The Recognition of Identical Ligands by Unrelated Proteins

Comprehensive detection and characterization of human druggable pockets through binding site descriptors

Druggable pockets are protein regions that have the ability to bind organic small molecules, and their characterization is essential in target-based drug discovery. However, deriving pocket descriptors is challenging and existing strategies are often limited in applicability. We introduce PocketVec, an approach to generate pocket descriptors via inverse virtual screening of lead-like molecules. PocketVec performs comparably to leading methodologies while addressing key limitations. Additionally, we systematically search for druggable pockets in the human proteome, using experimentally determined structures and AlphaFold2 models, identifying over 32,000 binding sites across 20,000 protein domains. We then generate PocketVec descriptors for each site and conduct an extensive similarity search, exploring over 1.2 billion pairwise comparisons. Our results reveal druggable pocket similarities not detected by structure- or sequence-based methods, uncovering clusters of similar pockets in proteins lacking crystallized inhibitors and opening the door to strategies for prioritizing chemical probe development to explore the druggable space.

SiteMine: Large-scale binding site similarity searching in protein structure databases

Drug discovery and design challenges, such as drug repurposing, analyzing protein-ligand and protein-protein complexes, ligand promiscuity studies, or function prediction, can be addressed by protein binding site similarity analysis. Although numerous tools exist, they all have individual strengths and drawbacks with regard to run time, provision of structure superpositions, and applicability to diverse application domains. Here, we introduce SiteMine, an all-in-one database-driven, alignment-providing binding site similarity search tool to tackle the most pressing challenges of binding site comparison. The performance of SiteMine is evaluated on the ProSPECCTs benchmark, showing a promising performance on most of the data sets. The method performs convincingly regarding all quality criteria for reliable binding site comparison, offering a novel state-of-the-art approach for structure-based molecular design based on binding site comparisons. In a SiteMine showcase, we discuss the high structural similarity between cathepsin L and calpain 1 binding sites and give an outlook on the impact of this finding on structure-based drug design. SiteMine is available at https://uhh.de/naomi.

Promiscuity in drug discovery on the verge of the structural revolution: recent advances and future chances

INTRODUCTION

Promiscuity denotes the ability of ligands and targets to specifically interact with multiple binding partners. Despite negative aspects like side effects, promiscuity is receiving increasing attention in drug discovery as it can enhance drug efficacy and provides a molecular basis for drug repositioning. The three-dimensional structure of ligand-target complexes delivers exclusive insights into the molecular mechanisms of promiscuity and structure-based methods enable the identification of promiscuous interactions. With the recent breakthrough in protein structure prediction, novel possibilities open up to reveal unknown connections in ligand-target interaction networks.

AREAS COVERED

This review highlights the significance of structure in the identification and characterization of promiscuity and evaluates the potential of protein structure prediction to advance our knowledge of drug-target interaction networks. It discusses the definition and relevance of promiscuity in drug discovery and explores different approaches to detecting promiscuous ligands and targets.

EXPERT OPINION

Examination of structural data is essential for understanding and quantifying promiscuity. The recent advancements in structure prediction have resulted in an abundance of targets that are well-suited for structure-based methods like docking. In silico approaches may eventually completely transform our understanding of drug-target networks by complementing the millions of predicted protein structures with billions of predicted drug-target interactions.

Mining the Protein Data Bank to inspire fragment library design

The fragment approach has emerged as a method of choice for drug design, as it allows difficult therapeutic targets to be addressed. Success lies in the choice of the screened chemical library and the biophysical screening method, and also in the quality of the selected fragment and structural information used to develop a drug-like ligand. It has recently been proposed that promiscuous compounds, i.e., those that bind to several proteins, present an advantage for the fragment approach because they are likely to give frequent hits in screening. In this study, we searched the Protein Data Bank for fragments with multiple binding modes and targeting different sites. We identified 203 fragments represented by 90 scaffolds, some of which are not or hardly present in commercial fragment libraries. By contrast to other available fragment libraries, the studied set is enriched in fragments with a marked three-dimensional character (download at 10.5281/zenodo.7554649).

Nonsteroidal anti-inflammatory drugs as antipyretics and modulators of a molecular clock(s) in the appendix of Sauromatum venosum inflorescence

The appendix of the Sauromatum senosum inflorescence is a striking example of thermogenesis in plants. On the day of opening, the Sauromatum appendix becomes hot, reaching up to 32 °C. Aspirin, salicylic acid and 2,6-dihydroxybenzoic acid, a subclass of NSAIDs, induce a temperature rise from three mitochondrial sources: alternative oxidase, F₁ F_O -ATP synthase and adenine nucleotide translocator. This temperature rise is synchronized and compounded under various light/dark regimes. We studied the effect of different subgroups of NSAIDs on the temperature rise. Tissue slices of appendix of Sauromatum and Arum italicum inflorescences at a pre-mature stage were treated with the three inducers in combination with one NSAID under constant light or darkness and under different photoperiods. Temperature rise generated by the three heat sources in the presence of inducers and different non-selective NSAIDs were not compounded and occurred at three different times. Under constant light, DuP-697, ibuprofen, flurbiprofen, acetaminophen and diclofenac suppressed the temperature rise induced by the three salicylates. Desynchronization and delayed temperature rise were detected with 6/42-h light/ dark and 15/33-h light/dark regimes in the presence of celecoxib and ibuprofen. With a 24/24-h light/dark regime, temperature rise was suppressed in the presence of ibuprofen. There were differences in response to individual NSAIDs between appendix tissue of A. italicum and S. venosum. Mitochondrial energy balance is affected by NSAIDs. There is an interaction between light/dark regime and temperature rise and a relationship between timing mechanism and temperature rise.

Classification of Protein-Binding Sites Using a Spherical Convolutional Neural Network

The analysis and comparison of protein-binding sites aid various applications in the drug discovery process, e.g., hit finding, drug repurposing, and polypharmacology. Classification of binding sites has been a hot topic for the past 30 years, and many different methods have been published. The rapid development of machine learning computational algorithms, coupled with the large volume of publicly available protein-ligand 3D structures, makes it possible to apply deep learning techniques in binding site comparison. Our method uses a cutting-edge spherical convolutional neural network based on the DeepSphere architecture to learn global representations of protein-binding sites. The model was trained on TOUGH-C1 and TOUGH-M1 data and validated with the ProSPECCTs datasets. Our results show that our model can (1) perform well in protein-binding site similarity and classification tasks and (2) learn and separate the physicochemical properties of binding sites. Lastly, we tested the model on a set of kinases, where the results show that it is able to cluster the different kinase subfamilies effectively. This example demonstrates the method's promise for lead hopping within or outside a protein target, directly based on binding site information.

