves as an electronic buffer to compensate for the electronic changes at Ni and Fe1 during the catalytic cycle. The small structural changes of cluster C agree well with the low reorganization energy expected for a reaction with turnover rates of 31,000 s–1 and a ratio of kcat (catalytic rate constant)/Km (Michaelis constant) of 1.7 × 109 M–1 s–1 at +70°C (22).

Fig. 3.

Structure-based mechanism of CO oxidation at cluster C. (I) The –320 mV state has been used as a model for Cred1, the state of cluster C competent of CO oxidation. (II) The proposed transition state of the reaction in which CO binds to the Ni2+ ion and reacts with the Fe1-bound OH group. (III) The –600 mV+CO2 state is used as a model for the stabilization of the metal carboxylate state. (IV) The –600 mV state is used as a model for the Cred2 state, which is supposed to contain two additional electrons compared with the Cred1 state, denoted as a formal change of the oxidation state of the Ni2+ ion.

The structure-based mechanism outlined agrees in all central aspects with the bimetallic mechanism proposed on the basis of electron paramagnetic resonance (EPR), electron nuclear double-resonance (ENDOR), and Mössbauer spectroscopy (14, 15).

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S6

Tables S1 and S2


References and Notes

Stay Connected to Science

Navigate This Article