Discovery of a structural class of antibiotics with explainable deep learning.

The discovery of novel structural classes of antibiotics is urgently needed to address the ongoing antibiotic resistance crisis1-9. Deep learning approaches have aided in exploring chemical spaces1,10-15; these typically use black box models and do not provide chemical insights. Here we reasoned that the chemical substructures associated with antibiotic activity learned by neural network models can be identified and used to predict structural classes of antibiotics. We tested this hypothesis by developing an explainable, substructure-based approach for the efficient, deep learning-guided exploration of chemical spaces. We determined the antibiotic activities and human cell cytotoxicity profiles of 39,312 compounds and applied ensembles of graph neural networks to predict antibiotic activity and cytotoxicity for 12,076,365 compounds. Using explainable graph algorithms, we identified substructure-based rationales for compounds with high predicted antibiotic activity and low predicted cytotoxicity. We empirically tested 283 compounds and found that compounds exhibiting antibiotic activity against Staphylococcus aureus were enriched in putative structural classes arising from rationales. Of these structural classes of compounds, one is selective against methicillin-resistant S. aureus (MRSA) and vancomycin-resistant enterococci, evades substantial resistance, and reduces bacterial titres in mouse models of MRSA skin and systemic thigh infection. Our approach enables the deep learning-guided discovery of structural classes of antibiotics and demonstrates that machine learning models in drug discovery can be explainable, providing insights into the chemical substructures that underlie selective antibiotic activity.

GSK3α, not GSK3ß, drives hippocampal NMDAR-dependent LTD via tau-mediated spine anchoring.

Glycogen synthase kinase-3 (GSK3) is an important signalling protein in the brain and modulates different forms of synaptic plasticity. Neuronal functions of GSK3 are typically attributed to one of its two isoforms, GSK3ß, simply because of its prevalent expression in the brain. Consequently, the importance of isoform-specific functions of GSK3 in synaptic plasticity has not been fully explored. We now directly address this question for NMDA receptor-dependent long-term depression (LTD) in the hippocampus. Here, we specifically target the GSK3 isoforms with shRNA knock-down in mouse hippocampus and with novel isoform-selective drugs to dissect their roles in LTD. Using electrophysiological and live imaging approaches, we find that GSK3α, but not GSK3ß, is required for LTD. The specific engagement of GSK3α occurs via its transient anchoring in dendritic spines during LTD induction. We find that the major GSK3 substrate, the microtubule-binding protein tau, is required for this spine anchoring of GSK3α and mediates GSK3α-induced LTD. These results link GSK3α and tau in a common mechanism for synaptic depression and rule out a major role for GSK3ß in this process.

Comprehensive characterization of amino acid positions in protein structures reveals molecular effect of missense variants.

Interpretation of the colossal number of genetic variants identified from sequencing applications is one of the major bottlenecks in clinical genetics, with the inference of the effect of amino acid-substituting missense variations on protein structure and function being especially challenging. Here we characterize the three-dimensional (3D) amino acid positions affected in pathogenic and population variants from 1,330 disease-associated genes using over 14,000 experimentally solved human protein structures. By measuring the statistical burden of variations (i.e., point mutations) from all genes on 40 3D protein features, accounting for the structural, chemical, and functional context of the variations' positions, we identify features that are generally associated with pathogenic and population missense variants. We then perform the same amino acid-level analysis individually for 24 protein functional classes, which reveals unique characteristics of the positions of the altered amino acids: We observe up to 46% divergence of the class-specific features from the general characteristics obtained by the analysis on all genes, which is consistent with the structural diversity of essential regions across different protein classes. We demonstrate that the function-specific 3D features of the variants match the readouts of mutagenesis experiments for BRCA1 and PTEN, and positively correlate with an independent set of clinically interpreted pathogenic and benign missense variants. Finally, we make our results available through a web server to foster accessibility and downstream research. Our findings represent a crucial step toward translational genetics, from highlighting the impact of mutations on protein structure to rationalizing the variants' pathogenicity in terms of the perturbed molecular mechanisms.

MISCAST: MIssense variant to protein StruCture Analysis web SuiTe.

