A quantum chemical interaction energy dataset for accurately modeling protein-ligand interactions.

Fast and accurate calculation of intermolecular interaction energies is desirable for understanding many chemical and biological processes, including the binding of small molecules to proteins. The Splinter ["Symmetry-adapted perturbation theory (SAPT0) protein-ligand interaction"] dataset has been created to facilitate the development and improvement of methods for performing such calculations. Molecular fragments representing commonly found substructures in proteins and small-molecule ligands were paired into >9000 unique dimers, assembled into numerous configurations using an approach designed to adequately cover the breadth of the dimers' potential energy surfaces while enhancing sampling in favorable regions. ~1.5 million configurations of these dimers were randomly generated, and a structurally diverse subset of these were minimized to obtain an additional ~80 thousand local and global minima. For all >1.6 million configurations, SAPT0 calculations were performed with two basis sets to complete the dataset. It is expected that Splinter will be a useful benchmark dataset for training and testing various methods for the calculation of intermolecular interaction energies.

Discovery of a Partial Glucokinase Activator Clinical Candidate: Diethyl ((3-(3-((5-(Azetidine-1-carbonyl)pyrazin-2-yl)oxy)-5-isopropoxybenzamido)-1H-pyrazol-1-yl)methyl)phosphonate (BMS-820132).

Glucokinase (GK) is a key regulator of glucose homeostasis, and its small-molecule activators represent a promising opportunity for the treatment of type 2 diabetes. Several GK activators have been advanced into clinical trials and have demonstrated promising efficacy; however, hypoglycemia represents a key risk for this mechanism. In an effort to mitigate this hypoglycemia risk while maintaining the efficacy of the GK mechanism, we have investigated a series of amino heteroaryl phosphonate benzamides as ''partial" GK activators. The structure-activity relationship studies starting from a "full GK activator" 11, which culminated in the discovery of the "partial GK activator" 31 (BMS-820132), are discussed. The synthesis and in vitro and in vivo preclinical pharmacology profiles of 31 and its pharmacokinetics (PK) are described. Based on its promising in vivo efficacy and preclinical ADME and safety profiles, 31 was advanced into human clinical trials.

CLIFF: A component-based, machine-learned, intermolecular force field.

Computation of intermolecular interactions is a challenge in drug discovery because accurate ab initio techniques are too computationally expensive to be routinely applied to drug-protein models. Classical force fields are more computationally feasible, and force fields designed to match symmetry adapted perturbation theory (SAPT) interaction energies can remain accurate in this context. Unfortunately, the application of such force fields is complicated by the laborious parameterization required for computations on new molecules. Here, we introduce the component-based machine-learned intermolecular force field (CLIFF), which combines accurate, physics-based equations for intermolecular interaction energies with machine-learning models to enable automatic parameterization. The CLIFF uses functional forms corresponding to electrostatic, exchange-repulsion, induction/polarization, and London dispersion components in SAPT. Molecule-independent parameters are fit with respect to SAPT2+(3)δMP2/aug-cc-pVTZ, and molecule-dependent atomic parameters (atomic widths, atomic multipoles, and Hirshfeld ratios) are obtained from machine learning models developed for C, N, O, H, S, F, Cl, and Br. The CLIFF achieves mean absolute errors (MAEs) no worse than 0.70 kcal mol-1 in both total and component energies across a diverse dimer test set. For the side chain-side chain interaction database derived from protein fragments, the CLIFF produces total interaction energies with an MAE of 0.27 kcal mol-1 with respect to reference data, outperforming similar and even more expensive methods. In applications to a set of model drug-protein interactions, the CLIFF is able to accurately rank-order ligand binding strengths and achieves less than 10% error with respect to SAPT reference values for most complexes.

Electron-Passing Neural Networks for Atomic Charge Prediction in Systems with Arbitrary Molecular Charge.

Atomic charges are critical quantities in molecular mechanics and molecular dynamics, but obtaining these quantities requires heuristic choices based on atom typing or relatively expensive quantum mechanical computations to generate a density to be partitioned. Most machine learning efforts in this domain ignore total molecular charges, relying on overfitting and arbitrary rescaling in order to match the total system charge. Here, we introduce the electron-passing neural network (EPNN), a fast, accurate neural network atomic charge partitioning model that conserves total molecular charge by construction. EPNNs predict atomic charges very similar to those obtained by partitioning quantum mechanical densities but at such a small fraction of the cost that they can be easily computed for large biomolecules. Charges from this method may be used directly for molecular mechanics, as features for cheminformatics, or as input to any neural network potential.

