Journal of Computer-Aided Molecular Design最新文献

Molecules with bioactivity towards G protein-coupled receptors represent a subset of the vast space of small drug-like molecules. Here, we compare machine learning models, including dilated graph convolutional networks, that conduct binary classification to quickly identify molecules with activity towards G protein-coupled receptors. The models are trained and validated using a large set of over 600,000 active, inactive, and decoy compounds. The best performing machine learning model, dubbed GPCRLigNet, was a surprisingly simple feedforward dense neural network mapping from Morgan fingerprints to activity. Incorporation of GPCRLigNet into a high-throughput virtual screening workflow is demonstrated with molecular docking towards a particular G protein-coupled receptor, the pituitary adenylate cyclase-activating polypeptide receptor type 1. Through rigorous comparison of docking scores for molecules selected with and without using GPCRLigNet, we demonstrate an enrichment of potentially potent molecules using GPCRLigNet. This work provides a proof of principle that GPCRLigNet can effectively hone the chemical search space towards ligands with G protein-coupled receptor activity.

Aqueous solubility is the most important physicochemical property for agrochemical and drug candidates and a prerequisite for uptake, distribution, transport, and finally the bioavailability in living species. We here present the first-ever direct machine learning models for pH-dependent solubility in water. For this, we combined almost 300000 data points from 11 solubility assays performed over 24 years and over one million data points from lipophilicity and melting point experiments. Data were split into three pH-classes − acidic, neutral and basic − , representing the conditions of stomach and intestinal tract for animals and humans, and phloem and xylem for plants. We find that multi-task neural networks using ECFP-6 fingerprints outperform baseline random forests and single-task neural networks on the individual tasks. Our final model with three solubility tasks using the pH-class combined data from different assays and five helper tasks results in root mean square errors of 0.56 log units overall (acidic 0.61; neutral 0.52; basic 0.54) and Spearman rank correlations of 0.83 (acidic 0.78; neutral 0.86; basic 0.86), making it a valuable tool for profiling of compounds in pharmaceutical and agrochemical research. The model allows for the prediction of compound pH profiles with mean and median RMSE per molecule of 0.62 and 0.56 log units.

Tuberculosis (TB) is one of the main causes of death from a single pathological agent, Mycobacterium tuberculosis (Mtb). In addition, the emergence of drug-resistant TB strains has exacerbated even further the treatment outcome of TB patients. It is thus needed the search for new therapeutic strategies to improve the current treatment and to circumvent the resistance mechanisms of Mtb. The shikimate kinase (SK) is the fifth enzyme of the shikimate pathway, which is essential for the survival of Mtb. The shikimate pathway is absent in humans, thereby indicating SK as an attractive target for the development of anti-TB drugs. In this work, a combination of in silico and in vitro techniques was used to identify potential inhibitors for SK from Mtb (MtSK). All compounds of our in-house database (Centro de Pesquisas em Biologia Molecular e Funcional, CPBMF) were submitted to in silico toxicity analysis to evaluate the risk of hepatotoxicity. Docking experiments were performed to identify the potential inhibitors of MtSK according to the predicted binding energy. In vitro inhibitory activity of MtSK-catalyzed chemical reaction at a single compound concentration was assessed. Minimum inhibitory concentration values for in vitro growth of pan-sensitive Mtb H37Rv strain were also determined. The mixed approach implemented in this work was able to identify five compounds that inhibit both MtSK and the in vitro growth of Mtb.

Chagas disease, also known as American trypanosomiasis, is a neglected tropical disease caused by the protozoa Trypanosoma cruzi, affecting nearly 7 million people only in the Americas. Polyamines are essential compounds for parasite growth, survival, and differentiation. However, because trypanosomatids are auxotrophic for polyamines, they must be obtained from the host by specific transporters. In this investigation, an ensemble of QSAR classifiers able to identify polyamine analogs with trypanocidal activity was developed. Then, a multi-template homology model of the dimeric polyamine transporter of T. cruzi, TcPAT12, was created with Rosetta, and then refined by enhanced sampling molecular dynamics simulations. Using representative snapshots extracted from the trajectory, a docking model able to discriminate between active and inactive compounds was developed and validated. Both models were applied in a parallel virtual screening campaign to repurpose known drugs as anti-trypanosomal compounds inhibiting polyamine transport in T. cruzi. Montelukast, Quinestrol, Danazol, and Dutasteride were selected for in vitro testing, and all of them inhibited putrescine uptake in biochemical assays, confirming the predictive ability of the computational models. Furthermore, all the confirmed hits proved to inhibit epimastigote proliferation, and Quinestrol and Danazol were able to inhibit, in the low micromolar range, the viability of trypomastigotes and the intracellular growth of amastigotes.

Graphical abstract

Binding free energy calculations are increasingly used in drug discovery research to predict protein-ligand binding affinities and to prioritize candidate drug molecules accordingly. It has taken decades of collective effort to transform this academic concept into a technology adopted by the pharmaceutical and biotech industry. Having personally witnessed and taken part in this transformation, here I recount the (incomplete) list of problems that had to be solved to make this computational tool practical and suggest areas of future development.