Human genome sequencing efforts have greatly expanded, and a plethora of missense variants identified both in patients and in the general population is now publicly accessible. Interpretation of the molecular-level effect of missense variants, however, remains challenging and requires a particular investigation of amino acid substitutions in the context of protein structure and function. Answers to questions like 'Is a variant perturbing a site involved in key macromolecular interactions and/or cellular signaling?', or 'Is a variant changing an amino acid located at the protein core or part of a cluster of known pathogenic mutations in 3D?' are crucial. Motivated by these needs, we developed MISCAST (missense variant to protein structure analysis web suite; http://miscast.broadinstitute.org/). MISCAST is an interactive and user-friendly web server to visualize and analyze missense variants in protein sequence and structure space. Additionally, a comprehensive set of protein structural and functional features have been aggregated in MISCAST from multiple databases, and displayed on structures alongside the variants to provide users with the biological context of the variant location in an integrated platform. We further made the annotated data and protein structures readily downloadable from MISCAST to foster advanced offline analysis of missense variants by a wide biological community.

Multimodal small-molecule screening for human prion protein binders.

Prion disease is a rapidly progressive neurodegenerative disorder caused by misfolding and aggregation of the prion protein (PrP), and there are currently no therapeutic options. PrP ligands could theoretically antagonize prion formation by protecting the native protein from misfolding or by targeting it for degradation, but no validated small-molecule binders have been discovered to date. We deployed a variety of screening methods in an effort to discover binders of PrP, including 19F-observed and saturation transfer difference (STD) NMR spectroscopy, differential scanning fluorimetry (DSF), DNA-encoded library selection, and in silico screening. A single benzimidazole compound was confirmed in concentration-response, but affinity was very weak (Kd > 1 mm), and it could not be advanced further. The exceptionally low hit rate observed here suggests that PrP is a difficult target for small-molecule binders. Whereas orthogonal binder discovery methods could yield high-affinity compounds, non-small-molecule modalities may offer independent paths forward against prion disease.

Roles of glycogen synthase kinase 3 alpha and calcineurin in regulating the ability of sperm to fertilize eggs.

Glycogen synthase kinase 3 (GSK3) was identified as an enzyme regulating sperm protein phosphatase. The GSK3α paralog, but not GSK3ß, is essential for sperm function. Sperm lacking GSK3α display altered motility and are unable to undergo hyperactivation, which is essential for fertilization. Male mice lacking sperm-specific calcineurin (PP2B), a calcium regulated phosphatase, in testis and sperm, are also infertile. Loss of PP2B results in impaired epididymal sperm maturation and motility. The phenotypes of GSK3α and PP2B knockout mice are similar, prompting us to examine the interrelationship between these two enzymes in sperm. High calcium levels must exist to permit catalytically active calcineurin to function during epididymal sperm maturation. Total and free calcium levels are high in immotile compared to motile epididymal sperm. Inhibition of calcineurin by FK506 results in an increase in the net phosphorylation and a consequent decrease in catalytic activity of sperm GSK3. The inhibitor FK506 and an isoform-selective inhibitor of GSK3α, BRD0705, also inhibited fertilization of eggs in vitro. Interrelated functions of GSK3α and sperm PP2B are essential during epididymal sperm maturation and during fertilization. Our results should enable the development of male contraceptives targeting one or both enzymes.

Multigram Preparation of BRD4780 Enantiomers and Assignment of Absolute Stereochemistry.

The development of a multigram synthesis of 3-exo-isopropylbicyclo[2.2.1]heptan-2-endo-amine hydrochloride (1) (also known as BRD4780 and AGN-192403) is described. The process involves protection of the amine as 4-nitrobenzyl carbamate, pNZ, which enables chiral SFC chromatography. The absolute configuration (AC) of the individual enantiomers has been determined by Mosher's amide method, VCD spectroscopy, and X-ray crystallography. We highlight the VCD approach as a rapid and effective means of AC determination that can be deployed directly on the target compounds.

Entropy as a Driver of Selectivity for Inhibitor Binding to Histone Deacetylase 6.