Estimating the Similarity between Protein Pockets

With the exponential increase in publicly available protein structures, the comparison of protein binding sites naturally emerged as a scientific topic to explain observations or generate hypotheses for ligand design, notably to predict ligand selectivity for on- and off-targets, explain polypharmacology, and design target-focused libraries. The current review summarizes the state-of-the-art computational methods applied to pocket detection and comparison as well as structural druggability estimates. The major strengths and weaknesses of current pocket descriptors, alignment methods, and similarity search algorithms are presented. Lastly, an exhaustive survey of both retrospective and prospective applications in diverse medicinal chemistry scenarios illustrates the capability of the existing methods and the hurdle that still needs to be overcome for more accurate predictions.

Discovery of Kinase and Carbonic Anhydrase Dual Inhibitors by Machine Learning Classification and Experiments

A multi-target small molecule modulator is advantageous for treating complicated diseases such as cancers. However, the strategy and application for discovering a multi-target modulator have been less reported. This study presents the dual inhibitors for kinase and carbonic anhydrase (CA) predicted by machine learning (ML) classifiers, and validated by biochemical and biophysical experiments. ML trained by CA I and CA II inhibitor molecular fingerprints predicted candidates from the protein-specific bioactive molecules approved or under clinical trials. For experimental tests, three sulfonamide-containing kinase inhibitors, 5932, 5946, and 6046, were chosen. The enzyme assays with CA I, CA II, CA IX, and CA XII have allowed the quantitative comparison in the molecules’ inhibitory activities. While 6046 inhibited weakly, 5932 and 5946 exhibited potent inhibitions with 100 nM to 1 μM inhibitory constants. The ML screening was extended for finding CAs inhibitors of all known kinase inhibitors. It found XMU-MP-1 as another potent CA inhibitor with an approximate 30 nM inhibitory constant for CA I, CA II, and CA IX. Differential scanning fluorimetry confirmed the direct interaction between CAs and small molecules. Cheminformatics studies, including docking simulation, suggest that each molecule possesses two separate functional moieties: one for interaction with kinases and the other with CAs.

Set-Theoretic Formalism for Treating Ligand-Target Datasets

Data on ligand–target (LT) interactions has played a growing role in drug research for several decades. Even though the amount of data has grown significantly in size and coverage during this period, most datasets remain difficult to analyze because of their extreme sparsity, as there is no activity data whatsoever for many LT pairs. Even within clusters of data there tends to be a lack of data completeness, making the analysis of LT datasets problematic. The current effort extends earlier works on the development of set-theoretic formalisms for treating thresholded LT datasets. Unlike many approaches that do not address pairs of unknown interaction, the current work specifically takes account of their presence in addition to that of active and inactive pairs. Because a given LT pair can be in any one of three states, the binary logic of classical set-theoretic methods does not strictly apply. The current work develops a formalism, based on ternary set-theoretic relations, for treating thresholded LT datasets. It also describes an extension of the concept of data completeness, which is typically applied to sets of ligands and targets, to the local data completeness of individual ligands and targets. The set-theoretic formalism is applied to the analysis of simple and joint polypharmacologies based on LT activity profiles, and it is shown that null pairs provide a means for determining bounds to these values. The methodology is applied to a dataset of protein kinase inhibitors as an illustration of the method. Although not dealt with here, work is currently underway on a more refined treatment of activity values that is based on increasing the number of activity classes.

Ligand Binding Site Comparison - LiBiSCo - a web-based tool for analyzing interactions between proteins and ligands to explore amino acid specificity within active sites

Interaction between protein and ligands are ubiquitous in a biological cell, and understanding these interactions at the atom level in protein-ligand complexes is crucial for structural bioinformatics and drug discovery. Here, we present a web-based protein-ligand interaction application named Ligand Binding Site Comparison (LiBiSCo) for comparing the amino acid residues interacting with atoms of a ligand molecule between different protein-ligand complexes available in the Protein Data Bank (PDB) database. The comparison is performed at the ligand atom level irrespectively of having binding site similarity or not between the protein structures of interest. The input used in LiBiSCo is one or several PDB IDs of protein-ligand complex(es) and the tool returns a list of identified interactions at ligand atom level including both bonded and non-bonded interactions. A sequence profile for the interaction for each ligand atoms is provided as a WebLogo. The LiBiSco is useful in understanding ligand binding specificity and structural promiscuity among families that are structurally unrelated. The LiBiSCo tool can be accessed through https://albiorix.bioenv.gu.se/LiBiSCo/HomePage.py.

Site2Vec: a reference frame invariant algorithm for vector embedding of protein–ligand binding sites

Protein–ligand interactions are one of the fundamental types of molecular interactions in living systems. Ligands are small molecules that interact with protein molecules at specific regions on their surfaces called binding sites. Binding sites would also determine ADMET properties of a drug molecule. Tasks such as assessment of protein functional similarity and detection of side effects of drugs need identification of similar binding sites of disparate proteins across diverse pathways. To this end, methods for computing similarities between binding sites are still evolving and is an active area of research even today. Machine learning methods for similarity assessment require feature descriptors of binding sites. Traditional methods based on hand engineered motifs and atomic configurations are not scalable across several thousands of sites. In this regard, deep neural network algorithms are now deployed which can capture very complex input feature space. However, one fundamental challenge in applying deep learning to structures of binding sites is the input representation and the reference frame. We report here a novel algorithm, Site2Vec, that derives reference frame invariant vector embedding of a protein–ligand binding site. The method is based on pairwise distances between representative points and chemical compositions in terms of constituent amino acids of a site. The vector embedding serves as a locality sensitive hash function for proximity queries and determining similar sites. The method has been the top performer with more than 95% quality scores in extensive benchmarking studies carried over 10 data sets and against 23 other site comparison methods in the field. The algorithm serves for high throughput processing and has been evaluated for stability with respect to reference frame shifts, coordinate perturbations and residue mutations. We also provide the method as a standalone executable and a web service hosted at (http://services.iittp.ac.in/bioinfo/home).

ChemBoost: A Chemical Language Based Approach for Protein - Ligand Binding Affinity Prediction

Identification of high affinity drug-target interactions is a major research question in drug discovery. Proteins are generally represented by their structures or sequences. However, structures are available only for a small subset of biomolecules and sequence similarity is not always correlated with functional similarity. We propose ChemBoost, a chemical language based approach for affinity prediction using SMILES syntax. We hypothesize that SMILES is a codified language and ligands are documents composed of chemical words. These documents can be used to learn chemical word vectors that represent words in similar contexts with similar vectors. In ChemBoost, the ligands are represented via chemical word embeddings, while the proteins are represented through sequence-based features and/or chemical words of their ligands. Our aim is to process the patterns in SMILES as a language to predict protein-ligand affinity, even when we cannot infer the function from the sequence. We used eXtreme Gradient Boosting to predict protein-ligand affinities in KIBA and BindingDB data sets. ChemBoost was able to predict drug-target binding affinity as well as or better than state-of-the-art machine learning systems. When powered with ligand-centric representations, ChemBoost was more robust to the changes in protein sequence similarity and successfully captured the interactions between a protein and a ligand, even if the protein has low sequence similarity to the known targets of the ligand.

