Early Stages Of Drug Discovery Research Articles

Prediction and Screening Model for Products Based on Fusion Regression and XGBoost Classification.

Performance prediction based on candidates and screening based on predicted performance value are the core of product development. For example, the performance prediction and screening of equipment components and parts are an important guarantee for the reliability of equipment products. The prediction and screening of drug bioactivity value and performance are the keys to pharmaceutical product development. The main reasons for the failure of pharmaceutical discovery are the low bioactivity of the candidate compounds and the deficiencies in their efficacy and safety, which are related to the absorption, distribution, metabolism, excretion, and toxicity (ADMET) of the compounds. Therefore, it is very necessary to quickly and effectively perform systematic bioactivity value prediction and ADMET property evaluation for candidate compounds in the early stage of drug discovery. In this paper, a data-driven pharmaceutical products screening prediction model is proposed to screen drug candidates with higher bioactivity value and better ADMET properties. First, a quantitative prediction method for bioactivity value is proposed using the fusion regression of LGBM and neural network based on backpropagation (BP-NN). Then, the ADMET properties prediction method is proposed using XGBoost. According to the predicted bioactivity value and ADMET properties, the BVAP method is defined to screen the drug candidates. And the screening model is validated on the dataset of antagonized Erα active compounds, in which the mean square error (MSE) of fusion regression is 1.1496, the XGBoost prediction accuracy of ADMET properties are 94.0% for Caco-2, 95.7% for CYP3A4, 89.4% for HERG, 88.6% for hob, and 96.2% for Mn. Compared with the commonly used methods for ADMET properties such as SVM, RF, KNN, LDA, and NB, the XGBoost in this paper has the highest prediction accuracy and AUC value, which has better guiding significance and can help screen pharmaceutical product candidates with good bioactivity, pharmacokinetic properties, and safety.

Protein-Ligand Docking in the Machine-Learning Era.

Molecular docking plays a significant role in early-stage drug discovery, from structure-based virtual screening (VS) to hit-to-lead optimization, and its capability and predictive power is critically dependent on the protein–ligand scoring function. In this review, we give a broad overview of recent scoring function development, as well as the docking-based applications in drug discovery. We outline the strategies and resources available for structure-based VS and discuss the assessment and development of classical and machine learning protein–ligand scoring functions. In particular, we highlight the recent progress of machine learning scoring function ranging from descriptor-based models to deep learning approaches. We also discuss the general workflow and docking protocols of structure-based VS, such as structure preparation, binding site detection, docking strategies, and post-docking filter/re-scoring, as well as a case study on the large-scale docking-based VS test on the LIT-PCBA data set.

Mass Spectrometry Detection and Imaging of a Non-Covalent Protein-Drug Complex in Tissue from Orally Dosed Rats.

Here, we demonstrate detection by mass spectrometry of an intact protein–drug complex directly from liver tissue from rats that had been orally dosed with the drug. The protein–drug complex comprised fatty acid binding protein 1, FABP1, non‐covalently bound to the small molecule therapeutic bezafibrate. Moreover, we demonstrate spatial mapping of the [FABP1+bezafibrate] complex across a thin section of liver by targeted mass spectrometry imaging. This work is the first demonstration of in situ mass spectrometry analysis of a non‐covalent protein–drug complex formed in vivo and has implications for early stage drug discovery by providing a route to target‐drug characterization directly from the physiological environment.

Mass Spectrometry Detection and Imaging of a Non-Covalent Protein-Drug Complex in Tissue from Orally Dosed Rats.

Here, we demonstrate detection by mass spectrometry of an intact protein-drug complex directly from liver tissue from rats that had been orally dosed with the drug. The protein-drug complex comprised fatty acid binding protein 1, FABP1, non-covalently bound to the small molecule therapeutic bezafibrate. Moreover, we demonstrate spatial mapping of the [FABP1+bezafibrate] complex across a thin section of liver by targeted mass spectrometry imaging. This work is the first demonstration of in situ mass spectrometry analysis of a non-covalent protein-drug complex formed in vivo and has implications for early stage drug discovery by providing a route to target-drug characterization directly from the physiological environment.

