SRPK1 and SEM1 as Key Polyamine Metabolism-Related Genes in Non-Small Cell Lung Cancer: Integrated Bioinformatics Analysis, Mendelian Randomization, and In Vitro Validation

Background Non-small cell lung cancer (NSCLC) is the most common type of lung cancer, and studies have shown that polyamine metabolism plays a significant role in its pathogenesis. However, the molecular mechanisms remain unclear. This study integrated bulk transcriptomic profiling, Mendelian randomization (MR), and in vitro validation to identify polyamine metabolism-associated key genes in NSCLC and to characterize their diagnostic, prognostic, and functional relevance. Methods NSCLC transcriptomic data from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) were analyzed. Differentially expressed genes (DEGs) and weighted gene co-expression network analysis (WGCNA) hub genes were intersected with polyamine metabolism genes to define candidate genes. After enrichment analyses, four machine-learning algorithms (XGBoost, Random Forest, LASSO, SVM-RFE) identified key genes, whose diagnostic and prognostic value was evaluated by receiver operating characteristic (ROC) curves and Kaplan-Meier analyses. Immune infiltration (CIBERSORTx), competing endogenous RNA (ceRNA) and transcription factor networks, Connectivity Map (CMap) drug prediction, molecular docking, and two-sample MR were then performed. Finally, qRT-PCR, Western blot, CCK-8, wound healing, and flow cytometric assays in A549, H1299, and BEAS-2B cells validated key gene expression and assessed SRPK1 knockdown effects on cell behavior and on ODC1 and SAT1. Results 49 candidate genes were obtained, enriched in cell cycle, transcriptional misregulation in cancer, and p53 signaling pathways. Machine learning converged on SRPK1 and SEM1 as key genes, both significantly upregulated in NSCLC and associated with poor survival. Key genes correlated with CD4 memory resting T cells, eosinophils, and M1 macrophages. A ceRNA network centered on KCNQ1OT1/hsa-miR-106a-5p/SRPK1 was constructed, and IRF2 was identified as a common upstream transcription factor. Ten candidate drugs were identified, including chelerythrine and diosmin. MR analysis confirmed positive causal associations between key gene expression and NSCLC risk. In vitro, SRPK1 and SEM1 were upregulated in A549 and H1299 cells, and SRPK1 silencing suppressed proliferation and migration, promoted apoptosis, downregulated ODC1, and upregulated SAT1. Conclusion This integrated study identifies SRPK1 and SEM1 as polyamine metabolism-related key genes in NSCLC, with SRPK1 mechanistically linked to ODC1/SAT1-mediated polyamine homeostasis, providing promising biomarkers and therapeutic targets for NSCLC.