QM/MM computations reveal details of the acetyl-CoA synthase catalytic center

https://doi.org/10.1016/j.bbagen.2020.129579Get rights and content

Highlights

  • QM/MM computations reveal atomic details of Aopen and Aclosed catalytic centers in acetyl-CoA synthase (ACS)

  • Geometry of one-electron reduced A-cluster with CO ligand and tetrahedral proximal Ni atom agrees with the crystal structure

  • Rotation of the side chain ring of Phe512 appears to serve as a structural gate for ligand binding

  • Aopen with square planar proximal nickel does not correspond to the ligand-free, oxidized [Fe4S4]2+--Nip2+--Nid2+ state

Abstract

The “open” (Aopen) and “closed” (Aclosed) A-clusters of the acteyl-CoA synthase (ACS) enzyme from Moorella thermoacetica have been studied using a combined quantum mechanical (QM)/molecular mechanical (MM) approach. Geometry optimizations of the oxidized, one- and two-electron reduced Aopen state have been carried out for the fully solvated ACS enzyme, and the CO ligand has been modeled in the reduced models. Using a combination of both αopen and αclosed protein scaffolds and the positions of metal atoms in these structures, we have been able to piece together critical parts of the catalytic cycle of ACS. We have replaced the unidentified exogenous ligand in the crystal structure with CO using both a square planar and tetrahedral proximal Ni atom. A one-electron reduced A-cluster that is characterized by a proximal Ni atom in a tetrahedral coordination pattern observed in both the Aopen (lower occupancy proximal Ni) and Aclosed (proximal Zn atom) geometries with three cysteine thiolates and a modeled CO ligand demonstrates excellent agreement with the crystal structure atomic positions, particularly with the displacement of the side chain ring of Phe512 which appears to serve as a structural gate for ligand binding. The QM/MM optimized geometry of the A-cluster of ACS with an uncoordinated, oxidized proximal nickel atom in a square planar geometry demonstrates poor agreement with the atomic coordinates taken from the crystal structure. Based on these calculations, we conclude that the square planar proximal nickel coordination that has been captured in the Aopen structure does not correspond to the ligand-free, oxidized [Fe4S4]2+ − Nip2+ − Nid2+ state. Overall, these computations shed further light on the mechanistic details of protein conformational changes and electronic transitions involved in the ACS catalytic cycle.

Introduction

The nickel-dependent acetyl-CoA synthase (ACS) and carbon monoxide dehydrogenase (CODH) constitute the central enzyme complex involved in the formation of acetyl-CoA, also known as the Wood-Ljungdahl pathway [1,2]. This bifunctional enzyme, ACS/CODH (Fig. 1) catalyzes the reversible reduction of CO2 to CO (CODH activity) according to2H++CO2+2eCO+H2O,

thereby fixing CO2 [3]. Following reduction of CO2 to CO in the CODH subunit, CO travels through a 70 Å tunnel to ACS [4]. There, CO, CoA, and a methyl cation donated by the cobalt-containing corrinoid iron‑sulfur protein (CoFeSP) are converted to acetyl-CoA (ACS activity) [4]:CO+CoA+CH3CoFeSPCH3COCoA+CoFeSP.

Forming a tight complex, the ACS/CODH enzyme from acetogenic bacterium Moorella thermoacetica is a 310 kDa α2β2 tetramer that contains four metalloclusters (A, B, C, and D) that span the 200 Å length of the protein [3] (Fig. 1A). Each β subunit contains a [Fe4S4] B-cluster and an active site consisting of a [Ni-Fe4-S4−or−5 C-cluster, while a single [Fe4S4] D-cluster bridges the two β subunits [3]. Each α subunit contains a single Fe4S4Ni2 center, the A-cluster, which is responsible for the ACS activity [3](Fig. 1B).

In the structure of ACS/CODH reported in 2003, the two α subunits contain A-clusters with different metal content and coordination geometries [3]. In one α subunit, the protein is observed in a “closed” conformation (αclosed), similar to a structure reported by Doukov et al. [4], whereas the other α subunit appears in an “open” (αopen) conformation, allowing for greater solvent accessibility to the A-cluster [3] (Fig. 1A). In the closed A-cluster (Aclosed), the [Fe4S4] cubane cluster is bridged by a cysteine thiolate to a proximal Zn ion (replacing Cu in the structure reported by Doukov et al. [4]) that connects to a distal Ni ion (Nid) [3]. The Zn ion is in a distorted tetrahedral geometry, further coordinated by an unknown exogenous ligand, likely heterodiatomic in nature. The Nid has a S2N2 square planar coordination geometry that includes the two thiolates of Cys595 and Cys597, as well as the two main chain N atoms from Gly596 and Cys597.

