Protein-Ligand Binding Free-Energy Calculations with ARROW─A Purely First-Principles Parameterized Polarizable Force Field

Protein solvation: Site-specific hydrophilicity, hydrophobicity, counter ions, and interaction entropy

Solvation free energy is a driving force that plays an important role in the stability of biomolecular conformations. Currently, the implicit solvent model is widely used to calculate solvation energies of biomolecules such as proteins. However, for proteins, the implicit solvent calculation does not provide much detailed information since a protein is highly inhomogeneous on its surface. In this study, we develop an explicit solvent approach to protein solvation, which allows us to investigate detailed site-specific hydrophilicity and hydrophobicity, including the role of counter ions and intra-protein interactions. This approach facilitates the analysis of specific residue interactions with solvent molecules, extending the understanding of protein solubility to the energetic impacts of site-specific residue-solvent interactions. Our study showed that specific residue-solvent interactions are strongly influenced by the electrostatic environment created by its nearby residues, especially charged residues. In particular, charged residues on the protein surface are mainly responsible for the heterogeneity of the electrostatic environment of the protein surface, and they significantly affect the local distribution of water. In addition, counter ions change the local electrostatic environment and alter specific residue-water interactions. Neutral residues also interact with water, with polar residues being more prominent than nonpolar ones but contributing less to solvation energy than charged residues. This study illustrates an explicit solvent approach to protein solvation, which gives residue-specific contributions to protein solvation and provides detailed information on site-specific hydrophilicity and hydrophobicity.

Effect of Protein-Polarized Ligand Charges on Relative Protein Ligand Binding Affinities

A major challenge in computer-aided drug design is predicting relative binding energies of different molecules to a target protein using fast and accurate free-energy calculation methods. Free-energy calculations are primarily computed by utilizing classical molecular dynamics simulations based on all-atom force fields (FF) to model the interactions in the system. The present standard classical all-atom FFs contain fixed partial charges on the atoms, and hence electrostatic interactions are modeled between them. The parametrization process to determine these partial charges usually relies on quantum mechanics or semiempirical calculations of the molecule in the gas phase or homogeneous water surrounding. These present standard parametrization schemes of the partial charges neglect, therefore, polarization effects from the protein surrounding. The absence of protein polarization effects can lead to significant errors in free-energy calculations in proteins. We present a parametrization scheme for the partial charges of ligands, named protein-induced polarization (PIP) charges, which account for the electrostatic polarization due to the protein surrounding. The scheme involves single-point quantum mechanics/molecular mechanics calculations of the ligand charges in the protein/water surrounding. Using PIP ligand partial charges, we have calculated the relative binding free energies (RBFEs) of well-studied protein-ligand systems. We show here that RBFEs computed with PIP charges are either significantly improved or at least comparable to those computed with nonpolarized standard GAFF charges. Overall, we present a simple-to-use parametrization scheme to include protein polarization in any type of binding free-energy calculations. The parametrization scheme increases the accuracy in RBFE calculations, while it does not add significant computation time to standard parametrization procedures.

Quantum mechanical-based strategies in drug discovery: Finding the pace to new challenges in drug design

The expansion of the chemical space to tangible libraries containing billions of synthesizable molecules opens exciting opportunities for drug discovery, but also challenges the power of computer-aided drug design to prioritize the best candidates. This directly hits quantum mechanics (QM) methods, which provide chemically accurate properties, but subject to small-sized systems. Preserving accuracy while optimizing the computational cost is at the heart of many efforts to develop high-quality, efficient QM-based strategies, reflected in refined algorithms and computational approaches. The design of QM-tailored physics-based force fields and the coupling of QM with machine learning, in conjunction with the computing performance of supercomputing resources, will enhance the ability to use these methods in drug discovery. The challenge is formidable, but we will undoubtedly see impressive advances that will define a new era.

FeNNol: An efficient and flexible library for building force-field-enhanced neural network potentials

Neural network interatomic potentials (NNPs) have recently proven to be powerful tools to accurately model complex molecular systems while bypassing the high numerical cost of ab initio molecular dynamics simulations. In recent years, numerous advances in model architectures as well as the development of hybrid models combining machine-learning (ML) with more traditional, physically motivated, force-field interactions have considerably increased the design space of ML potentials. In this paper, we present FeNNol, a new library for building, training, and running force-field-enhanced neural network potentials. It provides a flexible and modular system for building hybrid models, allowing us to easily combine state-of-the-art embeddings with ML-parameterized physical interaction terms without the need for explicit programming. Furthermore, FeNNol leverages the automatic differentiation and just-in-time compilation features of the Jax Python library to enable fast evaluation of NNPs, shrinking the performance gap between ML potentials and standard force-fields. This is demonstrated with the popular ANI-2x model reaching simulation speeds nearly on par with the AMOEBA polarizable force-field on commodity GPUs (graphics processing units). We hope that FeNNol will facilitate the development and application of new hybrid NNP architectures for a wide range of molecular simulation problems.