In Silico Drug Repositioning for Chagas Disease

Chagas disease is an infectious tropical disease included within the group of neglected tropical diseases. Though historically endemic to Latin America, it has lately spread to high-income countries due to human migration. At present, there are only two available drugs, nifurtimox and benznidazole, approved for this treatment, both with considerable side-effects (which often result in treatment interruption) and limited efficacy in the chronic stage of the disease in adults. Drug repositioning involves finding novel therapeutic indications for known drugs, including approved, withdrawn, abandoned and investigational drugs. It is today a broadly applied approach to develop innovative medications, since indication shifts are built on existing safety, ADME and manufacturing information, thus greatly shortening development timeframes. Drug repositioning has been signaled as a particularly interesting strategy to search for new therapeutic solutions for neglected and rare conditions, which traditionally present limited commercial interest and are mostly covered by the public sector and not-for-profit initiatives and organizations. Here, we review the applications of computer-aided technologies as systematic approaches to drug repositioning in the field of Chagas disease. In silico screening represents the most explored approach, whereas other rational methods such as network-based and signature-based approximations have still not been applied.

Repurposing approved drugs on the pathway to novel therapies

The time and cost of developing new drugs have led many groups to limit their search for therapeutics to compounds that have previously been approved for human use. Many "repurposed" drugs, such as derivatives of thalidomide, antibiotics, and antivirals have had clinical success in treatment areas well beyond their original approved use. These include applications in treating antibiotic-resistant organisms, viruses, cancers and to prevent burn scarring. The major theoretical justification for reusing approved drugs is that they have known modes of action and controllable side effects. Coadministering antibiotics with inhibitors of bacterial toxins or enzymes that mediate multidrug resistance can greatly enhance their activity. Drugs that control host cell pathways, including inflammation, tumor necrosis factor, interferons, and autophagy, can reduce the "cytokine storm" response to injury, control infection, and aid in cancer therapy. An active compound, even if previously approved for human use, will be a poor clinical candidate if it lacks specificity for the new target, has poor solubility or can cause serious side effects. Synergistic combinations can reduce the dosages of the individual components to lower reactivity. Preclinical analysis should take into account that severely ill patients with comorbidities will be more sensitive to side effects than healthy trial subjects. Once an active, approved drug has been identified, collaboration with medicinal chemists can aid in finding derivatives with better physicochemical properties, specificity, and efficacy, to provide novel therapies for cancers, emerging and rare diseases.

The Thiol-polyamine Metabolism of Trypanosoma cruzi: Molecular Targets and Drug Repurposing Strategies

Chagas´ disease continues to be a challenging and neglected public health problem in many American countries. The etiologic agent, Trypanosoma cruzi, develops intracellularly in the mammalian host, which hinders treatment efficacy. Progress in the knowledge of parasite biology and host-pathogen interaction has not been paralleled by the development of novel, safe and effective therapeutic options. It is then urgent to seek for novel therapeutic candidates and to implement drug discovery strategies that may accelerate the discovery process. The most appealing targets for pharmacological intervention are those essential for the pathogen and, whenever possible, absent or significantly different from the host homolog. The thiol-polyamine metabolism of T. cruzi offers interesting candidates for a rational design of selective drugs. In this respect, here we critically review the state of the art of the thiolpolyamine metabolism of T. cruzi and the pharmacological potential of its components. On the other hand, drug repurposing emerged as a valid strategy to identify new biological activities for drugs in clinical use, while significantly shortening the long time and high cost associated with de novo drug discovery approaches. Thus, we also discuss the different drug repurposing strategies available with a special emphasis in their applications to the identification of drug candidates targeting essential components of the thiol-polyamine metabolism of T. cruzi.

Identification of Target Associations for Polypharmacology from Analysis of Crystallographic Ligands of the Protein Data Bank

The design of a chemical entity that potently and selectively binds to a biological target of therapeutic relevance has dominated the scene of drug discovery so far. However, recent findings suggest that multitarget ligands may be endowed with superior efficacy and be less prone to drug resistance. The Protein Data Bank (PDB) provides experimentally validated structural information about targets and bound ligands. Therefore, it represents a valuable source of information to help identifying active sites, understanding pharmacophore requirements, designing novel ligands, and inferring structure-activity relationships. In this study, we performed a large-scale analysis of the PDB by integrating different ligand-based and structure-based approaches, with the aim of identifying promising target associations for polypharmacology based on reported crystal structure information. First, the 2D and 3D similarity profiles of the crystallographic ligands were evaluated using different ligand-based methods. Then, activity data of pairs of similar ligands binding to different targets were inspected by comparing structural information with bioactivity annotations reported in the ChEMBL, BindingDB, BindingMOAD, and PDBbind databases. Afterward, extensive docking screenings of ligands in the identified cross-targets were made in order to validate and refine the ligand-based results. Finally, the therapeutic relevance of the identified target combinations for polypharmacology was evaluated from comparison with information on therapeutic targets reported in the Therapeutic Target Database (TTD). The results led to the identification of several target associations with high therapeutic potential for polypharmacology.

Computational methods and tools for binding site recognition between proteins and small molecules: from classical geometrical approaches to modern machine learning strategies

In the current "genomic era" the number of identified genes is growing exponentially. However, the biological function of a large number of the corresponding proteins is still unknown. Recognition of small molecule ligands (e.g., substrates, inhibitors, allosteric regulators, etc.) is pivotal for protein functions in the vast majority of the cases and knowledge of the region where these processes take place is essential for protein function prediction and drug design. In this regard, computational methods represent essential tools to tackle this problem. A significant number of software tools have been developed in the last few years which exploit either protein sequence information, structure information or both. This review describes the most recent developments in protein function recognition and binding site prediction, in terms of both freely-available and commercial solutions and tools, detailing the main characteristics of the considered tools and providing a comparative analysis of their performance.