In the open A-cluster (Aopen), the [Fe4S4] cubane is bridged by a cysteine thiolate to a proximal, labile Ni ion, Nip, that in turn connects to a distal non-labile Ni ion (Nid) through two bridging cysteine residues (Fig. 1B). As in Aclosed, the Nid has a S2N2 square planar coordination geometry whereas the Nip is in a square planar coordination with three bridging cysteine thiolates (Cys509, Cys595, Cys597). Interestingly, the X-ray crystallographic structure of Darnault et al. also identifies an exogenous ligand of unknown character in Aopen, although, based on their electron densities, this ligand is believed to be different than the ligand in Aclosed [3].

Another striking feature of the ACS/CODH complex is the existence of a long hydrophobic tunnel that connects the C-clusters to each other and to the A-cluster in Aclosed [3,5,6]; in the open form, the tunnel ends about 20 Å from the [Fe4S4]-Nip-Nid center due to significant conformational changes in protein structure (Fig. 1A). Thus, the diffusion of CO is believed to occur via the tunnel in the closed form of ACS [3,4]. However, since the A-cluster of ACS is only accessible in an open conformation to coenzyme A (CoA) and the methyl-donating protein, the enzyme must rearrange to catalyze the subsequent transfer of the methyl group. Importantly, the binding of CoA and CoFeSP-CH3 may rely on different conformational changes in ACS. One such rearrangement, described by a rigid body movement of domain 3, has been observed in a truncated form of ACS [7]. The binding of CoFeSP-CH3 is believed to stabilize the Aopen state of ACS such that binding of the CH3 cation takes place with the proximal nickel atom in a square planar geometry [5].

Three central questions regarding the catalytic activity of ACS can be formulated as: 1) Which electronic states of the metal ions are responsible for providing two electrons required to form the metal-methyl bond? 2) What conformational changes in the protein structure accompany these electronic transitions? 3) Which substrate (CO or CH3) binds first? These questions have been addressed through structural analyses and spectroscopic investigations.

A-clusters are known to be stable in at least three redox states: 1) S = 1/2 for a one-electron, CO-bound state (Ared-CO); 2) an oxidized Aox state (S = 0 or integer); 3) a two-electron reduced state in which CH3 is bound (S = 0 or integer) [8,9]. The reduction of an oxidized A-cluster to a CO-bound species is mediated by one electron that is likely generated from the C-cluster [9]. Regarding the methylation of the A-cluster from CH3-Co3+FeSP, the lack of EPR signal associated with this process suggests a two-electron oxidation-reduction with S = 0 or integer in both states [9]. The [Fe4S4]2+ cubane cluster is believed to be an unlikely candidate for the reactive site as it reduces at a rate 200-fold slower than methyl transfer [3]. A reduced spin-coupled Nip1+–Nid1+ unit could oxidize to Nip2+–Nid2+ upon methylation. However, as it is less exposed to solvent, Nid is a better candidate for the non-labile Ni and likely does not change its oxidation state during catalysis [3]. This postulation has been confirmed by X-ray absorption spectroscopy spectra indicating that Nid in a square-planar coordination with N2S2 from the protein side chains is non-labile, such that Nid2+ will not reduce to +1 [10]. Previous density functional theory (DFT) computations have also suggested that the proximal Ni ion is the reactive atom in a two-electron activated species, e.g. Nip0 − Nid2+ and not Nip1+ − Nid1+ exists [11].

A proximal Ni atom, on the other hand, would be at the base of the pocket and exposed on the molecular surface, coordinated by three protein thiolates (Cys509, Cys595, Cys597); furthermore, the proximal metal is known to change its coordination between the αopen (square-planar) and αclosed (distorted tetrahedral) conformations. These features indicate that the proximal metal is the most likely candidate for serving as the labile Ni [10]. Over the past two decades, this hypothesis has been further supported by evidence indicating that only the NiNi combination in A-cluster is active and other dimetal combinations (ZnNi and CuNi) are not [5,[12], [13], [14], [15], [16]]. Nonetheless, a crystal structure that captures a proximal nickel in a distorted tetrahedral structure remains elusive. Darnault et al. propose that the active A-cluster is composed of [Fe4S4]-Nip-Nid in which only Nip is active and exists in the Ni0/Ni2+ redox pair; in the absence of reductants, Nip is proposed to be in the 2+ state [3]. Interestingly, in vitro, Nip2+ can be reduced by one electron and bind CO, yielding Ni1+–CO and the Ared-CO state [3]. A different redox combination, namely Ni1+/Ni3+, has been proposed by Gencic and Grahame [13]. However, this postulation has been disputed by Darnault et al. based on the existing structural and spectroscopic evidence [3].

