Single-Cell Transcriptomic Profiling Reveals the Heterogeneity of Epithelial Cells in Lung Adenocarcinoma Lymph Node Metastasis and Develops a Prognostic Signature

QiuQiao Mu¹Han Zhang²Ying Shi³Mengli Xue¹Jingxian Wang¹Yun Ding⁴Tan Lin⁵Hui Yuan⁶Xin Li^3*Daqiang Sun^3*

• ¹Tianjin Medical University, Tianjin, China
• ²Tianjin University, Tianjin, China
• ³Tianjin Chest Hospital, Tianjin, China
• ⁴Fujian Provincial Hospital, Fuzhou, China
• ⁵Qingdao Municipal Hospital Group, Qingdao, China
• ⁶Haihe Laboratory of Cell Ecosystem, Tianjin, China

Background: Lymph node metastasis markedly worsens prognosis in lung adenocarcinoma (LUAD); however, the evolutionary dynamics and regulatory mechanisms underlying the heterogeneity of malignant epithelial cells during this process remain poorly understood and warrant comprehensive investigation.We performed a comprehensive single-cell transcriptomic analysis of epithelial cells from 18 samples comprising normal lung tissue and lymph node metastases. Malignant epithelial cells were identified via inferred copy number variation (CNV) profiles. Key malignant subpopulations were further characterized through trajectory inference, cell-cell communication mapping, gene set variation analysis (GSVA), and reconstruction of transcription factor regulatory networks. To assess clinical relevance, we developed and validated a prognostic model-termed the EAS score-based on the transcriptional signatures of malignant epithelial subsets, using integrated data from multiple TCGA and GEO cohorts. The functional role of the hub gene SELENBP1 was experimentally validated through quantitative PCR (qPCR), Western blotting, immunohistochemistry (IHC), Transwell migration assays, colony formation assays, flow cytometry, ROS quantification, and subcutaneous tumorigenesis assays in vivo.Results: Single-cell transcriptomic analysis identified four distinct malignant epithelial subtypes (Clusters 0-3), each characterized by unique patterns of CNV.Leveraging these defined cellular subpopulations, we constructed a highly accurate 4 model for prognostication in LUAD, enabling reliable classification of patients based on clinical outcomes. Through detailed comparisons between groups with divergent prognostic risks, the study revealed notable differences across the tumor microenvironment (TME), including alterations in pathway activity, gene enrichment distributions, mutation profiles, and anticipated responses to immune checkpoint blockade. In addition, functional validation experiments confirmed that SELENBP1 plays a tumor-suppressive role, further supporting its relevance as a potential intervention target in LUAD.This research provides insights into the evolutionary complexity and heterogeneity of malignant epithelial populations in lymph node metastatic sites of LUAD. It also presents a scoring system based on prognostic indicators, which serves as a reliable tool for forecasting patient survival outcomes. Moreover, the discovery of SELENBP1 as a candidate tumor suppressor emphasizes its importance in guiding both clinical risk categorization and the design of personalized treatment strategies for individuals classified as high-risk LUAD cases.