High Impact: The Role of Promiscuous Binding Sites in Polypharmacology

The literature focuses on drug promiscuity, which is a drug’s ability to bind to several targets, because it plays an essential role in polypharmacology. However, little work has been completed regarding binding site promiscuity, even though its properties are now recognized among the key factors that impact drug promiscuity. Here, we quantified and characterized the promiscuity of druggable binding sites from protein-ligand complexes in the high quality Mother Of All Databases while using statistical methods. Most of the sites (80%) exhibited promiscuity, irrespective of the protein class. Nearly half were highly promiscuous and able to interact with various types of ligands. The corresponding pockets were rather large and hydrophobic, with high sulfur atom and aliphatic residue frequencies, but few side chain atoms. Consequently, their interacting ligands can be large, rigid, and weakly hydrophilic. The selective sites that interacted with one ligand type presented less favorable pocket properties for establishing ligand contacts. Thus, their ligands were highly adaptable, small, and hydrophilic. In the dataset, the promiscuity of the site rather than the drug mainly explains the multiple interactions between the drug and target, as most ligand types are dedicated to one site. This underlines the essential contribution of binding site promiscuity to drug promiscuity between different protein classes.

Binding site characterization - similarity, promiscuity, and druggability

The elucidation of non-obvious binding site similarities has provided useful indications for the establishment of polypharmacology, the identification of potential off-targets, or the repurposing of known drugs. The concept underlying all of these approaches is promiscuous binding which can be analyzed from a ligand-based or a binding site-based perspective. Herein, we applied methods for the automated analysis and comparison of protein binding sites to study promiscuous binding on a novel dataset of sites in complex with ligands sharing common shape and physicochemical properties. We show the suitability of this dataset for the benchmarking of novel binding site comparison methods. Our investigations also reveal promising directions for further in-depth analyses of promiscuity and druggability in a pocket-centered manner. Drawbacks concerning binding site similarity assessment and druggability prediction are outlined, enabling researchers to avoid the typical pitfalls of binding site analyses.

DeepDrug3D: Classification of ligand-binding pockets in proteins with a convolutional neural network

Comprehensive characterization of ligand-binding sites is invaluable to infer molecular functions of hypothetical proteins, trace evolutionary relationships between proteins, engineer enzymes to achieve a desired substrate specificity, and develop drugs with improved selectivity profiles. These research efforts pose significant challenges owing to the fact that similar pockets are commonly observed across different folds, leading to the high degree of promiscuity of ligand-protein interactions at the system-level. On that account, novel algorithms to accurately classify binding sites are needed. Deep learning is attracting a significant attention due to its successful applications in a wide range of disciplines. In this communication, we present DeepDrug3D, a new approach to characterize and classify binding pockets in proteins with deep learning. It employs a state-of-the-art convolutional neural network in which biomolecular structures are represented as voxels assigned interaction energy-based attributes. The current implementation of DeepDrug3D, trained to detect and classify nucleotide- and heme-binding sites, not only achieves a high accuracy of 95%, but also has the ability to generalize to unseen data as demonstrated for steroid-binding proteins and peptidase enzymes. Interestingly, the analysis of strongly discriminative regions of binding pockets reveals that this high classification accuracy arises from learning the patterns of specific molecular interactions, such as hydrogen bonds, aromatic and hydrophobic contacts. DeepDrug3D is available as an open-source program at https://github.com/pulimeng/DeepDrug3D with the accompanying TOUGH-C1 benchmarking dataset accessible from https://osf.io/enz69/.

Privileged Structures and Polypharmacology within and between Protein Families

Polypharmacology is often a key contributor to the efficacy of a drug, but is also a potential risk. We investigated two hits discovered via a cell-based phenotypic screen, the CDK9 inhibitor CCT250006 (1) and the pirin ligand CCT245232 (2), to establish methodology to elucidate their secondary protein targets. Using computational pocket-based analysis, we discovered intrafamily polypharmacology for our kinase inhibitor, despite little overall sequence identity. The interfamily polypharmacology of 2 with B-Raf was used to discover a novel pirin ligand from a very small but privileged compound library despite no apparent ligand or binding site similarity. Our data demonstrates that in areas of drug discovery where intrafamily polypharmacology is often an issue, ligand dissimilarity cannot necessarily be used to assume different off-target profiles and that understanding interfamily polypharmacology will be important in the future to reduce the risk of idiopathic toxicity and in the design of screening libraries.

A benchmark driven guide to binding site comparison: An exhaustive evaluation using tailor-made data sets (ProSPECCTs)

The automated comparison of protein-ligand binding sites provides useful insights into yet unexplored site similarities. Various stages of computational and chemical biology research can benefit from this knowledge. The search for putative off-targets and the establishment of polypharmacological effects by comparing binding sites led to promising results for numerous projects. Although many cavity comparison methods are available, a comprehensive analysis to guide the choice of a tool for a specific application is wanting. Moreover, the broad variety of binding site modeling approaches, comparison algorithms, and scoring metrics impedes this choice. Herein, we aim to elucidate strengths and weaknesses of binding site comparison methodologies. A detailed benchmark study is the only possibility to rationalize the selection of appropriate tools for different scenarios. Specific evaluation data sets were developed to shed light on multiple aspects of binding site comparison. An assembly of all applied benchmark sets (ProSPECCTs–Protein Site Pairs for the Evaluation of Cavity Comparison Tools) is made available for the evaluation and optimization of further and still emerging methods. The results indicate the importance of such analyses to facilitate the choice of a methodology that complies with the requirements of a specific scientific challenge.

Binding site similarities are useful in the context of promiscuity prediction, drug repurposing, the analysis of protein-ligand and protein-protein complexes, function prediction, and further fields of general interest in chemical biology and biochemistry. Many years of research have led to the development of a multitude of methods for binding site analysis and comparison. On the one hand, their availability supports research. On the other hand, the huge number of methods hampers the efficient selection of a specific tool. Our research is dedicated to the analysis of different cavity comparison tools. We use several binding site data sets to establish guidelines which can be applied to ensure a successful application of comparison methods by circumventing potential pitfalls.

The Recognition of Unrelated Ligands by Identical Proteins

Unrelated ligands, often found in drug discovery campaigns, can bind to the same receptor, even with the same protein residues. To investigate how this might occur, and whether it might be typically possible to find unrelated ligands for the same drug target, we sought examples of topologically unrelated ligands that bound to the same protein in the same site. Seventy-six pairs of ligands, each bound to the same protein (152 complexes total), were considered, classified into three groups. In the first (31 pairs of complexes), unrelated ligands interacted largely with the same pocket residues through different functional groups. In the second group (39 pairs), the unrelated ligand in each pair engaged different residues, though still within the same pocket. The smallest group (6 pairs) contained ligands with different scaffolds but with shared functional groups interacting with the same residues. We found that there are multiple chemically unrelated but physically similar functional groups that can complement any given local protein pocket; when these functional group substitutions are combined within a single molecule, they lead to topologically unrelated ligands that can each well-complement a site. It may be that many active and orthosteric sites can recognize topologically unrelated ligands.

