Large-Scale Pretraining Improves Sample Efficiency of Active Learning-Based Virtual Screening.

Virtual screening of large compound libraries to identify potential hit candidates is one of the earliest steps in drug discovery. As the size of commercially available compound collections grows exponentially to the scale of billions, active learning and Bayesian optimization have recently been proven as effective methods of narrowing down the search space. An essential component of those methods is a surrogate machine learning model that predicts the desired properties of compounds. An accurate model can achieve high sample efficiency by finding hits with only a fraction of the entire library being virtually screened. In this study, we examined the performance of a pretrained transformer-based language model and graph neural network in a Bayesian optimization active learning framework. The best pretrained model identifies 58.97% of the top-50,000 compounds after screening only 0.6% of an ultralarge library containing 99.5 million compounds, improving 8% over the previous state-of-the-art baseline. Through extensive benchmarks, we show that the superior performance of pretrained models persists in both structure-based and ligand-based drug discovery. Pretrained models can serve as a boost to the accuracy and sample efficiency of active learning-based virtual screening.

Machine Learning Methods as a Cost-Effective Alternative to Physics-Based Binding Free Energy Calculations.

The rank ordering of ligands remains one of the most attractive challenges in drug discovery. While physics-based in silico binding affinity methods dominate the field, they still have problems, which largely revolve around forcefield accuracy and sampling. Recent advances in machine learning have gained traction for protein-ligand binding affinity predictions in early drug discovery programs. In this article, we perform retrospective binding free energy evaluations for 172 compounds from our internal collection spread over four different protein targets and five congeneric ligand series. We compared multiple state-of-the-art free energy methods ranging from physics-based methods with different levels of complexity and conformational sampling to state-of-the-art machine-learning-based methods that were available to us. Overall, we found that physics-based methods behaved particularly well when the ligand perturbations were made in the solvation region, and they did not perform as well when accounting for large conformational changes in protein active sites. On the other end, machine-learning-based methods offer a good cost-effective alternative for binding free energy calculations, but the accuracy of their predictions is highly dependent on the experimental data available for training the model.

cMolGPT: A Conditional Generative Pre-Trained Transformer for Target-Specific De Novo Molecular Generation.

Deep generative models applied to the generation of novel compounds in small-molecule drug design have attracted a lot of attention in recent years. To design compounds that interact with specific target proteins, we propose a Generative Pre-Trained Transformer (GPT)-inspired model for de novo target-specific molecular design. By implementing different keys and values for the multi-head attention conditional on a specified target, the proposed method can generate drug-like compounds both with and without a specific target. The results show that our approach (cMolGPT) is capable of generating SMILES strings that correspond to both drug-like and active compounds. Moreover, the compounds generated from the conditional model closely match the chemical space of real target-specific molecules and cover a significant portion of novel compounds. Thus, the proposed Conditional Generative Pre-Trained Transformer (cMolGPT) is a valuable tool for de novo molecule design and has the potential to accelerate the molecular optimization cycle time.

Prospective Validation of Machine Learning Algorithms for Absorption, Distribution, Metabolism, and Excretion Prediction: An Industrial Perspective.

Absorption, distribution, metabolism, and excretion (ADME), which collectively define the concentration profile of a drug at the site of action, are of critical importance to the success of a drug candidate. Recent advances in machine learning algorithms and the availability of larger proprietary as well as public ADME data sets have generated renewed interest within the academic and pharmaceutical science communities in predicting pharmacokinetic and physicochemical endpoints in early drug discovery. In this study, we collected 120 internal prospective data sets over 20 months across six ADME in vitro endpoints: human and rat liver microsomal stability, MDR1-MDCK efflux ratio, solubility, and human and rat plasma protein binding. A variety of machine learning algorithms in combination with different molecular representations were evaluated. Our results suggest that gradient boosting decision tree and deep learning models consistently outperformed random forest over time. We also observed better performance when models were retrained on a fixed schedule, and the more frequent retraining generally resulted in increased accuracy, while hyperparameters tuning only improved the prospective predictions marginally.

Controlled sulfur-based engineering confers mouldability to phosphorothioate antisense oligonucleotides.