Intracellular Absolute Quantification of Oligonucleotide Therapeutics by NanoSIMS.

Antisense oligonucleotide (ASO)-based therapeutics hold great potential for the treatment of a variety of diseases. Therefore, a better understanding of cellular delivery, uptake, and trafficking mechanisms of ASOs is highly important for early-stage drug discovery. In particular, understanding the biodistribution and quantifying the abundance of ASOs at the subcellular level are needed to fully characterize their activity. Here, we used a combination of electron microscopy and NanoSIMS to assess the subcellular concentrations of a 34S-labeled GalNAc-ASO and a naked ASO in the organelles of primary human hepatocytes. We first cross-validated the method by including a 127I-labeled ASO, finding that the absolute concentration of the lysosomal ASO using two independent labeling strategies gave matching results, demonstrating the strength of our approach. This work also describes the preparation of external standards for absolute quantification by NanoSIMS. For both the 34S and 127I approaches used for our quantification methodology, we established the limit of detection (5 and 2 μM, respectively) and the lower limit of quantification (14 and 5 μM, respectively).

Identifying SARS-CoV-2 main protease inhibitors by a novel sandwich-like fluorescence polarization screening assay

SARS-CoV-2 main protease (Mpro) is responsible for polyprotein cleavage to release non-structural proteins (nsps) for viral genomic RNA replication, and its homologues are absent in human cells. Therefore, Mpro has been regarded as one of the ideal drug targets for the treatment of coronavirus disease 2019 (COVID-19). In this study, we first combined the fluorescence polarization (FP) technique with biotin-avidin system (BAS) to develop a novel sandwich-like FP screening assay for quick discovery of SARS-CoV-2 Mpro inhibitors from a natural product library. With this screening assay, anacardic acid (AA) and 1, 2, 3, 4, 6-O-pentagalloylglucose (PGG) were found to be the competitive inhibitor and mixed-type inhibitor targeting Mpro, respectively. Importantly, our results showed that the majority of the reported Mpro inhibitors are promiscuous cysteine inhibitors that are not specific to Mpro. In summary, this novel sandwich-like FP screening assay is simple, sensitive, and robust, which is ideal for large-scale screening. Natural products AA and PGG will be the promising lead compounds for generating more potent antiviral agents targeting Mpro, and the stringent hit validation at the early stage of drug discovery is urgently needed.

Abstract 1880: High throughput screening of 30 PDX cell lines in a 3D ECM hydrogel platform, incorporating tumor, stroma and immune components to demonstrate simultaneous investigation of multiple anti-tumor modalities

Abstract High throughput screening offers tangible benefits towards rapidly testing various permutations of novel or existing therapeutic agents. In particular, tumor panels that cover a range of histotypes and molecular subtypes have been previously developed, such as the NCI-60, however they utilize cell lines and, in some cases, a 2D cell culture format, which limit their translatability to preclinical and clinical trials. Moreover, the biological complexity of the tumor microenvironment (TME) has revealed a need for more translatable 3D in vitro tumor models that reflect the in vivo physiological outcome to therapies, particularly with the explosion of immunotherapy programs in drug discovery which target the immune compartment of the TME. Here, we describe for the first time a 3D in vitro PDX panel comprising 30 distinct PDX models in coculture with fibroblasts and PBMCs in engineered extracellular matrix hydrogels that display distinct similarities to the three compartments of the TME - tumor, stroma, and immune cells. Histotypes in the panel include, amongst others, Non-small cell lung, colorectal, breast, pancreatic, gastric, melanoma, and renal cancers. The panel is constructed in a high throughput 96-well format and rapidly assays tumor growth delay and other endpoints such as apoptosis in a dose-dependent manner across various drug modalities such as small molecules, biologics and cell therapy. The panel has been tested against targeted therapy (Docetaxel, Cisplatin) and immunomodulatory agents (Pembrolizumab, Atezolizumab) and the results correlate to the corresponding PDX in vivo data. Interestingly, cisplatin treatment displayed an average size reduction of >30% for melanoma but a varied response in NSCLC tumors. And, several NSCLC responded more favorably to checkpoint inhibitors, however, the lung epidermoid PDX demonstrated no significant response. Moreover, subsequent cytokine analysis and immunofluorescence staining of several models revealed protein signatures of cancer-associated fibroblasts and CD3+ sequestration in the tumor stroma in some 3D models, suggesting the fibroblasts’ critical role in regulating the immune response. In short, the 30-PDX Panel described here represents a large step forward towards achieving translatable efficacy data at the earliest stages of drug discovery where little is known about the mechanism of action for a particular therapeutic agent or combination of agents. Citation Format: Kolin C. Hribar, Bin Xue, Christopher Harrod, Timothy Jensen, Julia Schüler. High throughput screening of 30 PDX cell lines in a 3D ECM hydrogel platform, incorporating tumor, stroma and immune components to demonstrate simultaneous investigation of multiple anti-tumor modalities [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 1880.

