A Hybrid Structure-Based Machine Learning Approach for Predicting Kinase Inhibition by Small Molecules.

Kinases have been the focus of drug discovery programs for three decades leading to over 70 therapeutic kinase inhibitors and biophysical affinity measurements for over 130,000 kinase-compound pairs. Nonetheless, the precise target spectrum for many kinases remains only partly understood. In this study, we describe a computational approach to unlocking qualitative and quantitative kinome-wide binding measurements for structure-based machine learning. Our study has three components: (i) a Kinase Inhibitor Complex (KinCo) data set comprising in silico predicted kinase structures paired with experimental binding constants, (ii) a machine learning loss function that integrates qualitative and quantitative data for model training, and (iii) a structure-based machine learning model trained on KinCo. We show that our approach outperforms methods trained on crystal structures alone in predicting binary and quantitative kinase-compound interaction affinities; relative to structure-free methods, our approach also captures known kinase biochemistry and more successfully generalizes to distant kinase sequences and compound scaffolds.

Cyp1 Inhibition Prevents Doxorubicin-Induced Cardiomyopathy in a Zebrafish Heart-Failure Model.

Doxorubicin is a highly effective chemotherapy agent used to treat many common malignancies. However, its use is limited by cardiotoxicity, and cumulative doses exponentially increase the risk of heart failure. To identify novel heart failure treatment targets, a zebrafish model of doxorubicin-induced cardiomyopathy was previously established for small-molecule screening. Using this model, several small molecules that prevent doxorubicin-induced cardiotoxicity both in zebrafish and in mouse models have previously been identified. In this study, exploration of doxorubicin cardiotoxicity is expanded by screening 2271 small molecules from a proprietary, target-annotated tool compound collection. It is found that 120 small molecules can prevent doxorubicin-induced cardiotoxicity, including 7 highly effective compounds. Of these, all seven exhibited inhibitory activity towards cytochrome P450 family 1 (CYP1). These results are consistent with previous findings, in which visnagin, a CYP1 inhibitor, also prevents doxorubicin-induced cardiotoxicity. Importantly, genetic mutation of cyp1a protected zebrafish against doxorubicin-induced cardiotoxicity phenotypes. Together, these results provide strong evidence that CYP1 is an important contributor to doxorubicin-induced cardiotoxicity and highlight the CYP1 pathway as a candidate therapeutic target for clinical cardioprotection.

Representing high throughput expression profiles via perturbation barcodes reveals compound targets.

High throughput mRNA expression profiling can be used to characterize the response of cell culture models to perturbations such as pharmacologic modulators and genetic perturbations. As profiling campaigns expand in scope, it is important to homogenize, summarize, and analyze the resulting data in a manner that captures significant biological signals in spite of various noise sources such as batch effects and stochastic variation. We used the L1000 platform for large-scale profiling of 978 representative genes across thousands of compound treatments. Here, a method is described that uses deep learning techniques to convert the expression changes of the landmark genes into a perturbation barcode that reveals important features of the underlying data, performing better than the raw data in revealing important biological insights. The barcode captures compound structure and target information, and predicts a compound's high throughput screening promiscuity, to a higher degree than the original data measurements, indicating that the approach uncovers underlying factors of the expression data that are otherwise entangled or masked by noise. Furthermore, we demonstrate that visualizations derived from the perturbation barcode can be used to more sensitively assign functions to unknown compounds through a guilt-by-association approach, which we use to predict and experimentally validate the activity of compounds on the MAPK pathway. The demonstrated application of deep metric learning to large-scale chemical genetics projects highlights the utility of this and related approaches to the extraction of insights and testable hypotheses from big, sometimes noisy data.

Synthesis of Complex Druglike Molecules by the Use of Highly Functionalized Bench-Stable Organozinc Reagents.

The reactivity of a representative set of 17 organozinc pivalates with 18 polyfunctional druglike electrophiles (informers) in Negishi cross-coupling reactions was evaluated by high-throughput experimentation protocols. The high-fidelity scaleup of successful reactions in parallel enabled the isolation of sufficient material for biological testing, thus demonstrating the high value of these new solid zinc reagents in a drug-discovery setting and potentially for many other applications in chemistry. Principal component analysis (PCA) clearly defined the independent roles of the zincates and the informers toward druggable-space coverage.

Stitched α-helical peptides via bis ring-closing metathesis.