Phosphorothioates (PS) have proven their effectiveness in the area of therapeutic oligonucleotides with applications spanning from cancer treatment to neurodegenerative disorders. Initially, PS substitution was introduced for the antisense oligonucleotides (PS ASOs) because it confers an increased nuclease resistance meanwhile ameliorates cellular uptake and in-vivo bioavailability. Thus, PS oligonucleotides have been elevated to a fundamental asset in the realm of gene silencing therapeutic methodologies. But, despite their wide use, little is known on the possibly different structural changes PS-substitutions may provoke in DNA·RNA hybrids. Additionally, scarce information and significant controversy exists on the role of phosphorothioate chirality in modulating PS properties. Here, through comprehensive computational investigations and experimental measurements, we shed light on the impact of PS chirality in DNA-based antisense oligonucleotides; how the different phosphorothioate diastereomers impact DNA topology, stability and flexibility to ultimately disclose pro-Sp S and pro-Rp S roles at the catalytic core of DNA Exonuclease and Human Ribonuclease H; two major obstacles in ASOs-based therapies. Altogether, our results provide full-atom and mechanistic insights on the structural aberrations PS-substitutions provoke and explain the origin of nuclease resistance PS-linkages confer to DNA·RNA hybrids; crucial information to improve current ASOs-based therapies.

Critical Assessment of State-of-the-Art Ligand-Based Virtual Screening Methods.

The availability of large chemical libraries containing hundreds of millions to billions of diverse drug-like molecules combined with an almost unlimited amount of compute power to achieve scientific calculations has led investors and researchers to have a renewed interest in virtual screening (VS) methods to identify biologically active compounds. The number of in silico screening tools and software which employ the knowledge of the protein target or known bioactive ligands is increasing at a rapid pace, creating a crowded computational landscape where it has become difficult to assess the real advantages and disadvantages in terms of accuracy and efficiency of each individual VS technology. In the current work, we evaluate the performance of several state-of-the-art commercial software for 3D ligand-based VS against well-known 2D methods using an internally curated benchmarking data set. Our results show that the best individual methods can differ significantly based on the data set, and that combining them using data fusion techniques results in improved enrichment in the top 1 % of retrieved hits. Although 2D methods alone can already provide a significant enrichment in the number of predicted active compounds, the combination of data-fused 2D results with just one out of the best 3D methods (ROCS, FLAP or Blaze) further improves early enrichment and the likelihood of identifying additional chemotypes.

Harnessing Preclinical Data as a Predictive Tool for Human Brain Tissue Targeting.

One of the objectives within the medicinal chemistry discipline is to design tissue targeting molecules. The objective of tissue specificity can be either to gain drug access to the compartment of interest (e.g., the CNS) for Neuroscience targets or to restrict drug access to the CNS for all other therapeutic areas. Both neuroscience and non-neuroscience therapeutic areas have struggled to quantitatively estimate brain penetration or the lack thereof with compounds that are substrates of efflux transport proteins such as P-glycoprotein (P-gp) and breast cancer resistant protein (BCRP) that are key components of the blood-brain barrier (BBB). It has been well established that drug candidates with high efflux ratios (ER) of these transporters have poor penetration into brain tissue. In the current work, we outline a parallel analysis to previously published models for the prediction of brain penetration that utilize an alternate MDR1-MDCK cell line as a better predictor of brain penetration and whether a correlation between in vitro, rodent data, non-human primate (NHP), and human in vivo brain penetration data could be established. Analysis of structural and physicochemical properties in conjunction with in vitro parameters and preclinical in vivo data has been highlighted in this manuscript as a continuation of the previously published work.

PFRED: A computational platform for siRNA and antisense oligonucleotides design.

PFRED a software application for the design, analysis, and visualization of antisense oligonucleotides and siRNA is described. The software provides an intuitive user-interface for scientists to design a library of siRNA or antisense oligonucleotides that target a specific gene of interest. Moreover, the tool facilitates the incorporation of various design criteria that have been shown to be important for stability and potency. PFRED has been made available as an open-source project so the code can be easily modified to address the future needs of the oligonucleotide research community. A compiled version is available for downloading at https://github.com/pfred/pfred-gui/releases/tag/v1.0 as a java Jar file. The source code and the links for downloading the precompiled version can be found at https://github.com/pfred.

Photoaffinity Labeling and Quantitative Chemical Proteomics Identify LXRß as the Functional Target of Enhancers of Astrocytic apoE.