Discovery and structure activity relationships of 7-benzyl triazolopyridines as stable, selective, and reversible inhibitors of myeloperoxidase.

Myeloperoxidase (MPO) is a heme peroxidase found in neutrophils, monocytes and macrophages that efficiently catalyzes the oxidation of endogenous chloride into hypochlorous acid for antimicrobial activity. Chronic MPO activation can lead to indiscriminate protein modification causing tissue damage, and has been associated with chronic inflammatory diseases, atherosclerosis, and acute cardiovascular events. Triazolopyrimidine 5 is a reversible MPO inhibitor; however it suffers from poor stability in acid, and is an irreversible inhibitor of the DNA repair protein methyl guanine methyl transferase (MGMT). Structure-based drug design was employed to discover benzyl triazolopyridines with improved MPO potency, as well as acid stability, no reactivity with MGMT, and selectivity against thyroid peroxidase (TPO). Structure-activity relationships, a crystal structure of the MPO-inhibitor complex, and acute in vivo pharmacodynamic data are described herein.

AP-Net: An atomic-pairwise neural network for smooth and transferable interaction potentials.

Intermolecular interactions are critical to many chemical phenomena, but their accurate computation using ab initio methods is often limited by computational cost. The recent emergence of machine learning (ML) potentials may be a promising alternative. Useful ML models should not only estimate accurate interaction energies but also predict smooth and asymptotically correct potential energy surfaces. However, existing ML models are not guaranteed to obey these constraints. Indeed, systemic deficiencies are apparent in the predictions of our previous hydrogen-bond model as well as the popular ANI-1X model, which we attribute to the use of an atomic energy partition. As a solution, we propose an alternative atomic-pairwise framework specifically for intermolecular ML potentials, and we introduce AP-Net-a neural network model for interaction energies. The AP-Net model is developed using this physically motivated atomic-pairwise paradigm and also exploits the interpretability of symmetry adapted perturbation theory (SAPT). We show that in contrast to other models, AP-Net produces smooth, physically meaningful intermolecular potentials exhibiting correct asymptotic behavior. Initially trained on only a limited number of mostly hydrogen-bonded dimers, AP-Net makes accurate predictions across the chemically diverse S66x8 dataset, demonstrating significant transferability. On a test set including experimental hydrogen-bonded dimers, AP-Net predicts total interaction energies with a mean absolute error of 0.37 kcal mol-1, reducing errors by a factor of 2-5 across SAPT components from previous neural network potentials. The pairwise interaction energies of the model are physically interpretable, and an investigation of predicted electrostatic energies suggests that the model "learns" the physics of hydrogen-bonded interactions.

Approaches for machine learning intermolecular interaction energies and application to energy components from symmetry adapted perturbation theory.

Accurate prediction of intermolecular interaction energies is a fundamental challenge in electronic structure theory due to their subtle character and small magnitudes relative to total molecular energies. Symmetry adapted perturbation theory (SAPT) provides rigorous quantum mechanical means for computing such quantities directly and accurately, but for a computational cost of at least O(N5), where N is the number of atoms. Here, we report machine learned models of SAPT components with a computational cost that scales asymptotically linearly, O(N). We use modified multi-target Behler-Parrinello neural networks and specialized intermolecular symmetry functions to address the idiosyncrasies of the intermolecular problem, achieving 1.2 kcal mol-1 mean absolute errors on a test set of hydrogen bound complexes including structural data extracted from the Cambridge Structural Database and Protein Data Bank, spanning an interaction energy range of 20 kcal mol-1. Additionally, we recover accurate predictions of the physically meaningful SAPT component energies, of which dispersion and induction/polarization were the easiest to predict and electrostatics and exchange-repulsion are the most difficult.

Improving Accuracy, Diversity, and Speed with Prime Macrocycle Conformational Sampling.

A novel method for exploring macrocycle conformational space, Prime macrocycle conformational sampling (Prime-MCS), is introduced and evaluated in the context of other available algorithms (Molecular Dynamics, LowModeMD in MOE, and MacroModel Baseline Search). The algorithms were benchmarked on a data set of 208 macrocycles which was curated for diversity from the Cambridge Structural Database, the Protein Data Bank, and the Biologically Interesting Molecule Reference Dictionary. The algorithms were evaluated in terms of accuracy (ability to reproduce the crystal structure), diversity (coverage of conformational space), and computational speed. Prime-MCS most reliably reproduced crystallographic structures for RMSD thresholds >1.0 Å, most often produced the most diverse conformational ensemble, and was most often the fastest algorithm. Detailed analysis and examination of both typical and outlier cases were performed to reveal characteristics, shortcomings, expected performance, and complementarity of the methods.