Regarding the sequence of CO and CH3 binding, varying opinions can be found in the literature. Barondeau and Lindahl favor a sequence in which methyl binds prior to CO since they observe a “nonfunctional Ared-CO” reduced state [9]. Using steady-state kinetic experiments, Ragsdale and co-workers have argued that the order of methyl and CO binding is random and that the CO-bound state is catalytically competent [17,18]. DFT computations from Amara et al. favor a “gating mechanism” in which CO binds first in the closed form [19].

Indeed, several DFT computational studies have been carried out on Aopen using a truncated active site of the enzyme from both the bacterium Moorella thermoacetica [19,20] as well as from Carboxydothermus hydrogenoformans [15]. In the study by Schenker and Brunold [20], the nature of the proximal metal was investigated for two active site compositions, [Fe4S4]-Cup+-Nid and [Fe4S4]-Nip+-Nid2+, based on the 2.2 Å structure of Doukov et al. [4]. Calculations were compared with extended X-ray absorption fine structure (EXAFS), Mössbauer [21], and electron nuclear double resonance (ENDOR) data [22]. The computational results, particularly the isotropic hyperfine parameters, indicate that the CO-bound Nisingle bondFe center is best described by [Fe4S4]2+–Nip+-CO–Nid2+ [20].

The DFT study by Chmielowska et al. (based on the 2.2 Å structure from Svetlitchnyi et al. [16]), carried out with a polarized continuum model, indicates that the oxidation and ligand state of the proximal nickel atom influence structural changes in the enzyme, particularly the Fe–S6–Nip valence angle and the Np–S2–Nid folding angle [15]. In the one-electron reduced model, with and without ligands, the largest spin density is localized on the proximal nickel atom [15]. This distribution of spin density is also observed in the calculations of Amara et al. when they considered a [Fe4S4]2+-Nip+-Nid2+ configuration [19]. In their computations, the distal nickel atom remains in Nid2+ and the environment around Nid does not change, consistent with X-ray models of the open and closed protein conformations and confirming the reactivity of the proximal nickel atom [19].

Interestingly, all X-ray structures containing an NiNi pair and [Fe4S4] cluster indicate the presence of four ligands coordinated to the proximal nickel atom [3,16]. Three of the ligands are cysteine thiolates while the fourth ligand, located approximately 2.5 Å from the metal, is often of unknown origin, such as in the structure of Darnault et al. [3]; in the structure of Svetlitchny et al. the fourth ligand was identified to be water [16]. Within the structure file (PDB ID: 1OAO), Aopen lists the atomic coordinates of the proximal nickel atom with two occupancies, one with 70% (Nip,70, square planar coordination geometry) and the other with 30% (Nip,30, tetrahedral geometry) [3]. Overlap of Aopen with Aclosed from the same crystal structure indicates that the low occupancy proximal Ni atom in the open structure is in a very similar position to the (proximal) Zn atom in Aclosed (Fig. 2). In other words, the tetrahedral position of the low occupancy proximal Ni atom may, in fact, be a residual geometry from a closed conformation present in another catalytic intermediate. Indeed, a ligand coordinated to this lower-occupancy metal position (Nip,30) would have a separation of ~2.6 Å from the higher occupancy proximal metal, as the crystal structure of Darnault et al. exhibits [3].

To explore this possibility of the lower occupancy metal ion being part of the catalytic cycle, we have explored computationally the Aopen and Aclosed protein structures, with a CO ligand modeled at the proximal Ni, in both the square planar and tetrahedral positions. Our goal is to use clues provided by the crystal structure to glean an understanding of the mechanistic steps involved in the CODH/ACS catalytic cycle. To our knowledge, until now no QM/MM study of the αopen and αclosed subunits of ACS with the entire surrounding protein environment has been carried out. As the protein scaffold is a critical aspect of the enzyme's activity, we have performed QM/MM computations of the enzyme in the fully solvated, ACS protein environment. Preliminary pKa computations of the ACS enzyme with and without the [Fe4S4] cluster indicated that the side chain of the nearby His516 residue is highly sensitive to the electrostatic environment. Therefore, in our models we have investigated the effect of the protonation state of His516, in a neutral state as well as in a protonated state (charge +1), which we denote throughout the text as Hsp516. In the following section we will discuss the preparation of the models and computational details before we present the results of these calculations.