Protein-ligand binding site detection as an alternative route to molecular docking and drug repurposing

After the onset of the genomic era, the detection of ligand binding sites in proteins has emerged over the last few years as a powerful tool for protein function prediction. Several approaches, both sequence and structure based, have been developed, but the full potential of the corresponding tools has not been exploited yet. Here, we describe the development and classification of a large, almost exhaustive, collection of protein-ligand binding sites to be used, in conjunction with the Ligand Binding Site Recognition Application Web Application developed in our laboratory, as an alternative to virtual screening through molecular docking simulations to identify novel lead compounds for known targets. Ligand binding sites derived from the Protein Data Bank have been clustered according to ligand similarity, and given a known ligand, the binding mode of related ligands to the same target can be predicted. The collection of ligand binding sites contains more than 200,000 sites corresponding to more than 20,000 different ligands. Furthermore, the ligand binding sites of all Food and Drug Administration-approved drugs have been classified as well, allowing to investigate the possible binding of each of them (and related compounds) to a given target for drug repurposing and redesign initiatives. Sample usage cases are also described to demonstrate the effectiveness of this approach.

Docking Screens for Dual Inhibitors of Disparate Drug Targets for Parkinson's Disease

Modulation of multiple biological targets with a single drug can lead to synergistic therapeutic effects and has been demonstrated to be essential for efficient treatment of CNS disorders. However, rational design of compounds that interact with several targets is very challenging. Here, we demonstrate that structure-based virtual screening can guide the discovery of multi-target ligands of unrelated proteins relevant for Parkinson's disease. A library with 5.4 million molecules was docked to crystal structures of the A_2A adenosine receptor (A_2AAR) and monoamine oxidase B (MAO-B). Twenty-four compounds that were among the highest ranked for both binding sites were evaluated experimentally, resulting in the discovery of four dual-target ligands. The most potent compound was an A_2AAR antagonist with nanomolar affinity ( K_i = 19 nM) and inhibited MAO-B with an IC₅₀ of 100 nM. Optimization guided by the predicted binding modes led to the identification of a second potent dual-target scaffold. The two discovered scaffolds were shown to counteract 6-hydroxydopamine-induced neurotoxicity in dopaminergic neuronal-like SH-SY5Y cells. Structure-based screening can hence be used to identify ligands with specific polypharmacological profiles, providing new avenues for drug development against complex diseases.

How to Achieve Better Results Using PASS-Based Virtual Screening: Case Study for Kinase Inhibitors

Discovery of new pharmaceutical substances is currently boosted by the possibility of utilization of the Synthetically Accessible Virtual Inventory (SAVI) library, which includes about 283 million molecules, each annotated with a proposed synthetic one-step route from commercially available starting materials. The SAVI database is well-suited for ligand-based methods of virtual screening to select molecules for experimental testing. In this study, we compare the performance of three approaches for the analysis of structure-activity relationships that differ in their criteria for selecting of "active" and "inactive" compounds included in the training sets. PASS (Prediction of Activity Spectra for Substances), which is based on a modified Naïve Bayes algorithm, was applied since it had been shown to be robust and to provide good predictions of many biological activities based on just the structural formula of a compound even if the information in the training set is incomplete. We used different subsets of kinase inhibitors for this case study because many data are currently available on this important class of drug-like molecules. Based on the subsets of kinase inhibitors extracted from the ChEMBL 20 database we performed the PASS training, and then applied the model to ChEMBL 23 compounds not yet present in ChEMBL 20 to identify novel kinase inhibitors. As one may expect, the best prediction accuracy was obtained if only the experimentally confirmed active and inactive compounds for distinct kinases in the training procedure were used. However, for some kinases, reasonable results were obtained even if we used merged training sets, in which we designated as inactives the compounds not tested against the particular kinase. Thus, depending on the availability of data for a particular biological activity, one may choose the first or the second approach for creating ligand-based computational tools to achieve the best possible results in virtual screening.

Backbone Brackets and Arginine Tweezers delineate Class I and Class II aminoacyl tRNA synthetases

The origin of the machinery that realizes protein biosynthesis in all organisms is still unclear. One key component of this machinery are aminoacyl tRNA synthetases (aaRS), which ligate tRNAs to amino acids while consuming ATP. Sequence analyses revealed that these enzymes can be divided into two complementary classes. Both classes differ significantly on a sequence and structural level, feature different reaction mechanisms, and occur in diverse oligomerization states. The one unifying aspect of both classes is their function of binding ATP. We identified Backbone Brackets and Arginine Tweezers as most compact ATP binding motifs characteristic for each Class. Geometric analysis shows a structural rearrangement of the Backbone Brackets upon ATP binding, indicating a general mechanism of all Class I structures. Regarding the origin of aaRS, the Rodin-Ohno hypothesis states that the peculiar nature of the two aaRS classes is the result of their primordial forms, called Protozymes, being encoded on opposite strands of the same gene. Backbone Brackets and Arginine Tweezers were traced back to the proposed Protozymes and their more efficient successors, the Urzymes. Both structural motifs can be observed as pairs of residues in contemporary structures and it seems that the time of their addition, indicated by their placement in the ancient aaRS, coincides with the evolutionary trace of Proto- and Urzymes.

Aminoacyl tRNA synthetases (aaRS) are primordial enzymes essential for interpretation and transfer of genetic information. Understanding the origin of the peculiarities observed with aaRS can explain what constituted the earliest life forms and how the genetic code was established. The increasing amount of experimentally determined three-dimensional structures of aaRS opens up new avenues for high-throughput analyses of molecular mechanisms. In this study, we present an exhaustive structural analysis of ATP binding motifs. We unveil an oppositional implementation of enzyme substrate binding in each aaRS Class. While Class I binds via interactions mediated by backbone hydrogen bonds, Class II uses a pair of arginine residues to establish salt bridges to its ATP ligand. We show how nature realized the binding of the same ligand species with completely different mechanisms. In addition, we demonstrate that sequence or even structure analysis for conserved residues may miss important functional aspects which can only be revealed by ligand interaction studies. Additionally, the placement of those key residues in the structure supports a popular hypothesis, which states that prototypic aaRS were once coded on complementary strands of the same gene.

Redesign of LAOBP to bind novel l-amino acid ligands

Computational protein design is still a challenge for advancing structure-function relationships. While recent advances in this field are promising, more information for genuine predictions is needed. Here, we discuss different approaches applied to install novel glutamine (Gln) binding into the Lysine/Arginine/Ornithine binding protein (LAOBP) from Salmonella typhimurium. We studied the ligand binding behavior of two mutants: a binding pocket grafting design based on a structural superposition of LAOBP to the Gln binding protein QBP from Escherichia coli and a design based on statistical coupled positions. The latter showed the ability to bind Gln even though the protein was not very stable. Comparison of both approaches highlighted a nonconservative shared point mutation between LAOBP_graft and LAOBP_sca. This context dependent L117K mutation in LAOBP turned out to be sufficient for introducing Gln binding, as confirmed by different experimental techniques. Moreover, the crystal structure of LAOBP_L117K in complex with its ligand is reported.