Utilizing a phenotypic screen, we identified chemical matter that increased astrocytic apoE secretion in vitro. We designed a clickable photoaffinity probe based on a pyrrolidine lead compound and carried out probe-based quantitative chemical proteomics in human astrocytoma CCF-STTG1 cells to identify liver x receptor ß (LXRß) as the target. Binding of the small molecule ligand stabilized LXRß, as shown by cellular thermal shift assay (CETSA). In addition, we identified a probe-modified peptide by mass spectrometry and proposed a model where the photoaffinity probe is bound in the ligand-binding pocket of LXRß. Taken together, our findings demonstrated that the lead chemical matter bound directly to LXRß, and our results highlight the power of chemical proteomic approaches to identify the target of a phenotypic screening hit. Additionally, the LXR photoaffinity probe and lead compound described herein may serve as valuable tools to further evaluate the LXR pathway.

A Novel Lipidomics-Based Approach to Evaluating the Risk of Clinical Hepatotoxicity Potential of Drugs in 3D Human Microtissues.

The importance of adsorption, distribution, metabolism, excretion, and toxicity (ADMET) analysis is expected to grow substantially due to recent failures in detecting severe toxicity issues of new chemical entities during preclinical/clinical development. Traditionally, safety risk assessment studies for humans have been conducted in animals during advanced preclinical or clinical phase of drug development. However, potential drug toxicity in humans now needs to be detected in the drug discovery process as soon as possible without reliance on animal studies. The "omics", such as genomics, proteomics, and metabolomics, have recently entered pharmaceutical research in both drug discovery and drug development, but to the best of our knowledge, no applications in high-throughput safety risk assessment have been attempted so far. This paper reports an innovative method to anticipate adverse drug effects in an early discovery phase based on lipid fingerprints using human three-dimensional microtissues. The risk of clinical hepatotoxicity potential was evaluated for a data set of 22 drugs belonging to five different therapeutic chemical classes and with various drug-induced liver injury effect. The treatment of microtissues with repeated doses of each drug allowed collecting lipid fingerprints for five time points (2, 4, 7, 9, and 11 days), and multivariate statistical analysis was applied to search for correlations with the hepatotoxic effect. The method allowed clustering of the drugs based on their hepatotoxic effect, and the observed lipid impairments for a number of drugs was confirmed by literature sources. Compared to traditional screening methods, here multiple interconnected variables (lipids) are measured simultaneously, providing a snapshot of the cellular status from the lipid perspective at a molecular level. Applied here to hepatotoxicity, the proposed workflow can be applied to several tissues, being tridimensional microtissues from various origins.

Understanding allosteric interactions in hMLKL protein that modulate necroptosis and its inhibition.

Mixed Lineage Kinase domain-Like (MLKL), a key player in necroptosis, is a multi-domain protein with an N-terminal 4 helical bundle (4HB) and a pseudokinase domain (PsK) connected by brace helices. Phosphorylation of PsK domain of MLKL is a key step towards oligomerization of 4HB domain that causes cell death. Necrosulfonamide (NSA) binds to the 4HB domain of MLKL to inhibit necroptosis. To understand the molecular details of MLKL function and it's inhibition, we have performed a molecular dynamic study on hMLKL protein in apo, phosphorylated and NSA-bound states for a total 3 µs simulation time. Our simulations show increased inter-domain flexibility, increased rigidification of the activation loop and increased alpha helical content in the brace helix region revealing a form of monomeric hMLKL necessary for oligomerization upon phosphorylation as compared to apo state. NSA binding disrupts this activated form and causes two main effects on hMLKL conformation: (1) locking of the relative orientation of 4HB and PsK domains by the formation of several new interactions and (2) prevention of key 4HB residues to participate in cross-linking for oligomer formation. This new understanding of the effect of hMLKL conformations on phosphorylation and NSA binding suggest new avenues for designing effective allosteric inhibitors of hMLKL.

From Target to Drug: Generative Modeling for the Multimodal Structure-Based Ligand Design.

Chemical space is impractically large, and conventional structure-based virtual screening techniques cannot be used to simply search through the entire space to discover effective bioactive molecules. To address this shortcoming, we propose a generative adversarial network to generate, rather than search, diverse three-dimensional ligand shapes complementary to the pocket. Furthermore, we show that the generated molecule shapes can be decoded using a shape-captioning network into a sequence of SMILES enabling directly the structure-based de novo drug design. We evaluate the quality of the method by both structure- (docking) and ligand-based [quantitative structure-activity relationship (QSAR)] virtual screening methods. For both evaluation approaches, we observed enrichment compared to random sampling from initial chemical space of ZINC drug-like compounds.