Conformationally stabilized α-helical peptides are capable of inhibiting disease-relevant intracellular or extracellular protein-protein interactions in vivo. We have previously reported that the employment of ring-closing metathesis to introduce a single all-hydrocarbon staple along one face of an α-helical peptide greatly increases α-helical content, binding affinity to a target protein, cell penetration through active transport, and resistance to proteolytic degradation. In an effort to improve upon this technology for stabilizing a peptide in a bioactive α-helical conformation, we report the discovery of an efficient and selective bis ring-closing metathesis reaction leading to peptides bearing multiple contiguous staples connected by a central spiro ring junction. Circular dichroism spectroscopy, NMR, and computational analyses have been used to investigate the conformation of these "stitched" peptides, which are shown to exhibit remarkable thermal stabilities. Likewise, trypsin proteolysis assays confirm the achievement of a structural rigidity unmatched by peptides bearing a single staple. Furthermore, fluorescence-activated cell sorting (FACS) and confocal microscopy assays demonstrate that stitched peptides display superior cell penetrating ability compared to their stapled counterparts, suggesting that this technology may be useful not only in the context of enhancing the drug-like properties of α-helical peptides but also in producing potent agents for the intracellular delivery of proteins and oligonucleotides.

Pocket Crafter: a 3D generative modeling based workflow for the rapid generation of hit molecules in drug discovery.

We present a user-friendly molecular generative pipeline called Pocket Crafter, specifically designed to facilitate hit finding activity in the drug discovery process. This workflow utilized a three-dimensional (3D) generative modeling method Pocket2Mol, for the de novo design of molecules in spatial perspective for the targeted protein structures, followed by filters for chemical-physical properties and drug-likeness, structure-activity relationship analysis, and clustering to generate top virtual hit scaffolds. In our WDR5 case study, we acquired a focused set of 2029 compounds after a targeted searching within Novartis archived library based on the virtual scaffolds. Subsequently, we experimentally profiled these compounds, resulting in a novel chemical scaffold series that demonstrated activity in biochemical and biophysical assays. Pocket Crafter successfully prototyped an effective end-to-end 3D generative chemistry-based workflow for the exploration of new chemical scaffolds, which represents a promising approach in early drug discovery for hit identification.

Enhancing the Small-Scale Screenable Biological Space beyond Known Chemogenomics Libraries with Gray Chemical Matter─Compounds with Novel Mechanisms from High-Throughput Screening Profiles.

Phenotypic assays have become an established approach to drug discovery. Greater disease relevance is often achieved through cellular models with increased complexity and more detailed readouts, such as gene expression or advanced imaging. However, the intricate nature and cost of these assays impose limitations on their screening capacity, often restricting screens to well-characterized small compound sets such as chemogenomics libraries. Here, we outline a cheminformatics approach to identify a small set of compounds with likely novel mechanisms of action (MoAs), expanding the MoA search space for throughput limited phenotypic assays. Our approach is based on mining existing large-scale, phenotypic high-throughput screening (HTS) data. It enables the identification of chemotypes that exhibit selectivity across multiple cell-based assays, which are characterized by persistent and broad structure activity relationships (SAR). We validate the effectiveness of our approach in broad cellular profiling assays (Cell Painting, DRUG-seq, and Promotor Signature Profiling) and chemical proteomics experiments. These experiments revealed that the compounds behave similarly to known chemogenetic libraries, but with a notable bias toward novel protein targets. To foster collaboration and advance research in this area, we have curated a public set of such compounds based on the PubChem BioAssay dataset and made it available for use by the scientific community.

Mapping targetable sites on human telomerase RNA pseudoknot/template domain using 2'-OMe RNA-interacting polynucleotide (RIPtide) microarrays.

Most cellular RNAs engage in intrastrand base-pairing that gives rise to complex three-dimensional folds. This self-pairing presents an impediment toward binding of the RNA by nucleic acid-based ligands. An important step in the discovery of RNA-targeting ligands is therefore to identify those regions in a folded RNA that are accessible toward the nucleic acid-based ligand. Because the folding of RNA targets can involve interactions between nonadjacent regions and employ both Watson-Crick and non-Watson-Crick base-pairing, screening of candidate binder ensembles is typically necessary. Microarray-based screening approaches have shown great promise in this regard and have suggested that achieving complete sequence coverage would be a valuable attribute of a next generation system. Here, we report a custom microarray displaying a library of RNA-interacting polynucleotides comprising all possible 2'-OMe RNA sequences from 4- to 8-nucleotides in length. We demonstrate the utility of this array in identifying RNA-interacting polynucleotides that bind tightly and specifically to the highly conserved, functionally essential template/pseudoknot domain of human telomerase RNA and that inhibit telomerase function in vitro.

