Challenges in Protein QM/MM Simulations with Intra-Backbone Link Atoms

Abstract

Hybrid quantum mechanical/molecular mechanical (QM/MM) simulations fuel discoveries in many fields of science including computational biochemistry and enzymology. Development of more convenient tools leads to an increase in the number of works in which mechanical insights into enzymes’ mode of operation are obtained. Most commonly, these tools feature hydrogen-capping (link atom) approach to provide coupling between QM and MM subsystems across a covalent bond. Extensive studies were conducted to provide a solid foundation for the correctness of such an approach when a bond to a nonpolar MM atom is considered. However, not every task may be accomplished this way. Certain scenarios of using QM/MM in computational enzymology encourage or even necessitate the incorporation of backbone atoms into the QM region. Two out of three backbone atoms are polar, and in QM/MM with electrostatic embedding, a neighboring link atom will be hyperpolarized. Several schemes to mitigate this effect were previously proposed alongside a rigorous assessment of quantitative effects on model systems. However, it was not clear whether they may translate into qualitatively different results and how link atom hyperpolarization may manifest itself in a real-life enzymological scenario. Here, we show that the consequences of such an artifact may be severe and may completely overturn the conclusions drawn from the simulations. Our case advocates for the use of charge redistribution schemes whenever intra-backbone QM/MM boundaries are considered. Moreover, we addressed how different boundary types and charge redistribution schemes influence backbone dynamics. We showed that the results are heavily dependent on which boundary MM terms are retained, with charge alteration being of secondary importance. In the worst case, only three intra-backbone boundaries may be used with relative confidence in the adequacy of resulting simulations, irrespective of the hyperpolarization mitigation scheme. Thus, advances in the field are certainly needed to fuel new discoveries. As of now, we believe that issues raised in this work might encourage authors in the field to report what boundaries, boundary MM terms, and charge redistribution schemes they are using, so their results may be correctly interpreted.

Analytic Gradients for the Electrostatic Embedding QM/MM Model in Periodic Boundary Conditions Using Particle-Mesh Ewald Sums and Electrostatic Potential Fitted Charge Operators

“…As visible from the figure, we could not compute the free-energy profile at the QM/MM level in the interval between 0 and 70°. In such cases, we observed an overpolarization of the quantum density around the frontier region, which is a well-known limitation of the link-atom approach applied to the backbone of peptides . Consequently, as the experimental data reported in the literature, refer to the cis -to-trans isomerization, a direct comparison with the experiment is currently not possible, as we could not assess the relative free energy of the cis isomer.…”

“…In such cases, we observed an overpolarization of the quantum density around the frontier region, which is a wellknown limitation of the link-atom approach applied to the backbone of peptides. 119 Consequently, as the experimental data reported in the literature, 120 refer to the cis-to-trans isomerization, a direct comparison with the experiment is currently not possible, as we could not assess the relative free energy of the cis isomer. However, from the comparison with the profile computed at the MM level, which in turn is in good agreement with the experiments, it is possible to see that the ab initio calculations correctly reproduce the prediction on the position and height of the trans-to-cis barrier obtained with modified parameters (ΔΔG ⧧ = 0.17 ± 0.08 kcal•mol −1 ), thus suggesting that the QM/MM model is appropriate to predict the isomerization behavior.…”

Dynamic interchange between two protonation states is characteristic of active sites of cholinesterases

“…To prevent link atom hyperpolarization, a “charge shifting” scheme was used. The effect of non‐integer charge of the QM system on PME was alleviated with the Amber scheme as discussed before (Zlobin et al, 2023 ). The QM region was described with DFTB3 (Gaus et al, 2012 ) with 3ob‐3‐1 parameter set (Gaus et al, 2013 ) similarly to what was used before in various enzymological tasks (García‐Meseguer et al, 2023 ; Mokrushina et al, 2020 ; Roston et al, 2019 ; Terekhov et al, 2020 ; Zlobin and Golovin, 2022 ; A. Zlobin et al, 2018 ; A.S. Zlobin et al, 2018 ).…”

Computer Modeling of the Mechanisms of Enzymatic Reactions: Lessons From 20 Years of Practice

“…Это связано с тем, что некоторые молекулярные фрагменты, участвующие в химической реакции -аминокислотные остатки, являющиеся частью белковой макромолекулы. Границы КМи ММ-частей рекомендуется проводить по наименее поляризованным в этих системах связям С-С, желательно, алифатическим [34]. Например, если в КМ-часть необходимо добавить боковую цепь какого-либо аминокислотного остатка, то разумно провести границу между С α -и С β -атомами.…”