High-Throughput Screening

Streamlining Computational Fragment-Based Drug Discovery through Evolutionary Optimization Informed by Ligand-Based Virtual Prescreening

Recent advances in computational methods provide the promise of dramatically accelerating drug discovery. While mathematical modeling and machine learning have become vital in predicting drug-target interactions and properties, there is untapped potential in computational drug discovery due to the vast and complex chemical space. This paper builds on our recently published computational fragment-based drug discovery (FBDD) method called fragment databases from screened ligand drug discovery (FDSL-DD). FDSL-DD uses in silico screening to identify ligands from a vast library, fragmenting them while attaching specific attributes based on predicted binding affinity and interaction with the target subdomain. In this paper, we further propose a two-stage optimization method that utilizes the information from prescreening to optimize computational ligand synthesis. We hypothesize that using prescreening information for optimization shrinks the search space and focuses on promising regions, thereby improving the optimization for candidate ligands. The first optimization stage assembles these fragments into larger compounds using genetic algorithms, followed by a second stage of iterative refinement to produce compounds with enhanced bioactivity. To demonstrate broad applicability, the methodology is demonstrated on three diverse protein targets found in human solid cancers, bacterial antimicrobial resistance, and the SARS-CoV-2 virus. Combined, the proposed FDSL-DD and a two-stage optimization approach yield high-affinity ligand candidates more efficiently than other state-of-the-art computational FBDD methods. We further show that a multiobjective optimization method accounting for drug-likeness can still produce potential candidate ligands with a high binding affinity. Overall, the results demonstrate that integrating detailed chemical information with a constrained search framework can markedly optimize the initial drug discovery process, offering a more precise and efficient route to developing new therapeutics.

Advances in high throughput cell culture technologies for therapeutic screening and biological discovery applications

Cellular phenotypes and functional responses are modulated by the signals present in their microenvironment, including extracellular matrix (ECM) proteins, tissue mechanical properties, soluble signals and nutrients, and cell-cell interactions. To better recapitulate and analyze these complex signals within the framework of more physiologically relevant culture models, high throughput culture platforms can be transformative. High throughput methodologies enable scientists to extract increasingly robust and broad datasets from individual experiments, screen large numbers of conditions for potential hits, better qualify and predict responses for preclinical applications, and reduce reliance on animal studies. High throughput cell culture systems require uniformity, assay miniaturization, specific target identification, and process simplification. In this review, we detail the various techniques that researchers have used to face these challenges and explore cellular responses in a high throughput manner. We highlight several common approaches including two-dimensional multiwell microplates, microarrays, and microfluidic cell culture systems as well as unencapsulated and encapsulated three-dimensional high throughput cell culture systems, featuring multiwell microplates, micromolds, microwells, microarrays, granular hydrogels, and cell-encapsulated microgels. We also discuss current applications of these high throughput technologies, namely stem cell sourcing, drug discovery and predictive toxicology, and personalized medicine, along with emerging opportunities and future impact areas.

MELLODDY: Cross-pharma Federated Learning at Unprecedented Scale Unlocks Benefits in QSAR without Compromising Proprietary Information

Federated multipartner machine learning has been touted as an appealing and efficient method to increase the effective training data volume and thereby the predictivity of models, particularly when the generation of training data is resource-intensive. In the landmark MELLODDY project, indeed, each of ten pharmaceutical companies realized aggregated improvements on its own classification or regression models through federated learning. To this end, they leveraged a novel implementation extending multitask learning across partners, on a platform audited for privacy and security. The experiments involved an unprecedented cross-pharma data set of 2.6+ billion confidential experimental activity data points, documenting 21+ million physical small molecules and 40+ thousand assays in on-target and secondary pharmacodynamics and pharmacokinetics. Appropriate complementary metrics were developed to evaluate the predictive performance in the federated setting. In addition to predictive performance increases in labeled space, the results point toward an extended applicability domain in federated learning. Increases in collective training data volume, including by means of auxiliary data resulting from single concentration high-throughput and imaging assays, continued to boost predictive performance, albeit with a saturating return. Markedly higher improvements were observed for the pharmacokinetics and safety panel assay-based task subsets.

