Oxidative Stress Associated LncRNAs as Potential Biomarkers for Prognosis and Immune Responses in Lung Squamous Cell Carcinoma Patients

Summary

This study identifies a 10-lncRNA oxidative stress–related signature in lung squamous cell carcinoma (LUSC) using data from The Cancer Genome Atlas. Although validation showed limited clinical predictive performance, the signature differentiated immune microenvironment patterns, and in vitro assays supported LINC01615 as a migration-invasion-related candidate for further mechanistic investigation.

Abstract

Long-chain non-coding RNA (lncRNA) significantly influences lung squamous cell carcinoma's (LUSC) prognostic value and immune infiltration. This study aimed to demonstrate how oxidative stress-related lncRNAs impact lung squamous cell carcinoma (SCC). The Cancer Genome Atlas (TCGA) dataset gathered transcriptome information and related clinical data for LUSC. To build a prognostic model, 10 prognostic-related genes were identified using a series of bioinformatics analyses that compared the OS gene's aberrant expression in tumor and healthy tissues, as well as its association with malignancy. Subjects were stratified into high- and low-risk groups based on the median risk score derived from the 10-gene signature. While the mathematical risk model demonstrated limited independent predictive performance in the validation cohort (AUC ~ 0.5), functional and immunological evaluations revealed significant differences in the tumor microenvironment (TME) across risk strata. Specifically, high-risk patients exhibited distinct immune infiltration profiles and altered immunological scores relative to their low-risk counterparts. Therefore, rather than serving as a direct clinical prediction tool, this oxidative stress-related lncRNA signature provides valuable biological insights into the immune landscape of LUSC and highlights potential therapeutic targets for further mechanistic investigation.

Introduction

Lung cancer demonstrates the highest global incidence and mortality among malignancies¹. Within the non-small cell lung cancer (NSCLC) histological variants, lung squamous cell carcinoma (LUSC) is a predominant subtype, often correlating with unfavorable clinical outcomes and a lack of highly effective targeted treatments compared to lung adenocarcinoma². While primary therapeutic approaches encompass surgical intervention, radiotherapy, and chemotherapy, patients with advanced LUSC frequently exhibit therapeutic resistance³. Consequently, there is an urgent need to identify robust molecular biomarkers to improve prognostic stratification and guide exploratory therapeutic strategies for LUSC⁴.

Oxidative stress, characterized by the pathological accumulation of reactive oxygen species (ROS), plays a critical role in tumor initiation, growth, and metastasis. In the absence of efficient ROS clearance mechanisms, elevated ROS levels induce DNA damage and genotoxicity, ultimately resulting in various genomic alterations that contribute to cancer development. Notably, studies have consistently demonstrated an elevation in ROS levels and oxidative stress specifically within the pathogenesis of lung cancer^5,6. Recent research has further highlighted that clusters of genes linked to oxidative stress may function as prospective prognostic biomarkers in pulmonary malignancies⁷. Long noncoding RNAs (lncRNAs) have emerged as crucial regulators of gene expression and represent promising prognostic biomarkers in various malignancies. While a previous notable study⁷ successfully established an oxidative stress-related lncRNA prognostic model for lung adenocarcinoma (LUAD), the specific prognostic landscape and immune implications of such lncRNAs in LUSC remain largely unexplored.

In this investigation, the authors systematically analyzed TCGA datasets to construct a novel oxidative stress-related lncRNA signature specifically tailored for LUSC. Furthermore, unlike purely bioinformatic analyses, the present study incorporates in vitro experimental validation to confirm the biological functional role of key identified lncRNAs (e.g., Linc01615). Rather than claiming immediate clinical applicability for diagnostic and treatment decisions, this research primarily seeks to explore the potential mechanisms of oxidative stress in LUSC and its correlation with the tumor immune microenvironment. These findings are intended to provide valuable biological insights and highlight potential therapeutic targets for future mechanistic investigations.

