Reductive Cleavage Mechanism of Co−C Bond in Cobalamin

Reductive Cleavage Mechanism of Co−C Bond in Cobalamin-Dependent Methionine Synthase ... Fax: (502) 852-8149. e-mail: [email protected]., †...
0 downloads 0 Views 3MB Size
J. Phys. Chem. B 2010, 114, 12965–12971

12965

Reductive Cleavage Mechanism of Co-C Bond in Cobalamin-Dependent Methionine Synthase Mercedes Alfonso-Prieto,† Xevi Biarne´s, Manoj Kumar,§ Carme Rovira,†,‡,⊥ and Pawel M. Kozlowski*,§ Computer Simulation and Modeling Laboratory (CoSMoLab), Parc Científic de Barcelona, Baldiri Reixac 10-12, 08028 Barcelona, Spain, Institut de Quı´mica Teo`rica i Computacional (IQTCUB), Department of Chemistry, UniVersity of LouisVille, LouisVille, Kentucky 40292, Institucio´ Catalana de Recerca i Estudis AVanc¸ats (ICREA), Passeig Lluı´s Companys 23, 08018 Barcelona, Spain ReceiVed: May 13, 2010; ReVised Manuscript ReceiVed: July 26, 2010

The key step in the catalytic cycle of methionine synthase (MetH) is the transfer of a methyl group from the methylcobalamin (MeCbl) cofactor to homocysteine (Hcy). This mechanism has been traditionally viewed as an SN2-type reaction, but a different mechanism based on one-electron reduction of the cofactor (reductive cleavage) has been recently proposed. In this work, we analyze whether this mechanism is plausible from a theoretical point of view. By means of a combination of gas-phase as well as hybrid QM/MM calculations, we show that cleavage of the Co-C bond in a MeCbl · · · Hcy complex (Hcy ) methylthiolate substrate (MeS-), a structural mimic of deprotonated homocysteine) proceeds via a [CoIII(corrin•-)]-Me · · · •S-Me diradical configuration, involving electron transfer (ET) from a π*corrin-type state to a σ*Co-C one, and the methyl transfer displays an energy barrier e8.5 kcal/mol. This value is comparable to the one previously computed for the alternative SN2 reaction pathway (10.5 kcal/mol). However, the ET-based reductive cleavage pathway does not impose specific geometrical and distance constraints with respect to substrate and cofactor, as does the SN2 pathway. This might be advantageous from the enzymatic point of view because in that case, a methyl group can be transferred efficiently at longer distances. 1. Introduction B12-dependent enzymes catalyze a variety of complex chemical transformations in which breaking and forming a cobaltcarbon (Co-C) bond constitutes the key step.1-18 There are two cofactors (Figure 1) employed by B12-dependent enzymes, which support distinctive chemical pathways: enzymes utilizing adenosylcobalamin (AdoCbl) catalyze organic rearrangement reactions,7,8,11,12 in which the first step involves homolytic cleavage of the Co-C bond, whereas enzymes employing methylcobalamin (MeCbl) catalyze methyl transfer reactions,9,15,17 in which the Co-C bond is formally cleaved in a heterolytic manner.19 Among different corrinoid-dependent methyltransferases, methionine synthase (MetH) represents the most well studied enzyme that catalyzes the transfer of a methyl group from methyltetrahydrofolate (CH3-H4folate) to homocysteine (Hcy), forming methionine (Scheme 1). The key step in the catalytic cycle of MetH is the transfer of a methyl group from the MeCbl cofactor to the Hcy. It is generally believed that the enzyme operates via an SN2-type nucleophilic displacement9 (Figure 2), in which the S atom of Hcy attacks the methyl group, leading to the heterolytic cleavage of the Co-C bond. However, a related mechanism, referred to as reductiVe cleaVage (Figure 2) has been recently suggested.20 According to this mechanism, the cofactor is initially reduced by transfer of an electron from the deprotonated Hcy, forming a π-corrin based radical, followed by Co-C bond cleavage. In * Corresponding author. Phone: (502) 852-6609. Fax: (502) 852-8149. e-mail: [email protected]. † Parc Cientific de Barcelona. ‡ IQTCUB. § University of Louisville. ⊥ ICREA.

