The intratumoral microbiome plays a crucial role in cancer progression, prompting the development of therapies targeting it. However, due to the heterogeneous effects of intratumoral microbes, designing treatments tailored to the unique microecological characteristics of individual tumors poses a significant challenge. Here, we found significant variations in the abundance of five bacterial genera—Lysinibacillus, Stenotrophomonas, Weissella, Comamonas, and Aeromonas—between lung adenocarcinoma (LUAD) and normal tissues by analyzing single-cell transcriptomic datasets. These specific bacterial clusters were significantly associated with immune infiltrates in the tumor microenvironment (TME). After confirming their effects in mouse models, these bacterial taxa were identified as potential therapeutic targets. Through in vitro drug screening assays, berberine was identified as a promising agent that selectively inhibits harmful bacteria while sparing beneficial ones. To address berberine’s low solubility and tumor targeting issues, it was encapsulated into tumor cell-derived extracellular vesicles (EV-ber). Feature analysis demonstrated that EV-ber shifted the intratumoral microbiome profile toward an anti-tumor phenotype and enhanced anti-tumor immunity in the TME. Furthermore, EV-ber administration inhibited LUAD growth, impaired LUAD metastatic ability, and boosted the effectiveness of anti-PD-L1 immunotherapy in mouse models. In conclusion, this work demonstrates the potential of personalized intratumoral microbial re-education strategies in LUAD therapy.

Intratumoral microbiome  /  Personalized medicine  /  Antibiotics  /  Tumor microenvironment  /  Tumor cell-derived extracellular vesicles  /  Berberine  /  Lung cancer  /  Immunotherapy