From intuition to AI: evolution of small molecule representations in drug discovery.

Within drug discovery, the goal of AI scientists and cheminformaticians is to help identify molecular starting points that will develop into safe and efficacious drugs while reducing costs, time and failure rates. To achieve this goal, it is crucial to represent molecules in a digital format that makes them machine-readable and facilitates the accurate prediction of properties that drive decision-making. Over the years, molecular representations have evolved from intuitive and human-readable formats to bespoke numerical descriptors and fingerprints, and now to learned representations that capture patterns and salient features across vast chemical spaces. Among these, sequence-based and graph-based representations of small molecules have become highly popular. However, each approach has strengths and weaknesses across dimensions such as generality, computational cost, inversibility for generative applications and interpretability, which can be critical in informing practitioners' decisions. As the drug discovery landscape evolves, opportunities for innovation continue to emerge. These include the creation of molecular representations for high-value, low-data regimes, the distillation of broader biological and chemical knowledge into novel learned representations and the modeling of up-and-coming therapeutic modalities.

Phenonaut: multiomics data integration for phenotypic space exploration.

SUMMARY: Data integration workflows for multiomics data take many forms across academia and industry. Efforts with limited resources often encountered in academia can easily fall short of data integration best practices for processing and combining high-content imaging, proteomics, metabolomics, and other omics data. We present Phenonaut, a Python software package designed to address the data workflow needs of migration, control, integration, and auditability in the application of literature and proprietary techniques for data source and structure agnostic workflow creation. AVAILABILITY AND IMPLEMENTATION: Source code: https://github.com/CarragherLab/phenonaut, Documentation: https://carragherlab.github.io/phenonaut, PyPI package: https://pypi.org/project/phenonaut/.

CLAffinity: A Software Tool for Identification of Optimum Ligand Affinity for Competition-Based Primary Screens.

A simplistic assumption in setting up a competition assay is that a low affinity labeled ligand can be more easily displaced from a target protein than a high affinity ligand, which in turn produces a more sensitive assay. An often-cited paper correctly rallies against this assumption and recommends the use of the highest affinity ligand available for experiments aiming to determine competitive inhibitor affinities. However, we have noted this advice being applied incorrectly to competition-based primary screens where the goal is optimum assay sensitivity, enabling a clear yes/no binding determination for even low affinity interactions. The published advice only applies to secondary, confirmatory assays intended for accurate affinity determination of primary screening hits. We demonstrate that using very high affinity ligands in competition-based primary screening can lead to reduced assay sensitivity and, ultimately, the discarding of potentially valuable active compounds. We build on techniques developed in our PyBindingCurve software for a mechanistic understanding of complex biological interaction systems, developing the "CLAffinity tool" for simulating competition experiments using protein, ligand, and inhibitor concentrations common to drug screening campaigns. CLAffinity reveals optimum labeled ligand affinity ranges based on assay parameters, rather than general rules to optimize assay sensitivity. We provide the open source CLAffinity software toolset to carry out assay simulations and a video summarizing key findings to aid in understanding, along with a simple lookup table allowing identification of optimal dynamic ranges for competition-based primary screens. The application of our freely available software and lookup tables will lead to the consistent creation of more performant competition-based primary screens identifying valuable hit compounds, particularly for difficult targets.

PyBindingCurve, Simulation, and Curve Fitting to Complex Binding Systems at Equilibrium.

Understanding multicomponent binding interactions in protein-ligand, protein-protein, and competition systems is essential for fundamental biology and drug discovery. Hand-deriving equations quickly become unfeasible when the number of components is increased, and direct analytical solutions only exist to a certain complexity. To address this problem and allow easy access to simulation, plotting, and parameter fitting to complex systems at equilibrium, we present the Python package PyBindingCurve. We apply this software to explore homodimer and heterodimer formations culminating in the discovery that under certain conditions, homodimers are easier to break with an inhibitor than heterodimers and may also be more readily depleted. This is a potentially valuable and overlooked phenomenon of great importance to drug discovery. PyBindingCurve may be expanded to operate on any equilibrium binding system and allows definition of custom systems using a simple syntax. PyBindingCurve is available under the MIT license at https://github.com/stevenshave/pybindingcurve as the Python source code accompanied by examples and as an easily installable package within the Python Package Index.