Comparative assessment of strategies to identify similar ligand-binding pockets in proteins

Background

Detecting similar ligand-binding sites in globally unrelated proteins has a wide range of applications in modern drug discovery, including drug repurposing, the prediction of side effects, and drug-target interactions. Although a number of techniques to compare binding pockets have been developed, this problem still poses significant challenges.

Results

We evaluate the performance of three algorithms to calculate similarities between ligand-binding sites, APoc, SiteEngine, and G-LoSA. Our assessment considers not only the capabilities to identify similar pockets and to construct accurate local alignments, but also the dependence of these alignments on the sequence order. We point out certain drawbacks of previously compiled datasets, such as the inclusion of structurally similar proteins, leading to an overestimated performance. To address these issues, a rigorous procedure to prepare unbiased, high-quality benchmarking sets is proposed. Further, we conduct a comparative assessment of techniques directly aligning binding pockets to indirect strategies employing structure-based virtual screening with AutoDock Vina and rDock.

Conclusions

Thorough benchmarks reveal that G-LoSA offers a fairly robust overall performance, whereas the accuracy of APoc and SiteEngine is satisfactory only against easy datasets. Moreover, combining various algorithms into a meta-predictor improves the performance of existing methods to detect similar binding sites in unrelated proteins by 5–10%. All data reported in this paper are freely available at https://osf.io/6ngbs/.

Discovery of new GPCR ligands to illuminate new biology

Although a plurality of drugs target G-protein-coupled receptors (GPCRs), most have emerged from classical medicinal chemistry and pharmacology programs and resemble one another structurally and functionally. Though effective, these drugs are often promiscuous. With the realization that GPCRs signal via multiple pathways, and with the emergence of crystal structures for this family of proteins, there is an opportunity to target GPCRs with new chemotypes and confer new signaling modalities. We consider structure-based and physical screening methods that have led to the discovery of new reagents, focusing particularly on the former. We illustrate their use against previously untargeted or orphan GPCRs, against allosteric sites, and against classical orthosteric sites that selectively activate one downstream pathway over others. The ligands that emerge are often chemically novel, which can lead to new biological effects.

Computational design of environmental sensors for the potent opioid fentanyl

We describe the computational design of proteins that bind the potent analgesic fentanyl. Our approach employs a fast docking algorithm to find shape complementary ligand placement in protein scaffolds, followed by design of the surrounding residues to optimize binding affinity. Co-crystal structures of the highest affinity binder reveal a highly preorganized binding site, and an overall architecture and ligand placement in close agreement with the design model. We use the designs to generate plant sensors for fentanyl by coupling ligand binding to design stability. The method should be generally useful for detecting toxic hydrophobic compounds in the environment.

Many small molecules, including toxins and some medicines, have flexible structures, which makes it difficult to detect and/or neutralize them. The pain medication fentanyl, for example, can rotate to adopt many shapes. In recent years, fentanyl drug abuse has become increasingly common, and the drug is often illegally produced. The number of deaths caused by fentanyl has risen greatly, which provides a strong reason to find new ways to detect this and other drugs.

Now, Baker et al. have created new sensors that are able to detect fentanyl. First, the 11 most likely shapes that fentanyl could adopt were identified based on known information about the structure of the molecule. Then, a computer program was used to design proteins that were predicted to strongly bind to these most common shapes. Next, genes that coded for these proteins were synthesized in the laboratory and introduced into bacteria, which read the genes to build the proteins.

Similar to a well-fitted lock and key, the shape of the newly designed protein had to complement a likely shape of the fentanyl molecule. Baker et al. used a technique called X-ray crystallography to visualize the proteins in atomic detail and confirm that these fentanyl-binders matched their corresponding computational models. Those proteins that bound fentanyl best were then engineered into plant cells, and later into whole plants, together with reporter systems that gave signals when the sensors detected fentanyl.

In future, these specifically synthesized proteins could be integrated into entire panels of plants or other systems to detect toxins and other harmful chemicals. Such systems would be of interest in a medical setting and for detecting environmental contamination.

A Simple Representation of Three-Dimensional Molecular Structure

Statistical and machine learning approaches predict drug-to-target relationships from 2D small-molecule topology patterns. One might expect 3D information to improve these calculations. Here we apply the logic of the extended connectivity fingerprint (ECFP) to develop a rapid, alignment-invariant 3D representation of molecular conformers, the extended three-dimensional fingerprint (E3FP). By integrating E3FP with the similarity ensemble approach (SEA), we achieve higher precision-recall performance relative to SEA with ECFP on ChEMBL20 and equivalent receiver operating characteristic performance. We identify classes of molecules for which E3FP is a better predictor of similarity in bioactivity than is ECFP. Finally, we report novel drug-to-target binding predictions inaccessible by 2D fingerprints and confirm three of them experimentally with ligand efficiencies from 0.442-0.637 kcal/mol/heavy atom.

Choose and Use Your Chemical Probe Wisely to Explore Cancer Biology

Small-molecule chemical probes or tools have become progressively more important in recent years as valuable reagents to investigate fundamental biological mechanisms and processes causing disease, including cancer. Chemical probes have also achieved greater prominence alongside complementary biological reagents for target validation in drug discovery. However, there is evidence of widespread continuing misuse and promulgation of poor-quality and insufficiently selective chemical probes, perpetuating a worrisome and misleading pollution of the scientific literature. We discuss current challenges with the selection and use of chemical probes, and suggest how biologists can and should be more discriminating in the probes they employ.

Multi-target Fragments Display Versatile Binding Modes

Promiscuity is an interesting concept in fragment-based drug design as fragments with low specificity can be advantageous for finding many screening hits. We present a PDB-wide analysis of multi-target fragments and their binding mode conservation. Focussing on multi-target fragments, we found that the majority shows non-conserved binding modes, even if they bind in a similar conformation or similar protein targets. Surprisingly, fragment properties alone are not able to predict whether a fragment will exhibit a versatile or conserved binding mode, emphasizing the interplay between protein and fragment features during a binding event and the importance of structure-based modelling.

