I provide updates on activity-based DEL technology, which interfaces solid-phase “OBOC” libraries with HTS-style activity assay. Our recent efforts focus on three thematic areas: 1) the discovery of translation modulators as a general strategy for interrogating the proteome via one universal biochemical activity assay, 2) development of pharmacokinetic assays for analyzing “beyond Rule of 5” libraries, and 3) novel 3D tissue culture strategies to enable cell-based DEL screening.

Phosphodiesterases (PDEs) play critical roles in cellular signaling, making them desirable therapeutic targets. Finding isoform-specific inhibitors can be essential to limiting adverse side effects in drug discovery. DEL platforms allow for simultaneous interrogation of numerous conditions in parallel making it an ideal platform to find selectivity. This presentation will highlight how we leveraged the Valo DEL platform to find potent and selective inhibitors for a PDE program.

At Anagenex, we are successfully developing methods via both machine learning and in-lab experimental design. Herein, we will discuss how our new methods have enabled us to leverage both active and inactive molecules towards the drug discovery pipeline as well as establish binding compounds for a wide range of protein classes including DNA-binding proteins.

Targeted protein degradation has emerged as a promising new modality for addressing previously undruggable targets. DNA-Encoded Libraries offer significant and specific advantages for identifying ligands for bifunctional degraders. Nurix has designed and deployed an integrated DEL discovery pipeline to harness novel ligases and degrade both traditional and challenging targets.

Molecular glues are an established modality for therapeutic intervention. There is high interest in identifying putative molecular glues to modulate protein-protein interactions involved in protein degradation, cellular localization, or activity. DNA-encoded libraries allow for the screening of billions of small molecules as potential molecular glues. These molecular glues are identified via binding enrichment in the presence of both proteins over-enrichment in selections with a single protein.

We have designed and synthesized a series of covalent libraries for the identification of the covalent inhibitors. We describe three different proteins for which we identified inhibitors by heterocycle scaffold-based DELs (https://pubs.acs.org/doi/abs/10.1021/acsmedchemlett.2c00127). 

This talk will highlight the discovery efforts from mRNA display screening hits to a tricyclic peptide PCSK9 inhibitor drug candidate, which demonstrated its pharmacodynamic effects similar to the FDA-approved, parenterally dosed anti-PCSK9 mAb, with the advantage of oral administration using lipidic dosing vehicle Labrasol.

I will discuss my lab's research which touches on macrocycles (DELs) and a strategy we patented for creating super pure DELs which should work out much better in DEL selections by improving signal:noise.

We design libraries of macrocycles aiming at generating better cell membrane permeability and apply to the RaPID selection against protein target of interest.

An approach to estimating data noise level has been developed by including replicate samples in DEL selections. The noise level is seen to be dependent on sequencing depth and specific selection conditions. The estimated noise cutoff can be used to remove compounds with low sequence reads for DEL hit triage, which greatly reduces (>100-fold) challenges encountered in DEL data analysis without impacting interpretation of the results.

DEL screening has typically classified enriched compounds as binder/not binder, with a highly variable confirmation rate after off-DNA synthesis. A better understanding of the impacts of selection variables on the output of DEL could improve off-DNA confirmation rates and provide more useful datasets for ML. Here, tool compounds of known different affinities are used to validate selection techniques to improve the quantitative prediction of on-DNA affinity from DEL selections.

Automated structure-activity relationship (SAR) analysis has historically relied on clustering, Quantitative-SAR models, or machine learning to identify potent medicines from high-throughput screening assays. We propose a new straightforward scaffold-based approach for identifying enriched chemical series. Our technique picks chemotypes by breaking down hits into a network of scaffold and scaffold fragments, then uses rank-choice voting to select which scaffolds best represent their exemplars and confer enrichment.