Real-time tracking of complex ubiquitination cascades using a fluorescent confocal on-bead assay.

BACKGROUND: The ubiquitin-proteasome system (UPS) controls the stability, localization and/or activity of the proteome. However, the identification and characterization of complex individual ubiquitination cascades and their modulators remains a challenge. Here, we report a broadly applicable, multiplexed, miniaturized on-bead technique for real-time monitoring of various ubiquitination-related enzymatic activities. The assay, termed UPS-confocal fluorescence nanoscanning (UPS-CONA), employs a substrate of interest immobilized on a micro-bead and a fluorescently labeled ubiquitin which, upon enzymatic conjugation to the substrate, is quantitatively detected on the bead periphery by confocal imaging. RESULTS: UPS-CONA is suitable for studying individual enzymatic activities, including various E1, E2, and HECT-type E3 enzymes, and for monitoring multi-step reactions within ubiquitination cascades in a single experimental compartment. We demonstrate the power of the UPS-CONA technique by simultaneously following ubiquitin transfer from Ube1 through Ube2L3 to E6AP. We applied this multi-step setup to investigate the selectivity of five ubiquitination inhibitors reportedly targeting different classes of ubiquitination enzymes. Using UPS-CONA, we have identified a new activity of a small molecule E2 inhibitor, BAY 11-7082, and of a HECT E3 inhibitor, heclin, towards the Ube1 enzyme. CONCLUSIONS: As a sensitive, quantitative, flexible, and reagent-efficient method with a straightforward protocol, UPS-CONA constitutes a powerful tool for interrogation of ubiquitination-related enzymatic pathways and their chemical modulators, and is readily scalable for large experiments.

High-throughput chemical screening for antivirulence developmental phenotypes in Trypanosoma brucei.

In the bloodstream of mammalian hosts, the sleeping sickness parasite, Trypanosoma brucei, exists as a proliferative slender form or a nonproliferative, transmissible, stumpy form. The transition between these developmental forms is controlled by a density-dependent mechanism that is important for the parasite's infection dynamics, immune evasion via ordered antigenic variation, and disease transmissibility. However, stumpy formation has been lost in most laboratory-adapted trypanosome lines, generating monomorphic parasites that proliferate uncontrolled as slender forms in vitro and in vivo. Nonetheless, these forms are readily amenable to cell culture and high-throughput screening for trypanocidal lead compounds. Here, we have developed and exploited a high-throughput screen for developmental phenotypes using a transgenic monomorphic cell line expressing a reporter under the regulation of gene control signals from the stumpy-specific molecule PAD1. Using a whole-cell fluorescence-based assay to screen over 6,000 small molecules from a kinase-focused compound library, small molecules able to activate stumpy-specific gene expression and proliferation arrest were assayed in a rapid assay format. Independent follow-up validation identified one hit able to induce modest, yet specific, changes in mRNA expression indicative of a partial differentiation to stumpy forms in monomorphs. Further, in pleomorphs this compound induced a stumpy-like phenotype, entailing growth arrest, morphological changes, PAD1 expression, and enhanced differentiation to procyclic forms. This not only provides a potential tool compound for the further understanding of stumpy formation but also demonstrates the use of high-throughput screening in the identification of compounds able to induce specific phenotypes, such as differentiation, in African trypanosomes.

Identification and X-ray co-crystal structure of a small-molecule activator of LFA-1-ICAM-1 binding.

Stabilization of protein-protein interactions by small molecules is a concept with few examples reported to date. Herein we describe the identification and X-ray co-crystal structure determination of IBE-667, an ICAM-1 binding enhancer for LFA-1. IBE-667 was designed based on the SAR information obtained from an on-bead screen of tagged one-bead one-compound combinatorial libraries by confocal nanoscanning and bead picking (CONA). Cellular assays demonstrate the activity of IBE-667 in promoting the binding of LFA-1 on activated immune cells to ICAM-1.