The Ukrainian Factor in Early-Stage Drug Discovery in the Context of Russian Invasion: The Case of Enamine Ltd.

Ukrainian companies occupy an important niche in the global drug discovery process; however, before the Russian invasion, the role of Ukraine was not obvious. The biggest Ukrainian fine chemical supplier, Enamine Ltd, had to stop operation for more than a month, which significantly affected various early-stage drug discovery projects. The role of Enamine in drug discovery and the company's past and future in the context of the Russian invasion are described in this Viewpoint.

Differential proteomics analysis of JEG-3 and JAR placental cell models and the effect of androgen treatment

The placenta is a vital fetal organ that plays an important role in maintaining fetal sex hormone homeostasis. Xenobiotics can alter placental sex-steroidogenic enzymes and transporters, including enzymes such as aromatase (CYP19A1) and the hydroxysteroid dehydrogenases (HSDs) but studying how compounds disrupt in vivo placental metabolism is complex. Utilizing high-throughput in vitro models is critical to predict the disruption of placental sex-steroidogenic enzymes and transporters, particularly by drug candidates in the early stages of drug discovery. JAR and JEG-3 cells are the most common, simple, and cost-effective placental cell models that are capable of high-throughput screening, but how well they express the sex-steroidogenic enzymes and transporters is not well known. Here, we compared the proteomes of JAR and JEG-3 cells in the presence and absence of physiologically relevant concentrations of dehydroepiandrosterone (DHEA, 8 µM) and testosterone (15 nM) to aid the characterization of sex-steroidogenic enzymes and transporters in these cell models. Global proteomics analysis detected 2931 and 3449 proteins in JAR cells and JEG-3 cells, respectively. However, dramatic differences in sex-steroidogenic enzymes and transporters were observed between these cells. In particular, the basal expression of steroid sulfatase (STS), HSD17B1, and HSD17B7 were unique to JEG-3 cells. JEG-3 cells also showed significantly higher protein levels of aldo-keto reductase (AKR) 1A1 and AKR1B1, while JAR cells showed significantly higher levels of HSD17B4 and HSDB12. Aldehyde dehydrogenase (ALDH) 3A2 and HSD17B11 enzymes as well as the transporters sterol O-acyltransferase (SOAT) 1 and ATP binding cassette subfamily G2 (ABCG2) were comparable between the cell lines, whereas sulfotransferases (SULTs) were uniquely present within JAR cells. Androgen treatments significantly lowered HSD17B11, HSD17B4, HSD17B12, and ALDH3A2 levels in JAR cells. DHEA treatment significantly raised the level of HSD17B1 by 51 % in JEG-3 cells, whereas CYP19A1 was increased to significant levels in both JAR and JEG-3 cells after androgen treatments. The proteomics data were supported by a complementary targeted metabolomics analysis of culture media in the DHEA (8 µM) and testosterone (15 nM) treated groups. This study has indicated that untreated JEG-3 cells express more sex-steroidogenic enzymes and transporters. Nevertheless, JEG-3 and JAR cells are unique and their respective proteomics data can be used to select the best model depending on the hypothesis.

