Hybrid Kinetic Monte Carlo/Molecular Dynamics Simulations of Bond Scissions in Proteins

Abstract

Protonation states of ionizable amino acid residues such as histidine depend on pH and define the spatial distribution of the electrostatic potential inside protein molecules. They thus play a key role in many biological processes, such as enzyme catalysis, protein dynamics and inter-and intra-molecular interactions. However, standard force fields used in molecular dynamics (MD) simulations do not allow protonation state changes, which can occur in response to a conformational change in a protein. Therefore, many so called constant pH methods for dynamic protonation have been developed [1-3], in which a proton appears or disappears on titratable amino acid residues during MD simulations. Here, we compare the pKa values obtained via our recent constant pH MD implementation to measured NMR titration curves in order to assess the accuracy of our method. We chose a set of pentapeptides with challenging charge compositions yet are small enough to ensure exhaustive sampling in our MD simulations. Both measured and calculated titration curves and pKa values agree well with each other, differing by less than 0.25 pKa units.

Collagen breaks at weak sacrificial bonds taming its mechanoradicals

“…We use our previously developed simulation scheme Ki netic M onte Carlo / M olecular Dy namics (KIM-MDY) [10] to enable bond scissions on the molecular level. In short, it takes the bond elongations from the MD simulations as input for a Bell-type model to calculate rupture rates.…”

“…Rupture propensity is determined not only by the thermodynamic strength of the bonds, as calculated above, but also by the actual force that is acting upon that bond. To directly predict relative rupture counts within a stretched collagen fibril, we utilized KIMMDY [10], a hybrid simulation scheme that invokes bond rupture in kinetic Monte Carlo steps during atomistic MD simulations. Bond rupture rates are calculated based on the bond's strength given by its BDE and its mechanical weakening by the force distribution through the 5 protein in the MD simulation.…”

“…After the aforementioned 100 ns pulling MD simulations, 5,000 bond elongations, taken from another 1 ns of simulation now using a PLUMED-patched [55] Gromacs version, per breakage candidate were averaged to generate the bond rupture rates as described previously [10]. For each of these ensembles of rates, the Kinetic Monte Carlo step was invoked 10,000 times to sample the rupture propensities properly.…”

“…In the simplest model, the Bell-Evans model, force acting on a bond exponentially increases its rate of rupture [7][8][9]. Our previous work identified crosslinks as well as adjacent backbone bonds as the most strongly loaded links [2,10]. Similarly, crosslinks have been reported as responding to stress in smaller scale molecular simulations [11] as well as in coarse-grained models [12], while recent semi-empirical calculations identified C α -C bonds of prolines at X-positions as rupture candidates [13].…”

Drug Discovery by Automated Adaptation of Chemical Structure and Identity

“…112−114 For biological systems, MC methods have been used to perform constant-pH simulations, 115 calculations of absolute binding free energies, 116 enhanced sampling of protein conformations, 116,117 and protein folding. 118,119 Furthermore, coupled MCMD has been used to study protonation dynamics, 120−124 protein bond cleavage, 125 and configurational properties of systems containing metals, 126,127 among other applications. 128 As a model system for MCMD demonstration and evaluation, we use the L99A mutant of T4 lysozyme, which contains an internal ligand-binding pocket.…”