Machine Learning‐Based Screening of Optimal miRNA Targets Enabled by a Cytosolic Delivery Carrier for Augmenting Cancer Therapy
Hang Xu, Pengxiang Min, Yanbin Fu, Ge Sun, Hui Wu, Mengqiao Pan, Chaoxian Zhao, Jiatong Li, Ying Wu, Ying Xu, Bin Gu, Jiahua Pan, Lin Zhang, Peifeng LiuABSTRACT
Nucleic acid therapeutics offer promise for precision cancer treatment, but are limited by inaccurate target selection and inefficient delivery. This work presents a therapeutic platform combining dual machine learning models (LASSO and RSF)‐based miRNA target screening with a modular biomimetic nanocarrier system. Systematic screening and experimental validation identify miR‐873 as a previously uncharacterized tumor‐suppressive miRNA in prostate cancer, and miR‐20a as oncogenic. Mechanistic studies reveal that they regulate HNRNPK and ABL2, modulating pathways involved in proliferation and metastasis. Then, the biomimetic nanocarrier, engineered with prostate cancer cell‐derived endoplasmic reticulum membrane (ERM), is successfully developed for cytosolic co‐delivery of miR‐873 mimic and miR‐20a inhibitor in prostate cancer, achieving significant and synergistic suppression of tumor proliferation and metastasis both in vitro and in vivo. Besides, co‐loading with docetaxel further enhances antitumor effects, underscoring its potential for combination therapies to inhibit cancer progression. These findings highlight both the therapeutic synergy of rationally selected miRNA combinations and the platform's versatility for precision cancer treatment.