Figure 1. (a) Molecular structure of B12 cofactors (R ) Me or Ado) R1 ) CH2CONH2, R2 ) CH2CH2CONH2, R3 ) (CH2)2CONHCH2CH(CH3)OPO3-. (b) Structural model used in the present work.

fact, electrochemical experiments on methylcobalamin21-24 and gas-phase DFT calculations20 have demonstrated that one electron reduction of the cofactor weakens the Co-C bond due

10.1021/jp1043738  2010 American Chemical Society Published on Web 09/20/2010

12966

J. Phys. Chem. B, Vol. 114, No. 40, 2010

SCHEME 1

Alfonso-Prieto et al. SCHEME 2

SCHEME 3

to the effective population of the antibonding cobalt-carbon (σ*Co-C) orbital. However, previous calculations20 were performed using gas-phase models and in the absence of the Hcy. Therefore it is still to be demonstrated theoretically whether the Hcy can act as a reducing agent to cause the one-electron reduction of the cofactor and whether this mechanism is feasible inside the enzyme. Support for the reductive cleavage of the Co-C bond also comes from the recent computational studies on the related AdoCbl-dependent mutases, where an active site residue (namely, tyrosine) reduces the cofactor via protoncoupled electron transfer.25,26 However, it has also been argued that the biological reducing agents might not be strong enough to reduce the cofactor in MetH.27,28 Therefore, the feasibility of the reductive cleavage mechanism remains to be proved. In the present contribution, the cleavage of the Co-C bond in the MeCbl-dependent MetH is further investigated considering the Hcy substrate explicitly, modeled as Me-S-. It will be shown that Hcy can induce the formation of a diradical species [CoIII(corrin•-)]-Me · · · •S-Me, which preactivates the reactant complex and facilitates the Co-C bond dissociation. It will also be shown that the same diradical species is formed inside the enzyme. 2. Computational Details 2.1. Initial Models. Gas-phase calculations of the Co-C bond cleavage were performed on a model of the cofactor, including the corrin ring and the methyl ligand (Figure 1b). The axial base was not included, since previous experimental24 and theoretical20 studies have demonstrated that the reductive cleavage of Co-C bond preferably occurs from the base-off

mode. Moreover, it is known that the displacement of the axial base in the one-electron reduced cofactor (i.e., for an anion radical corrin system, that is involved in the diradical species [CoIII(corrin•-)]-Me · · · •S-Me) is easier than in the case of the neutral cofactor.20 For that very reason, we explored the cleavage of the Co-C bond in the base-off structural complex (Scheme 2). The structure of the corrin ring and the methyl group were taken from our previous work at the TD-DFT/BP86/6-31G(d) level of theory.20 To investigate how the presence of the substrate affects the properties of the Co-C bond, another model was built by placing a methylthiolate molecule (-S-Me) (a structural mimic of Hcy) above the methyl group, with the S atom in line with the Co-C bond (Scheme 3). Quantum Mechanics/ Molecular Mechanics (QM/MM) calculations of the enzyme were based on the best available X-ray structure of the MeCbl binding domain of MetH (PDB code: 1BMT, at 3 Å resolution).29 This structure includes the entire cofactor, but only one crystal water molecule. The protein was solvated in a 128.8 × 128.1 × 104.37 Å3 box of waters, and the system was equilibrated with classical MD before the QM/ MM simulations were initiated. The AMBER force field30 was used for the protein, and parameters for the MeCbl cofactor were adapted from Marques and Brown.31 The following procedure was used to obtain the initial structure of the

Figure 2. SN2 vs reductive cleavage mechanisms for Co-C bond dissociation in cobalamin-dependent methionine synthase (MetH).