Early Feasibility Assessment: A Method for Accurately Predicting Biotherapeutic Dosing to Inform Early Drug Discovery Decisions.

The application of model-informed drug discovery and development (MID3) approaches in the early stages of drug discovery can help determine feasibility of drugging a target, prioritize between targets, or define optimal drug properties for a target product profile (TPP). However, applying MID3 in early discovery can be challenging due to the lack of pharmacokinetic (PK) and pharmacodynamic (PD) data at this stage. Early Feasibility Assessment (EFA) is the application of mechanistic PKPD models, built from first principles, and parameterized by data that is readily available early in drug discovery to make effective dose predictions. This manuscript demonstrates the ability of EFA to make accurate predictions of clinical effective doses for nine approved biotherapeutics and outlines the potential of extending this approach to novel therapeutics to impact early drug discovery decisions.

Application of Pharmacokinetic Prediction Platforms in the Design of Optimized Anti-Cancer Drugs.

Cancer is the second most common cause of death in the United States, accounting for 602,350 deaths in 2020. Cancer-related death rates have declined by 27% over the past two decades, partially due to the identification of novel anti-cancer drugs. Despite improvements in cancer treatment, newly approved oncology drugs are associated with increased toxicity risk. These toxicities may be mitigated by pharmacokinetic optimization and reductions in off-target interactions. As such, there is a need for early-stage implementation of pharmacokinetic (PK) prediction tools. Several PK prediction platforms exist, including pkCSM, SuperCypsPred, Pred-hERG, Similarity Ensemble Approach (SEA), and SwissADME. These tools can be used in screening hits, allowing for the selection of compounds were reduced toxicity and/or risk of attrition. In this short commentary, we used PK prediction tools in the optimization of mitogen activated extracellular signal-related kinase kinase 1 (MEK1) inhibitors. In doing so, we identified MEK1 inhibitors with retained activity and optimized predictive PK properties, devoid of hERG inhibition. These data support the use of publicly available PK prediction platforms in early-stage drug discovery to design safer drugs.

Surface Plasmon Resonance Screening to Identify Active and Selective Adenosine Receptor Binding Fragments.

Surface plasmon resonance (SPR) is a widely used method to study ligand-protein interactions. The throughput and sensitivity of SPR has made it an important technology for measuring low-affinity, ultralow weight fragments (<200 Da) in the early stages of drug discovery. However, the biochemistry of membrane proteins, such as G-protein-coupled receptors (GPCRs), makes their SPR fragment screening particularly challenging, especially for native/wild-type, nonthermostabilized mutant receptors. In this study, we demonstrate the use of SPR-based biosensors to study the entire human family of adenosine receptors and present biologically active novel binders with a range of selectivity to human adenosine 2a receptor (hA2AR) from an ultralow weight fragment library and the public GlaxoSmithKline (GSK) kinase library. Thus, we demonstrate the ability of SPR to screen ultra-low-affinity fragments and identify biologically meaningful chemical equity and that SPR campaigns are highly effective "chemical filters" for screening small building block fragments that can be used to enable drug discovery programs.

Solvatochromic peptidic binder obtained via extended phage display acts as a fluororeporter for fragment-based drug discovery (FBDD).

We have previously established a selection system to obtain a solvatochromic protein binder from a peptidic fluoroprobe library via the extended T7 phage display. Here, we use the peptidic binder as a fluororeporter in this proof-of-concept study of fragment-based screening approach to drug discovery. The binder is released from the target protein on mixing with an appropriate lead compound, thereby altering its fluorescence color/intensity under 365nm ultraviolet wavelength irradiation. By this instant screening outcome, the affinity of the lead compound is apparent to the naked eye, and quantified with a portable microvolume fluorophotometer. We envision that our simple and affordable screening system will provideopportunities for early stage drug discovery, especially for non-experts in academia and education because expensive hardware is not required for qualifying the measurements.