Using Generative Modeling to Endow with Potency Initially Inert Compounds with Good Bioavailability and Low Toxicity

In the early stages of drug development, large chemical libraries are typically screened to identify compounds of promising potency against the chosen targets. Often, however, the resulting hit compounds tend to have poor drug metabolism and pharmacokinetics (DMPK), with negative developability features that may be difficult to eliminate. Therefore, starting the drug discovery process with a "null library", compounds that have highly desirable DMPK properties but no potency against the chosen targets, could be advantageous. Here, we explore the opportunities offered by machine learning to realize this strategy in the case of the inhibition of α-synuclein aggregation, a process associated with Parkinson's disease. We apply MolDQN, a generative machine learning method, to build an inhibitory activity against α-synuclein aggregation into an initial inactive compound with good DMPK properties. Our results illustrate how generative modeling can be used to endow initially inert compounds with desirable developability properties.

Drug Release from Surfactant-Containing Amorphous Solid Dispersions: Mechanism and Role of Surfactant in Release Enhancement

PURPOSE

To understand how surfactants affect drug release from ternary amorphous solid dispersions (ASDs), and to investigate different mechanisms of release enhancement.

METHODS

Ternary ASDs containing ritonavir (RTV), polyvinylpyrrolidone/vinyl acetate (PVPVA) and a surfactant (sodium dodecyl sulfate (SDS), Tween 80, Span 20 or Span 85) were prepared with rotary evaporation. Release profiles of ternary ASDs were measured with surface normalized dissolution. Phase separation morphologies of ASD compacts during hydration/dissolution were examined in real-time with a newly developed confocal fluorescence microscopy method. The water ingress rate of different formulations was measured with dynamic vapor sorption. Microscopy was employed to check for matrix crystallization during release studies.

RESULTS

All surfactants improved drug release at 30% DL, while only SDS and Tween 80 improved drug release at higher DLs, although SDS promoted matrix crystallization. The dissolution rate of neat polymer increased when SDS and Tween 80 were present. The water ingress rate also increased in the presence of all surfactants. Surfactant-incorporation affected both the kinetic and thermodynamics factors governing phase separation of RTV-PVPVA-water system, modifying the phase morphology during ASD dissolution. Importantly, SDS increased the miscibility of RTV-PVPVA-water system, whereas other surfactants mainly affected the phase separation kinetics/drug-rich barrier persistence.

CONCLUSION

Incorporation of surfactants enhanced drug release from RTV-PVPVA ASDs compared to the binary system. Increased drug-polymer-water miscibility and disruption of the drug-rich barrier at the gel-solvent interface via plasticization are highlighted as two key mechanisms underlying surfactant impacts based on direct visualization of the phase separation process upon hydration and release.

Tissue-Engineered 3D In Vitro Disease Models for High-Throughput Drug Screening

During high-throughput drug screening, in vitro models are fabricated and the effects of therapeutics on the models evaluated in high throughput-for example, with automated liquid handling systems and microplate reader-based high-throughput screening (HTS) assays. The most frequently-used model systems for HTS, 2D models, do not adequately model the in vivo 3D microenvironment-an important aspect of which is the extracellular matrix-and therefore, 2D models may not be appropriate for drug screening. Instead, tissue-engineered 3D models with extracellular matrix-mimicking components are destined to become the preferred in vitro systems for HTS. However, for 3D models, such as 3D cell-laden hydrogels and scaffolds, cell sheets, and spheroids as well as 3D microfluidic and organ-on-a-chip systems, to replace 2D models in HTS, they must be compatible with high-throughput fabrication schemes and evaluation methods. In this review, we summarize HTS in 2D models and discuss recent studies that have successfully demonstrated HTS-compatible 3D models of high-impact diseases, such as cancers or cardiovascular diseases.