Author Correction: Dopamine D3 receptor antagonist reveals a cryptic pocket in aminergic GPCRs.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has not been fixed in the paper.

Discovering New Casein Kinase 1d Inhibitors with an Innovative Molecular Dynamics Enabled Virtual Screening Workflow.

The value of including protein flexibility in structure-based drug design (SBDD) is widely documented, and currently, molecular dynamics (MD) simulations represent a powerful tool to investigate protein dynamics. Yet, the inclusion of MD-derived information in pre-existing SBDD workflows is still far from trivial. We recently published an integrated MD-FLAP (Fingerprints for Ligands and Proteins) approach combining MD, clustering and Linear Discriminant Analysis (LDA) for enhancing accuracy, efficacy, and for protein conformational selection in virtual screening (VS) campaigns. Here we prospectively applied the MD-FLAP workflow to discover novel chemotypes inhibiting the Casein Kinase 1 delta (CSNK1D) enzyme. We first obtained a VS model able to separate active from inactive compounds, with a global AUC of 0.9 and a partial ROC enrichment at 0.5% of 0.18, and use it to mine the internal Pfizer screening database. Seven active molecules sharing a phenyl-indazole scaffold, not yet reported among CSNK1D inhibitors, were found. The most potent inhibitor showed an IC50 of 134 nM.

DeltaDelta neural networks for lead optimization of small molecule potency.

The capability to rank different potential drug molecules against a protein target for potency has always been a fundamental challenge in computational chemistry due to its importance in drug design. While several simulation-based methodologies exist, they are hard to use prospectively and thus predicting potency in lead optimization campaigns remains an open challenge. Here we present the first machine learning approach specifically tailored for ranking congeneric series based on deep 3D-convolutional neural networks. Furthermore we prove its effectiveness by blindly testing it on datasets provided by Janssen, Pfizer and Biogen totalling over 3246 ligands and 13 targets as well as several well-known openly available sets, representing one the largest evaluations ever performed. We also performed online learning simulations of lead optimization using the approach in a predictive manner obtaining significant advantage over experimental choice. We believe that the evaluation performed in this study is strong evidence of the usefulness of a modern deep learning model in lead optimization pipelines against more expensive simulation-based alternatives.

Dopamine D3 receptor antagonist reveals a cryptic pocket in aminergic GPCRs.

The recent increase in the number of X-ray crystal structures of G-protein coupled receptors (GPCRs) has been enabling for structure-based drug design (SBDD) efforts. These structures have revealed that GPCRs are highly dynamic macromolecules whose function is dependent on their intrinsic flexibility. Unfortunately, the use of static structures to understand ligand binding can potentially be misleading, especially in systems with an inherently high degree of conformational flexibility. Here, we show that docking a set of dopamine D3 receptor compounds into the existing eticlopride-bound dopamine D3 receptor (D3R) X-ray crystal structure resulted in poses that were not consistent with results obtained from site-directed mutagenesis experiments. We overcame the limitations of static docking by using large-scale high-throughput molecular dynamics (MD) simulations and Markov state models (MSMs) to determine an alternative pose consistent with the mutation data. The new pose maintains critical interactions observed in the D3R/eticlopride X-ray crystal structure and suggests that a cryptic pocket forms due to the shift of a highly conserved residue, F6.52. Our study highlights the importance of GPCR dynamics to understand ligand binding and provides new opportunities for drug discovery.

Hybrid peptide-small molecule oxytocin analogs are potent and selective agonists of the oxytocin receptor.

Oxytocin (OT) is a peptide hormone agonist of the oxytocin receptor (OTR) that has been proposed as a therapeutic to treat a number of social and emotional disorders in addition to its current clinical use to induce labor and treat postpartum bleeding. OT is administered intravenously and intranasally rather than orally, in part because its low passive permeability causes low oral bioavailability. Non-peptidic OTR agonists have also been reported, but none with the exquisite potency of the peptide based agonists. In this report, we describe the OTR agonist activity and exposed polarity of a set of truncated OT analogs as well as hybrid peptide-small molecule analogs of OT. Examples of both truncated analogs and peptide-small molecule hybrid analogs are potent and selective OTR agonists. Hybrid agonist 13, which is 232 Da smaller than OT, still retains subnanomolar potency, full agonist activity, and selectivity over V1a. While these compounds were designed to address the low permeability of OT and other full length analogs, we found that reduction in molecular weight and the removal or replacement of the three amino acid tail of OT did not have a significant effect on passive permeability.