On the Frustration to Predict Binding Affinities from Protein-Ligand Structures with Deep Neural Networks.

Accurate prediction of binding affinities from protein-ligand atomic coordinates remains a major challenge in early stages of drug discovery. Using modular message passing graph neural networks describing both the ligand and the protein in their free and bound states, we unambiguously evidence that an explicit description of protein-ligand noncovalent interactions does not provide any advantage with respect to ligand or protein descriptors. Simple models, inferring binding affinities of test samples from that of the closest ligands or proteins in the training set, already exhibit good performances, suggesting that memorization largely dominates true learning in the deep neural networks. The current study suggests considering only noncovalent interactions while omitting their protein and ligand atomic environments. Removing all hidden biases probably requires much denser protein-ligand training matrices and a coordinated effort of the drug design community to solve the necessary protein-ligand structures.

HelixADMET: a robust and endpoint extensible ADMET system incorporating self-supervised knowledge transfer.

Accurate ADMET (an abbreviation for 'absorption, distribution, metabolism, excretion and toxicity') predictions can efficiently screen out undesirable drug candidates in the early stage of drug discovery. In recent years, multiple comprehensive ADMET systems that adopt advanced machine learning models have been developed, providing services to estimate multiple endpoints. However, those ADMET systems usually suffer from weak extrapolation ability. First, due to the lack of labelled data for each endpoint, typical machine learning models perform frail for the molecules with unobserved scaffolds. Second, most systems only provide fixed built-in endpoints and cannot be customized to satisfy various research requirements. To this end, we develop a robust and endpoint extensible ADMET system, HelixADMET (H-ADMET). H-ADMET incorporates the concept of self-supervised learning to produce a robust pre-trained model. The model is then fine-tuned with a multi-task and multi-stage framework to transfer knowledge between ADMET endpoints, auxiliary tasks and self-supervised tasks. Our results demonstrate that H-ADMET achieves an overall improvement of 4%, compared with existing ADMET systems on comparable endpoints. Additionally, the pre-trained model provided by H-ADMET can be fine-tuned to generate new and customized ADMET endpoints, meeting various demands of drug research and development requirements. H-ADMET is freely accessible at https://paddlehelix.baidu.com/app/drug/admet/train. Supplementary data are available at Bioinformatics online.

Development of Simple and Accurate in Silico Ligand-Based Models for Predicting ABCG2 Inhibition.

The ATP binding cassette transporter ABCG2 is a physiologically important drug transporter that has a central role in determining the ADMET (absorption, distribution, metabolism, elimination, and toxicity) profile of therapeutics, and contributes to multidrug resistance. Thus, development of predictive in silico models for the identification of ABCG2 inhibitors is of great interest in the early stage of drug discovery. In this work, by exploiting a large public dataset, a number of ligand-based classification models were developed using partial least squares-discriminant analysis (PLS-DA) with molecular interaction field- and fingerprint-based structural description methods, regarding physicochemical and fragmental properties related to ABCG2 inhibition. An in-house dataset compiled from recently experimental studies was used to rigorously validated the model performance. The key molecular properties and fragments favored to inhibitor binding were discussed in detail, which was further explored by docking simulations. A highly informative chemical property was identified as the principal determinant of ABCG2 inhibition, which was utilized to derive a simple rule that had a strong capability for differentiating inhibitors from non-inhibitors. Furthermore, the incorporation of the rule into the best PLS-DA model significantly improved the classification performance, particularly achieving a high prediction accuracy on the independent in-house set. The integrative model is simple and accurate, which could be applied to the evaluation of drug-transporter interactions in drug development. Also, the dominant molecular features derived from the models may help medicinal chemists in the molecular design of novel inhibitors to circumvent ABCG2-mediated drug resistance.