The Determination of Free Energy of Hydration of Water Ions from First Principles

We model the autoionization of water by determining the free energy of hydration of the major intermediate species of water ions. We represent the smallest ions─the hydroxide ion OH-, the hydronium ion H3O+, and the Zundel ion H5O2+─by bonded models and the more extended ionic structures by strong nonbonded interactions (e.g., the Eigen H9O4+ = H3O+ + 3(H2O) and the Stoyanov H13O6+ = H5O2+ + 4(H2O)). Our models are faithful to the precise QM energies and their components to within 1% or less. Using the calculated free energies and atomization energies, we compute the pKa of pure water from first principles as a consistency check and arrive at a value within 1.3 log units of the experimental one. From these calculations, we conclude that the hydronium ion, and its hydrated state, the Eigen cation, are the dominant species in the water autoionization process.

Fourteen immunomodulatory alkaloids and two prenylated phenylpropanoids with dual therapeutic approach for COVID-19: molecular docking and dynamics studies

The pandemic outbreak of COVID-19 caused by the new severe acute respiratory syndrome coronavirus (SARS-CoV-2) is a global health burden. To date, there is no highly effective antiviral therapy to eradicate the virus; as a result, researchers are racing to introduce new potential therapeutic agents. Alternatively, traditional immunity boosters and symptomatic treatment based on natural bioactive compounds are also an option. The 3-chymotrypsin-like protease (3CLpro) crystal structure, the main proteolytic enzyme of SARS-CoV-2, has been unraveled, allowing the development of effective protease inhibitors via in silico and biological studies. In COVID-19 infected patients, the loss of lung function, and mortality are reported to be linked to several inflammatory mediators and cytokines. In this context, the approach of introducing immunomodulatory agents may be considered a dual lifesaving strategy in combination with antiviral drugs. This study aims to provide immunomodulatory natural products exhibiting potential protease inhibitory activities. Selected groups of alkaloids of different classes and two prenylated phenylpropanoids from the Brazilian green propolis were in silico screened for their ability to inhibit COVID-19 3CLpro protease. Results showed that compounds exhibited binding energy scores with values ranging from -6.96 to -3.70 compared to the reference synthetic protease inhibitor O6K with a binding energy score of -7.57. O6K binding energy was found comparable with lead phytochemicals in our study, while their toxicity and drug-likeness criteria are better than that of O6K. The activities of these molecules are mainly ascribed to their ability to form hydrogen bonding with 3CLpro crucial amino acid residues of the catalytic site. In addition, the molecular dynamics simulations further showed that some of these compounds formed stable complexes as evidenced by the occupancy fraction measurements. The study suggested that the major immunomodulators 3β, 20α-diacetamido-5α-pregnane, (20S)-(benzamido)-3β-(N,N-dimethyamino)-pregnane, and baccharin are 3CLpro inhibitors. Biological screenings of these phytochemicals will be valuable to experimentally validate and consolidate the results of this study before a rigid conclusion is reached, which may pave the way for the development of efficient modulatory bioactive compounds with dual bioactions in COVID-19 intervention. Communicated by Ramaswamy H. Sarma.

Neural Network Corrections to Intermolecular Interaction Terms of a Molecular Force Field Capture Nuclear Quantum Effects in Calculations of Liquid Thermodynamic Properties

We incorporate nuclear quantum effects (NQE) in condensed matter simulations by introducing short-range neural network (NN) corrections to the ab initio fitted molecular force field ARROW. Force field NN corrections are fitted to average interaction energies and forces of molecular dimers, which are simulated using the Path Integral Molecular Dynamics (PIMD) technique with restrained centroid positions. The NN-corrected force field allows reproduction of the NQE for computed liquid water and methane properties such as density, radial distribution function (RDF), heat of evaporation (HVAP), and solvation free energy. Accounting for NQE through molecular force field corrections circumvents the need for explicit computationally expensive PIMD simulations in accurate calculations of the properties of chemical and biological systems. The accuracy and locality of pairwise NN NQE corrections indicate that this approach could be applicable to complex heterogeneous systems, such as proteins.

Combining Force Fields and Neural Networks for an Accurate Representation of Bonded Interactions

We present a formalism of a neural network encoding bonded interactions in molecules. This intramolecular encoding is consistent with the models of intermolecular interactions previously designed by this group. Variants of the encoding fed into a corresponding neural network may be used to economically improve the representation of torsional degrees of freedom in any force field. We test the accuracy of the reproduction of the ab initio potential energy surface on a set of conformations of two dipeptides, methyl-capped ALA and ASP, in several scenarios. The encoding, either alone or in conjunction with an analytical potential, improves agreement with ab initio energies that are on par with those of other neural network-based potentials. Using the encoding and neural nets in tandem with an analytical model places the agreements firmly within "chemical accuracy" of ±0.5 kcal/mol.

