A Deep Learning Approach to Antibiotic Discovery.

Due to the rapid emergence of antibiotic-resistant bacteria, there is a growing need to discover new antibiotics. To address this challenge, we trained a deep neural network capable of predicting molecules with antibacterial activity. We performed predictions on multiple chemical libraries and discovered a molecule from the Drug Repurposing Hub-halicin-that is structurally divergent from conventional antibiotics and displays bactericidal activity against a wide phylogenetic spectrum of pathogens including Mycobacterium tuberculosis and carbapenem-resistant Enterobacteriaceae. Halicin also effectively treated Clostridioides difficile and pan-resistant Acinetobacter baumannii infections in murine models. Additionally, from a discrete set of 23 empirically tested predictions from >107 million molecules curated from the ZINC15 database, our model identified eight antibacterial compounds that are structurally distant from known antibiotics. This work highlights the utility of deep learning approaches to expand our antibiotic arsenal through the discovery of structurally distinct antibacterial molecules.

Nucleosome Turnover Regulates Histone Methylation Patterns over the Genome.

Recent studies have indicated that nucleosome turnover is rapid, occurring several times per cell cycle. To access the effect of nucleosome turnover on the epigenetic landscape, we investigated H3K79 methylation, which is produced by a single methyltransferase (Dot1l) with no known demethylase. Using chemical-induced proximity (CIP), we find that the valency of H3K79 methylation (mono-, di-, and tri-) is determined by nucleosome turnover rates. Furthermore, propagation of this mark is predicted by nucleosome turnover simulations over the genome and accounts for the asymmetric distribution of H3K79me toward the transcriptional unit. More broadly, a meta-analysis of other conserved histone modifications demonstrates that nucleosome turnover models predict both valency and chromosomal propagation of methylation marks. Based on data from worms, flies, and mice, we propose that the turnover of modified nucleosomes is a general means of propagation of epigenetic marks and a determinant of methylation valence.

Systematic molecular evolution enables robust biomolecule discovery.

Evolution occurs when selective pressures from the environment shape inherited variation over time. Within the laboratory, evolution is commonly used to engineer proteins and RNA, but experimental constraints have limited the ability to reproducibly and reliably explore factors such as population diversity, the timing of environmental changes and chance on outcomes. We developed a robotic system termed phage- and robotics-assisted near-continuous evolution (PRANCE) to comprehensively explore biomolecular evolution by performing phage-assisted continuous evolution in high-throughput. PRANCE implements an automated feedback control system that adjusts the stringency of selection in response to real-time measurements of each molecular activity. In evolving three distinct types of biomolecule, we find that evolution is reproducibly altered by both random chance and the historical pattern of environmental changes. This work improves the reliability of protein engineering and enables the systematic analysis of the historical, environmental and random factors governing biomolecular evolution.

Enabling high-throughput biology with flexible open-source automation.

Our understanding of complex living systems is limited by our capacity to perform experiments in high throughput. While robotic systems have automated many traditional hand-pipetting protocols, software limitations have precluded more advanced maneuvers required to manipulate, maintain, and monitor hundreds of experiments in parallel. Here, we present Pyhamilton, an open-source Python platform that can execute complex pipetting patterns required for custom high-throughput experiments such as the simulation of metapopulation dynamics. With an integrated plate reader, we maintain nearly 500 remotely monitored bacterial cultures in log-phase growth for days without user intervention by taking regular density measurements to adjust the robotic method in real-time. Using these capabilities, we systematically optimize bioreactor protein production by monitoring the fluorescent protein expression and growth rates of a hundred different continuous culture conditions in triplicate to comprehensively sample the carbon, nitrogen, and phosphorus fitness landscape. Our results demonstrate that flexible software can empower existing hardware to enable new types and scales of experiments, empowering areas from biomanufacturing to fundamental biology.

Proximity labeling of endogenous protein interactions enabled by directed evolution.

Enhanced ascorbate peroxidase 2 (APEX2) is a protein generated with directed evolution by Lam et al. that has transformed our understanding of subcellular entities and phenomena. The rapid kinetics of this engineered protein highlights the power of directed evolution to expand the molecular toolkit for biologists.

High-throughput approaches to uncover synergistic drug combinations in leukemia.

