Integrative single-cell and bulk transcriptomics uncovers a lipid metabolism-related lncRNA signature alongside the ELFN1-AS1/LYPLA1 axis driving osteosarcoma progression

Abstract

Background Osteosarcoma is the most prevalent primary malignant bone tumor in adolescents, featuring prominent intratumoral heterogeneity, early pulmonary metastasis and stagnant survival improvement for metastatic/chemoresistant patients. Altered lipid metabolism represents as a core adaptive hallmark of malignancy, whereas the prognostic implication and molecular regulatory functions of lipid metabolism-related long non-coding RNAs (LMRLs) remain poorly elucidated in osteosarcoma. Methods Single-cell RNA-seq data from GSE152048 and bulk transcriptomic and clinical data from TARGET-OS were integrated. Lipid metabolic activity, malignant cell-state transitions, and intercellular communication were evaluated using AUCell, Monocle, and CellChat, respectively. Lipid metabolism-related genes were identified by integrating single-cell differential expression analysis with WGCNA. A prognostic lncRNA signature was constructed using Pearson correlation, Cox regression, and LASSO analyses. Immune infiltration, tumor mutational burden, immune checkpoint expression, and drug sensitivity were further assessed. scTenifoldKnk-based virtual knockout screening was performed to predict downstream target genes of ELFN1-AS1, followed by comprehensive in vitro and in vivo functional validation of the ELFN1-AS1/LYPLA1 regulatory axis. Results Single-cell analysis revealed marked lipid-metabolic heterogeneity across osteosarcoma cell populations, with dynamic remodeling along malignant cell-state transitions and enhanced communication between high-lipid-metabolism osteosarcoma cells and stromal or vascular compartments. By integrating single-cell and bulk transcriptomic evidence, 22 lipid metabolism-related candidate genes were identified, from which a five-LMRL signature comprising AL133410.1, AL596247.1, ELFN1-AS1, IL10RB-DT, and NECTIN3-AS1 was developed. This signature effectively stratified patients into prognostically distinct risk groups and remained an independent predictor of overall survival. Exploratory computational analyses indicated risk subgroups exhibited divergent immune landscapes, mutational profiles and predicted drug responsiveness. Virtual knockout screening prioritized LYPLA1 as the key downstream effector of ELFN1-AS1. Among the signature lncRNAs, ELFN1-AS1 was prominently upregulated in osteosarcoma tissues and cells and was associated with poor prognosis. Mechanistically, ELFN1-AS1 drives osteosarcoma progression by triggering LYPLA1-dependent lipid accumulation-associated phenotype. Pharmacological blockade of de-novo lipogenesis using orlistat (selective FASN inhibitor) attenuated ELFN1-AS1-driven oncogenic phenotypes. Conclusion This study establishes an LMRL-based for osteosarcoma prognostic stratification that reflects immune, genomic and therapeutic heterogeneity. Combining virtual knockout screening and experimental validation, we characterize the oncogenic ELFN1-AS1/LYPLA1 regulatory cascade as a potential biomarker and therapeutic target for precision management of osteosarcoma.