Enforced presentation of an extrahelical guanine to the lesion recognition pocket of human 8-oxoguanine glycosylase, hOGG1.

A poorly understood aspect of DNA repair proteins is their ability to identify exceedingly rare sites of damage embedded in a large excess of nearly identical undamaged DNA, while catalyzing repair only at the damaged sites. Progress toward understanding this problem has been made by comparing the structures and biochemical behavior of these enzymes when they are presented with either a target lesion or a corresponding undamaged nucleobase. Trapping and analyzing such DNA-protein complexes is particularly difficult in the case of base extrusion DNA repair proteins because of the complexity of the repair reaction, which involves extrusion of the target base from DNA followed by its insertion into the active site where glycosidic bond cleavage is catalyzed. Here we report the structure of a human 8-oxoguanine (oxoG) DNA glycosylase, hOGG1, in which a normal guanine from DNA has been forcibly inserted into the enzyme active site. Although the interactions of the nucleobase with the active site are only subtly different for G versus oxoG, hOGG1 fails to catalyze excision of the normal nucleobase. This study demonstrates that even if hOGG1 mistakenly inserts a normal base into its active site, the enzyme can still reject it on the basis of catalytic incompatibility.

Structure of the stapled p53 peptide bound to Mdm2.

Mdm2 is a major negative regulator of the tumor suppressor p53 protein, a protein that plays a crucial role in maintaining genome integrity. Inactivation of p53 is the most prevalent defect in human cancers. Inhibitors of the Mdm2-p53 interaction that restore the functional p53 constitute potential nongenotoxic anticancer agents with a novel mode of action. We present here a 2.0 Å resolution structure of the Mdm2 protein with a bound stapled p53 peptide. Such peptides, which are conformationally and proteolytically stabilized with all-hydrocarbon staples, are an emerging class of biologics that are capable of disrupting protein-protein interactions and thus have broad therapeutic potential. The structure represents the first crystal structure of an i, i + 7 stapled peptide bound to its target and reveals that rather than acting solely as a passive conformational brace, a staple can intimately interact with the surface of a protein and augment the binding interface.

Chemotography for multi-target SAR analysis in the context of biological pathways.

The increasing amount of chemogenomics data, that is, activity measurements of many compounds across a variety of biological targets, allows for better understanding of pharmacology in a broad biological context. Rather than assessing activity at individual biological targets, today understanding of compound interaction with complex biological systems and molecular pathways is often sought in phenotypic screens. This perspective poses novel challenges to structure-activity relationship (SAR) assessment. Today, the bottleneck of drug discovery lies in the understanding of SAR of rich datasets that go beyond single targets in the context of biological pathways, potential off-targets, and complex selectivity profiles. To aid in the understanding and interpretation of such complex SAR, we introduce Chemotography (chemotype chromatography), which encodes chemical space using a color spectrum by combining clustering and multidimensional scaling. Rich biological data in our approach were visualized using spatial dimensions traditionally reserved for chemical space. This allowed us to analyze SAR in the context of target hierarchies and phylogenetic trees, two-target activity scatter plots, and biological pathways. Chemotography, in combination with the Kyoto Encyclopedia of Genes and Genomes (KEGG), also allowed us to extract pathway-relevant SAR from the ChEMBL database. We identified chemotypes showing polypharmacology and selectivity-conferring scaffolds, even in cases where individual compounds have not been tested against all relevant targets. In addition, we analyzed SAR in ChEMBL across the entire Kinome, going beyond individual compounds. Our method combines the strengths of chemical space visualization for SAR analysis and graphical representation of complex biological data. Chemotography is a new paradigm for chemogenomic data visualization and its versatile applications presented here may allow for improved assessment of SAR in biological context, such as phenotypic assay hit lists.

Discovery of an Anion-Dependent Farnesyltransferase Inhibitor from a Phenotypic Screen.