Mimicking bioactive conformations of peptide segments involved in the formation of protein-protein interfaces with small molecules is thought to represent a promising strategy for the design of protein-protein interaction (PPI) inhibitors. For compound design, the use of three-dimensional (3D) scaffolds rich in sp3-centers makes it possible to precisely mimic bioactive peptide conformations. Herein, we introduce DeepCubist, a molecular generator for designing peptidomimetics based on 3D scaffolds. Firstly, enumerated 3D scaffolds are superposed on a target peptide conformation to identify a preferred template structure for designing peptidomimetics. Secondly, heteroatoms and unsaturated bonds are introduced into the template via a deep generative model to produce candidate compounds. DeepCubist was applied to design peptidomimetics of exemplary peptide turn, helix, and loop structures in pharmaceutical targets engaging in PPIs.

G protein-coupled receptors (GPCRs) are membrane proteins constituting the largest family of drug targets. The activated GPCR binds either the heterotrimeric G proteins or arrestin through its activation cycle. Water molecules have been reported to play a role in GPCR activation. Nevertheless, reported studies are focused on the hydrophobic helical bundle region. How water molecules function in GPCR bound either G protein or arrestin is rarely studied. To address this issue, we carried out computational studies on water molecules in both GPCR/G protein complexes and GPCR/arrestin complexes. Using inhomogeneous fluid theory (IFT), we locate all possible hydration sites in GPCRs binding either to G protein or arrestin. We observe that the number of water molecules on the interaction surface between GPCRs and signal proteins are correlated with the insertion depths of the α5-helix from G-protein or “finger loop” from arrestin in GPCRs. In three out of the four simulation pairs, the interfaces of Rhodopsin, M₂R and NTSR1 in the G protein-associated systems show more water-mediated hydrogen-bond networks when compared to these in arrestin-associated systems. This reflects that more functionally relevant water molecules may probably be attracted in G protein-associated structures than that in arrestin-associated structures. Moreover, we find the water-mediated interaction networks throughout the NPxxY region and the orthosteric pocket, which may be a key for GPCR activation. Reported studies show that non-biased agonist, which can trigger both GPCR-G protein and GPCR-arrestin activation signal, can result in pharmacologically toxicities. Our comprehensive studies of the hydration sites in GPCR/G protein complexes and GPCR/arrestin complexes may provide important insights in the design of G-protein biased agonists.

Cellulose/collagen composites have been widely used in biomedicine and tissue engineering. Interfacial interactions are crucial in determining the final properties of cellulose/collagen composite. Molecular dynamics simulations were carried out to gain insights into the interactions between cellulose and collagen. It has been found that the structure of collagen remained intact during adsorption. The results derived from umbrella sampling showed that (110) and ((1bar{1}0)) faces exhibited the strongest affinity with collagen (100) face came the second and (010) the last, which could be attributed to the surface roughness and hydrogen-bonding linkers involved water molecules. Cellulose planes with flat surfaces and the capability to form hydrogen-bonding linkers produce stronger affinity with collagen. The occupancy of hydrogen bonds formed between cellulose and collagen was low and not significantly contributive to the binding affinity. These findings provided insights into the interactions between cellulose and collagen at the molecular level, which may guide the design and fabrication of cellulose/collagen composites.

Graphical abstract

The Hepatitis C Virus (HCV) NS3/4A is an attractive target for the treatment of Hepatitis C infection. Herein, we present an investigation of HCV NS3/4A inhibitors based on a sulfonamidobenzamide scaffold. Inhibitor interactions with HCV NS3/4A were explored by molecular docking, molecular dynamics simulations, and MM/PBSA binding free energy calculations. All of the inhibitors adopt similar molecular docking poses in the catalytic site of the protease that are stabilized by hydrogen bond interactions with G137 and the catalytic S139, which are known to be important for potency and binding stability. The quantitative assessments of binding free energies from MM/PBSA correlate well with the experimental results, with a high coefficient of determination, R² of 0.92. Binding free energy decomposition analyses elucidate the different contributions of Q41, F43, H57, R109, K136, G137, S138, S139, A156, M485, and Q526 in binding different inhibitors. The importance of these sidechain contributions was further confirmed by computational alanine scanning mutagenesis. In addition, the sidechains of K136 and S139 show crucial but distinct contributions to inhibitor binding with HCV NS3/4A. The structural basis of the potency has been elucidated, demonstrating the importance of the R155 sidechain conformation. This extensive exploration of binding energies and interactions between these compounds and HCV NS3/4A at the atomic level should benefit future antiviral drug design.

Fragment spaces are an efficient way to model large chemical spaces using a handful of small fragments and a few connection rules. The development of Enamine’s REAL Space has shown that large spaces of readily available compounds may be created this way. These are several orders of magnitude larger than previous libraries. So far, searching and navigating these spaces is mostly limited to topological approaches. A way to overcome this limitation is optimization via metaheuristics which can be combined with arbitrary scoring functions. Here we present Galileo, a novel Genetic Algorithm to sample fragment spaces. We showcase Galileo in combination with a novel pharmacophore mapping approach, called Phariety, enabling 3D searches in fragment spaces. We estimate the effectiveness of the approach with a small fragment space. Furthermore, we apply Galileo to two pharmacophore searches in the REAL Space, detecting hundreds of compounds fulfilling a HSP90 and a FXIa pharmacophore.