Co-C Bond Cleavage Mechanism

J. Phys. Chem. B, Vol. 114, No. 40, 2010 12967

Figure 3. Spin density distribution with respect to the Co-C distance (in Å) in MeCbl•-, modeled as [CoIII(corrin)-Me]•-. Spin isodensity surfaces at 0.005 e Å-3 are plotted in orange. The hydrogen atoms of the cofactor are omitted for generating a better view.

enzyme-substrate complex. Since there is no structural information available for the protein-protein complex between the MeCbl and Hcy binding domains, the Hcy was manually inserted into the QM/MM optimized structure of the enzyme (see section 2.3 for details of the QM/MM calculations), following the same criteria as in gas-phase models (i.e., the substrate is placed in a cavity above the methyl group, with the S atom at ∼4.00 Å distance from the methyl carbon atom). The enzyme-substrate complex was then subjected to a 20 ps classical MD in which the C-S distance was fixed to accommodate the substrate in the cavity. The charges of the water molecules within 5 Å of the corrin-substrate complex were set to zero to mimic the hydrophobic active site formed in the actual MeCbl-Hcy protein-protein complex.32 2.2. Gas-Phase Calculations. Gas-phase calculations were carried out within the DFT-based Car-Parrinello molecular dynamics (CPMD) framework.33 The Kohn-Sham orbitals were expanded in a plane wave (PW) with a kinetic energy cutoff of 70 Ry, as previously used in cobalt-containing macrocycles.34,35 Periodic boundary conditions (PBC) were used, except for the calculations on charged systems, which were performed on an isolated cell, using the Tuckerman and Martina’s Poisson solver.36 Only valence electrons were explicitly included in the computations, and their interaction with the ionic cores was described by norm-conserving, ab initio pseudopotentials generated following the recipe of Troullier and Martins.37 The pseudopotential for Co was supplemented with nonlinear core corrections38 to improve its transferability with respect to spin state energies. Several theoretical reports dealing with BDE of Co-C bond in B12 cofactors have shown that the energy required to cleave this bond depends mainly on the type of functional used for the computations rather than on the structural simplifications of the cofactor.39,40 The B3LYP functional41 tends to underestimate the BDE by ∼10 kcal/mol, whereas BP8642,43 gives fairly reasonable values in the case of B12-dependent systems. For this reason, the BP86 functional was used in all computations reported in this work. Structure optimizations were performed by means of molecular dynamics with scaling of the nuclear velocities, using a time step of 3 au and a fictitious mass of the CP Lagrangian of 850 au. 2.3. QM/MM Calculations. QM/MM calculations were performed using the method developed by Laio et al.,44 which combines the first-principles MD method of Car and Parrinello33 with a force-field MD methodology (i.e., QM/MM CPMD). In this approach, the system is partitioned into QM and MM

regions. The dynamics of the atoms in the QM region depends on the electronic density, F(r), computed with density functional theory (DFT), whereas the dynamics of the atoms in the MM region is ruled by an empirical force field. The QM/MM interface is modeled by the use of capping hydrogen atoms that saturate the QM region. The electrostatic interactions between the QM and MM regions are handled via a fully Hamiltonian coupling scheme,44 in which the short-range electrostatic interactions between the QM and the MM regions were explicitly taken into account for all atoms. An appropriately modified Coulomb potential is used to ensure that no unphysical escape of the electronic density from the QM to the MM region occurs.44 The electrostatic interactions with the more distant MM atoms are treated via a multipole expansion. Bonded and van der Waals interactions between the QM and the MM regions are treated using the standard Amber force field.30 Long-range electrostatic interactions between MM atoms are described with P3M implementation,45 using a 64 × 64 × 64 mesh. An accurate description of energetic, dynamic, and structural features of biological systems has been previously obtained with this methodology, confirming its reliability.46,47 The QM region was confined in an isolated supercell of size 17 × 17 × 15 Å (enzyme) or 17 × 17 × 18 Å (enzyme-substrate complex). Figure S1 of the Supporting Information shows the QM-MM partition used. The QM region includes the cofactor with part of its side chains as well as the axial imidazole. The total number of atoms treated quantum mechanically is 106 and 111 atoms for the calculations with and without substrate, respectively. 3. Results 3.1. Reductive Cleavage of the Co-C Bond. We previously demonstrated that one-electron reduction of isolated MeCbl reduces the BDE of the Co-C bond significantly20 because the cleavage of the Co-C bond occurs efficiently from the [CoIII(corrin•-)]-Me configuration (Scheme 2). This can be rationalized by crossing between the energy of a half-filled π*corrin orbital and that of a σ*Co-C orbital upon Co-C stretching.20 As a first step in our investigation, we checked that our current computational setup (CPMD/BP86/PW) gives the same results as our previous TD-DFT/BP86/6-31 g(d) calculations. Indeed, the optimized structure of [CoIII(corrin•-)]-Me and the energy profile with respect to Co-C dissociation are very similar to the one obtained previously (Figure S2).