By employing a phenotypic screen, a set of compounds, exemplified by 1, were identified which potentiate the ability of histone deacetylase inhibitor vorinostat to reverse HIV latency. Proteome enrichment followed by quantitative mass spectrometric analysis employing a modified analogue of 1 as affinity bait identified farnesyl transferase (FTase) as the primary interacting protein in cell lysates. This ligand-FTase binding interaction was confirmed via X-ray crystallography and temperature dependent fluorescence studies, despite 1 lacking structural and binding similarity to known FTase inhibitors. Although multiple lines of evidence established the binding interaction, these ligands exhibited minimal inhibitory activity in a cell-free biochemical FTase inhibition assay. Subsequent modification of the biochemical assay by increasing anion concentration demonstrated FTase inhibitory activity in this novel class. We propose 1 binds together with the anion in the active site to inhibit farnesyl transferase. Implications for phenotypic screening deconvolution and HIV reactivation are discussed.

Targeting RNA with Small Molecules: Identification of Selective, RNA-Binding Small Molecules Occupying Drug-Like Chemical Space.

Although the potential value of RNA as a target for new small molecule therapeutics is becoming increasingly credible, the physicochemical properties required for small molecules to selectively bind to RNA remain relatively unexplored. To investigate the druggability of RNAs with small molecules, we have employed affinity mass spectrometry, using the Automated Ligand Identification System (ALIS), to screen 42 RNAs from a variety of RNA classes, each against an array of chemically diverse drug-like small molecules (~50,000 compounds) and functionally annotated tool compounds (~5100 compounds). The set of RNA-small molecule interactions that was generated was compared with that for protein-small molecule interactions, and naïve Bayesian models were constructed to determine the types of specific chemical properties that bias small molecules toward binding to RNA. This set of RNA-selective chemical features was then used to build an RNA-focused set of ~3800 small molecules that demonstrated increased propensity toward binding the RNA target set. In addition, the data provide an overview of the specific physicochemical properties that help to enable binding to potential RNA targets. This work has increased the understanding of the chemical properties that are involved in small molecule binding to RNA, and the methodology used here is generally applicable to RNA-focused drug discovery efforts.

Molecular Profiling Reveals a Common Metabolic Signature of Tissue Fibrosis.

Fibrosis, or the accumulation of extracellular matrix, is a common feature of many chronic diseases. To interrogate core molecular pathways underlying fibrosis, we cross-examine human primary cells from various tissues treated with TGF-ß, as well as kidney and liver fibrosis models. Transcriptome analyses reveal that genes involved in fatty acid oxidation are significantly perturbed. Furthermore, mitochondrial dysfunction and acylcarnitine accumulation are found in fibrotic tissues. Substantial downregulation of the PGC1α gene is evident in both in vitro and in vivo fibrosis models, suggesting a common node of metabolic signature for tissue fibrosis. In order to identify suppressors of fibrosis, we carry out a compound library phenotypic screen and identify AMPK and PPAR as highly enriched targets. We further show that pharmacological treatment of MK-8722 (AMPK activator) and MK-4074 (ACC inhibitor) reduce fibrosis in vivo. Altogether, our work demonstrate that metabolic defect is integral to TGF-ß signaling and fibrosis.

All-atom model for stabilization of alpha-helical structure in peptides by hydrocarbon staples.

Recent work has shown that the incorporation of an all-hydrocarbon "staple" into peptides can greatly increase their alpha-helix propensity, leading to an improvement in pharmaceutical properties such as proteolytic stability, receptor affinity, and cell permeability. Stapled peptides thus show promise as a new class of drugs capable of accessing intractable targets such as those that engage in intracellular protein-protein interactions. The extent of alpha-helix stabilization provided by stapling has proven to be substantially context dependent, requiring cumbersome screening to identify the optimal site for staple incorporation. In certain cases, a staple encompassing one turn of the helix (attached at residues i and i+4) furnishes greater helix stabilization than one encompassing two turns (i,i+7 staple), which runs counter to expectation based on polymer theory. These findings highlight the need for a more thorough understanding of the forces that underlie helix stabilization by hydrocarbon staples. Here we report all-atom Monte Carlo folding simulations comparing unmodified peptides derived from RNase A and BID BH3 with various i,i+4 and i,i+7 stapled versions thereof. The results of these simulations were found to be in quantitative agreement with experimentally determined helix propensities. We also discovered that staples can stabilize quasi-stable decoy conformations, and that the removal of these states plays a major role in determining the helix stability of stapled peptides. Finally, we critically investigate why our method works, exposing the underlying physical forces that stabilize stapled peptides.

CHEMGENIE: integration of chemogenomics data for applications in chemical biology.

