There is considerable interest in the discovery of novel molecular glues that induce proximity between oncology targets and their modulators. Induced proximity can be used to degrade proteins by a range of mechanisms, including ubiquitination and trafficking to the proteasome. DNA-Encoded Chemical Library screening can readily be adapted to the discovery of glues, and we will describe case studies that exemplify how carefully designed screening campaigns with billions of encoded compounds can find rare examples of compounds able to induce proximity. Case studies will include a small molecule discovered in a DEL screen against ATAD2 that is able to inappropriately dimerize ATAD2 thereby reducing its ability to bind to acetylated histones and an activator of the integrated stress response pathway that only binds to the eIF2b complex, not its isolated components, thereby stabilizing it.

DNA-encoded chemical library (DEL) technology has become a powerful tool for hit identification, enabling the efficient screening of large collections of encoded compounds with advantages over traditional high-throughput screening in terms of library size, cost, and resource efficiency. The high chemical diversity and encoding fidelity of DELs yields extensive data that can support computational approaches such as machine learning (ML), 3D pharmacophore generation and structure-based discovery. This DEL-chemomics approach has been applied to the discovery of novel E3 ligase receptor ligands, expanding the repertoire of E3 ligands beyond the traditional CRBN and VHL for degradation of drug targets. A case study of the discovery of DCAF1 ligand and its application in PROTACs that targeted WDR5 for degradation will be presented.

DNA-encoded library (DEL) screening with encoded combinatorial chemical synthesis has emerged as a powerful approach to discover novel ligands for targets of interest. Here, we describe a complementary “digitization” workflow that encodes pre-synthesized (non-combinatorial) compounds. In this approach, a bifunctional linker is photoactivated to generate a carbene that directly inserts into the putative ligand, followed by a [3+2] cycloaddition to conjugate a unique DNA barcode. The resulting DNA-tagged compounds are pooled and screened in parallel against protein targets, and active ligands are enriched.

DNA-encoded library (DEL) screening has historically been used to identify hits from libraries of hundreds of billions of diverse molecules. We have developed novel methods to access and leverage the DEL screens into actionable SAR. Thompson sampling is method that has been re-implemented in our hands to harvest DEL data, both positive and negative. We have also developed novel pharmacophore methods that summarize the salient features of thousands of chemically and mechanistically related DEL compounds to allow for their broader use in Med Chem design. The combined access between the two methods produces data that facilitates not just hit identification, but into datasets that facilitate rapid acceleration of Medicinal Chemistry efforts.

Macrocyclic DNA-encoded libraries (DELs) offer a powerful approach for discovering novel therapeutics, especially against challenging biological targets. Previously, we identified potent, selective macrocyclic inhibitors of Bcl-2 family proteins, optimized to low-nanomolar potency with favorable drug-like properties. Building on this success, our next-generation macrocyclic DEL features diverse ring sizes, rich stereochemistry, and proprietary chemistries, enabling efficient exploration of complex targets and accelerating the path to high-quality leads.

Covalent drug discovery has reshaped pharma, expanding targetable residues and warhead diversity. While GSH half-life remains key for intrinsic reactivity, consensus on optimal reactivity and potency metrics is lacking. This evolving field brings new warheads, mechanisms, and kinetics. This poster highlights novel kinetic parameters for GSH reactivity and X-Chem’s approach to de-risk covalent leads by measuring phospholipid/albumin binding and reduction potentials.

DNA-encoded chemical library (DEL) technology enables the efficient screening of large collections of encoded compounds and has emerged as a powerful tool for hit identification. De Novo design and synthesis of non-traditional encoded libraries such as a reversible covalent compound library and beyond Lipinski’s rule of five (bRo5) macrocyclic (MC) libraries have significantly expanded the chemical space available for DEL screening, and further enhances the feasibility of discovering highly selective ligands for challenging oncology targets. We will present examples of affinity-based ligand discovery targeting protein-protein interactions within the Bcl-2 family of proteins, specifically focusing on MCL1 and BFL1. We highlight the multiplexed DEL selection strategy employed and elucidate the molecular details around the uniqueness of the ligand interactions.

Training machine-learned models using DNA-Encoded Chemical Library screening data “DEL-ML” and using these models to rank compounds in virtual catalogs has been successful for hit identification for a range of individual oncology targets including ER alpha and c-KIT [McCloskey et al 2020]. Here we describe the importance of library quality and diversity, protein quality, screen design, appropriate profile choice, training set preparation and evaluation, promiscuity labeling, chemical representations, modeling algorithms and their tuning in the context of a systematic target class-based evaluation for a range of oncology targets including DCAF1, WDR5 and WDR12. This approach yielded first-in-class, drug-like ligands for multiple targets that were characterized using a range of biophysical approaches including x-ray crystallography.

Discovering RNA-binding small molecules is a formidable challenge with immense therapeutic potential but few reported successes. Here we report the discovery of specific small molecule ligands for expansion repeat, splice site, and riboswitch targets via affinity-mediated selection of DNA-encoded libraries (DELs) against immobilized mirror-image RNA. In a model system, we demonstrate that the mirror-image strategy eliminates false enrichment caused by DNA:RNA hybridization (a major issue in previous DEL:RNA studies) and enhances the enrichment of a DNA-tagged known ligand, consistent with the well-precedented assumption that DNA:RNA hybridization only occurs between homochiral strands. For three therapeutically relevant RNA targets, we employed our DEL deck containing more than 100 billion compounds, discovered novel chemical matter, synthesized the mirrored versions of the DEL hits, and confirmed their affinity to natural RNA target in orthogonal biophysical assays. Our method makes RNA as amenable to DEL screens as other target classes and unleashes the unparalleled throughput of DEL for RNA-targeted small molecule drug discovery.

X-Chem’s DNA-encoded libraries (DELs) contain hundreds of billions of molecules spanning diverse chemical space. When screened against biological targets, positive and negative structure-activity relationship (SAR) trends within this chemical space emerge, yielding rich datasets on which machine learning thrives. Here, we report the discovery of first-in-class ligands for WD40 repeat-containing protein 91 (WDR91) by applying machine learning models trained on X-Chem’s DEL screening data to synthetically accessible commercial compound catalogues. Initial predictions resulted in the discovery of a WDR91-selective compound with a K_D of 6±2 μM, machine learning-driven SAR expansion of the primary hits identified an additional 11 compounds with K_D ≤ 25 μM and %R_max > 60%, and structure-based drug design (SBDD) enabled the discovery of 2 covalent analogues with KD ≤ 45 μM and %R_max > 50%, for which covalent adduct formation was confirmed by intact mass liquid chromatography−mass spectrometry.

This poster was presented at the 12^th International Symposium on DNA-encoded chemical libraries