The conventional approach to drug discovery relies on the assumption that defects in the activity of a single protein are primarily responsible for the disease pathology; the end goal is to correct this activity, typically through small molecules that bind to a structured pocket of the protein, to change its structure and associated function.

Condensates are membraneless organelles that compartmentalize and concentrate communities of biomolecules (proteins and nucleic acids) and the respective biochemical processes they are involved in. They often function as integrating nodes for multiple biochemical processes, playing the role of regulator of multicomponent biological systems. We now understand that many conventional targets that, due to their lack of three-dimensional structure or high dynamic properties, have been considered ‘undruggable’, concentrate and perform their key functions inside condensates.

Reframing the definition of a target from a single protein to a condensate opens new opportunities to leverage druggable features that are not accessible for an individual biomolecule, such as the emergent properties of the community, and context-dependent conformations that are adopted preferentially inside the condensate. This new perspective on drug discovery also refocuses the end goal from correcting the activity of a single protein to correcting the activity of a complex biological system. Learn more about this topic here.

Dewpoint developed an AI-powered, end-to-end platform that is broadly applicable to the identification of new targets and discovery of condensate modifying drugs (c-mods). Specific insights about disease biology are used to select the most relevant model systems and disease-specific markers when applying the process for a specific indication. Our drug discovery approach is targeted to the biological pathway driving pathology, but agnostic to a unique biomolecule target and allows for identification of hit compounds with a wide variety of mechanisms of action. The platform can be applied to identify broad-acting drugs by selecting condensate markers that are shared between and associated with pathophysiology in multiple indications.

Because condensates serve as central nodes of dysfunction in disease, some c-mods exhibit efficacy across multiple indications. For example, our first development candidate c-mod, targeting aberrant beta catenin transcriptional condensates, is broadly active across a range of diverse Wnt-associated cancers. Learn more about the mechanism of action here.

Similarly, a c-mod developed for the ALS indication, shows activity in mouse models of ALS as well as traumatic brain injury, which share a TDP-43 condensatopathy.

Dewpoint uses condensate phenotypes as readouts to identify and optimize condensate modulators (c-mods). As with any phenotypic drug discovery approach, the elucidation of the mechanism of action is an important step. We efficiently optimize c-mods by combining distinctive assay stacks, a high-throughput DMTA cycle, and AI/ML capabilities.

We employ a combination of traditional target engagement approaches (e.g., CRISPR, CETSA, chemical pulldown, proteomics, transcriptomics, functional assays, etc.) and condensate specific methods including advanced, proprietary AI/ML and knowledge graph approaches to discover, validate and/or prioritize hypotheses. Efforts to characterize how the drug works are designed and incorporated along the drug discovery value chain.

Our target discovery approach uniquely focuses on identifying condensate targets that encode disease states. Using our omics and knowledge-graph based platform, we nominate condensate targets at scale.

We employ a holistic approach that integrates multiomics, bioinformatics, AI/ML and experimental phenotypic and functional assays. Using proprietary knowledge graphs that integrate disease, mechanistic and sequence information available in public databases, along with proprietary experimental data, we identify and prioritize condensate-associated biomolecules (e.g., proteins, long-noncoding RNA) likely to be driving disease and define the condensate hypothesis. Validation of the condensate hypothesis is performed by quantifying the correlation between the aberrant condensate phenotype in healthy vs disease cellular models and/or tissue, relative to validated biomarkers associated with the disease indication.

Dewpoint’s library is focused on drug-like, non-reactive compounds and is built to balance chemical diversity and ability to evolve SAR.