How the Destabilization of a Reaction Intermediate Affects Enzymatic Efficiency: The Case of Human Transketolase

Abstract

Atomic resolution X-ray crystallography has shown that an intermediate (the X5P-ThDP adduct) of the catalytic cycle of transketolase (TK) displays a significant, putatively highly energetic, out-of-plane distortion in a sp ² carbon adjacent to a lytic bond, suggested to lower the barrier of the subsequent step, and thus was postulated to embody a clear-cut demonstration of the intermediate destabilization effect. The lytic bond of the subsequent rate-limiting step was very elongated in the X-ray structure (1.61 Å), which was proposed to be a consequence of the out-of-plane distortion. Here we use high-level QM and QM/MM calculations to study the intermediate destabilization effect. We show that the intrinsic energy penalty for the observed distortion is small (0.2 kcal·mol^-1) and that the establishment of a favorable hydrogen bond within X5P-ThDP, instead of enzyme steric strain, was found to be the main cause for the distortion. As the net energetic effect of the distortion is small, the establishment of the internal hydrogen bond (-0.6 kcal·mol^-1) offsets the associated penalty. This makes the distorted structure more stable than the nondistorted one. Even though the energy contributions determined here are close to the accuracy of the computational methods in estimating penalties for geometric distortions, our data show that the intermediate destabilization effect provides a small contribution to the observed reaction rate and does not represent a catalytic effect that justifies the many orders of magnitude which enzymes accelerate reaction rates. The results help to understand the intrinsic enzymatic machinery behind enzyme's amazing proficiency.

Scheme 1. Catalytic Mechanism of the Human Transketolase

Figure 1

Covalent X5P-ThDP adduct in the high-resolution hTK X-ray structure 4KXV. The atoms highlighted mark the dihedral angle whose distortion brings C2x out-of-plane. The bond to be broken in the subsequent step involves the C2x and the C3x atoms. The black triangles pointing to the atoms represent the rest of the adduct X5P-ThDP, not represented in the figure for simplicity.

Figure 2

QM/MM model adopted starting from X-ray PDB code 4KXV. (Left) The entire protein, treated at the MM level of theory (chain A in cyan and chain B in red), and the DFT region are depicted on the left and on the right, respectively. (Right) Amino acid residues and water molecules are depicted in ball and stick while the adduct ThDP-X5P is represented in stick.

Figure 3

Small molecular models used to investigate the origin of the abnormal C2x–C3x bond length. Species A represents the ThDP-X5P adduct. Species B and C represent analogues of the same adduct but without the electron-withdrawing inductive/mesomeric/both effects on C2x and C3x.

Figure 4

(Top left) Gibbs free energy profile for the conversion of X5P into G3P. INT1 is the lowest free energy state of this stage, and very close to the absolute free energy minimum of the whole cycle (at 2.3 kcal·mol^–1 from INT4, according to an earlier study) that, according to the energy span model, makes the reaction rate to depend on this state. Therefore, preventing INT1 from becoming too stable would have a catalytic effect in the overall reaction rate. Remaining panels: the geometry of the stationary states through the reaction of formation of the X5P-ThDP covalent intermediate and its transformation into G3P. Distances (Å) are reported in black and in red for bonds and geometrical parameters of hydrogen bond, respectively. The imaginary frequency values (cm^–1) for TS1 and TS2 are reported.

Figure 5

Out-of-plane distortion of the C2–C2x bond in the two used models. The models show a 9°–12° distortion when optimized at the B3LYP/6-31G(d). The ThDP-X5P system is free to rotate around the C2–C2x bond and avoid the repulsion between O2 and N4′ without clashing with the (absent) protein scaffold, eliminating the out-of-plane distortion, but the distorted geometry is still more favorable because it allows for the establishment of an internal O2–N4′ hydrogen bond. For clarity, residues retained during the QMMM calculations and hydrogen have not been shown. On the right, the superposition of the two models (in blue, the QMMM adduct).

Figure 6

Energy profile as a function of out-of-plane distortion of the C2–C2x bond, from 0° to 22°, in the QM/MM model (blue), in the ThDP-X5P adduct (green), and in the same model but with the N4′ group replaced by hydrogen (red), which eliminates the putative clash and the internal hydrogen bond.

Figure 7

C2–C2x, C2x–C3x, and O2–N4’ distances as a function of the C5–S–C2-C2x dihedral, from 0° to 22° calculated at the B3LYP/6-31G(d) level.