We report a comprehensive drug synergy study in acute myeloid leukemia (AML). In this work, we investigate a panel of cell lines spanning both MLL-rearranged and non-rearranged subtypes. The work comprises a resource for the community, with many synergistic drug combinations that could not have been predicted a priori, and open source code for automation and analyses. We base our definitions of drug synergy on the Chou-Talalay method, which is useful for visualizations of synergy experiments in isobolograms, and median-effects plots, among other representations. Our key findings include drug synergies affecting the chromatin state, specifically in the context of regulation of the modification state of histone H3 lysine-27. We report open source high throughput methodology such that multidimensional drug screening can be accomplished with equipment that is accessible to most laboratories. This study will enable preclinical investigation of new drug combinations in a lethal blood cancer, with data analysis and automation workflows freely available to the community.

RNA-responsive elements for eukaryotic translational control.

The ability to control translation of endogenous or exogenous RNAs in eukaryotic cells would facilitate a variety of biotechnological applications. Current strategies are limited by low fold changes in transgene output and the size of trigger RNAs (trRNAs). Here we introduce eukaryotic toehold switches (eToeholds) as modular riboregulators. eToeholds contain internal ribosome entry site sequences and form inhibitory loops in the absence of a specific trRNA. When the trRNA is present, eToeholds anneal to it, disrupting the inhibitory loops and allowing translation. Through optimization of RNA annealing, we achieved up to 16-fold induction of transgene expression in mammalian cells. We demonstrate that eToeholds can discriminate among viral infection status, presence or absence of gene expression and cell types based on the presence of exogenous or endogenous RNA transcripts.

Chemical Inhibitors of a Selective SWI/SNF Function Synergize with ATR Inhibition in Cancer Cell Killing.

SWI/SNF (BAF) complexes are a diverse family of ATP-dependent chromatin remodelers produced by combinatorial assembly that are mutated in and thought to contribute to 20% of human cancers and a large number of neurologic diseases. The gene-activating functions of BAF complexes are essential for viability of many cell types, limiting the development of small molecule inhibitors. To circumvent the potential toxicity of SWI/SNF inhibition, we identified small molecules that inhibit the specific repressive function of these complexes but are relatively nontoxic and importantly synergize with ATR inhibitors in killing cancer cells. Our studies suggest an avenue for therapeutic enhancement of ATR/ATM inhibition and provide evidence for chemical synthetic lethality of BAF complexes as a therapeutic strategy in cancer.

Chemically induced proximity in biology and medicine.

Proximity, or the physical closeness of molecules, is a pervasive regulatory mechanism in biology. For example, most posttranslational modifications such as phosphorylation, methylation, and acetylation promote proximity of molecules to play deterministic roles in cellular processes. To understand the role of proximity in biologic mechanisms, chemical inducers of proximity (CIPs) were developed to synthetically model biologically regulated recruitment. Chemically induced proximity allows for precise temporal control of transcription, signaling cascades, chromatin regulation, protein folding, localization, and degradation, as well as a host of other biologic processes. A systematic analysis of CIPs in basic research, coupled with recent technological advances utilizing CRISPR, distinguishes roles of causality from coincidence and allows for mathematical modeling in synthetic biology. Recently, induced proximity has provided new avenues of gene therapy and emerging advances in cancer treatment.

Rapid and reversible epigenome editing by endogenous chromatin regulators.

Understanding the causal link between epigenetic marks and gene regulation remains a central question in chromatin biology. To edit the epigenome we developed the FIRE-Cas9 system for rapid and reversible recruitment of endogenous chromatin regulators to specific genomic loci. We enhanced the dCas9-MS2 anchor for genome targeting with Fkbp/Frb dimerizing fusion proteins to allow chemical-induced proximity of a desired chromatin regulator. We find that mSWI/SNF (BAF) complex recruitment is sufficient to oppose Polycomb within minutes, leading to activation of bivalent gene transcription in mouse embryonic stem cells. Furthermore, Hp1/Suv39h1 heterochromatin complex recruitment to active promoters deposits H3K9me3 domains, resulting in gene silencing that can be reversed upon washout of the chemical dimerizer. This inducible recruitment strategy provides precise kinetic information to model epigenetic memory and plasticity. It is broadly applicable to mechanistic studies of chromatin in mammalian cells and is particularly suited to the analysis of endogenous multi-subunit chromatin regulator complexes.Understanding the link between epigenetic marks and gene regulation requires the development of new tools to directly manipulate chromatin. Here the authors demonstrate a Cas9-based system to recruit chromatin remodelers to loci of interest, allowing rapid, reversible manipulation of epigenetic states.

Dynamics of BAF-Polycomb complex opposition on heterochromatin in normal and oncogenic states.

