An endoplasmic reticulum stress- and Golgi apparatus-related signature reveals immune microenvironment remodeling and MUC16-driven PI3K/AKT activation in lung adenocarcinoma

Background Lung adenocarcinoma (LUAD), the most prevalent histological subtype of non-small cell lung cancer (NSCLC), is characterized by substantial clinical heterogeneity and a frequent propensity to acquire resistance to targeted therapies. Accumulating evidence indicates that both endoplasmic reticulum stress and Golgi apparatus dysfunction play critical roles in reshaping the tumor microenvironment (TME). However, their coordinated contribution to LUAD progression remains poorly understood. Against this background, we developed a machine learning-based prognostic framework focused on endoplasmic reticulum stress- and Golgi apparatus-related genes (EGRGs) to improve prognostic stratification and explore their potential therapeutic implications in LUAD. Methods Publicly available transcriptomic profiles and corresponding clinical data of LUAD patients were collected from TCGA and GEO. DEGs identified in the TCGA-LUAD cohort were intersected with endoplasmic reticulum stress- and Golgi apparatus-related genes (EGRGs) to obtain differentially expressed EGRGs in LUAD. A machine learning-based prognostic signature was constructed in the TCGA-LUAD cohort and validated in external GEO datasets. Patients were stratified according to the calculated risk score, after which tumor mutational burden, immune microenvironment characteristics, and drug sensitivity were compared between risk groups. In addition, the key gene MUC16 was further evaluated through in vitro experiments. Results A 17-gene prognostic signature was established based on 133 differentially expressed endoplasmic reticulum stress- and Golgi apparatus-related genes. The prognostic performance of this signature was further supported in additional independent LUAD cohorts, and the defined risk groups differed significantly in tumor microenvironment features, immune infiltration patterns, mutational landscapes, and predicted therapeutic responses. Subsequent bioinformatic analyses and in vitro validation highlighted MUC16 as a markedly upregulated gene in LUAD, whose higher expression was associated with poorer patient outcomes. Functional and mechanistic experiments further suggested that MUC16 may promote malignant phenotypes in LUAD cells, at least in part, through FAK-mediated activation of PI3K/AKT signaling. Conclusion We developed an EGRG-based prognostic signature for LUAD, which may provide a useful reference for risk stratification and therapeutic decision-making. Further evidence suggested that MUC16 may contribute to LUAD progression, at least in part, through FAK-mediated activation of PI3K/AKT signaling, supporting its potential value as a prognostic biomarker and a candidate for further therapeutic investigation.