Related Experiment Video
Updated: Aug 12, 2026
In Vitro Model for Studying Differentiation and Changes of Multi-Omics on Murine Airway Epithelial Cells Stimulated with Cigarette Smoke Extract
Published on: July 12, 2024
Integrative machine learning and single-cell analysis identifies nicotine-related diagnostic genes and myeloid
Wei Zhang1, Hongmei Sheng1, Yueyue Shi2
1Department of Respiratory and Critical Care Medicine, Chest Hospital, Tianjin University, Tianjin, China.
Background:
Cigarette smoke is a major risk factor for chronic obstructive pulmonary disease (COPD), but its toxicity results from multiple harmful constituents. Nicotine is a chemically defined and biologically active tobacco-related compound, yet its COPD-related molecular features remain unclear. This study aimed to identify nicotine-related transcriptomic signatures in COPD and explore their immune and single-cell relevance.
Methods:
Nicotine-related targets were collected from PubChem, ChEMBL, and SwissTargetPrediction. COPD transcriptomic data from GSE57148 were used to identify differentially expressed genes, which were intersected with nicotine-related targets to obtain candidate genes. CIBERSORT was applied to evaluate immune infiltration. LASSO regression and multiple machine learning algorithms were used to construct diagnostic models. Molecular subtypes were defined using consensus clustering based on SVM-derived important genes, followed by GSVA and GSEA. Single-cell RNA-seq data from GSE173896 were analyzed to characterize cell composition, gene set activity, cell-cell communication, and myeloid-cell trajectories. Molecular docking, scTenifoldKnk-based virtual knockout, and RT-qPCR validation were further performed.
Results:
A total of 132 nicotine-related target genes were obtained. Nicotine-related differentially expressed genes showed distinct expression patterns between COPD and control samples and were associated with immune cell infiltration. Machine learning analysis identified an SVM model with the best diagnostic performance, achieving an AUC of 0.923. Five key genes, including CHKA, EBP, RPS6KA5, NCSTN, and CHRM3, were ranked as the most important features. COPD samples were further divided into two molecular subtypes with different immune infiltration and pathway activity. Single-cell analysis showed that nicotine-related diagnostic gene activity was enriched mainly in epithelial, stromal, endothelial, and myeloid-lineage cells. Myeloid-cell reclustering revealed inflammatory, antigen-presenting, and lipid-metabolic states. Pseudotime analysis linked SVM-prioritized genes to myeloid-cell state transitions. Molecular docking suggested potential interactions between nicotine and key target proteins. RT-qPCR confirmed decreased expression of CHKA and EBP, and increased expression of CHRM3, NCSTN, and RPS6KA5 in COPD samples.
Conclusion:
This study identified a nicotine-related diagnostic gene panel associated with immune infiltration, molecular subtypes, and myeloid-cell remodeling in COPD. These findings provide insight into nicotine-related molecular alterations, while indicating that nicotine-related signals represent only one component of cigarette smoke-associated COPD biology.