Increasing amounts of biological data are accumulating in the pharmaceutical industry and academic institutions. However, data does not equal actionable information, and guidelines for appropriate data capture, harmonization, integration, mining, and visualization need to be established to fully harness its potential. Here, we describe ongoing efforts at Merck & Co. to structure data in the area of chemogenomics. We are integrating complementary data from both internal and external data sources into one chemogenomics database (Chemical Genetic Interaction Enterprise; CHEMGENIE). Here, we demonstrate how this well-curated database facilitates compound set design, tool compound selection, target deconvolution in phenotypic screening, and predictive model building.

Linking High-Throughput Screens to Identify MoAs and Novel Inhibitors of Mycobacterium tuberculosis Dihydrofolate Reductase.

Though phenotypic and target-based high-throughput screening approaches have been employed to discover new antibiotics, the identification of promising therapeutic candidates remains challenging. Each approach provides different information, and understanding their results can provide hypotheses for a mechanism of action (MoA) and reveal actionable chemical matter. Here, we describe a framework for identifying efficacy targets of bioactive compounds. High throughput biophysical profiling against a broad range of targets coupled with machine learning was employed to identify chemical features with predicted efficacy targets for a given phenotypic screen. We validate the approach on data from a set of 55 000 compounds in 24 historical internal antibacterial phenotypic screens and 636 bacterial targets screened in high-throughput biophysical binding assays. Models were built to reveal the relationships between phenotype, target, and chemotype, which recapitulated mechanisms for known antibacterials. We also prospectively identified novel inhibitors of dihydrofolate reductase with nanomolar antibacterial efficacy against Mycobacterium tuberculosis. Molecular modeling provided structural insight into target-ligand interactions underlying selective killing activity toward mycobacteria over human cells.

Iterative Focused Screening with Biological Fingerprints Identifies Selective Asc-1 Inhibitors Distinct from Traditional High Throughput Screening.

N-methyl-d-aspartate receptors (NMDARs) mediate glutamatergic signaling that is critical to cognitive processes in the central nervous system, and NMDAR hypofunction is thought to contribute to cognitive impairment observed in both schizophrenia and Alzheimer's disease. One approach to enhance the function of NMDAR is to increase the concentration of an NMDAR coagonist, such as glycine or d-serine, in the synaptic cleft. Inhibition of alanine-serine-cysteine transporter-1 (Asc-1), the primary transporter of d-serine, is attractive because the transporter is localized to neurons in brain regions critical to cognitive function, including the hippocampus and cortical layers III and IV, and is colocalized with d-serine and NMDARs. To identify novel Asc-1 inhibitors, two different screening approaches were performed with whole-cell amino acid uptake in heterologous cells stably expressing human Asc-1: (1) a high-throughput screen (HTS) of 3 M compounds measuring 35S l-cysteine uptake into cells attached to scintillation proximity assay beads in a 1536 well format and (2) an iterative focused screen (IFS) of a 45 000 compound diversity set using a 3H d-serine uptake assay with a liquid scintillation plate reader in a 384 well format. Critically important for both screening approaches was the implementation of counter screens to remove nonspecific inhibitors of radioactive amino acid uptake. Furthermore, a 15 000 compound expansion step incorporating both on- and off-target data into chemical and biological fingerprint-based models for selection of additional hits enabled the identification of novel Asc-1-selective chemical matter from the IFS that was not identified in the full-collection HTS.

Prospective Assessment of Virtual Screening Heuristics Derived Using a Novel Fusion Score.

High-throughput screening (HTS) is a widespread method in early drug discovery for identifying promising chemical matter that modulates a target or phenotype of interest. Because HTS campaigns involve screening millions of compounds, it is often desirable to initiate screening with a subset of the full collection. Subsequently, virtual screening methods prioritize likely active compounds in the remaining collection in an iterative process. With this approach, orthogonal virtual screening methods are often applied, necessitating the prioritization of hits from different approaches. Here, we introduce a novel method of fusing these prioritizations and benchmark it prospectively on 17 screening campaigns using virtual screening methods in three descriptor spaces. We found that the fusion approach retrieves 15% to 65% more active chemical series than any single machine-learning method and that appropriately weighting contributions of similarity and machine-learning scoring techniques can increase enrichment by 1% to 19%. We also use fusion scoring to evaluate the tradeoff between screening more chemical matter initially in lieu of replicate samples to prevent false-positives and find that the former option leads to the retrieval of more active chemical series. These results represent guidelines that can increase the rate of identification of promising active compounds in future iterative screens.