Fluorescent probe for the identification of potent inhibitors of the macrophage infectivity potentiator (Mip) protein of Burkholderia pseudomallei

The macrophage infectivity potentiator (Mip) protein belongs to the immunophilin superfamily. This class of enzymes catalyzes the interconversion between the cis and trans configuration of proline-containing peptide bonds. Mip has been shown to be important for the virulence of a wide range of pathogenic microorganisms, including the Gram-negative bacterium Burkholderia pseudomallei. Small molecules derived from the natural product rapamycin, lacking its immunosuppression-inducing moiety, inhibit Mip's peptidyl-prolyl cis-trans isomerase (PPIase) activity and lead to a reduction in pathogen load in vitro. Here, a fluorescence polarization assay (FPA) to enable the screening and effective development of BpMip inhibitors was established. A fluorescent probe was prepared, derived from previous pipecolic scaffold Mip inhibitors labeled with fluorescein. This probe showed moderate affinity for BpMip and enabled a highly robust FPA suitable for screening large compound libraries with medium- to high-throughput (Z factor ∼ 0.89) to identify potent new inhibitors. The FPA results are consistent with data from the protease-coupled PPIase assay. Analysis of the temperature dependence of the probe's binding highlighted that BpMip's ligand binding is driven by enthalpic rather than entropic effects. This has considerable consequences for the use of low-temperature kinetic assays.

Improving drug discovery through parallelism

Compound identification in ligand-based virtual screening is limited by two key issues: the quality and the time needed to obtain predictions. In this sense, we designed OptiPharm, an algorithm that obtained excellent results in improving the sequential methods in the literature. In this work, we go a step further and propose its parallelization. Specifically, we propose a two-layer parallelization. Firstly, an automation of the molecule distribution process between the available nodes in a cluster, and secondly, a parallelization of the internal methods (initialization, reproduction, selection and optimization). This new software, called pOptiPharm, aims to improve the quality of predictions and reduce experimentation time. As the results show, the performance of the proposed methods is good. It can find better solutions than the sequential OptiPharm, all while reducing its computation time almost proportionally to the number of processing units considered.

DenovoProfiling: A webserver for de novo generated molecule library profiling

Various deep learning-based architectures for molecular generation have been proposed for de novo drug design. The flourish of the de novo molecular generation methods and applications has created a great demand for the visualization and functional profiling for the de novo generated molecules. An increasing number of publicly available chemogenomic databases sets good foundations and creates good opportunities for comprehensive profiling of the de novo library. In this paper, we present DenovoProfiling, a webserver dedicated to de novo library visualization and functional profiling. Currently, DenovoProfiling contains six modules: (1) identification & visualization module for chemical structure visualization and identify the reported structures, (2) chemical space module for chemical space exploration using similarity maps, principal components analysis (PCA), drug-like properties distribution, and scaffold-based clustering, (3) ADMET prediction module for predicting the ADMET properties of the de novo molecules, (4) molecular alignment module for three dimensional molecular shape analysis, (5) drugs mapping module for identifying structural similar drugs, and (6) target & pathway module for identifying the reported targets and corresponding functional pathways. DenovoProfiling could provide structural identification, chemical space exploration, drug mapping, and target & pathway information. The comprehensive annotated information could give users a clear picture of their de novo library and could guide the further selection of candidates for chemical synthesis and biological confirmation. DenovoProfiling is freely available at http://denovoprofiling.xielab.net.

Parallelized Raman Difference Spectroscopy for the Investigation of Chemical Interactions

Raman spectroscopy provides an extremely high chemical selectivity. Raman difference spectroscopy is a technique to reveal even the smallest differences that occur due to weak interactions between substances and changes in the molecular structure. To enable parallelized and highly sensitive Raman difference spectroscopy in a microtiter-array, a diffractive optical element, a lens array, and a fiber bundle were integrated into a Raman spectroscopy setup in a unique fashion. The setup was evaluated with a microtiter-array containing pyridine-water complexes, and subwavenumber changes below the spectrometer's resolution could be resolved. The spectral changes were emphasized with two-dimensional correlation analysis. Density functional theory calculation and "atoms in molecule" analysis were performed to simulate the intermolecular long-range interactions between water and pyridine molecules and to get insight into the involved noncovalent interactions, respectively. It was found that by the addition of pyridine, the energy portion of hydrogen bonds to the total complexation energy between pyridine and water reduces. These results demonstrate the unique abilities of the new setup to investigate subtle changes due to biochemically important molecular interactions and opens new avenues to perform drug binding assays and to monitor highly parallelized chemical reactions.