A simple mathematical approach to the analysis of polypharmacology and polyspecificity data

There many possible types of drug-target interactions, because there are a surprising number of ways in which drugs and their targets can associate with one another.  These relationships are expressed as polypharmacology and polyspecificity.  Polypharmacology is the capability of a given drug to exhibit activity with respect to multiple drug targets, which are not necessarily in the same activity class. Adverse drug reactions ('side effects') are its principal manifestation, but polypharmacology is also playing a role in the repositioning of existing drugs for new therapeutic indications.  Polyspecificity, on the other hand, is the capability of a given target to exhibit activity with respect to multiple, structurally dissimilar drugs.  That these concepts are closely related to one another is, surprisingly, not well known.  It will be shown in this work that they are, in fact, mathematically related to one another and are in essence 'two sides of the same coin'.  Hence, information on polypharmacology provides equivalent information on polyspecificity, and vice versa. Networks are playing an increasingly important role in biological research. Drug-target networks, in particular, are made up of drug nodes that are linked to specific target nodes if a given drug is active with respect to that target.  Such networks provide a graphic depiction of polypharmacology and polyspecificity.  However, by their very nature they can obscure information that may be useful in their interpretation and analysis.  This work will show how such latent information can be used to determine bounds for the degrees of polypharmacology and polyspecificity, and how to estimate other useful features associated with the lack of completeness of most drug-target datasets.

In silico design of novel probes for the atypical opioid receptor MRGPRX2

The primate-exclusive MRGPRX2 G protein-coupled receptor (GPCR) has been suggested to modulate pain and itch. Despite putative peptide and small molecule MRGPRX2 agonists, selective nanomolar potency probes have not yet been reported. To identify a MRGPRX2 probe, we first screened 5,695 small molecules and found many opioid compounds activated MRGPRX2, including (−)- and (+)-morphine, hydrocodone, sinomenine, dextromethorphan and the prodynorphin-derived peptides, dynorphin A, dynorphin B, and α- and β-neoendorphin. We used these to select for mutagenesis-validated homology models and docked almost 4 million small molecules. From this docking, we predicted ZINC-3573, which represents a potent MRGPRX2-selective agonist, showing little activity against 315 other GPCRs and 97 representative kinases, and an essentially inactive enantiomer. ZINC-3573 activates endogenous MRGPRX2 in a human mast cell line inducing degranulation and calcium release. MRGPRX2 is a unique atypical opioid-like receptor important for modulating mast cell degranulation, which can now be specifically modulated with ZINC-3573.

Identification and analysis of promiscuity cliffs formed by bioactive compounds and experimental implications

For three promiscuity cliffs (enclosed), cliff compounds, their promiscuity degrees (PDs), and color-coded substitution sites are shown. Comparison of these cliffs suggests the design of a new analog to further explore promiscuity.

Evolutionary studies of ligand binding sites in proteins

Biological processes at their most fundamental molecular aspects are defined by molecular interactions with ligand-protein interactions in particular at the core of cellular functions such as metabolism and signalling. Divergent and convergent processes shape the evolution of ligand binding sites. The competition between similar ligands and binding sites across protein families create evolutionary pressures that affect the specificity and selectivity of interactions. This short review showcases recent studies of the evolution of ligand binding-sites and methods used to detect binding-site similarities.

Integrated Computational Approach for Virtual Hit Identification against Ebola Viral Proteins VP35 and VP40

The Ebola virus (EBOV) has been recognised for nearly 40 years, with the most recent EBOV outbreak being in West Africa, where it created a humanitarian crisis. Mortalities reported up to 30 March 2016 totalled 11,307. However, up until now, EBOV drugs have been far from achieving regulatory (FDA) approval. It is therefore essential to identify parent compounds that have the potential to be developed into effective drugs. Studies on Ebola viral proteins have shown that some can elicit an immunological response in mice, and these are now considered essential components of a vaccine designed to protect against Ebola haemorrhagic fever. The current study focuses on chemoinformatic approaches to identify virtual hits against Ebola viral proteins (VP35 and VP40), including protein binding site prediction, drug-likeness, pharmacokinetic and pharmacodynamic properties, metabolic site prediction, and molecular docking. Retrospective validation was performed using a database of non-active compounds, and early enrichment of EBOV actives at different false positive rates was calculated. Homology modelling and subsequent superimposition of binding site residues on other strains of EBOV were carried out to check residual conformations, and hence to confirm the efficacy of potential compounds. As a mechanism for artefactual inhibition of proteins through non-specific compounds, virtual hits were assessed for their aggregator potential compared with previously reported aggregators. These systematic studies have indicated that a few compounds may be effective inhibitors of EBOV replication and therefore might have the potential to be developed as anti-EBOV drugs after subsequent testing and validation in experiments in vivo.

Comparing atom-based with residue-based descriptors in predicting binding site similarity: do backbone atoms matter?

AIM

We question the level of detail required in protein 3D-representation to detect site similarity which is relevant for polypharmacology prediction.

RESULTS

We modified the in-house program SiteAlign to replace generic pharmacophoric descriptors of cavity-lining amino acids by descriptors accounting for solvent exposure. Benchmarking the novel, atom-based, method (SiteAlign2) revealed no global improvement of performance. However, in the rare cases of no sequence or global structure similarities between the compared proteins, SiteAlign2 was more successful if backbone atoms are key determinants of ligand binding.

CONCLUSION

SiteAlign suits the comparison of binding sites for close or distant homologs. SiteAlign2 provides a better insight into the physical model of site similarity between nonhomologs, but at the expense of an increased sensitivity to atomic coordinates.

Hydrogen Bonding of 1,2-Azaborines in the Binding Cavity of T4 Lysozyme Mutants: Structures and Thermodynamics

Protein crystallography and calorimetry were used to characterize the binding of 1,2-azaborines to model cavities in T4 lysozyme in direct comparison to their carbonaceous counterparts. In the apolar L99A cavity, affinity for Ab dropped only slightly versus benzene. In the cavity designed to accommodate a single hydrogen bond (L99A/M102Q), Gln102═O···H-N hydrogen bonding for Ab and BEtAb was observed in the crystallographic complexes. The strength of the hydrogen bonding was estimated as 0.94 and 0.64 kcal/mol for Ab and BEtAb, respectively. This work unambiguously demonstrates that 1,2-azaborines can be readily accommodated in classic aryl recognition pockets and establishes one of 1,2-azaborine's distinguishing features from its carbonaceous isostere benzene: its ability to serve as an NH hydrogen bond donor in a biological setting.

Rule of five in 2015 and beyond: Target and ligand structural limitations, ligand chemistry structure and drug discovery project decisions

The rule of five (Ro5), based on physicochemical profiles of phase II drugs, is consistent with structural limitations in protein targets and the drug target ligands. Three of four parameters in Ro5 are fundamental to the structure of both target and drug binding sites. The chemical structure of the drug ligand depends on the ligand chemistry and design philosophy. Two extremes of chemical structure and design philosophy exist; ligands constructed in the medicinal chemistry synthesis laboratory without input from natural selection and natural product (NP) metabolites biosynthesized based on evolutionary selection. Exceptions to Ro5 are found mostly among NPs. Chemistry chameleon-like behavior of some NPs due to intra-molecular hydrogen bonding as exemplified by cyclosporine A is a strong contributor to NP Ro5 outliers. The fragment derived, drug Navitoclax is an example of the extensive expertise, resources, time and key decisions required for the rare discovery of a non-NP Ro5 outlier.