Protocol

Data acquisition and preprocessing

All data analyzed in this study were obtained from the publicly accessible TCGA database listed in the Table of Materials. This study strictly adhered to the National Institutes of Health (NIH) Genomic Data Sharing Policy and the publication guidelines provided by TCGA. Because the datasets consist of de-identified, publicly available clinical and multi-omics data, this study was exempt from further Institutional Review Board (IRB) approval. Transcriptome data and corresponding clinical information for LUSC were downloaded from the Cancer Genome Atlas (TCGA) database. Initially, the cohort comprised 551 samples (502 tumor and 49 normal). To ensure the robustness of this prognostic model, patients lacking complete overall survival (OS) data, survival status, or essential clinical characteristics were excluded from the downstream survival analysis. Following this strict filtering process, a final analytic cohort of 470 LUSC patients was established. These 470 patients were subsequently randomly divided into a training set (n = 235) and a testing set (n = 235) for risk model construction and validation. A total of 792 oxidative stress-related genes were gathered via the GeneCards database, listed in the Table of Materials, utilizing an inclusion criterion of a relevance score ≥ 7. The limma package (version 3.66.0) was utilized to screen for differentially expressed oxidative stress genes. Volcano plots were produced using the ggplot2 package (version 4.0.2). Between the training group and the testing group, there was no appreciable variation in clinical characteristics (p > 0.05). The training set was used for model construction, and the testing set was used for internal validation.

Construction and verification of the risk model

To identify lncRNAs associated with oxidative stress, a Pearson co-expression analysis was conducted. The authors evaluated the correlation between the expression profiles of the identified differentially expressed oxidative stress-related genes and all annotated lncRNAs within the TCGA-LUSC dataset. The correlation criteria for selecting co-expressed lncRNAs were set at a correlation coefficient |R| > 0.4 and a p-value < 0.001. Based on these thresholds, a total of 6,088 candidate oxidative stress-related lncRNAs were derived and extracted. Subsequently, these 6,088 candidate lncRNAs were evaluated through LASSO regression and complementary analytical approaches to construct the prognostic risk model. Utilizing the processed data, patient participants were categorized into high- and low-risk strata.

Independent factors and ROC curves

Univariate and multivariate Cox regression models were utilized to identify independent prognostic variables. Specifically, to control the family-wise error rate during the multiple univariate Cox regression analyses (n = 236 tests), the Bonferroni correction was applied, setting the stringently adjusted significance threshold at p < 0.00021 (calculated as 0.05 / 236).

Survival analysis and principal component analysis

The survival package (version 3.8-6) was used to compute overall survival (OS) rates, and principal component analysis (PCA) was used to assess the risk model’s robustness.

Nomogram

The fit index was evaluated using the rms package in R to assess the reliability of the nomogram. Tumor mutational burden (TMB) was evaluated and visualized using the maftools package (version 2.22.0). Algorithms, including ssGSEA, were used to detect immune infiltration. Additionally, the ggpubr package (version 0.6.3) was used to compare immune checkpoints between risk groups. The ESTIMATE algorithm was used to characterize the tumor microenvironment (TME).

Functional analysis

The clusterProfiler package (version 4.14.6) in R was used for enrichment analysis. To explore functional pathways further, GSEA analysis was performed. Using Cytoscape, a co-expression network was created for observation.

Experimental methods

Cell culture conditions and grouping

The NCI-H520 human lung squamous cell carcinoma cell line (RRID: CVCL_1566; listed in the Table of Materials) was cultured in RPMI-1640 medium supplemented with 10% (v/v) fetal bovine serum and 1% penicillin-streptomycin. Log-phase cells were seeded into 6-well culture plates. Upon successful adherence, the cells were randomly allocated into three distinct experimental groups: a blank control group maintained under standard culture conditions, a negative control group (si-NC) transfected with non-targeting siRNA, and an experimental knockdown group (si-LINC01615) transfected with LINC01615-specific siRNA.

Cell transfection

The si-LINC01615 and si-NC reagents were thawed on ice. For the transfection complex preparation, 95 µL of serum-free RPMI-1640 medium was pipetted into sterile centrifuge tubes, followed by the sequential addition of 3 µg of the respective siRNA and 5 µL of Lipo3000 transfection reagent. The identical procedure was applied for both the siLINC01615 and siNC groups. The solutions were gently mixed and incubated at room temperature for 5 min to facilitate complex formation, resulting in a total volume of approximately 200 µL per tube. Subsequently, this mixture was uniformly distributed into the designated culture wells containing 800 µL of RPMI-1640 basal medium. Following a 6 h incubation period at 37 °C, the transfection medium was carefully aspirated and replaced with fresh complete culture medium.

Wound-healing (scratch) assay

For the wound-healing assay, logarithmically growing NCI-H520 cells (RRID: CVCL_1566) were seeded into 6-well culture plates at a density of 5.6 × 10⁵ cells/well. Upon reaching 95–100% confluence, a linear wound was artificially created by scratching the cell monolayer vertically across the center of the well using a sterile 200 µL pipette tip. The wells were subsequently washed three times with PBS to remove detached cells and cellular debris, and then incubated in serum-free RPMI-1640 medium. Images of the identical wounded areas were captured at 0, 24, and 48 h utilizing an inverted microscope. The wound-healing area was quantified by measuring the gap distance using ImageJ (RRID: SCR_003070; listed in the Table of Materials), and the migration rate was calculated relative to the initial wound area at 0 h.