The opposition between Polycomb repressive complexes (PRCs) and BAF (mSWI/SNF) complexes has a critical role in both development and disease. Mutations in the genes encoding BAF subunits contribute to more than 20% of human malignancies, yet the underlying mechanisms remain unclear, owing largely to a lack of assays to assess BAF function in living cells. To address this, we have developed a widely applicable recruitment assay system through which we find that BAF opposes PRC by rapid, ATP-dependent eviction, leading to the formation of accessible chromatin. The reversal of this process results in reassembly of facultative heterochromatin. Surprisingly, BAF-mediated PRC eviction occurs in the absence of RNA polymerase II (Pol II) occupancy, transcription, and replication. Further, we find that tumor-suppressor and oncogenic mutant BAF complexes have different effects on PRC eviction. The results of these studies define a mechanistic sequence underlying the resolution and formation of facultative heterochromatin, and they demonstrate that BAF opposes PRC on a minute-by-minute basis to provide epigenetic plasticity.

Structure-guided DOT1L probe optimization by label-free ligand displacement.

The DOT1L lysine methyltransferase has emerged as a validated therapeutic target in MLL-rearranged (MLLr) acute leukemias. Although S-adenosylmethionine competitive inhibitors have demonstrated pharmacological proof-of-principle in MLLr-leukemia, these compounds require further optimization to improve cellular potency and pharmacokinetic stability. Limiting DOT1L inhibitor discovery and ligand optimization have been complex biochemical methods often using radionucleotides and cellular methods requiring prolonged culture. We therefore developed a new suite of assay technologies that allows comparative assessment of chemical tools for DOT1L in a miniaturized format. Coupling these assays with structural information, we developed new insights into DOT1L ligand binding and identified several functionalized probes with increased cellular potency (IC50 values ∼10 nM) and excellent selectivity for DOT1L. Together these assay technologies define a platform capability for discovery and optimization of small-molecule DOT1L inhibitors.

p38 signaling and receptor recycling events in a microfluidic endothelial cell adhesion assay.

Adhesion-based microfluidic cell separation has proven to be very useful in applications ranging from cancer diagnostics to tissue engineering. This process involves functionalizing microchannel surfaces with a capture molecule. High specificity and purity capture can be achieved using this method. Despite these advances, little is known about the mechanisms that govern cell capture within these devices and their relationships to basic process parameters such as fluid shear stress and the presence of soluble factors. This work examines how the adhesion of human endothelial cells (ECs) is influenced by a soluble tetrapeptide, Arg-Glu-Asp-Val (REDV) and fluidic shear stress. The ability of these ECs to bind within microchannels coated with REDV is shown to be governed by shear- and soluble-factor mediated changes in p38 mitogen-activated protein kinase expression together with recycling of adhesion receptors from the endosome.

Separation of two phenotypically similar cell types via a single common marker in microfluidic channels.

To isolate clinically and biologically relevant cell types from a heterogeneous population, fluorescent or magnetic tagging together with knowledge of surface biomarker profiles represents the state of the art. To date, it remains exceedingly difficult to separate phenotypically and physically similar cell types from a mixed population. We report a microfluidic platform engineered to separate two highly similar cell types using a single antibody by taking advantage of subtle variations in surface receptor density and cell size. This platform utilizes antibody-conjugated surfaces in microfluidic channels together with precise modulation of fluid shear stresses to accomplish selective fractionation in a continuous flow process. Antibody conjugation density variation on the adhesive surfaces is achieved by covalently immobilizing an antibody in the presence of poly(ethylene glycol). This platform is used to demonstrate separation of two CD31 positive cell types, human umbilical vein endothelial cells and human micro vascular endothelial cells.

Catalytic site remodelling of the DOT1L methyltransferase by selective inhibitors.

Selective inhibition of protein methyltransferases is a promising new approach to drug discovery. An attractive strategy towards this goal is the development of compounds that selectively inhibit binding of the cofactor, S-adenosylmethionine, within specific protein methyltransferases. Here we report the three-dimensional structure of the protein methyltransferase DOT1L bound to EPZ004777, the first S-adenosylmethionine-competitive inhibitor of a protein methyltransferase with in vivo efficacy. This structure and those of four new analogues reveal remodelling of the catalytic site. EPZ004777 and a brominated analogue, SGC0946, inhibit DOT1L in vitro and selectively kill mixed lineage leukaemia cells, in which DOT1L is aberrantly localized via interaction with an oncogenic MLL fusion protein. These data provide important new insight into mechanisms of cell-active S-adenosylmethionine-competitive protein methyltransferase inhibitors, and establish a foundation for the further development of drug-like inhibitors of DOT1L for cancer therapy.