CSBB-ConeExclusion, adapting structure based solution virtual screening to libraries on solid support.

Combinatorial chemical libraries produced on solid support offer fast and cost-effective access to a large number of unique compounds. If such libraries are screened directly on-bead, the speed at which chemical space can be explored by chemists is much greater than that addressable using solution based synthesis and screening methods. Solution based screening has a large supporting body of software such as structure-based virtual screening tools which enable the prediction of protein-ligand complexes. Use of these techniques to predict the protein bound complexes of compounds synthesized on solid support neglects to take into account the conjugation site on the small molecule ligand. This may invalidate predicted binding modes, the linker may be clashing with protein atoms. We present CSBB-ConeExclusion, a methodology and computer program which provides a measure of the applicability of solution dockings to solid support. Output is given in the form of statistics for each docking pose, a unique 2D visualization method which can be used to determine applicability at a glance, and automatically generated PyMol scripts allowing visualization of protein atom incursion into a defined exclusion volume. CSBB-ConeExclusion is then exemplarically used to determine the optimum attachment point for a purine library targeting cyclin-dependent kinase 2 CDK2.

EDULISS: a small-molecule database with data-mining and pharmacophore searching capabilities.

We present the relational database EDULISS (EDinburgh University Ligand Selection System), which stores structural, physicochemical and pharmacophoric properties of small molecules. The database comprises a collection of over 4 million commercially available compounds from 28 different suppliers. A user-friendly web-based interface for EDULISS (available at http://eduliss.bch.ed.ac.uk/) has been established providing a number of data-mining possibilities. For each compound a single 3D conformer is stored along with over 1600 calculated descriptor values (molecular properties). A very efficient method for unique compound recognition, especially for a large scale database, is demonstrated by making use of small subgroups of the descriptors. Many of the shape and distance descriptors are held as pre-calculated bit strings permitting fast and efficient similarity and pharmacophore searches which can be used to identify families of related compounds for biological testing. Two ligand searching applications are given to demonstrate how EDULISS can be used to extract families of molecules with selected structural and biophysical features.

Evolution and impact of high content imaging.

The field of high content imaging has steadily evolved and expanded substantially across many industry and academic research institutions since it was first described in the early 1990's. High content imaging refers to the automated acquisition and analysis of microscopic images from a variety of biological sample types. Integration of high content imaging microscopes with multiwell plate handling robotics enables high content imaging to be performed at scale and support medium- to high-throughput screening of pharmacological, genetic and diverse environmental perturbations upon complex biological systems ranging from 2D cell cultures to 3D tissue organoids to small model organisms. In this perspective article the authors provide a collective view on the following key discussion points relevant to the evolution of high content imaging: • Evolution and impact of high content imaging: An academic perspective • Evolution and impact of high content imaging: An industry perspective • Evolution of high content image analysis • Evolution of high content data analysis pipelines towards multiparametric and phenotypic profiling applications • The role of data integration and multiomics • The role and evolution of image data repositories and sharing standards • Future perspective of high content imaging hardware and software.

Discovery of a new class of inhibitors for the protein arginine deiminase type 4 (PAD4) by structure-based virtual screening.

BACKGROUND: Rheumatoid arthritis (RA) is an autoimmune disease with unknown etiology. Anticitrullinated protein autoantibody has been documented as a highly specific autoantibody associated with RA. Protein arginine deiminase type 4 (PAD4) is the enzyme responsible for catalyzing the conversion of peptidylarginine into peptidylcitrulline. PAD4 is a new therapeutic target for RA treatment. In order to search for inhibitors of PAD4, structure-based virtual screening was performed using LIDAEUS (Ligand discovery at Edinburgh university). Potential inhibitors were screened experimentally by inhibition assays. RESULTS: Twenty two of the top-ranked water-soluble compounds were selected for inhibitory screening against PAD4. Three compounds showed significant inhibition of PAD4 and their IC50 values were investigated. The structures of the three compounds show no resemblance with previously discovered PAD4 inhibitors, nor with existing drugs for RA treatment. CONCLUSION: Three compounds were discovered as potential inhibitors of PAD4 by virtual screening. The compounds are commercially available and can be used as scaffolds to design more potent inhibitors against PAD4.

