Focusing on building blocks and properties is the best way to design better DNA-Encoded Libraries (DELs)

By Matt Clark, Ph.D., CEO

It seems to be a common experience. Your new drug discovery project has accessed a DEL screen and you’re eagerly anticipating the results. The hit list arrives and you excitedly peruse the active compounds. Your heart sinks as your eyes scan a long list of peptides and peptidomimetics, and other compounds with high molecular weights and lipophilicities. You turn to your colleague for advice. “Well don’t you know, that’s what you get with DEL,” they say.

Your colleague voiced a common misconception in the DEL space: that screening ultra-large libraries requires lowering one’s standards for a high quality hit. It’s an understandable perspective, given what’s out there in the literature. Many groups continue to publish DEL-derived compounds that, while interesting from an academic perspective, are poorly suited to drug discovery.

I’m happy to report that DEL screening doesn’t need to lead to difficult-to-optimize starting points. With proper library design, highly attractive and developable compounds can arise from DEL. What is the most important factor in library design? It is to minimize the proportion of the structure that is invariant and maximize the proportion that comes from the building blocks.

Guidelines around physicochemical properties create an atomic budget

Lipinski Rules and related guidelines tell us that drug development is more difficult when our compounds are too big and too lipophilic. In effect these rules set “atom budgets” for hits, leads, and drugs. At X-Chem we’ve realized that accessing the incredible diversity and complexity of modern building block classes puts us at odds with that atom budget. Diverse and complex building blocks tend to have higher molecular weights, so that once we’ve combined even two of them together, it doesn’t leave us much more room in the “budget”. In order to accommodate these larger building blocks, we need to minimize the non-building block components of the library, such as scaffolds or cores. We find four or five atoms to be an ambitious but attainable goal.

What does this look like in practice?

As an example, we were able to develop methods for preparation of aminooxadiazole libraries, created from primary amines and either aldehydes or carboxylates. We used vast and diverse sets of building blocks in these libraries, which necessarily carried high molecular weights. But because these libraries only contain four non-building block atoms, they still produce very attractive compounds even though the building blocks are large. The average MW for these libraries is under 400, yet they still contain over 7.5 million structures.  These libraries have resulted in hits to many targets, from kinases to E3 ligases.

We continue to apply this logic to other ring systems, including aliphatic azacycles. While the synthetic chemistry in this case is more difficult, the payoff is a library with similarly attractive MW and cLogP profiles and a better Fsp³ profile.

What does this mean for the future?

At X-Chem we put as much effort in developing novel reagents for DEL creation as we do on new reaction development.  While new reactions are beneficial to generating on-DNA chemical diversity, novel reagents allow us to expand the set of cores we can construct. We have been focusing on small, densely functionalized reagents that generate interesting scaffolds when exposed to diverse building blocks. We take our inspiration from highly designed reagents like Danishefsky’s diene or TosMIC, that exploit mechanistic understanding to trigger useful reaction cascades, forming attractive products. If these reagents only leave a residue of 4-5 atoms, then we can fully leverage available building blocks without worrying about property inflation.

For more information, see our opinion piece in Bioorganic & Medicinal Chemistry which presents the technical data from some of X-Chem’s building block-guided DEL designs.

Back to Blog