Among the metal-dependent histone deacetylases, the class IIb isozyme HDAC6 is remarkable because of its role in the regulation of microtubule dynamics in the cytosol. Selective inhibition of HDAC6 results in microtubule hyperacetylation, leading to cell cycle arrest and apoptosis, which is a validated strategy for cancer chemotherapy and the treatment of other disorders. HDAC6 inhibitors generally consist of a Zn2+-binding group such as a hydroxamate, a linker, and a capping group; the capping group is a critical determinant of isozyme selectivity. Surprisingly, however, even "capless" inhibitors exhibit appreciable HDAC6 selectivity. To probe the chemical basis for this selectivity, we now report high-resolution crystal structures of HDAC6 complexed with capless cycloalkyl hydroxamate inhibitors 1-4. Each inhibitor hydroxamate group coordinates to the catalytic Zn2+ ion with canonical bidentate geometry. Additionally, the olefin moieties of compounds 2 and 4 bind in an aromatic crevice between the side chains of F583 and F643. Reasoning that similar binding could be achieved in the representative class I isozyme HDAC8, we employed isothermal titration calorimetry to study the thermodynamics of inhibitor binding. These measurements indicate that the entropy of inhibitor binding is generally positive for binding to HDAC6 and negative for binding to HDAC8, resulting in ≤313-fold selectivity for binding to HDAC6 relative to HDAC8. Thus, favorable binding entropy contributes to HDAC6 selectivity. Notably, cyclohexenyl hydroxamate 2 represents a promising lead for derivatization with capping groups that may further enhance its impressive 313-fold thermodynamic selectivity for HDAC6 inhibition.

Isoform-selective inhibitor of histone deacetylase 3 (HDAC3) limits pancreatic islet infiltration and protects female nonobese diabetic mice from diabetes.

Preservation of insulin-secreting ß-cells is an important goal for therapies aimed at restoring normoglycemia in patients with diabetes. One approach, the inhibition of histone deacetylases (HDACs), has been reported to suppress pancreatic islet inflammation and ß-cell apoptosis in vitro In this report, we demonstrate the efficacy of HDAC inhibitors (HDACi) in vivo We show that daily administration of BRD3308, an isoform-selective HDAC3 inhibitor, for 2 weeks to female nonobese diabetic (NOD) mice, beginning at 3 weeks of age, followed by twice-weekly injections until age 25 weeks, protects the animals from diabetes. The preservation of ß-cells was because of a significant decrease in islet infiltration of mononuclear cells. Moreover, the BRD3308 treatment increased basal insulin secretion from islets cultured in vitro All metabolic tissues tested in vehicle- or BRD3308-treated groups showed virtually no sign of immune cell infiltration, except minimal infiltration in white adipose tissue in animals treated with the highest BRD3308 dose (10 mg/kg), providing additional evidence of protection from immune attack in the treated groups. Furthermore, pancreata from animals treated with 10 mg/kg BRD3308 exhibited significantly decreased numbers of apoptotic ß-cells compared with those treated with vehicle or low-dose BRD3308. Finally, animals treated with 1 or 10 mg/kg BRD3308 had enhanced ß-cell proliferation. These in vivo results point to the potential use of selective HDAC3 inhibitors as a therapeutic approach to suppress pancreatic islet infiltration and prevent ß-cell death with the long-term goal of limiting the progression of type 1 diabetes.

Overlapping and Divergent Actions of Structurally Distinct Histone Deacetylase Inhibitors in Cardiac Fibroblasts.

Inhibitors of zinc-dependent histone deacetylases (HDACs) profoundly affect cellular function by altering gene expression via changes in nucleosomal histone tail acetylation. Historically, investigators have employed pan-HDAC inhibitors, such as the hydroxamate trichostatin A (TSA), which simultaneously targets members of each of the three zinc-dependent HDAC classes (classes I, II, and IV). More recently, class- and isoform-selective HDAC inhibitors have been developed, providing invaluable chemical biology probes for dissecting the roles of distinct HDACs in the control of various physiologic and pathophysiological processes. For example, the benzamide class I HDAC-selective inhibitor, MGCD0103 [N-(2-aminophenyl)-4-[[(4-pyridin-3-ylpyrimidin-2-yl)amino]methyl] benzamide], was shown to block cardiac fibrosis, a process involving excess extracellular matrix deposition, which often results in heart dysfunction. Here, we compare the mechanisms of action of structurally distinct HDAC inhibitors in isolated primary cardiac fibroblasts, which are the major extracellular matrix-producing cells of the heart. TSA, MGCD0103, and the cyclic peptide class I HDAC inhibitor, apicidin, exhibited a common ability to enhance histone acetylation, and all potently blocked cardiac fibroblast cell cycle progression. In contrast, MGCD0103, but not TSA or apicidin, paradoxically increased expression of a subset of fibrosis-associated genes. Using the cellular thermal shift assay, we provide evidence that the divergent effects of HDAC inhibitors on cardiac fibroblast gene expression relate to differential engagement of HDAC1- and HDAC2-containing complexes. These findings illustrate the importance of employing multiple compounds when pharmacologically assessing HDAC function in a cellular context and during HDAC inhibitor drug development.

