Abstract PR004: Targeting aberrant condensate formation in fusion-positive cancers with an integrated discovery platform

Abstract Fusion oncogenes drive malignant transformation by altering transcriptional programs, disrupting cellular homeostasis, and frequently rewiring protein condensate networks. Emerging evidence suggests that aberrant condensate formation is a key mechanism through which these fusion proteins exert oncogenic control. Specifically, by amplifying transcriptional output, sequestering cofactors, and enforcing pathological gene expression states. Despite their genetic validation as oncogenic drivers, many fusion proteins remain pharmacologically intractable due to the absence of catalytic domains or well-defined binding pockets. To overcome this, we developed an integrated discovery platform that directly assesses the biophysical and functional properties of fusion protein-driven condensates and enables identification of small molecules that disrupt these oncogenic assemblies. Using protein language models and in silico predictors of phase separation, we systematically mapped condensate-forming potential across recurrent oncogenic fusions catalogued in TCGA. This analysis revealed that most fusion proteins (including EML4-ALK and EWS-FLI1) harbor extensive intrinsically disordered regions (IDRs) that promote abnormal condensation and transcriptional dysregulation. We further identified disease-specific variants that alter condensate formation relative to wild-type proteins, providing mechanistic insight into oncogenic gain- and loss-of-function events. To experimentally validate and therapeutically exploit these findings, we established PhaseScan™, a high-throughput droplet microfluidics screening platform capable of quantitatively profiling condensate dynamics and screening thousands of compounds against reconstituted fusion condensates to identify therapeutically relevant small molecule inhibitors. Across multiple fusion proteins, PhaseScan™ identified small molecules that selectively disrupt aberrant condensates while sparing physiological assemblies such as nuclear speckles and stress granules. Using EML4-ALK as a representative model, we demonstrated that compound-mediated condensate dissolution disrupts oncogenic ALK signaling and reverses oncogenic phenotypes in cell-based systems. Our results highlight condensate biology as a generalizable vulnerability across fusion-driven cancers and demonstrate the feasibility of discovering small molecules against fusion proteins traditionally considered undruggable. By integrating AI-driven predictive modeling, and on-target molecular high-throughput screening, our platform provides a roadmap for rationally identifying and drugging fusion protein condensates. This approach establishes condensate-targeted therapeutics as a novel class of precision medicines for patients with fusion-positive malignancies. Citation Format: Andrew Seeber, William E. Arter, Seema Qamar, Cinzia Sgambato, Amal Alex, Kadi Saar, Julia Doh, Martin Kulander, Tuomas Knowles, Shilpi Arora. Targeting aberrant condensate formation in fusion-positive cancers with an integrated discovery platform [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Fusion-Positive Cancer: From Discovery to Therapy; 2026 Jan 13-15; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86(1_Suppl):Abstract nr PR004.