Matthew Clark, CEO of X-Chem, looks at what the future could hold for DNA-encoded library (DEL) technology.
In the 13 years since it was first described, DNA-encoded library (DEL) technology has become a firmly established approach to small molecule hit generation in drug discovery. Most major pharma companies operate DEL platforms, and several service providers offer DEL capabilities to companies that outsource their DEL activities. Reports of successful DEL screening continue to multiply, as do publications regarding improvements to DEL technology, such as better library chemistry and selection methods. Most commonly, DELs are used to supply high-quality starting points to medicinal chemistry projects for a specific biological target. Often, these starting points are inhibitors of enzymes (e.g., protease or kinase inhibitors) or protein-protein interactions (e.g., cytokine-receptor disrupters). These use cases remain valuable applications for DELs, likely constituting the majority of DEL usage across the industry today, and they will continue to do so into the future. But due to its flexibility of screening and unbiased exploration of accessible chemical space, DEL technology is starting to find application in more novel areas of drug discovery — areas that will represent an increasing proportion of DEL screening activity, and drug discovery research generally, in the future.
Targeted protein degradation and related modalities
A new modality where DEL screening has had particular impact is targeted protein degradation (TPD), a modality that has generated rapidly increasing interest in recent years. The TPD approach relies on the use of bifunctional compounds, often called bispecifics or chimeras, to bring together a protein target and an E3 ligase. The E3 ligase, held in proximity to the target protein — referred to as the protein of interest (POI) — transfers ubiquitin onto a lysine of the POI, targeting it for proteosomal degradation. Thus, TPD technology repurposes the cell’s native protein housekeeping function to target proteins of therapeutic relevance.
The chimeras that hold the E3 and POI in proximity must have simultaneous binding interactions with both proteins. In practice, this is achieved by chemically linking ligands for each protein, generating a chimera. In most applications, the E3 ligand is already known, drawing from a growing toolkit of proven E3 engagers published over the years. As for the POI ligand, while it could be a repurposed inhibitor, its functional consequence on the target is secondary and superfluous. Binding alone is necessary and sufficient to induce degradation.
Since DEL screening, as it is commonly practiced, is fundamentally a binding-mediated selection, it is well suited for discovery of POI ligands. Furthermore, DEL hits are discovered while attached to their encoding DNA via a linker. This allows DEL hits to be easily linked to E3 engagers, since the placement of this linker is already informed by the DEL screen. This makes a TPD-DEL workflow quite efficient. DEL hits are resynthesised with reactive linkers, allowing rapid diversification with E3 ligands and variable linker lengths. The resulting array of chimeras is then screened for degradation activity.
DEL screening can also be applied to the E3 side of the chimera. While as many as 600 E3s may be present in the human genome, only a handful have been utilised for TPD. DEL screening offers a rapid approach to discovering ligands to novel E3s. The same benefits of binding-based discovery and linker placement that apply to POI ligand discovery also apply to E3 screening. As protein-protein interaction targets, E3 ligases are a fruitful class for DEL screening.
Furthermore, the TPD area has given rise to other emerging strategies that rely on induced proximity, including RIBOTACs (RNA-degradation), LYTACs (lysosomal degradation), CHAMPs (Hsp90 chaperone) and others. As in the TPD area, these emerging fields will benefit from DEL screening for the rapid discovery of novel and linkable ligands. DEL screening could be a foundational screening platform for a whole host of applications involving induced proximity. As new cellular pathways are repurposed for therapeutic effect through the action of chimeric molecules, DEL screening will stand ready as a source for novel ligands with known tolerated linkage chemistries on both sides of the interaction.
Membrane targets and cellular selection
In most DEL selections, the target is a highly purified, recombinant sample of the POI. While this approach has a long track record of success, it also can be limiting for proteins that are not amenable to purification, freeze-thaw cycles, and storage. Targets like transmembrane proteins, disordered proteins, multiprotein complexes, and others may not be available in a pure, stable format. In these situations, more advanced selection methods may be required.
An early example is GSK’s work with the G-protein coupled receptor (GPCR) NK3 (neurokinin 3 receptor), in which DEL selection was conducted against whole cells overexpressing the target. The extracellular domain of the receptor was accessible to the library in the selection incubation, and several families of high-affinity antagonists were discovered. While subsequent reports of whole-cell selection of GPCRs are scarce, approaches to stabilising purified receptors have come to the forefront. Using mutagenic stabilisation, PAR2 (protease-activated receptor 2) was purified and subjected to DEL selection, giving rise to several families of agonists and antagonists. Taking GPCR screening a step further, X-Chem has demonstrated the DEL selection of ligands for wild-type GPCRs from cell lysates. The success of this approach probably stems partly from the fact that the receptor is in a more native environment during the selection. Any binding partners or cofactors, as well as components of the plasma membrane, remain available to the target.
