Turning high-throughput structural biology into predictive inhibitor design.

A common challenge in drug design pertains to finding chemical modifications to a ligand that increases its affinity to the target protein. An underutilized advance is the increase in structural biology throughput, which has progressed from an artisanal endeavor to a monthly throughput of hundreds of different ligands against a protein in modern synchrotrons. However, the missing piece is a framework that turns high-throughput crystallography data into predictive models for ligand design. Here, we designed a simple machine learning approach that predicts protein-ligand affinity from experimental structures of diverse ligands against a single protein paired with biochemical measurements. Our key insight is using physics-based energy descriptors to represent protein-ligand complexes and a learning-to-rank approach that infers the relevant differences between binding modes. We ran a high-throughput crystallography campaign against the SARS-CoV-2 main protease (MPro), obtaining parallel measurements of over 200 protein-ligand complexes and their binding activities. This allows us to design one-step library syntheses which improved the potency of two distinct micromolar hits by over 10-fold, arriving at a noncovalent and nonpeptidomimetic inhibitor with 120 nM antiviral efficacy. Crucially, our approach successfully extends ligands to unexplored regions of the binding pocket, executing large and fruitful moves in chemical space with simple chemistry.

Design and synthesis of a library of C8-substituted sulfamidoadenosines to probe bacterial permeability.

Gram-negative bacteria pose a major challenge in antibiotic drug discovery because their cell envelope presents a permeability barrier that affords high intrinsic resistance to small-molecule drugs. The identification of correlations between chemical structure and Gram-negative permeability would thus enable development of predictive tools to facilitate antibiotic discovery. Toward this end, have advanced a library design paradigm in which various chemical scaffolds are functionalized at different regioisomeric positions using a uniform reagent set. This design enables decoupling of scaffold, regiochemistry, and substituent effects upon Gram-negative permeability of these molecules. Building upon our recent synthesis of a library of C2-substituted sulfamidoadenosines, we have now developed an efficient synthetic route to an analogous library of regioisomeric C8-substituted congeners. The C8 library samples a region of antibiotic-relevant chemical space that is similar to that addressed by the C2 library, but distinct from that sampled by a library of analogously substituted oxazolidinones. Selected molecules were tested for accumulation in Escherichia coli in a pilot analysis, setting the stage for full comparative evaluation of these libraries in the future.

Design and synthesis of a library of C2-substituted sulfamidoadenosines to probe bacterial permeability.

Antibiotic resistance is a major threat to public health, and Gram-negative bacteria pose a particular challenge due to their combination of a low permeability cell envelope and efflux pumps. Our limited understanding of the chemical rules for overcoming these barriers represents a major obstacle in antibacterial drug discovery. Several recent efforts to address this problem have involved screening compound libraries for accumulation in bacteria in order to understand the structural properties required for Gram-negative permeability. Toward this end, we used cheminformatic analysis to design a library of sulfamidoadenosines (AMSN) having diverse substituents at the adenine C2 position. An efficient synthetic route was developed with installation of a uniform cross-coupling reagent set using Sonogashira and Suzuki reactions of a C2-iodide. The potential utility of these compounds was demonstrated by pilot analysis of selected analogues for accumulation in Escherichia coli.

GAC inhibitors with a 4-hydroxypiperidine spacer: Requirements for potency.

Allosteric inhibitors of glutaminase (GAC), such as BPTES, CB-839 and UPGL00019, have great promise as inhibitors of cancer cell growth, but potent inhibitors with drug-like qualities have been difficult to achieve. Here, a small library of GAC inhibitors based on the UPGL00019 core is described. This set of derivatives was designed to assess if one or both of the phenylacetyl groups flanking the UPGL00019 core can be replaced by smaller simple aliphatic acyl groups without loss in potency. We found that one of the phenylacetyl moieties can be replaced by a set of small aliphatic moieties without loss in potency. We also found that enzymatic potency co-varies with the VDW volume or the maximum projection area of the groups used as replacements of the phenylacetyl moiety and used literature X-ray data to provide an explanation for this finding.