Kinetic and structural insights into the binding of histone deacetylase 1 and 2 (HDAC1, 2) inhibitors.

The structure-activity and structure-kinetic relationships of a series of novel and selective ortho-aminoanilide inhibitors of histone deacetylases (HDACs) 1 and 2 are described. Different kinetic and thermodynamic selectivity profiles were obtained by varying the moiety occupying an 11Å channel leading to the Zn(2+) catalytic pocket of HDACs 1 and 2, two paralogs with a high degree of structural similarity. The design of these novel inhibitors was informed by two ligand-bound crystal structures of truncated hHDAC2. BRD4884 and BRD7232 possess kinetic selectivity for HDAC1 versus HDAC2. We demonstrate that the binding kinetics of HDAC inhibitors can be tuned for individual isoforms in order to modulate target residence time while retaining functional activity and increased histone H4K12 and H3K9 acetylation in primary mouse neuronal cell culture assays. These chromatin modifiers, with tuned binding kinetic profiles, can be used to define the relation between target engagement requirements and the pharmacodynamic response of HDACs in different disease applications.

Signal-dependent repression of DUSP5 by class I HDACs controls nuclear ERK activity and cardiomyocyte hypertrophy.

Cardiac hypertrophy is a strong predictor of morbidity and mortality in patients with heart failure. Small molecule histone deacetylase (HDAC) inhibitors have been shown to suppress cardiac hypertrophy through mechanisms that remain poorly understood. We report that class I HDACs function as signal-dependent repressors of cardiac hypertrophy via inhibition of the gene encoding dual-specificity phosphatase 5 (DUSP5) DUSP5, a nuclear phosphatase that negatively regulates prohypertrophic signaling by ERK1/2. Inhibition of DUSP5 by class I HDACs requires activity of the ERK kinase, mitogen-activated protein kinase kinase (MEK), revealing a self-reinforcing mechanism for promotion of cardiac ERK signaling. In cardiac myocytes treated with highly selective class I HDAC inhibitors, nuclear ERK1/2 signaling is suppressed in a manner that is absolutely dependent on DUSP5. In contrast, cytosolic ERK1/2 activation is maintained under these same conditions. Ectopic expression of DUSP5 in cardiomyocytes results in potent inhibition of agonist-dependent hypertrophy through a mechanism involving suppression of the gene program for hypertrophic growth. These findings define unique roles for class I HDACs and DUSP5 as integral components of a regulatory signaling circuit that controls cardiac hypertrophy.

Discovery of a Highly Selective Glycogen Synthase Kinase-3 Inhibitor (PF-04802367) That Modulates Tau Phosphorylation in the Brain: Translation for PET Neuroimaging.

Glycogen synthase kinase-3 (GSK-3) regulates multiple cellular processes in diabetes, oncology, and neurology. N-(3-(1H-1,2,4-triazol-1-yl)propyl)-5-(3-chloro-4-methoxyphenyl)oxazole-4-carboxamide (PF-04802367 or PF-367) has been identified as a highly potent inhibitor, which is among the most selective antagonists of GSK-3 to date. Its efficacy was demonstrated in modulation of tau phosphorylation in vitro and in vivo. Whereas the kinetics of PF-367 binding in brain tissues are too fast for an effective therapeutic agent, the pharmacokinetic profile of PF-367 is ideal for discovery of radiopharmaceuticals for GSK-3 in the central nervous system. A (11) C-isotopologue of PF-367 was synthesized and preliminary PET imaging studies in non-human primates confirmed that we have overcome the two major obstacles for imaging GSK-3, namely, reasonable brain permeability and displaceable binding.