This concept of making the selection conditions as close as possible to the native situation also extends to cytosolic targets. Taking advantage of the large volumes of individual oocytes, researchers at Vipergen have developed an in-cell selection method. Other researchers have sought to use cross-linking to trap the transient interactions that arise when relatively low (native) concentrations of target protein are available. We believe that whole-cell, cell-lysate and cross-linking-based selections will become increasingly important in the DEL field going forward.
One of the advantages of DEL screening is the volume of data a selection experiment can produce. High-throughput DNA sequencing techniques can generate hundreds of millions of sequences in a single run. When the sequencing sample is a DEL selection output, many millions of chemical data points are generated. Traditionally, these data have been substantially filtered, clustered, and analysed to arrive at a compound list of tractable size. But, with the advent of artificial intelligence, it is now possible to take advantage of the full scope of the selection output to inform the training of predictive models. This unique ability of DEL technology to generate large volumes of data quickly is increasingly appreciated in the AI drug discovery community. The first report on the effectiveness of DEL data to drive machine learning (ML) model generation was published in 2020. Since then, several AI-focused biotechs have invested in DEL platforms, presumably to generate data at a sufficient scale to feed their internal AI engines.
The current application of DEL-derived ML models is for novel hit generation. A target-specific model based on DEL data can help predict new binders from virtual libraries or commercial catalogs, substantially increasing the scope of chemical matter relevant for the target of interest. This increases the likelihood that one of the starting points will have favourable properties for development into a clinical candidate.
Going forward, DEL-derived models can accomplish more than hit identification. Since DEL selection can be run on multiple targets in parallel, one can imagine using DEL to create target class- or pathway-specific “super models,” which could provide a comprehensive view of the chemical space relevant for particular disease pathways. Going further, models could be generated using DEL data across whole swaths of the proteome. Such a collection of models would be a “one-stop shop” for chemical matter for newly validated biological targets. We are approaching a paradigm where we can screen and build models faster than we can identify new targets to treat disease. In such a world, having models in hand for the targets of the future will greatly accelerate the development of new therapeutics. Newly identified targets can immediately be validated by compounds predicted from an encyclopaedia of DEL-derived models.
The DEL field continues to evolve and mature. As new drug modalities and data science technologies emerge, rapid, comprehensive surveying of chemical space will become increasingly important. Ultimately, most therapeutics will continue to rely on molecules interacting with biological systems. It is becoming increasingly clear that DEL technology is the preeminent technique to discover novel ligands capable of such interactions.
- Brown DG, et al. Agonists and Antagonists of Protease-Activated Receptor 2 Discovered within a DNA-Encoded Chemical Library Using Mutational Stabilization of the Target. SLAS Discovery (2018). https://doi.org/10.1177/2472555217749847
- Disch JS, et al. Bispecific Estrogen Receptor α Degraders Incorporating Novel Binders Identified Using DNA-Encoded Chemical Library Screening. J Med Chem (2021). https://doi.org/10.1021/acs.jmedchem.1c00127
- McCloskey K, et al. Machine Learning on DNA-Encoded Libraries: A New Paradigm for Hit Finding. J Med Chem (2020). https://doi.org/10.1021/acs.jmedchem.0c00452
- Petersen LK, et al. Screening of DNA-Encoded Small Molecule Libraries inside a Living Cell. J Am Chem Soc (2021). https://doi.org/10.1021/jacs.0c09213
- Wu Z, et al. Cell-Based Selection Expands the Utility of DNA-Encoded Small-Molecule Library Technology to Cell Surface Drug Targets: Identification of Novel Antagonists of the NK3 Tachykinin Receptor. ACS Comb Sci (2015). https://doi.org/10.1021/acscombsci.5b00124
About the author
Matthew Clark was part of X-Chem’s founding team and served as VP of chemistry and SVP of research prior to his appointment to CEO. Before joining X-Chem, Clark was director of chemistry at GlaxoSmithKline, where he led the group responsible for design and synthesis of early-iteration DELs. He began his professional career at Praecis Pharmaceuticals and received his BSc in Biochemistry from the University of California, San Diego, holds a PhD in Chemistry from Cornell University and conducted postdoctoral studies at the Massachusetts Institute of Technology.