Synergy between machine learning and natural products cheminformatics: Application to the lead discovery of anthraquinone derivatives

Cheminformatics utilizing machine learning (ML) techniques have opened up a new horizon in drug discovery. This is owing to vast chemical space expansion with rocketing numbers of expected hits and lead compounds that match druggable macromolecular targets, in particular from natural compounds. Due to the natural products' (NP) structural complexity, uniqueness, and diversity, they could occupy a bigger space in pharmaceuticals, allowing the industry to pursue more selective leads in the nanomolar range of binding affinity. ML is an essential part of each step of the drug design pipeline, such as target prediction, compound library preparation, and lead optimization. Notably, molecular mechanic and dynamic simulations, induced docking, and free energy perturbations are essential in predicting best binding poses, binding free energy values, and molecular mechanics force fields. Those applications have leveraged from artificial intelligence (AI), which decreases the computational costs required for such costly simulations. This review aimed to describe chemical space and compound libraries related to NPs. High-throughput screening utilized for fractionating NPs and high-throughput virtual screening and their strategies, and significance, are reviewed. Particular emphasis was given to AI approaches, ML tools, algorithms, and techniques, especially in drug discovery of macrocyclic compounds and approaches in computer-aided and ML-based drug discovery. Anthraquinone derivatives were discussed as a source of new lead compounds that can be developed using ML tools for diverse medicinal uses such as cancer, infectious diseases, and metabolic disorders. Furthermore, the power of principal component analysis in understanding relevant protein conformations, and molecular modeling of protein-ligand interaction were also presented. Apart from being a concise reference for cheminformatics, this review is a useful text to understand the application of ML-based algorithms to molecular dynamics simulation and in silico absorption, distribution, metabolism, excretion, and toxicity prediction.

Constructing Head-to-Tail Cyclic Peptide DNA-Encoded Libraries Using Two-Directional Synthesis Strategy

Macrocyclic peptides are an important class of therapeutic agents for the biological targets that are difficult to modulate by small-molecule compounds. Meanwhile, DNA-encoded library technology (DELT) provides a powerful platform for hits discovery. The unity of both fields has proven highly productive in finding cyclic peptide hits against diverse pharmaceutical proteins. Many researchers have extended the chemical toolbox for constructing head-to-tail macrocyclic DNA-encoded libraries with various ring sizes. However, the linear peptides of different lengths necessitate tuning the distance between closing sites and DNA-linked sites to perform the macrocyclization process, presumably due to the constrained conformation of linear precursors. To tackle this issue and streamline the synthetic workflow, we report a two-directional synthesis strategy. This method starts from a trifunctional reagent and prepares DNA-linked macrocyclic peptides of ring size between 15 (5-mer) and 24 (8-mer) via amide bond formation reaction, a common method to create macrocyclic peptides.

Substrate Specificity of SARS-CoV-2 Nsp10-Nsp16 Methyltransferase

The ongoing COVID-19 pandemic exemplifies the general need to better understand viral infections. The positive single-strand RNA genome of its causative agent, the SARS coronavirus 2 (SARS-CoV-2), encodes all viral enzymes. In this work, we focused on one particular methyltransferase (MTase), nsp16, which, in complex with nsp10, is capable of methylating the first nucleotide of a capped RNA strand at the 2′-O position. This process is part of a viral capping system and is crucial for viral evasion of the innate immune reaction. In light of recently discovered non-canonical RNA caps, we tested various dinucleoside polyphosphate-capped RNAs as substrates for nsp10-nsp16 MTase. We developed an LC-MS-based method and discovered four types of capped RNA (m⁷Gp₃A(G)- and Gp₃A(G)-RNA) that are substrates of the nsp10-nsp16 MTase. Our technique is an alternative to the classical isotope labelling approach for the measurement of 2′-O-MTase activity. Further, we determined the IC₅₀ value of sinefungin to illustrate the use of our approach for inhibitor screening. In the future, this approach may be an alternative technique to the radioactive labelling method for screening inhibitors of any type of 2′-O-MTase.