Hydroxamic acid-based histone deacetylase (HDAC) inhibitors can mediate neuroprotection independent of HDAC inhibition.

Histone deacetylase (HDAC) inhibition improves function and extends survival in rodent models of a host of neurological conditions, including stroke, and neurodegenerative diseases. Our understanding, however, of the contribution of individual HDAC isoforms to neuronal death is limited. In this study, we used selective chemical probes to assess the individual roles of the Class I HDAC isoforms in protecting Mus musculus primary cortical neurons from oxidative death. We demonstrated that the selective HDAC8 inhibitor PCI-34051 is a potent neuroprotective agent; and by taking advantage of both pharmacological and genetic tools, we established that HDAC8 is not critically involved in PCI-34051's mechanism of action. We used BRD3811, an inactive ortholog of PCI-34051, and showed that, despite its inability to inhibit HDAC8, it exhibits robust neuroprotective properties. Furthermore, molecular deletion of HDAC8 proved insufficient to protect neurons from oxidative death, whereas both PCI-34051 and BRD3811 were able to protect neurons derived from HDAC8 knock-out mice. Finally, we designed and synthesized two new, orthogonal negative control compounds, BRD9715 and BRD8461, which lack the hydroxamic acid motif and showed that they stably penetrate cell membranes but are not neuroprotective. These results indicate that the protective effects of these hydroxamic acid-containing small molecules are likely unrelated to direct epigenetic regulation via HDAC inhibition, but rather due to their ability to bind metals. Our results suggest that hydroxamic acid-based HDAC inhibitors may mediate neuroprotection via HDAC-independent mechanisms and affirm the need for careful structure-activity relationship studies when using pharmacological approaches.

Chromatin-targeting small molecules cause class-specific transcriptional changes in pancreatic endocrine cells.

Under the instruction of cell-fate-determining, DNA-binding transcription factors, chromatin-modifying enzymes mediate and maintain cell states throughout development in multicellular organisms. Currently, small molecules modulating the activity of several classes of chromatin-modifying enzymes are available, including clinically approved histone deacetylase (HDAC) and DNA methyltransferase (DNMT) inhibitors. We describe the genome-wide expression changes induced by 29 compounds targeting HDACs, DNMTs, histone lysine methyltransferases (HKMTs), and protein arginine methyltransferases (PRMTs) in pancreatic α- and ß-cell lines. HDAC inhibitors regulate several hundred transcripts irrespective of the cell type, with distinct clusters of dissimilar activity for hydroxamic acids and orthoamino anilides. In contrast, compounds targeting histone methyltransferases modulate the expression of restricted gene sets in distinct cell types. For example, we find that G9a/GLP methyltransferase inhibitors selectively up-regulate the cholesterol biosynthetic pathway in pancreatic but not liver cells. These data suggest that, despite their conservation across the entire genome and in different cell types, chromatin pathways can be targeted to modulate the expression of selected transcripts.

Studies toward the total synthesis of dihydrolycolucine. Preparation of AB and CEF ring fragments.

A strategy for the synthesis of the lycopodium alkaloid dihydrolycolucine (1) has been investigated. Synthetic routes were developed based on N-acylpyridinium salt chemistry to prepare target fragments 3 and 4 that could ultimately converge to the natural product. Key reactions include IMDA cycloadditions and retro-Mannich ring-openings to form both the AB and the EF ring fragments. The ring C precursor was prepared using pyridine substitution and directed lithiation chemistry. A Suzuki cross-coupling of rings C and EF led to the CEF ring fragment. Initial attempts at closure of the seven-membered D ring were unsuccessful.

A Fast-Binding, Functionally Reversible, COX-2 Radiotracer for CNS PET Imaging.