Combining Force Fields and Neural Networks for an Accurate Representation of Chemically Diverse Molecular Interactions

A key goal of molecular modeling is the accurate reproduction of the true quantum mechanical potential energy of arbitrary molecular ensembles with a tractable classical approximation. The challenges are that analytical expressions found in general purpose force fields struggle to faithfully represent the intermolecular quantum potential energy surface at close distances and in strong interaction regimes; that the more accurate neural network approximations do not capture crucial physics concepts, e.g., nonadditive inductive contributions and application of electric fields; and that the ultra-accurate narrowly targeted models have difficulty generalizing to the entire chemical space. We therefore designed a hybrid wide-coverage intermolecular interaction model consisting of an analytically polarizable force field combined with a short-range neural network correction for the total intermolecular interaction energy. Here, we describe the methodology and apply the model to accurately determine the properties of water, the free energy of solvation of neutral and charged molecules, and the binding free energy of ligands to proteins. The correction is subtyped for distinct chemical species to match the underlying force field, to segment and reduce the amount of quantum training data, and to increase accuracy and computational speed. For the systems considered, the hybrid ab initio parametrized Hamiltonian reproduces the two-body dimer quantum mechanics (QM) energies to within 0.03 kcal/mol and the nonadditive many-molecule contributions to within 2%. Simulations of molecular systems using this interaction model run at speeds of several nanoseconds per day.

Selectivity Mechanism of Pyrrolopyridone Analogues Targeting Bromodomain 2 of Bromodomain-Containing Protein 4 from Molecular Dynamics Simulations

Bromodomain and extra-terminal domain (BET) proteins play an important role in epigenetic regulation and are linked to several diseases; therefore, they are interesting targets. BET has two bromodomains: bromodomain 1 (BD1) and BD2. Selective targeting of BD1 or BD2 may produce different activities and greater effects than pan-BD inhibitors. However, the selective mechanism of the specific core must be studied at the atomic level. This study determined the effectiveness of pyrrolopyridone analogues to selectively inhibit BD2 using a pan-BD inhibitor (ABBV-075) and a selective-BD2 inhibitor (ABBV-744). Molecular dynamics simulations and calculations of binding free energies were used to systematically study the selectivity of BD2 inhibition by the pyrrolopyridone analogues. Overall, the pyrrolopyridone analogue inhibitors targeting BD2 interacted mainly with the following amino acid pairs between bromodomain-containing protein 4 (BRD4)-BD1 and BRD4-BD2 complexes: I146/V439, N140/N433, D144/H437, P82/P375, V87/V380, D88/D381, and Y139/Y432. The pyrrolopyridone analogues targeting BRD4-BD2 were divided into five regions based on selectivity mechanism. These results suggest that the R3 and R5 regions of pyrrolopyridone analogues can be modified to improve the selectivity between BRD4-BD1 and BRD4-BD2. The selectivity of BD2 inhibition by pyrrolopyridone analogues can be used to design novel BD2 inhibitors based on a pyrrolopyridone core.

Predicting the Effect of Single Mutations on Protein Stability and Binding with Respect to Types of Mutations

The development of methods and algorithms to predict the effect of mutations on protein stability, protein-protein interaction, and protein-DNA/RNA binding is necessitated by the needs of protein engineering and for understanding the molecular mechanism of disease-causing variants. The vast majority of the leading methods require a database of experimentally measured folding and binding free energy changes for training. These databases are collections of experimental data taken from scientific investigations typically aimed at probing the role of particular residues on the above-mentioned thermodynamic characteristics, i.e., the mutations are not introduced at random and do not necessarily represent mutations originating from single nucleotide variants (SNV). Thus, the reported performance of the leading algorithms assessed on these databases or other limited cases may not be applicable for predicting the effect of SNVs seen in the human population. Indeed, we demonstrate that the SNVs and non-SNVs are not equally presented in the corresponding databases, and the distribution of the free energy changes is not the same. It is shown that the Pearson correlation coefficients (PCCs) of folding and binding free energy changes obtained in cases involving SNVs are smaller than for non-SNVs, indicating that caution should be used in applying them to reveal the effect of human SNVs. Furthermore, it is demonstrated that some methods are sensitive to the chemical nature of the mutations, resulting in PCCs that differ by a factor of four across chemically different mutations. All methods are found to underestimate the energy changes by roughly a factor of 2.