The Discovery of a Novel Phosphodiesterase (PDE) 4B-Preferring Radioligand for Positron Emission Tomography (PET) Imaging.

As part of our effort in identifying phosphodiesterase (PDE) 4B-preferring inhibitors for the treatment of central nervous system (CNS) disorders, we sought to identify a positron emission tomography (PET) ligand to enable target occupancy measurement in vivo. Through a systematic and cost-effective PET discovery process, involving expression level (Bmax) and biodistribution determination, a PET-specific structure-activity relationship (SAR) effort, and specific binding assessment using a LC-MS/MS "cold tracer" method, we have identified 8 (PF-06445974) as a promising PET lead. Compound 8 has exquisite potency at PDE4B, good selectivity over PDE4D, excellent brain permeability, and a high level of specific binding in the "cold tracer" study. In subsequent non-human primate (NHP) PET imaging studies, [18F]8 showed rapid brain uptake and high target specificity, indicating that [18F]8 is a promising PDE4B-preferring radioligand for clinical PET imaging.

Dihedral Angle-Based Sampling of Natural Product Polyketide Conformations: Application to Permeability Prediction.

Macrocycles pose challenges for computer-aided drug design due to their conformational complexity. One fundamental challenge is identifying all low-energy conformations of the macrocyclic ring, which is important for modeling target binding, passive membrane permeation, and other conformation-dependent properties. Macrocyclic polyketides are medically and biologically important natural products characterized by structural and functional diversity. Advances in synthetic biology and semisynthetic methods may enable creation of an even more diverse set of non-natural product polyketides for drug discovery and other applications. However, the conformational sampling of these flexible compounds remains demanding. We developed and optimized a dihedral angle-based macrocycle conformational sampling method for macrocycles of arbitrary structure, and here we apply it to diverse polyketide natural products. First, we evaluated its performance using a data set of 37 polyketides with available crystal structures, with 9-22 rotatable bonds in the macrocyclic ring. Our optimized protocol was able to reproduce the crystal structure of polyketides' aglycone backbone within 0.50 Å RMSD for 31 out of 37 polyketides. Consistent with prior structural studies, our analysis suggests that polyketides tend to have multiple distinct low-energy structures, including the bioactive (target-bound) conformation as well as others of unknown significance. For this reason, we also introduce a strategy to improve both efficiency and accuracy of the conformational search by utilizing torsional restraints derived from NMR vicinal proton couplings to restrict the conformational search. Finally, as a first application of the method, we made blinded predictions of the passive membrane permeability of a diverse set of polyketides, based on their predicted structures in low- and high-dielectric media.

Systematic N-methylation of oxytocin: Impact on pharmacology and intramolecular hydrogen bonding network.

Oxytocin (OT) is a peptide hormone agonist of the OT receptor (OTR) that plays an important role in social behaviors such as pair bonding, maternal bonding and trust. The pharmaceutical development of OT as an oral peptide therapeutic has been hindered historically by its unfavorable physicochemical properties, including molecular weight, polarity and number of hydrogen bond donors, which determines poor cell permeability. Here we describe the first systematic study of single and multiple N-methylations of OT and their effect on physicochemical properties as well as potency at the OT receptor. The agonist EC50 and percent effect for OTR are reported and show that most N-methylations are tolerated but with some loss in potency compared to OT. The effect of N-methylation on exposed polarity is assessed through the EPSA chromatographic method and the results validated against NMR temperature coefficient experiments and the determination of NMR solution structures. We found that backbone methylation of residues not involved in IMHB and removal of the N-terminal amine can significantly reduce the exposed polarity of peptides, and yet retain a significant OTR agonist activity. The results of this study also expose the potential challenge of using the N-methylation strategy for the OT system; while exposed polarity is reduced, in some cases backbone methylation produces a significant conformational change that compromises agonist activity. The data presented provides useful insights on the SAR of OT and suggests future design strategies that can be used to develop more permeable OTR agonists based on the OT framework.