Cyclooxygenase-2 (COX-2) is an enzyme that plays a pivotal role in peripheral inflammation and pain via the prostaglandin pathway. In the central nervous system (CNS), COX-2 is implicated in neurodegenerative and psychiatric disorders as a potential therapeutic target and biomarker. However, clinical studies with COX-2 have yielded inconsistent results, partly due to limited mechanistic understanding of how COX-2 activity relates to CNS pathology. Therefore, developing COX-2 positron emission tomography (PET) radiotracers for human neuroimaging is of interest. This study introduces [11C]BRD1158, which is a potent and uniquely fast-binding, selective COX-2 PET radiotracer. [11C]BRD1158 was developed by prioritizing potency at COX-2, isoform selectivity over COX-1, fast binding kinetics, and free fraction in the brain. Evaluated through in vivo PET neuroimaging in rodent models with human COX-2 overexpression, [11C]BRD1158 demonstrated high brain uptake, fast target-engagement, functional reversibility, and excellent specific binding, which is advantageous for human imaging applications. Lastly, post-mortem samples from Huntington's disease (HD) patients and preclinical HD mouse models showed that COX-2 levels were elevated specifically in disease-affected brain regions, primarily from increased expression in microglia. These findings indicate that COX-2 holds promise as a novel clinical marker of HD onset and progression, one of many potential applications of [11C]BRD1158 human PET.

Synthesis and profiling of a diverse collection of azetidine-based scaffolds for the development of CNS-focused lead-like libraries.

The synthesis and diversification of a densely functionalized azetidine ring system to gain access to a wide variety of fused, bridged, and spirocyclic ring systems is described. The in vitro physicochemical and pharmacokinetic properties of representative library members are measured in order to evaluate the use of these scaffolds for the generation of lead-like molecules to be used in targeting the central nervous system. The solid-phase synthesis of a 1976-membered library of spirocyclic azetidines is also described.

Novel Fluorescence-Based High-Throughput FLIPR Assay Utilizing Membrane-Tethered Genetic Calcium Sensors to Identify T-Type Calcium Channel Modulators.

T-type voltage-gated Ca2+ channels have been implicated in many human disorders, and there has been increasing interest in developing highly selective and potent T-type Ca2+ channel modulators for potential clinical use. However, the unique biophysical properties of T-type Ca2+ channels are not conducive for developing high-throughput screening (HTS) assays to identify modulators, particularly potentiators. To illustrate, T-type Ca2+ channels are largely inactivated and unable to open to allow Ca2+ influx at -25 mV, the typical resting membrane potential of the cell lines commonly used in cellular screening assays. To address this issue, we developed cell lines that express Kir2.3 channels to hyperpolarize the membrane potential to -70 mV, thus allowing T-type channels to return to their resting state where they can be subsequently activated by membrane depolarization in the presence of extracellular KCl. Furthermore, to simplify the HTS assay and to reduce reagent cost, we stably expressed a membrane-tethered genetic calcium sensor, GCaMP6s-CAAX, that displays superior signal to the background compared to the untethered GCaMP6s or the synthetic Ca2+ sensor Fluo-4AM. Here, we describe a novel GCaMP6s-CAAX-based calcium assay utilizing a high-throughput fluorometric imaging plate reader (Molecular Devices, Sunnyvale, CA) format that can identify both activators and inhibitors of T-type Ca2+ channels. Lastly, we demonstrate the utility of this novel fluorescence-based assay to evaluate the activities of two distinct G-protein-coupled receptors, thus expanding the use of GCaMP6s-CAAX to a wide range of applications relevant for developing cellular assays in drug discovery.

Virtual screening for small-molecule pathway regulators by image-profile matching.

Identifying the chemical regulators of biological pathways is a time-consuming bottleneck in developing therapeutics and research compounds. Typically, thousands to millions of candidate small molecules are tested in target-based biochemical screens or phenotypic cell-based screens, both expensive experiments customized to each disease. Here, our uncustomized, virtual, profile-based screening approach instead identifies compounds that match to pathways based on the phenotypic information in public cell image data, created using the Cell Painting assay. Our straightforward correlation-based computational strategy retrospectively uncovered the expected, known small-molecule regulators for 32% of positive-control gene queries. In prospective, discovery mode, we efficiently identified new compounds related to three query genes and validated them in subsequent gene-relevant assays, including compounds that phenocopy or pheno-oppose YAP1 overexpression and kill a Yap1-dependent sarcoma cell line. This image-profile-based approach could replace many customized labor- and resource-intensive screens and accelerate the discovery of biologically and therapeutically useful compounds.