Impact of Binding Site Comparisons on Medicinal Chemistry and Rational Molecular Design

Modern rational drug design not only deals with the search for ligands binding to interesting and promising validated targets but also aims to identify the function and ligands of yet uncharacterized proteins having impact on different diseases. Additionally, it contributes to the design of inhibitors with distinct selectivity patterns and the prediction of possible off-target effects. The identification of similarities between binding sites of various proteins is a useful approach to cope with those challenges. The main scope of this perspective is to describe applications of different protein binding site comparison approaches to outline their applicability and impact on molecular design. The article deals with various substantial application domains and provides some outstanding examples to show how various binding site comparison methods can be applied to promote in silico drug design workflows. In addition, we will also briefly introduce the fundamental principles of different protein binding site comparison methods.

Docking Screens for Novel Ligands Conferring New Biology

It is now plausible to dock libraries of 10 million molecules against targets over several days or weeks. When the molecules screened are commercially available, they may be rapidly tested to find new leads. Although docking retains important liabilities (it cannot calculate affinities accurately nor even reliably rank order high-scoring molecules), it can often can distinguish likely from unlikely ligands, often with hit rates above 10%. Here we summarize the improvements in libraries, target quality, and methods that have supported these advances, and the open access resources that make docking accessible. Recent docking screens for new ligands are sketched, as are the binding, crystallographic, and in vivo assays that support them. Like any technique, controls are crucial, and key experimental ones are reviewed. With such controls, docking campaigns can find ligands with new chemotypes, often revealing the new biology that may be docking's greatest impact over the next few years.

Machine learning models identify molecules active against the Ebola virus in vitro

The search for small molecule inhibitors of Ebola virus (EBOV) has led to several high throughput screens over the past 3 years. These have identified a range of FDA-approved active pharmaceutical ingredients (APIs) with anti-EBOV activity in vitro and several of which are also active in a mouse infection model. There are millions of additional commercially-available molecules that could be screened for potential activities as anti-EBOV compounds. One way to prioritize compounds for testing is to generate computational models based on the high throughput screening data and then virtually screen compound libraries. In the current study, we have generated Bayesian machine learning models with viral pseudotype entry assay and the EBOV replication assay data. We have validated the models internally and externally. We have also used these models to computationally score the MicroSource library of drugs to select those likely to be potential inhibitors. Three of the highest scoring molecules that were not in the model training sets, quinacrine, pyronaridine and tilorone, were tested in vitro and had EC50 values of 350, 420 and 230 nM, respectively. Pyronaridine is a component of a combination therapy for malaria that was recently approved by the European Medicines Agency, which may make it more readily accessible for clinical testing. Like other known antimalarial drugs active against EBOV, it shares the 4-aminoquinoline scaffold. Tilorone, is an investigational antiviral agent that has shown a broad array of biological activities including cell growth inhibition in cancer cells, antifibrotic properties, α7 nicotinic receptor agonist activity, radioprotective activity and activation of hypoxia inducible factor-1. Quinacrine is an antimalarial but also has use as an anthelmintic. Our results suggest data sets with less than 1,000 molecules can produce validated machine learning models that can in turn be utilized to identify novel EBOV inhibitors in vitro.

Machine learning models identify molecules active against the Ebola virus in vitro

The search for small molecule inhibitors of Ebola virus (EBOV) has led to several high throughput screens over the past 3 years. These have identified a range of FDA-approved active pharmaceutical ingredients (APIs) with anti-EBOV activity in vitro and several of which are also active in a mouse infection model. There are millions of additional commercially-available molecules that could be screened for potential activities as anti-EBOV compounds. One way to prioritize compounds for testing is to generate computational models based on the high throughput screening data and then virtually screen compound libraries. In the current study, we have generated Bayesian machine learning models with viral pseudotype entry assay and the EBOV replication assay data. We have validated the models internally and externally. We have also used these models to computationally score the MicroSource library of drugs to select those likely to be potential inhibitors. Three of the highest scoring molecules that were not in the model training sets, quinacrine, pyronaridine and tilorone, were tested in vitro and had EC ₅₀ values of 350, 420 and 230 nM, respectively. Pyronaridine is a component of a combination therapy for malaria that was recently approved by the European Medicines Agency, which may make it more readily accessible for clinical testing. Like other known antimalarial drugs active against EBOV, it shares the 4-aminoquinoline scaffold. Tilorone, is an investigational antiviral agent that has shown a broad array of biological activities including cell growth inhibition in cancer cells, antifibrotic properties, α7 nicotinic receptor agonist activity, radioprotective activity and activation of hypoxia inducible factor-1. Quinacrine is an antimalarial but also has use as an anthelmintic. Our results suggest data sets with less than 1,000 molecules can produce validated machine learning models that can in turn be utilized to identify novel EBOV inhibitors in vitro.

Machine learning models identify molecules active against the Ebola virus in vitro

The search for small molecule inhibitors of Ebola virus (EBOV) has led to several high throughput screens over the past 3 years. These have identified a range of FDA-approved active pharmaceutical ingredients (APIs) with anti-EBOV activity in vitro and several of which are also active in a mouse infection model. There are millions of additional commercially-available molecules that could be screened for potential activities as anti-EBOV compounds. One way to prioritize compounds for testing is to generate computational models based on the high throughput screening data and then virtually screen compound libraries. In the current study, we have generated Bayesian machine learning models with viral pseudotype entry assay and the EBOV replication assay data. We have validated the models internally and externally. We have also used these models to computationally score the MicroSource library of drugs to select those likely to be potential inhibitors. Three of the highest scoring molecules that were not in the model training sets, quinacrine, pyronaridine and tilorone, were tested in vitro and had EC ₅₀ values of 350, 420 and 230 nM, respectively. Pyronaridine is a component of a combination therapy for malaria that was recently approved by the European Medicines Agency, which may make it more readily accessible for clinical testing. Like other known antimalarial drugs active against EBOV, it shares the 4-aminoquinoline scaffold. Tilorone, is an investigational antiviral agent that has shown a broad array of biological activities including cell growth inhibition in cancer cells, antifibrotic properties, α7 nicotinic receptor agonist activity, radioprotective activity and activation of hypoxia inducible factor-1. Quinacrine is an antimalarial but also has use as an anthelmintic. Our results suggest data sets with less than 1,000 molecules can produce validated machine learning models that can in turn be utilized to identify novel EBOV inhibitors in vitro.