12968

J. Phys. Chem. B, Vol. 114, No. 40, 2010

Alfonso-Prieto et al.

Figure 4. Spin density distribution with respect to the Co-C distance (in Å) in MeCbl · · · S, modeled as [CoIII(corrin)]-Me · · · S-Me. Spin isodensity surfaces at 0.005 e Å-3 are plotted in orange. The hydrogen atoms of the cofactor are omitted for clarity.

Figure 3 illustrates the changes in electronic structure upon stretching the Co-C bond. For Co-C bond lengths close to the equilibrium value, the extra electron is located on the corrin ring (i.e., on the LUMO π*corrin orbital). However, as the Co-C bond lengthens (Co-C ∼ 2.20 Å), the electron relocates to a σ*Co-C orbital and the cobalt changes oxidation state from III to II, forming [CoII(corrin)]-CH3, that is characterized by a three-electron (σ)2(σ*)1 Co-C bond. At Co-C ∼ 3.0 Å, the spin density on Co almost vanishes, and a methyl radical starts to form, consistent with the change to a CoI(corrin)] · · · Me• configuration.20 Therefore, as we previously observed,20 cleavage of the Co-C bond in the reduced cofactor proceeds via transfer from a π*corrin-type state to a σ*Co-C one. 3.2. Reductive Cleavage of Co-C Bond in the Presence of the Hcy. To elucidate whether the Hcy could act as a reducing agent (i.e., whether electron transfer from the HOMO of the substrate to the LUMO of the cofactor can be feasible or not), we repeated the calculations of Co-C cleavage in the presence of methylthiolate (-S-Me), which mimics the Hcy (it is expected that the Hcy is unprotonated in the E · S complex).9 The methylthiolate molecule was placed over the corrin ring, with its S atom aligned with the Co-C bond (Scheme 3), and the Co-C bond was systematically stretched. All degrees of freedom were optimized for every constrained Co-C bond distance. In the first step, to evaluate the effect of a proximal reducing agent on the dissociation energy, formation of the products (i.e., the C-S covalent bond; see Scheme 3) was avoided by fixing the methyl · · · thiolate (C · · · S) distance at 4.00 Å. The evolution of the spin density distribution of the system with the Co-C bond distance (Figure S3) is very similar to the one observed for the reduced cofactor (Figure 3). The energy required for Co-C bond cleavage is approximately equal to the energy required for Co-C bond cleavage when an external electron is added to the system. Therefore, we conclude that the reductive cleavage of the Co-C bond can take place in the presence of the methylthiolate. In the second step, we modeled the complete reaction of methyl transfer from the cofactor to the Hcy. To avoid the steric

clashes between the substrate and the cofactor, the distance between the cofactor and the substrate (Co · · · S in Scheme 3) was kept fixed at 6.00 Å. Although there is no structural information available regarding how the two protein domains containing the MeCbl cofactor and the Hcy substrate interact during the methyl transfer reaction, in the case of a similar type of methyl transfer reactions catalyzed by corrinoid iron sulfur proteins,48 it was suggested that the SN2 type reaction should occur at a distance of 3.00-4.00 Å, but the manual docking of two modules could not be achieved at a distance below 8.00 Å. In the case of methylcobalamin, DFT computations have predicted a distance of 2.80 Å for an SN2-type methyl transfer;49 thus, we expect that the proper interaction of the two modules in MetH should occur around a distance of 6.00 Å. Within the reductive cleavage mechanism, it is not necessary that the two moieties should be present in close proximity, as would be required in an SN2 mechanism. It is also important to note that the calculations do not impose a particular type of mechanism to take place (reductive cleavage or SN2); rather, the system evolves according to the lowest energy state, either an open-shell singlet (reductive cleavage) or closed-shell singlet (SN2), at each value of the Co-C distance. Figure 4 shows the spin density distribution associated with the cleavage of the Co-C bond (all degrees of freedom were optimized for every constrained Co-C bond distance). When the Co-C bond length is close to the equilibrium value (∼2.01 Å), the spin density is centered on a π*corrin orbital and on a S(p) orbital. Therefore, one electron transfer from the HOMO of the thiolate to the LUMO of the corrin takes place, leading to a [CoIII(corrin•-)]-Me · · · •S-Me diradical configuration. This is the same spin distribution as previously observed for the reduced cofactor (Figure 3). As the Co-C bond is stretched, the spin density shifts from the corrin plane into the axial Co-C bond, forming [CoII(corrin)]-Me · · · •S-Me (Figure 4). At Co-C ) 2.82 Å, the carbon atom attains its maximum spin density, with only some residual spin on the cobalt, i.e. [CoI(corrin)] · · · Me• · · · •S-Me. In other words, the Co-C bond cleaves homolytically, leading