Design and evaluation of novel glutaminase inhibitors.

A novel set of GAC (kidney glutaminase isoform C) inhibitors able to inhibit the enzymatic activity of GAC and the growth of the triple negative MDA-MB-231 breast cancer cells with low nanomolar potency is described. Compounds in this series have a reduced number of rotatable bonds, improved ClogPs, microsomal stability and ligand efficiency when compared to the leading GAC inhibitors BPTES and CB-839. Property improvements were achieved by the replacement of the flexible n-diethylthio or the n-butyl moiety present in the leading inhibitors by heteroatom substituted heterocycloalkanes.

Open science discovery of potent noncovalent SARS-CoV-2 main protease inhibitors.

We report the results of the COVID Moonshot, a fully open-science, crowdsourced, and structure-enabled drug discovery campaign targeting the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) main protease. We discovered a noncovalent, nonpeptidic inhibitor scaffold with lead-like properties that is differentiated from current main protease inhibitors. Our approach leveraged crowdsourcing, machine learning, exascale molecular simulations, and high-throughput structural biology and chemistry. We generated a detailed map of the structural plasticity of the SARS-CoV-2 main protease, extensive structure-activity relationships for multiple chemotypes, and a wealth of biochemical activity data. All compound designs (>18,000 designs), crystallographic data (>490 ligand-bound x-ray structures), assay data (>10,000 measurements), and synthesized molecules (>2400 compounds) for this campaign were shared rapidly and openly, creating a rich, open, and intellectual property-free knowledge base for future anticoronavirus drug discovery.

Synthesis of bicyclic ethers by a palladium-catalyzed oxidative cyclization-redox relay-π-allyl-Pd cyclization cascade reaction.

Bicyclic ether scaffolds are found in a variety of natural products and are of interest in probe and drug discovery. A palladium-catalyzed cascade reaction has been developed to provide efficient access to these scaffolds from readily available linear diene-diol substrates. A Pd redox-relay process is used strategically to transmit reactivity between an initial oxypalladative cyclization and a subsequent π-allyl-Pd cyclization at remote sites. The reaction affords a variety of bicyclic ether scaffolds with complete diastereoselectivity for cis-ring fusion.

O6-alkylguanine postlesion DNA synthesis is correct with the right complement of hydrogen bonding.

The ability of a DNA polymerase to replicate DNA beyond a mismatch containing a DNA lesion during postlesion DNA synthesis (PLS) can be a contributing factor to mutagenesis. In this study, we investigate the ability of Dpo4, a Y-family DNA polymerase from Sulfolobus solfataricus, to perform PLS beyond the pro-mutagenic DNA adducts O(6)-benzylguanine (O(6)-BnG) and O(6)-methylguanine (O(6)-MeG). Here, O(6)-BnG and O(6)-MeG were paired opposite artificial nucleosides that were structurally altered to systematically test the influence of hydrogen bonding and base pair size and shape on O(6)-alkylguanine PLS. Dpo4-mediated PLS was more efficient past pairs containing Benzi than pairs containing the other artificial nucleoside probes. Based on steady-state kinetic analysis, frequencies of mismatch extension were 7.4 × 10(-3) and 1.5 × 10(-3) for Benzi:O(6)-MeG and Benzi:O(6)-BnG pairs, respectively. Correct extension was observed when O(6)-BnG and O(6)-MeG were paired opposite the smaller nucleoside probes Benzi and BIM; conversely, Dpo4 did not extend past the larger nucleoside probes, Peri and Per, placed opposite O(6)-BnG and O(6)-MeG. Interestingly, Benzi was extended with high fidelity by Dpo4 when it was paired opposite O(6)-BnG and O(6)-MeG but not opposite G. These results indicate that hydrogen bonding is an important noncovalent interaction that influences the fidelity and efficiency of Dpo4 to perform high-fidelity O(6)-alkylguanine PLS.