Network-based discovery of tumor-checkpoint inverter drugs targeting pancreatic ductal adenocarcinoma cell states and macrophage reprogramming

Pancreatic ductal adenocarcinoma (PDAC) remains one of the most lethal malignancies, driven by extreme tumor heterogeneity and a profoundly immunosuppressive tumor microenvironment (TME). Distinct PDAC cell states-Gastrointestinal-like (GLS), Morphogenic (MOS), and Primitive (PLS)-coexist within individual tumors and are further stratified by MAPK activity (M⁺/M⁻), reflecting dynamic transcriptional programs sustained by Master Regulator (MR) proteins. These cell states are hypothesized to differentially modulate the tumor-associated macrophages in the TME. To investigate this, we established a co-culture system of THP-1-derived macrophages with PDAC cell lines representing each state and profiled macrophage transcriptional reprogramming. Macrophages co-cultured with distinct PDAC states exhibited differential activation of M2-like and TREM2⁺/APOE⁺/C1Q⁺ immunosuppressive phenotype, suggesting that PDAC cell states may uniquely influence macrophage phenotypes and immune evasion.To identify compounds capable of reprogramming these malignant states, we applied a network-based systems biology framework integrating ARACNe and VIPER to infer MR activity across PDAC states, OncoMatch to identify representative cell line models, and OncoTreat to predict small molecules capable of inverting tumor checkpoint-module activity. Cross-model validation identified state-specific candidate drugs, including Leuprolide, Vinblastine, and Mercaptopurine for GLS; Vindesine, Gossypol, and Binimetinib for MOS; and AT9283, Crizotinib, and Afatinib for PLS. Predicted MR-inversion scores correlated with experimental dose-response profiles in cell lines selected via OncoMatch.Together, these results establish a mechanistic link between tumor-intrinsic transcriptional states and macrophage immunosuppression, while identifying mutation-agnostic, state-specific drugs capable of reprogramming both tumor and immune compartments. This work provides a generalizable framework for network-based drug repurposing to overcome transcriptional plasticity and immune resistance in PDAC.