Machine learning-guided discovery of withangulatin A as a novel inducer of megakaryocytic differentiation through direct targeting of TYK2 in leukemia.

BACKGROUND: Leukemia is an aggressive hematological malignancy. Effective differentiation-inducing agents for non-acute promyelocytic leukemia (non-APL), especially those derived from natural products, remain unavailable. Withanolides isolated from Physalis angulata var. villosa possess diverse anticancer activities, yet their potential to induce leukemia cell differentiation has not been systematically investigated. AIM: To identify withanolides capable of inducing megakaryocytic differentiation in leukemia cells and elucidate their underlying mechanisms. METHODS: A library of 113 withanolides was screened using machine learning models. Candidate compounds were evaluated in K562 and HEL leukemia cells via cell viability assays, differentiation analyses, and RNA-seq. A syngeneic leukemia mouse model and primary human non-APL samples were used for biological validation. Target engagement was assessed by molecular docking, SPR, CETSA, and DARTS. RESULTS: Machine learning-guided screening identified withangulatin A (WA) as the most potent candidate. WA significantly inhibited leukemia cell proliferation and induced megakaryocytic differentiation, but not erythroid maturation. Transcriptomic analyses revealed activation of megakaryopoiesis-associated programs and enrichment of platelet activation and PKC signaling pathways. Mechanistically, WA coordinately activated both the JAK2/STAT3 and PKC/MAPK cascades. TYK2 was identified as a direct functional target of WA. Notably, pharmacological inhibition or genetic silencing of TYK2 attenuated WA-induced signaling activation and megakaryocytic differentiation. WA also promoted megakaryocytic differentiation in primary non-APL cells. In vivo, WA alleviated splenomegaly, promoted megakaryocytic differentiation, prolonged survival, and showed no detectable systemic toxicity. CONCLUSION: WA is a novel natural inducer of megakaryocytic differentiation via direct targeting and activation of TYK2, highlighting its potential for differentiation therapy in non-APL leukemia.