Large-Scale Virtual Screening for the Discovery of SARS-CoV-2 Papain-like Protease (PLpro) Non-covalent Inhibitors

Advances in the Search for SARS-CoV-2 Mpro and PLpro Inhibitors

SARS-CoV-2 is a spherical, positive-sense, single-stranded RNA virus with a large genome, responsible for encoding both structural proteins, vital for the viral particle's architecture, and non-structural proteins, critical for the virus's replication cycle. Among the non-structural proteins, two cysteine proteases emerge as promising molecular targets for the design of new antiviral compounds. The main protease (Mpro) is a homodimeric enzyme that plays a pivotal role in the formation of the viral replication-transcription complex, associated with the papain-like protease (PLpro), a cysteine protease that modulates host immune signaling by reversing post-translational modifications of ubiquitin and interferon-stimulated gene 15 (ISG15) in host cells. Due to the importance of these molecular targets for the design and development of novel anti-SARS-CoV-2 drugs, the purpose of this review is to address aspects related to the structure, mechanism of action and strategies for the design of inhibitors capable of targeting the Mpro and PLpro. Examples of covalent and non-covalent inhibitors that are currently being evaluated in preclinical and clinical studies or already approved for therapy will be also discussed to show the advances in medicinal chemistry in the search for new molecules to treat COVID-19.

Structure-based virtual screening of vast chemical space as a starting point for drug discovery

Structure-based virtual screening aims to find molecules forming favorable interactions with a biological macromolecule using computational models of complexes. The recent surge of commercially available chemical space provides the opportunity to search for ligands of therapeutic targets among billions of compounds. This review offers a compact overview of structure-based virtual screens of vast chemical spaces, highlighting successful applications in early drug discovery for therapeutically important targets such as G protein-coupled receptors and viral enzymes. Emphasis is placed on strategies to explore ultra-large chemical libraries and synergies with emerging machine learning techniques. The current opportunities and future challenges of virtual screening are discussed, indicating that this approach will play an important role in the next-generation drug discovery pipeline.

SchNetPack Hyperparameter Optimization for a More Reliable Top Docking Scores Prediction

Options to improve the extrapolation power of the neural network designed using the SchNetPack package with respect to top docking scores prediction are presented. It is shown that hyperparameter tuning of the atomistic model representation (in the schnetpack.representation) improves the prediction of the top scoring compounds, which have characteristically a low incidence in randomized data sets for training of machine learning models. The prediction robustness is evaluated according to the mean square error (MSE) and the entropy of the average loss landscape decrease. Admittedly, the improvement of the top scoring compounds' prediction accuracy comes with the penalty of worsening the overall prediction power. It is revealed that the most impactful hyperparameter is the cutoff (5 Å is reported as the optimal choice). Other parameters (e.g., number of radial basis functions, number of interaction layers of the neural network, feature vector size or its batch size) are found to not affect the prediction robustness of the top scoring compounds in any comparable way relative to the cutoff. The MSE of the best docking score prediction (below -13 kcal/mol) improves from ca. 3.5 to 0.9 kcal/mol, while the prediction of less potent compounds (-13 to -11 kcal/mol) shows a lesser improvement, i.e., a decrease of MSE from 1.6 to 1.3 kcal/mol. Additionally, oversampling and undersampling of the training set with respect to the top scoring compounds' abundance is presented. The results indicate that the cutoff choice performs better than over- or undersampling of the training set, with undersampling performing better than oversampling.

Integrating QSAR modelling and deep learning in drug discovery: the emergence of deep QSAR

Quantitative structure-activity relationship (QSAR) modelling, an approach that was introduced 60 years ago, is widely used in computer-aided drug design. In recent years, progress in artificial intelligence techniques, such as deep learning, the rapid growth of databases of molecules for virtual screening and dramatic improvements in computational power have supported the emergence of a new field of QSAR applications that we term 'deep QSAR'. Marking a decade from the pioneering applications of deep QSAR to tasks involved in small-molecule drug discovery, we herein describe key advances in the field, including deep generative and reinforcement learning approaches in molecular design, deep learning models for synthetic planning and the application of deep QSAR models in structure-based virtual screening. We also reflect on the emergence of quantum computing, which promises to further accelerate deep QSAR applications and the need for open-source and democratized resources to support computer-aided drug design.

