Catalytic mechanism and pH dependence of a methyltransferase ribozyme (MTR1) from computational enzymology

Abstract

A methyltransferase ribozyme (MTR1) was selected in vitro to catalyze alkyl transfer from exogenous O6-methylguanine (O6mG) to a target adenine N1, and recently, high-resolution crystal structures have become available. We use a combination of classical molecular dynamics, ab initio quantum mechanical/molecular mechanical (QM/MM) and alchemical free energy (AFE) simulations to elucidate the atomic-level solution mechanism of MTR1. Simulations identify an active reactant state involving protonation of C10 that hydrogen bonds with O6mG:N1. The deduced mechanism involves a stepwise mechanism with two transition states corresponding to proton transfer from C10:N3 to O6mG:N1 and rate-controlling methyl transfer (19.4 kcal·mol-1 barrier). AFE simulations predict the pKa for C10 to be 6.3, close to the experimental apparent pKa of 6.2, further implicating it as a critical general acid. The intrinsic rate derived from QM/MM simulations, together with pKa calculations, enables us to predict an activity-pH profile that agrees well with experiment. The insights gained provide further support for a putative RNA world and establish new design principles for RNA-based biochemical tools.

Graphical Abstract

Mechanistic study of a novel artificially engineered methyltransferase ribozyme (MTR1), catalyzing site-specific methylation of adenine, utilizing state-of-the-art computational tools.

Figure 1.

Active structure of MTR1 in solution. (A) Secondary structure of MTR1 colored by helical regions P1, P2 and P3 with loop regions in gray. (B) Average solution structure over the last 10 ns of simulation with O⁶mG (purple) bound. (C) Distribution of nucleophile–electrophile angle (O⁶mG:O6–Cm–A63:N1) and distance (O⁶mG:Cm–A63:N1) colored from least catalytically fit (blue) to most catalytically fit (red). Fitness scores were assigned such that the angle must be between 140° and 180° and distance between 3.50 and 2.75 Å (as denoted by the upper left box), otherwise a score of zero (blue) was given (see Supplementary Data for a detailed explanation of scoring). (D) Atomic fluctuations averaged per nucleotide, iterating over i nucleotides , derived from analysis of 5 × 100 ns MD simulations (red) and compared with fluctuations estimated from the crystallographic B-values (black) with correlation coefficient of 0.62. Helical arms and loop regions are highlighted on the horizontal axis, and shaded regions indicate crystal contacts. (E) Stereo view of the active site nucleotides of MTR1 in solution with coloring corresponding to panels (A) and (B).

Figure 2.

Results of ab initio QM/MM simulation of the reaction trajectory of MTR1 departing from the active state in solution. (A) 2D free energy landscape from ab initio QM/MM simulations with converged, stepwise path overlaid (black) and stationary points marked with crosses. (B) Free energy profiles (46) corresponding to the path indicated in panel (A) shown in black and resulting from the 1D reaction coordinate of methyl transfer in the absence of C10 protonation shown in gray. (C) Schematic of the active site depicting the transition states and corresponding mechanistic steps. (D) Average transition state structures calculated from additional 5 ps sampling at the stationary points found on the minimum free energy path. Bond lengths (Å) corresponding to reaction coordinates are indicated.

Figure 3.

Analysis of methyl transfer distances (Å) in the context of the free energy profile (black) from ab initio QM/MM simulations. The distances within O⁶mG:O6–A63:N1 (red), Cm–A63:N1 (blue) and O⁶mG:O6–Cm (green) were measured for the trajectory average structures of each umbrella window on the converged reaction path. Statistical values such as 95% confidence intervals (shaded), standard deviations (dark error bars) and range of observed values (vertical lines) are traced through the trajectories for each umbrella window. The distances in the crystal structure of the product state are indicated by dashed lines.

Figure 4.

Prediction of activity–pH profiles for C10H⁺ (WT) based on the free energy barrier from QM/MM and pK_a shift from AFE simulations. (A) Free energy profiles corresponding to QM/MM simulations of WT C10H⁺ (black) and C10U mutation (blue). (B) Predicted (black) and experimental (red) WT activity–pH profile calculated using the two-state model shown in Equation (3). Red points and error bars are those of Deng et al. (32). Dotted lines indicate the predicted and experimental C10 pK_a values, and the lower pK_a of 5.0 from experiment is assumed. The calculated C10U mutant results were below the detection limit of 10⁻⁶ min⁻¹ (in accord with experiment) and hence not visible on the scale of the plot. Computational error analysis of the free energy profiles is given in Supplementary Figure S6.