Chemical Interrogation of Nuclear Size Identifies Compounds with Cancer Cell Line-Specific Effects on Migration and Invasion.

Background: Lower survival rates for many cancer types correlate with changes in nuclear size/scaling in a tumor-type/tissue-specific manner. Hypothesizing that such changes might confer an advantage to tumor cells, we aimed at the identification of commercially available compounds to guide further mechanistic studies. We therefore screened for Food and Drug Administration (FDA)/European Medicines Agency (EMA)-approved compounds that reverse the direction of characteristic tumor nuclear size changes in PC3, HCT116, and H1299 cell lines reflecting, respectively, prostate adenocarcinoma, colonic adenocarcinoma, and small-cell squamous lung cancer. Results: We found distinct, largely nonoverlapping sets of compounds that rectify nuclear size changes for each tumor cell line. Several classes of compounds including, e.g., serotonin uptake inhibitors, cyclo-oxygenase inhibitors, ß-adrenergic receptor agonists, and Na+/K+ ATPase inhibitors, displayed coherent nuclear size phenotypes focused on a particular cell line or across cell lines and treatment conditions. Several compounds from classes far afield from current chemotherapy regimens were also identified. Seven nuclear size-rectifying compounds selected for further investigation all inhibited cell migration and/or invasion. Conclusions: Our study provides (a) proof of concept that nuclear size might be a valuable target to reduce cell migration/invasion in cancer treatment and (b) the most thorough collection of tool compounds to date reversing nuclear size changes specific to individual cancer-type cell lines. Although these compounds still need to be tested in primary cancer cells, the cell line-specific nuclear size and migration/invasion responses to particular drug classes suggest that cancer type-specific nuclear size rectifiers may help reduce metastatic spread.

A two-step resin based approach to reveal survivin-selective fluorescent probes.

The identification of modulators for proteins without assayable biochemical activity remains a challenge in chemical biology. The presented approach adapts a high-throughput fluorescence binding assay and functional chromatography, two protein-resin technologies, enabling the discovery and isolation of fluorescent natural product probes that target proteins independently of biochemical function. The resulting probes also suggest targetable pockets for lead discovery. Using human survivin as a model, we demonstrate this method with the discovery of members of the prodiginine family as fluorescent probes to the cancer target survivin.

Quantitative Microdialysis: Experimental Protocol and Software for Small Molecule Protein Affinity Determination and for Exclusion of Compounds with Poor Physicochemical Properties.

Quantitative microdialysis is a traditional biophysical affinity determination technique. In the development of the detailed experimental protocol presented, we used commercially available equipment, rapid equilibrium dialysis (RED) devices (ThermoFisher Scientific), which means that it is open to most laboratories. The target protein and test compound are incubated in a chamber partitioned to allow only small molecules to transition to a larger reservoir chamber, then reversed-phase high performance liquid chromatography (RP-HPLC) or liquid chromatography-mass spectrometry (LC-MS) is used to determine the abundance of compound in each chamber. A higher compound concentration measured in the chamber that contains the target protein indicates binding. As a novel, and differentiating contribution, we present a protocol for mathematical analysis of experimental data. We provide the equations and the software to yield dissociation constants for the test compound-target protein complex up to 0.5 mM KD, and we quantitatively discuss the limitations of affinities in relation to measured compound concentrations.

Extensive rewiring of the EGFR network in colorectal cancer cells expressing transforming levels of KRASG13D.

