Unveiling the crucial role of glycosylation modification in lung adenocarcinoma metastasis through artificial neural network-based spatial multi-omics single-cell analysis and Mendelian randomization

Abstract

Analysis at the single-cell level uncovered marked glycosylation modification expressions in metastatic tissues of lung adenocarcinoma. Moreover, tissues of lung adenocarcinoma with elevated expression of genes associated with glycosylation displayed enhanced differentiation and activation across signaling pathways including TGF-β, oxidative stress, and WNT. Through spatial transcriptomics, zones of intense glycosylation modification were pinpointed within tumor nests and proximate to tumor-associated blood vessels. An artificial neural network-derived prognostic model demonstrated outstanding predictive capability, with AUC scores achieving 0.84, 0.83, and 0.89 for 1, 3, and 5-year forecasts, respectively. The group identified as high-risk was characterized by pronounced immunosuppression and diminished responsiveness to immunotherapy. Mendelian randomization analysis pinpointed GLANT2 (OR = 1.3654, p < 0.05) and GYS1 (OR = 1.2668, p < 0.05) as genes contributing to the pathogenesis of lung adenocarcinoma. Cell assays have reaffirmed that the inhibition of GYS1 significantly reduces proliferation and invasion in lung adenocarcinoma cell lines, while also decreasing glycogen storage and the formation of glycosylation end products, indicating suppression of glycosylation processes. These findings identify GYS1 as a prospective glycosylation-linked biological target for lung adenocarcinoma therapy.