4-Benzothiazole-7-hydroxyindolinyl diaryl ureas are potent P2Y1 antagonists with favorable pharmacokinetics: low clearance and small volume of distribution.

Current antithrombotic discovery efforts target compounds that are highly efficacious in thrombus reduction with less bleeding liability than the standard of care. Preclinical data suggest that P2Y1 antagonists may have lower bleeding liabilities than P2Y12 antagonists while providing similar antithrombotic efficacy. This article describes our continuous SAR efforts in a series of 7-hydroxyindolinyl diaryl ureas. When dosed orally, 4-trifluoromethyl-7-hydroxy-3,3-dimethylindolinyl analogue 4 was highly efficacious in a model of arterial thrombosis in rats with limited bleeding. The chemically labile CF3 group in 4 was then transformed to various groups via a novel one-step synthesis, yielding a series of potent P2Y1 antagonists. Among them, the 4-benzothiazole-substituted indolines had desirable PK properties in rats, specifically, low clearance and small volume of distribution. In addition, compound 40 had high i.v. exposure and modest bioavailability, giving it the best overall profile.

Conformationally constrained ortho-anilino diaryl ureas: discovery of 1-(2-(1'-neopentylspiro[indoline-3,4'-piperidine]-1-yl)phenyl)-3-(4-(trifluoromethoxy)phenyl)urea, a potent, selective, and bioavailable P2Y1 antagonist.

Preclinical antithrombotic efficacy and bleeding models have demonstrated that P2Y1 antagonists are efficacious as antiplatelet agents and may offer a safety advantage over P2Y12 antagonists in terms of reduced bleeding liabilities. In this article, we describe the structural modification of the tert-butyl phenoxy portion of lead compound 1 and the subsequent discovery of a novel series of conformationally constrained ortho-anilino diaryl ureas. In particular, spiropiperidine indoline-substituted diaryl ureas are described as potent, orally bioavailable small-molecule P2Y1 antagonists with improved activity in functional assays and improved oral bioavailability in rats. Homology modeling and rat PK/PD studies on benchmark compound 3l will also be presented. Compound 3l was our first P2Y1 antagonist to demonstrate a robust oral antithrombotic effect with mild bleeding liability in the rat thrombosis and hemostasis models.

The role of tyrosine 71 in modulating the flap conformations of BACE1.

ß-Site amyloid precursor protein cleaving enzyme 1 (BACE1) is a potential target for treating Alzheimer's disease. BACE1's binding site is partially covered by a flexible loop on its N-terminal domain, known as the "flap," which has been found in several conformations in crystal structures of BACE1 and other aspartyl proteases. The side chain of the invariant residue Tyr71 on the flap adopts several rotameric orientations, leading to our hypothesis that the orientation of this residue dictates the movement and conformations available to the flap. We investigated this hypothesis by performing 220 ns of molecular dynamics simulations of bound and unbound wild-type BACE1 as well as the unbound Y71A mutant. Our findings indicate that the flap exhibits various degrees of mobility and adopts different conformations depending on the Tyr71 orientation. Surprisingly, the "self-inhibited" form is stable in our simulations, making it a reasonable target for drug design. The alanine mutant, lacking a large side chain at position 71, displays significant differences in flap dynamics from wild type, freely sampling very open and closed conformations. Our simulations show that Tyr71, in addition to its previously determined functions in catalysis and substrate binding, has the important role of modulating flap conformations in BACE1.

Voltage-dependent hydration and conduction properties of the hydrophobic pore of the mechanosensitive channel of small conductance.

A detailed picture of water and ion properties in small pores is important for understanding the behavior of biological ion channels. Several recent modeling studies have shown that small, hydrophobic pores exclude water and ions even if they are physically large enough to accommodate them, a mechanism called hydrophobic gating. This mechanism has been implicated in the gating of several channels, including the mechanosensitive channel of small conductance (MscS). Although the pore in the crystal structure of MscS is wide and was initially hypothesized to be open, it is lined by hydrophobic residues and may represent a nonconducting state. Molecular dynamics simulations were performed on MscS to determine whether or not the structure can conduct ions. Unlike previous simulations of hydrophobic nanopores, electric fields were applied to this system to model the transmembrane potential, which proved to be important. Although simulations without a potential resulted in a dehydrated, occluded pore, the application of a potential increased the hydration of the pore and resulted in current flow through the channel. The calculated channel conductance was in good agreement with experiment. Therefore, it is likely that the MscS crystal structure is closer to a conducting than a nonconducting state.