Protein-protein-interaction networks (PPINs) organize fundamental biological processes, but how oncogenic mutations impact these interactions and their functions at a network-level scale is poorly understood. Here, we analyze how a common oncogenic KRAS mutation (KRASG13D) affects PPIN structure and function of the Epidermal Growth Factor Receptor (EGFR) network in colorectal cancer (CRC) cells. Mapping >6000 PPIs shows that this network is extensively rewired in cells expressing transforming levels of KRASG13D (mtKRAS). The factors driving PPIN rewiring are multifactorial including changes in protein expression and phosphorylation. Mathematical modelling also suggests that the binding dynamics of low and high affinity KRAS interactors contribute to rewiring. PPIN rewiring substantially alters the composition of protein complexes, signal flow, transcriptional regulation, and cellular phenotype. These changes are validated by targeted and global experimental analysis. Importantly, genetic alterations in the most extensively rewired PPIN nodes occur frequently in CRC and are prognostic of poor patient outcomes.

PuLSE: Quality control and quantification of peptide sequences explored by phage display libraries.

The design of highly diverse phage display libraries is based on assumption that DNA bases are incorporated at similar rates within the randomized sequence. As library complexity increases and expected copy numbers of unique sequences decrease, the exploration of library space becomes sparser and the presence of truly random sequences becomes critical. We present the program PuLSE (Phage Library Sequence Evaluation) as a tool for assessing randomness and therefore diversity of phage display libraries. PuLSE runs on a collection of sequence reads in the fastq file format and generates tables profiling the library in terms of unique DNA sequence counts and positions, translated peptide sequences, and normalized 'expected' occurrences from base to residue codon frequencies. The output allows at-a-glance quantitative quality control of a phage library in terms of sequence coverage both at the DNA base and translated protein residue level, which has been missing from toolsets and literature. The open source program PuLSE is available in two formats, a C++ source code package for compilation and integration into existing bioinformatics pipelines and precompiled binaries for ease of use.

Diclofenac Identified as a Kynurenine 3-Monooxygenase Binder and Inhibitor by Molecular Similarity Techniques.

In this study, we apply a battery of molecular similarity techniques to known inhibitors of kynurenine 3-monooxygenase (KMO), querying each against a repository of approved, experimental, nutraceutical, and illicit drugs. Four compounds are assayed against KMO. Subsequently, diclofenac (also known by the trade names Voltaren, Voltarol, Aclonac, and Cataflam) has been confirmed as a human KMO protein binder and inhibitor in cell lysate with low micromolar KD and IC50, respectively, and low millimolar cellular IC50. Hit to drug hopping, as exemplified here for one of the most successful anti-inflammatory medicines ever invented, holds great promise for expansion into new disease areas and highlights the not-yet-fully-exploited potential of drug repurposing.

UFSRAT: Ultra-fast Shape Recognition with Atom Types--the discovery of novel bioactive small molecular scaffolds for FKBP12 and 11ßHSD1.

MOTIVATION: Using molecular similarity to discover bioactive small molecules with novel chemical scaffolds can be computationally demanding. We describe Ultra-fast Shape Recognition with Atom Types (UFSRAT), an efficient algorithm that considers both the 3D distribution (shape) and electrostatics of atoms to score and retrieve molecules capable of making similar interactions to those of the supplied query. RESULTS: Computational optimization and pre-calculation of molecular descriptors enables a query molecule to be run against a database containing 3.8 million molecules and results returned in under 10 seconds on modest hardware. UFSRAT has been used in pipelines to identify bioactive molecules for two clinically relevant drug targets; FK506-Binding Protein 12 and 11ß-hydroxysteroid dehydrogenase type 1. In the case of FK506-Binding Protein 12, UFSRAT was used as the first step in a structure-based virtual screening pipeline, yielding many actives, of which the most active shows a KD, app of 281 µM and contains a substructure present in the query compound. Success was also achieved running solely the UFSRAT technique to identify new actives for 11ß-hydroxysteroid dehydrogenase type 1, for which the most active displays an IC50 of 67 nM in a cell based assay and contains a substructure radically different to the query. This demonstrates the valuable ability of the UFSRAT algorithm to perform scaffold hops. AVAILABILITY AND IMPLEMENTATION: A web-based implementation of the algorithm is freely available at http://opus.bch.ed.ac.uk/ufsrat/.