Towards Profiling of the G-Quadruplex Targeting Drugs in the Living Human Cells Using NMR Spectroscopy

Recently, the 1H-detected in-cell NMR spectroscopy has emerged as a unique tool allowing the characterization of interactions between nucleic acid-based targets and drug-like molecules in living human cells. Here, we assess the application potential of 1H and 19F-detected in-cell NMR spectroscopy to profile drugs/ligands targeting DNA G-quadruplexes, arguably the most studied class of anti-cancer drugs targeting nucleic acids. We show that the extension of the original in-cell NMR approach is not straightforward. The severe signal broadening and overlap of 1H in-cell NMR spectra of polymorphic G-quadruplexes and their complexes complicate their quantitative interpretation. Nevertheless, the 1H in-cell NMR can be used to identify drugs that, despite strong interaction in vitro, lose their ability to bind G-quadruplexes in the native environment. The in-cell NMR approach is adjusted to a recently developed 3,5-bis(trifluoromethyl)phenyl probe to monitor the intracellular interaction with ligands using 19F-detected in-cell NMR. The probe allows dissecting polymorphic mixture in terms of number and relative populations of individual G-quadruplex species, including ligand-bound and unbound forms in vitro and in cellulo. Despite the probe's discussed limitations, the 19F-detected in-cell NMR appears to be a promising strategy to profile G-quadruplex-ligand interactions in the complex environment of living cells.

Imaging and Fluorescence Quantification in Caenorhabditis eleganswith Flow Vermimetry and Automated Microscopy

Gene activation and cellular biomarkers are commonly monitored using fluorescent signals from transgenic reporters or dyes. These quantifiable markers are critical for biological research and serve as an incredibly powerful tool, even more so when combined with high-throughput screening. Caenorhabditis elegans is a particularly useful model in this regard, as it is inexpensive to grow in vast numbers, has a rapid generation time, is optically transparent, and can readily fit within 384-well plates. However, fluorescence quantification in worms is often cumbersome. Quantification is frequently performed using laborious, low-throughput, bias-prone methods that measure fluorescence in a comparatively small number of individual worms. Here we describe two methods, flow vermimetry using a COPAS BioSorter and an automated imaging platform and analysis pipeline using a Cytation5 multimode plate reader and image analysis software, that enable high-throughput, high-content screening in C. elegans. Flow vermimetry provides a better signal-to-noise ratio with fewer processing steps, while the Cytation5 provides a convenient platform to image samples across time. Fluorescence values from the two methods show strong correlation. Either method can be easily extended to include other parameters, such as the measurement of various metabolites, worm viability, and other aspects of cell physiology. This broadens the utility of the system and allows it to be used for a wide range of molecular biological purposes.

Divergent On-DNA Transformations from DNA-Linked Piperidones

A group of highly efficient and divergent transformations for constructing multiple DNA-linked chemotypes based on a piperidone core were successfully developed. We reported the first procedure for the synthesis of a DNA-conjugated piperidine intermediate under basic conditions. Subsequently, this substructure was subjected to additional reactions to generate several privileged scaffolds, including 4-aminopiperidine, fused [1,2,4]triazolo[1,5-a]pyrimidine, and a quinoline derivative. These transformations paved the way for constructing focused scaffold-based DNA-encoded libraries with druglike properties.