Section snippets

Methods and calculation details

The structural models for ACS with and without the CO ligand were constructed based on the 1.9 Å structure (PDB ID:1OAO) obtained through X-ray crystallography [3]. From the tetrameric α2β2 structure, both the “open” and “closed” protein scaffolds (αopen and αclosed) and metal clusters (Aopen and Aclosed) were modeled. In Aclosed, the proximal metal (Zn) was replaced manually by a nickel atom. Within the structure file (1OAO.pdb), Aopen lists the atomic coordinates of the proximal nickel atom

αopen with CO-bound, tetrahedral Nip,30 in Aopen

Geometry optimizations of the CO-bound, one-electron reduced [Fe4S4]2+-Nip1+-Nid2+ state based on the distorted tetrahedral coordination geometry (lower occupancy Nip,30 position) in the αopen protein scaffold lead to a protein structure similar to the structure with higher Nip occupancy and very close to the active site geometry of the Aclosed protein structure containing Zn instead of Ni (see Fig. 4A-C). In both cases with His516 and protonated Hsp516, a rotation of the side chain of Phe512

Conclusion

The Aopen and Aclosed clusters of the αopen and αclosed ACS protein subunits have been investigated here with a QM/MM approach. The models have considered possible combinations of proteins matrices, metal coordination geometries, His516 protonation states, and electronic configurations that best reproduce the states observed in the crystal structure of ACS from acetogenic bacterium Moorella thermoacetica [3]. Our geometry optimizations and spin density analyses confirm that the proximal Ni atom

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

Insightful discussions with J. Kreibich are gratefully acknowledged. Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy – EXC 2008/1–390540038 – UniSysCat. Gefördert durch die Deutsche Forschungsgemeinschaft (DFG) im Rahmen der Exzellenzstrategie des Bundes und der Länder – EXC 2008/1–390540038.

References (37)

  • T.I. Doukov et al.

    A Ni-Fe-Cu center in a bifuctional carbon monoxide dehydrogenase/acetyl-CoA synthase

    Science

    (2002)
  • A. Volbeda et al.

    Crystallographic evidence for a CO/CO2 tunnel gating mechanism in the bifunctional carbon monoxide dehydrogenase/acetyl coenzyme A synthase from Moorella thermoacetica

    J. Biol. Inorg. Chem.

    (2004)
  • P. Wang et al.

    Uncovering a dynamically formed substrate access tunnel in carbon monoxide dehydrogenase/acetyl-CoA synthase

    J. Am. Chem. Soc.

    (2013)
  • A. Volbeda et al.

    Novel domain arrangement in the crystal structure of a truncated acetyl-CoA synthase from Moorella thermoacetica

    Biochemistry

    (2009)
  • W. Shin et al.

    Function and CO binding properties of the NiFe complex in carbon monoxide dehydrogenase from Clostridium thermoaceticum

    Biochemistry

    (1992)
  • D.P. Barondeau et al.

    Methylation of carbon monoxide dehydrogenase from Clostridium thermoaceticum and the mechanism of acetyl-CoA synthesis

    J. Am. Chem. Soc.

    (1997)
  • W.K. Russell et al.

    Spectroscopic, redox, and structural characterization of the Ni-labile and nonlabile forms of the acetyl-CoA synthase active site of carbon monoxide dehydrogenase

    J. Am. Chem. Soc.

    (1998)
  • C.E. Webster et al.

    Structures and energetics of models for the active site of acetyl-coenzyme a synthase: rolse of distale and proximal metals in catalysis

    J. Am. Chem. Soc.

    (2004)
  • Cited by (7)

    • The two-electron reduced A cluster in acetyl-CoA synthase: Preparation, characteristics and mechanistic implications

      2023, Journal of Inorganic Biochemistry
      Citation Excerpt :

      Square-planar geometry of Nip was found both in the structure of monomeric ACS purified from Carboxydothermus hydrogenoformans [4] and in one of the structures in [2], where the fourth ligand was tentatively assigned to a water molecule, or left unidentified, respectively. The possibility that substantial geometric changes occur at Nip during catalysis was suspected from the outset [2] and it is remarkable, but perhaps not surprising that, when geometric constraints are lifted in combined quantum mechanical (QM)/molecular mechanical (MM) approaches, the geometry-optimized structures do not always recapitulate the square-planar conformation of Nip held in the protein crystal structures [5,6]. Optimized geometries of Nip have been calculated for the A cluster bearing a variety of different ligands [5,7,8].

    • Editorial: Advances in computational molecular biophysics

      2021, Biochimica et Biophysica Acta - General Subjects
    • Engineering Clostridium ljungdahlii as the gas-fermenting cell factory for the production of biofuels and biochemicals

      2020, Current Opinion in Chemical Biology
      Citation Excerpt :

      This carbon fixation pathway consists of a methyl and carbonyl branch [5,17]. The Codh/Acs complex, responsible for the final step of the WLP, is actually a biofunctional α2β2 tetramer, in which the α- and β-subunits catalyze the formation of acetyl-CoA and the reduction of CO2 to CO, respectively [8,18,19]. Starting from acetyl-CoA, a variety of metabolites can be synthesized in C. ljungdahlii, in which the two dominant products are acetate and ethanol.

    View all citing articles on Scopus
    View full text