Inhibition of the ERCC1-XPF structure-specific endonuclease to overcome cancer chemoresistance.

ERCC1-XPF is a structure-specific endonuclease that is required for the repair of DNA lesions, generated by the widely used platinum-containing cancer chemotherapeutics such as cisplatin, through the Nucleotide Excision Repair and Interstrand Crosslink Repair pathways. Based on mouse xenograft experiments, where ERCC1-deficient melanomas were cured by cisplatin therapy, we proposed that inhibition of ERCC1-XPF could enhance the effectiveness of platinum-based chemotherapy. Here we report the identification and properties of inhibitors against two key targets on ERCC1-XPF. By targeting the ERCC1-XPF interaction domain we proposed that inhibition would disrupt the ERCC1-XPF heterodimer resulting in destabilisation of both proteins. Using in silico screening, we identified an inhibitor that bound to ERCC1-XPF in a biophysical assay, reduced the level of ERCC1-XPF complexes in ovarian cancer cells, inhibited Nucleotide Excision Repair and sensitised melanoma cells to cisplatin. We also utilised high throughput and in silico screening to identify the first reported inhibitors of the other key target, the XPF endonuclease domain. We demonstrate that two of these compounds display specificity in vitro for ERCC1-XPF over two other endonucleases, bind to ERCC1-XPF, inhibit Nucleotide Excision Repair in two independent assays and specifically sensitise Nucleotide Excision Repair-proficient, but not Nucleotide Excision Repair-deficient human and mouse cells to cisplatin.

11ß-Hydroxysteroid dehydrogenase type 1 contributes to the balance between 7-keto- and 7-hydroxy-oxysterols in vivo.

11ß-Hydroxysteroid dehydrogenase 1 (11ßHSD1; EC 1.1.1.146) generates active glucocorticoids from inert 11-keto metabolites. However, it can also metabolize alternative substrates, including 7ß-hydroxy- and 7-keto-cholesterol (7ßOHC, 7KC). This has been demonstrated in vitro but its consequences in vivo are uncertain. We used genetically modified mice to investigate the contribution of 11ßHSD1 to the balance of circulating levels of 7KC and 7ßOHC in vivo, and dissected in vitro the kinetics of the interactions between oxysterols and glucocorticoids for metabolism by the mouse enzyme. Circulating levels of 7KC and 7ßOHC in mice were 91.3±22.3 and 22.6±5.7 nM respectively, increasing to 1240±220 and 406±39 nM in ApoE(-/-) mice receiving atherogenic western diet. Disruption of 11ßHSD1 in mice increased (p<0.05) the 7KC/7ßOHC ratio in plasma (by 20%) and also in isolated microsomes (2 fold). The 7KC/7ßOHC ratio was similarly increased when NADPH generation was restricted by disruption of hexose-6-phosphate dehydrogenase. Reduction and oxidation of 7-oxysterols by murine 11ßHSD1 proceeded more slowly and substrate affinity was lower than for glucocorticoids. in vitro 7ßOHC was a competitive inhibitor of oxidation of corticosterone (Ki=0.9 µM), whereas 7KC only weakly inhibited reduction of 11-dehydrocorticosterone. However, supplementation of 7-oxysterols in cultured cells, secondary to cholesterol loading, preferentially slowed reduction of glucocorticoids, rather than oxidation. Thus, in mouse, 11ßHSD1 influenced the abundance and balance of circulating and tissue levels of 7ßOHC and 7KC, promoting reduction of 7KC. In health, 7-oxysterols are unlikely to regulate glucocorticoid metabolism. However, in hyperlipidaemia, 7-oxysterols may inhibit glucocorticoid metabolism and modulate signaling through corticosteroid receptors.