Two-peak Elution Profile of a Bispecific VHH-IgG Fusion Protein in Ion Exchange Purification Process caused by Atypical N-Glycosylation

In-depth analysis and thorough understanding of Nglycosylation is critical for therapeutic proteins. For IgG or IgG-like molecules, the conserved N-glycosylation at Asn297 has critical safety and efficacy implications and therefore is under close scrutiny. Any additional N-glycosylation that increases complexity and heterogeneity of the molecule, should be avoided at the early stage of drug discovery. However, unexpected glycosylation at atypical sites brings unforeseen challenges for antibody drug development process. Here, we observed a two-peak elution profile when developing ion exchange purification process for a novel bispecific VHH-IgG fusion protein. The charge and size properties of the two peaks were characterized by a variety of analytical methods. Two atypical N-glycosylation sites in the VHH domain were identified by means of mass spectrometry and confirmed by site-specific mutagenesis. Further characterization indicated that G0F was the predominant glycan species for both sites with varying occupancy. Interestingly, the addition of neutral glycans changed the charge behaviour of the fusion protein, leading to the unexpected ion exchange profile. These findings suggest that glycosylation may occur at non-canonical sites and may have significant effect on the process development and drug developability.

An academic–industrial collaboration delivering better starting points for drug discovery: new compound libraries driven by intelligent design

An academic–industrial collaboration delivering better starting points for drug discovery: new compound libraries driven by intelligent design Project iDesign’s primary research objective is to deliver better starting points for early-stage drug discovery. To this end, we have synthesised compound libraries comprising molecules that are demonstrably novel, structurally and functionally diverse, and lead-like in character. Designed with safety considerations from the outset, our libraries interrogate underexplored regions of chemical space and therefore complement existing diversity sets.

Pose Classification Using Three-Dimensional Atomic Structure-Based Neural Networks Applied to Ion Channel-Ligand Docking.

The identification of promising lead compounds showing pharmacological activities toward a biological target is essential in early stage drug discovery. With the recent increase in available small-molecule databases, virtual high-throughput screening using physics-based molecular docking has emerged as an essential tool in assisting fast and cost-efficient lead discovery and optimization. However, the best scored docking poses are often suboptimal, resulting in incorrect screening and chemical property calculation. We address the pose classification problem by leveraging data-driven machine learning approaches to identify correct docking poses from AutoDock Vina and Glide screens. To enable effective classification of docking poses, we present two convolutional neural network approaches: a three-dimensional convolutional neural network (3D-CNN) and an attention-based point cloud network (PCN) trained on the PDBbind refined set. We demonstrate the effectiveness of our proposed classifiers on multiple evaluation data sets including the standard PDBbind CASF-2016 benchmark data set and various compound libraries with structurally different protein targets including an ion channel data set extracted from Protein Data Bank (PDB) and an in-house KCa3.1 inhibitor data set. Our experiments show that excluding false positive docking poses using the proposed classifiers improves virtual high-throughput screening to identify novel molecules against each target protein compared to the initial screen based on the docking scores.

Knowledge-based BERT: a method to extract molecular features like computational chemists.

Molecular property prediction models based on machine learning algorithms have become important tools to triage unpromising lead molecules in the early stages of drug discovery. Compared with the mainstream descriptor- and graph-based methods for molecular property predictions, SMILES-based methods can directly extract molecular features from SMILES without human expert knowledge, but they require more powerful algorithms for feature extraction and a larger amount of data for training, which makes SMILES-based methods less popular. Here, we show the great potential of pre-training in promoting the predictions of important pharmaceutical properties. By utilizing three pre-training tasks based on atom feature prediction, molecular feature prediction and contrastive learning, a new pre-training method K-BERT, which can extract chemical information from SMILES like chemists, was developed. The calculation results on 15 pharmaceutical datasets show that K-BERT outperforms well-established descriptor-based (XGBoost) and graph-based (Attentive FP and HRGCN+) models. In addition, we found that the contrastive learning pre-training task enables K-BERT to 'understand' SMILES not limited to canonical SMILES. Moreover, the general fingerprints K-BERT-FP generated by K-BERT exhibit comparative predictive power to MACCS on 15 pharmaceutical datasets and can also capture molecular size and chirality information that traditional binary fingerprints cannot capture. Our results illustrate the great potential of K-BERT in the practical applications of molecular property predictions in drug discovery.