Docking and other computing tools in drug design against SARS-CoV-2

The use of computer simulation methods has become an indispensable component in identifying drugs against the SARS-CoV-2 coronavirus. There is a huge body of literature on application of molecular modelling to predict inhibitors against target proteins of SARS-CoV-2. To keep our review clear and readable, we limited ourselves primarily to works that use computational methods to find inhibitors and test the predicted compounds experimentally either in target protein assays or in cell culture with live SARS-CoV-2. Some works containing results of experimental discovery of corresponding inhibitors without using computer modelling are included as examples of a success. Also, some computational works without experimental confirmations are also included if they attract our attention either by simulation methods or by databases used. This review collects studies that use various molecular modelling methods: docking, molecular dynamics, quantum mechanics, machine learning, and others. Most of these studies are based on docking, and other methods are used mainly for post-processing to select the best compounds among those found through docking. Simulation methods are presented concisely, information is also provided on databases of organic compounds that can be useful for virtual screening, and the review itself is structured in accordance with coronavirus target proteins.

Targeting SARS-CoV-2 nonstructural protein 3: Function, structure, inhibition, and perspective in drug discovery

As a highly contagious human pathogen, severe acute respiratory syndrome-associated coronavirus-2 (SARS-CoV-2) has infected billions of people worldwide with more than 6 million deaths. With several effective vaccines and antiviral drugs now available, the SARS-CoV-2 pandemic been brought under control. However, a new pathogenic coronavirus could emerge in the future, given the zoonotic nature of this virus. Natural evolution and drug-induced mutations of SARS-CoV-2 also require continued efforts for new anti-coronavirus drugs. Nonstructural protein (nsp) 3 of CoVs is a large, multifunctional protein, containing a papain-like protease (PLpro) and a macrodomain (Mac1), which are essential for viral replication. Here, we provide a comprehensive review of the function, structure, and inhibition of SARS-CoV/-CoV-2 PLpro and Mac1. We also discuss advances in, and challenges to, the discovery of drugs against these targets.

Computational study of the binding orientation and affinity of noncovalent inhibitors of the papain-like protease (PLpro) from SARS-CoV-1 considering the protein flexibility by using molecular dynamics and cross-docking

The papain-like protease (PLpro) from zoonotic coronaviruses (CoVs) has been identified as a target with an essential role in viral respiratory diseases caused by Severe Acute Respiratory Syndrome-associated coronaviruses (SARS-CoVs). The design of PLpro inhibitors has been proposed as an alternative to developing potential drugs against this disease. In this work, 67 naphthalene-derived compounds as noncovalent PLpro inhibitors were studied using molecular modeling methods. Structural characteristics of the bioactive conformations of these inhibitors and their interactions at the SARS-CoV-1 PLpro binding site were reported here in detail, taking into account the flexibility of the protein residues. Firstly, a molecular docking protocol was used to obtain the orientations of the inhibitors. After this, the orientations were compared, and the recurrent interactions between the PLpro residues and ligand chemical groups were described (with LigRMSD and interaction fingerprints methods). In addition, efforts were made to find correlations between docking energy values and experimentally determined binding affinities. For this, the PLpro was sampled by using Gaussian Accelerated Molecular Dynamics (GaMD), generating multiple conformations of the binding site. Diverse protein conformations were selected and a cross-docking experiment was performed, yielding models of the 67 naphthalene-derived compounds adopting different binding modes. Representative complexes for each ligand were selected to obtain the highest correlation between docking energies and activities. A good correlation (R 2 = 0.948) was found when this flexible docking protocol was performed.