Co-C Bond Cleavage Mechanism

J. Phys. Chem. B, Vol. 114, No. 40, 2010 12969 the products of the reaction, [CoI(corrin)] and Me-S-Me (Figure 4). The energy barrier for the methyl transfer turns out to be 8.5 kcal/mol (Figure 5). This value should be considered as an upper bound, since zero point energy corrections were not included in the calculations. 3.4. QM/MM Calculations of the E · S Complex. To investigate whether the reductive cleavage mechanism could be feasible in the presence of a protein environment, it is necessary to demonstrate that the electron transfer from the substrate to the cofactor also occurs in the protein. Therefore, we included the enzyme in the calculations via the QM/MM approach. It should be pointed out that MetH is a modular machine involving interaction among four different domains with one of the steps of catalytic reaction involving the interaction between MeCbl and Hcy binding domains. Although, the X-ray crystal structure of individual domains has been well characterized,15 yet the crystal structure of the whole enzyme has not been resolved successfully, mainly owing to the very high degree of conformational flexibility affiliated with this enzyme. As a result, there is no structural information available that could be used as a guiding tool to prepare a structural model for studying the methylation of Hcy substrate. Taking into account that we are interested only in studying the role of the substrate as ET donor, it seems appropriate to place the methylthiolate, a structural mimic of the homocysteine substrate, over the upper face of the cofactor and optimize the complex. Figure 6 shows the optimized structure of the E · S complex (MetH-MeCbl · S) and its main structural parameters are listed in Table 1. The values obtained for the enzyme-cofactor system (MetH-MeCbl) and the isolated cofactor (MeCbl), computed with the same methodology, are also listed for comparison. In both cases (MetH-MeCbl and MeCbl), the computed values are in good agreement with the experimental crystal structures.29,50 They also reproduce the tendency that the Co-C bond distance is not affected by the presence of the enzyme,29 whereas the Co-Nax bond slightly lengthens. The fact that the Co-Nax bond is much weaker than the Co-C bond (32 vs 7 kcal/mol)34 probably explains the small deviation (0.03 Å) between the computed and experimental values in the enzyme-cofactor system. It has also been pointed out that this bond is very sensitive to changes in its local environment.35 As shown in Table 1, the presence of the substrate hardly affects the structure of the coenzyme as the structural parameters obtained for MetH-MeCbl are very similar to those for MetH-MeCbl · S, implying that the structural complex involving the reduced cofactor does not show major structural changes. In contrast, the substrate does induce a significant change in the electronic structure of the cofactor, via one-electron trans-

Figure 5. Energy barrier for the methyl transfer reaction in MeCbl · · · S shown in Figure 4.

Figure 6. QM/MM optimized structure of the E · S complex. QM atoms are shown in ball-and-stick representation. The spin density distribution on the QM region is shown in the inset.

to the formation of a methyl radical. Further elongation of the Co-C bond (distances >3.10 Å) leads to a closed-shell configuration, [CoI(corrin)] · · · Me-S-Me, due to the formation of the Me-S covalent bond. Finally, at Co-C ) 4.20 Å, the methyl group is completely transferred to the substrate, yielding

TABLE 1: Key Structural Parameters of the Computationally Optimized Systems Investigated (see also Figures 6 and S4), in Comparison with Experimental Dataa MeCbl

a

MetH-MeCbl

MetH-MeCbl · · · -S-Me

parameter

calcd34

X-ray50

calcd

X-ray29

calcd

Co-C Co-Nax Co-Nc Nc-C C1-C2