Transwell migration and invasion assays

Cellular migration and invasion capacities were evaluated utilizing 24-well Transwell chambers equipped with 8.0 µm pore-size polycarbonate membrane inserts listed in the Table of Materials. For the invasion assay, the apical chambers were pre-coated with 50 µL of Matrigel (diluted 1:8 in serum-free medium; listed in the Table of Materials) and incubated at 37 °C for 2 h to polymerize, whereas the migration assay utilized uncoated inserts. Following cell transfection and a 24 h serum starvation period, NCI-H520 cells were harvested, resuspended in serum-free RPMI-1640 medium, and seeded into the apical chambers at a density of 1.8 × 10⁵ cells/well in 200 µL. The basal chambers were filled with 600 µL of RPMI-1640 medium supplemented with 10% FBS as a chemoattractant. After a 48 h incubation at 37 °C with 5% CO₂, the inserts were removed and washed three times with PBS. Cells remaining on the apical surface of the membrane were gently swabbed away using a wet cotton swab. The cells that had migrated or invaded the basolateral surface were fixed with 4% paraformaldehyde for 20 min and stained with Giemsa solution listed in the Table of Materials for 30 min at room temperature. The migrated and invaded cells were visualized and enumerated across three randomly selected optical fields per insert using an inverted microscope at 40x magnification.

Statistical analysis

All statistical analyses and data visualizations were performed using R software (version 4.1.2; RRID: SCR_001905; listed in the Table of Materials). The specific version numbers for all R packages utilized in this study have been explicitly detailed in their respective methodological subsections. A significant value was conventionally defined as p < 0.05 unless otherwise reported. Where multiple comparisons were conducted (e.g., the multiple univariate Cox regressions), the Bonferroni method was stringently applied to maintain the family-wise error rate (FWER), with the exact adjusted threshold (e.g., p < 0.00021) detailed in the respective methodological subsection. Quantitative in vitro data were expressed as mean ± standard deviation and analyzed using Student's t-test or one-way ANOVA as appropriate.

Results

Identification of differentially expressed genes for oxidative stress

The workflow is shown in Figure 1. The clinical characteristics of the final analytic cohort comprising 470 LUSC patients are detailed in Table 1. Based on the defined threshold of a relevance score ≥ 7, the authors retrieved a total of 792 oxidative stress-related genes from the GeneCards database (Supplementary File 1). Between normal and LUSC...

Discussion

Lung squamous cell carcinoma (LUSC) is a predominant and highly aggressive histological subtype of non-small cell lung cancer, often characterized by grim survival rates and resistance to conventional therapeutic modalities^8,9. Emerging evidence underscores oxidative stress as a pivotal driver in the pathogenesis and progression of various malignancies. The pathological accumulation of reactive oxygen species (ROS) and the dysregulation of cellular antioxidant de...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
4% Paraformaldehyde	Beyotime	P0099	Used for fixing cells
Cytoscape	Cytoscape Consortium	https://cytoscape.org/	Used for co-expression network creation
Fetal Bovine Serum (FBS)	Gibco	10099-141	Final concentration 10% (v/v)
GeneCards database	GeneCards Human Gene Database	https://www.genecards.org/	Oxidative stress-related genes screening
Giemsa Stain	Sigma-Aldrich	GS500	Used for staining migrated/invaded cells
ImageJ	NIH	https://imagej.nih.gov/ij/	Used for wound-gap quantification
Lipo3000 Transfection Reagent	GLPBIO	GK20006	Used for siRNA transfection
Matrigel	Corning	356234	Diluted 1:8 in serum-free medium
NCI-H520 human lung squamous cell carcinoma	ATCC	HTB-182	Used for in vitro validation assays
PBS (Phosphate-Buffered Saline)	Gibco	10010-023	Used for cell washing
Penicillin-Streptomycin (1%)	Gibco	15140-122	Cell culture antibiotic supplement
R software (v4.1.2)	R Foundation	https://www.r-project.org/	Statistical analysis and visualization
RPMI-1640 Medium	Gibco	11875-093	Cell culture medium
The Cancer Genome Atlas (TCGA)	Genomic Data Commons	https://portal.gdc.cancer.gov/	Transcriptome & clinical data source
Transwell Chambers (8.0 μm)	Corning	3422	Used for migration and invasion assays