High-Throughput Screening Platforms in the Discovery of Novel Drugs for Neurodegenerative Diseases

Neurodegenerative diseases (NDDs) are incurable and debilitating conditions that result in progressive degeneration and/or death of nerve cells in the central nervous system (CNS). Identification of viable therapeutic targets and new treatments for CNS disorders and in particular, for NDDs is a major challenge in the field of drug discovery. These difficulties can be attributed to the diversity of cells involved, extreme complexity of the neural circuits, the limited capacity for tissue regeneration, and our incomplete understanding of the underlying pathological processes. Drug discovery is a complex and multidisciplinary process. The screening attrition rate in current drug discovery protocols mean that only one viable drug may arise from millions of screened compounds resulting in the need to improve discovery technologies and protocols to address the multiple causes of attrition. This has identified the need to screen larger libraries where the use of efficient high-throughput screening (HTS) becomes key in the discovery process. HTS can investigate hundreds of thousands of compounds per day. However, if fewer compounds could be screened without compromising the probability of success, the cost and time would be largely reduced. To that end, recent advances in computer-aided design, in silico libraries, and molecular docking software combined with the upscaling of cell-based platforms have evolved to improve screening efficiency with higher predictability and clinical applicability. We review, here, the increasing role of HTS in contemporary drug discovery processes, in particular for NDDs, and evaluate the criteria underlying its successful application. We also discuss the requirement of HTS for novel NDD therapies and examine the major current challenges in validating new drug targets and developing new treatments for NDDs.

Evolution of Novartis' Small Molecule Screening Deck Design

This article summarizes the evolution of the screening deck at the Novartis Institutes for BioMedical Research (NIBR). Historically, the screening deck was an assembly of all available compounds. In 2015, we designed a first deck to facilitate access to diverse subsets with optimized properties. We allocated the compounds as plated subsets on a 2D grid with property based ranking in one dimension and increasing structural redundancy in the other. The learnings from the 2015 screening deck were applied to the design of a next generation in 2019. We found that using traditional leadlikeness criteria (mainly MW, clogP) reduces the hit rates of attractive chemical starting points in subset screening. Consequently, the 2019 deck relies on solubility and permeability to select preferred compounds. The 2019 design also uses NIBR's experimental assay data and inferred biological activity profiles in addition to structural diversity to define redundancy across the compound sets.

A Flexible, Microfluidic, Dispensing System for Screening Drug Combinations

It is known that in many cases a combination of drugs is more effective than single-drug treatments both for reducing toxicity and increasing efficacy. With the advent of organoid screens, personalised medicine has become possible for many diseases. Automated pipetting to well plates is the pharmaceutical industry standard for drug screening, but this is relatively expensive and slow. Here, a rotary microfluidic system is presented that can test all possible drug combinations at speed with the use of droplets. For large numbers of combinations, it is shown how the experimental scale is reduced by considering drug dilutions and machine learning. As an example, two cases are considered; the first is a three-ring and three radii configuration and the second is a four ring and forty-eight radii configuration. Between these two, all other cases are shown to be possible. The proposed commercial instrument is shown to be flexible, the user choosing which wells to fill and which driver-computational sub-routine to select. The major issues addressed here are the programming theory of the instrument and the reduction of droplets to be generated by drug dilutions and machine learning.

Assessing molecular interactions with biophysical methods using the validation cross

There are numerous methods for studying molecular interactions. However, each method gives rise to false negative- or false positive binding results, stemming from artifacts of the scientific equipment or from shortcomings of the experimental format. To validate an initial positive binding result, additional methods need to be applied to cover the shortcomings of the primary experiment. The aim of such a validation procedure is to exclude as many artifacts as possible to confirm that there is a true molecular interaction that meets the standards for publishing or is worth investing considerable resources for follow-up activities in a drug discovery project. To simplify this validation process, a graphical scheme — the validation cross — can be used. This simple graphic is a powerful tool for identifying blind spots of a binding hypothesis, for selecting the most informative combination of methods to reveal artifacts and, in general, for understanding more thoroughly the nature of a validation process. The concept of the validation cross was originally introduced for the validation of protein–ligand interactions by NMR in drug discovery. Here, an attempt is made to expand the concept to further biophysical methods and to generalize it for binary molecular interactions.