Revealing the anti-tumor mechanisms of aromatic oil from Amomum villosum through integrated network pharmacology, bioinformatics, machine learning, single-cell sequencing, and cell experiments.

The dry fruits of Amomum villosum (Av) are a traditional Chinese medicine used for gastrointestinal disease. Aromatic oil has been reported to have anti-tumor properties. However, its therapeutic potential and molecular mechanisms remain unclear. Integrated approaches of response surface design, gas chromatography-mass spectrum-driven network pharmacology analysis, transcriptomics, machine learning, clinic correlation analysis, single-cell sequencing, molecular docking, and dynamics simulation to characterize bioactive compounds in Av and further uncover their potential mechanisms against gastric cancer. The extraction process of aromatic oil was optimized using Box-Behnken design, and the optimal yield of 2.85 % was achieved under the conditions of extraction time of 6.5 h, solution ratio of 16 mL/g, and mesh size of 30. Subsequently, 28 compounds were identified from aromatic oils using the NIST 14 database. Survival curves indicated that the intersection of compound targets and cancer targets was highly correlated with gastric cancer (GC). Machine learning algorithms screened out HMOX1 as the key target from the public dataset and validated the clinic relevance. Single-cell sequencing indicated that HMOX1 might be a crucial immune regulator for myeloid cells. In vitro cell experiments showed that both aromatic oil and its primary compound, bornyl acetate, significantly suppressed HGC27 and AGS cell proliferation and migration, while downregulating HMOX1; bornyl acetate exhibited superior efficacy. Finally, molecular docking and dynamics showed that bornyl acetate had good binding affinity with HMOX1. Aromatic oil and bornyl acetate exert anti-gastric cancer effects via HMOX1, which provides a molecular basis for developing Av-derived therapy in GC therapy.