Identification of potential diagnostic targets and therapeutic strategies for anoikis-related biomarkers in lung squamous cell carcinoma using machine learning and computational virtual screening

Abstract

Objective: Lung squamous cell carcinoma (LUSC) is a common subtype of non-small cell lung cancer (NSCLC) characterized by high invasiveness, high metastatic potential, and drug resistance, resulting in poor patient prognosis. Anoikis, a specific form of apoptosis triggered by cell detachment from the extracellular matrix (ECM), plays a crucial role in tumor metastasis. Resistance to anoikis is a key mechanism by which cancer cells acquire metastatic potential. Although several studies have identified biomarkers related to LUSC, the role of anoikis-related genes (ARGs) remains largely unexplored.

Methods: Anoikis-related genes were obtained from the Harmonizome and GeneCards databases, and 222 differentially expressed genes (DEGs) in LUSC were identified via differential expression analysis. Univariate Cox regression analysis identified 74 ARGs significantly associated with survival, and a prognostic model comprising 8 ARGs was developed using LASSO and multivariate Cox regression analyses. The model was internally validated using receiver operating characteristic (ROC) curves and Kaplan-Meier (K-M) survival curves. Differences in immune cell infiltration and gene expression between high- and low-risk groups were analyzed. Virtual drug screening and molecular dynamics simulations were performed to evaluate the therapeutic potential of CSNK2A1, a key gene in the model. Finally, in vitro experiments were conducted to validate the therapeutic effects of the identified drug on LUSC.

Results: The 8-gene prognostic model demonstrated excellent predictive performance and stability. Significant differences in immune cell infiltration and immune microenvironment characteristics were observed between the high- and low-risk groups, suggesting the critical role of ARGs in shaping the immune landscape of LUSC. Virtual drug screening identified Dihydroergotamine as having the highest binding affinity for CSNK2A1. Molecular dynamics simulations confirmed that the CSNK2A1-Dihydroergotamine complex exhibited strong binding stability. Further in vitro experiments demonstrated that Dihydroergotamine significantly inhibited LUSC cell viability, migration, and invasion, and downregulated CSNK2A1 expression.

Conclusion: This study is the first to construct an anoikis-related prognostic model for LUSC, highlighting its role in the tumor immune microenvironment and providing insights into personalized therapy. Dihydroergotamine exhibited significant anti-LUSC activity and holds promise as a potential therapeutic agent. CSNK2A1 emerged as a robust candidate for early diagnosis and a therapeutic target in LUSC.

Keywords: CSNK2A1; anoikis; lung squamous cell carcinoma; machine learning; virtual screening.

FIGURE 1

Flowchart of this study.

FIGURE 2

Anoikis-related differentially expressed genes and their associated regulatory factors in LUSC. (A) A total of 222 ARGs were identified from TCGA-LUSC cohort. (B) A forest plot showing the top 45 ARGs identified through univariate Cox regression analysis. (C) Copy number variations (CNVs) of the top 45 ARGs in the TCGA-LUSC. (D) Correlation network diagram among the top 45 ARGs. (E) Chromosomal region alterations of the ARGs.

FIGURE 3

LUSC subtypes associated with ARGs. (A–D) A consensus matrix for k = 2 was generated through consensus clustering. (C,D) The two subtypes were differentiated using UMAP and t-SNE based on the expression of ARGs.

FIGURE 4

LUSC subtypes associated with ARGs. (A) Overall survival analysis of the two subtypes (P < 0.001). (B) Differential expression of ARGs between the two subtypes. (C, D) Different KEGG pathway enrichment levels between the two subtypes.

FIGURE 5

Feasibility analysis of the two subtypes models.

FIGURE 6

Differences in gene expression and immune infiltration between the two subtypes. (A) Differential expression of ARGs across the two subtypes. (B) Immune infiltration profiles of the two subtypes.

FIGURE 7

Identification of ARGs prognostic signature. (A) LASSO analysis with cross-validation identified 11 prognostically relevant ARGs. (B) Coefficients of 11 prognostically relevant ARGs. (C, D) Kaplan-Meier curves for two subtype risk groups. (E, F) Time-dependent ROC curves for 1-, 3- and 5-year OS. (G) Risk score distribution of ARG clusters. (H) Alluvial diagram showing subtypes transitions and survival status.(I) Heatmap of the expression patterns of the 11 ARGs.

FIGURE 8

The immune microenvironment of LUSC. (A) Proportion of immune cell infiltration. (B) Correlation between risk scores with the proportion of M0 macrophage in LUSC.(C) Differences in immune cell populations between high-risk and low-risk groups. (D) Correlation analysis among immune cells.(E) Gene-immune cell correlation analysis. (F) Estimated scores for expression profiles of the two risk groups.

FIGURE 9

Nomogram for LUSC patients. (A) Nomogram constructed based on ARGs scores and clinicopathologic features.(B) Calibrated Nomogram. (C–E) DCA evaluation of LUSC patients prognosis.(F) Risk curves showing survival probability progression of over time.(G) Forest plot of multivariate Cox regression analysis, Illustrating the association between clinical characteristics and risk scores for LUAD patients.

FIGURE 10

Survival analysis of model genes.

FIGURE 11

Molecular docking and molecular dynamics simulation. (A) The docking result of the CSNK2A1-Dihydroergotamine complex. (B–F) The curve of CSNK2A1 protein and Dihydroergotamine complex: RMSD, RMSF, Rg, Hydrogen bond analysis, and SASA. Curves for the CSNK2A1 protein and the Dihydroergotamine complex. (G) Comparison of conformation of the complex at five different molecular dynamics simulation time points. (H) Free energy distribution. (I) Average binding free energy. (J) Contributions of amino acid residues involved in binding.

FIGURE 12

Effect of Dihydroergotamine on NCI-H2170 cells viability, CSNK2A1 expression, migration, invasion. (A) Inhibition of NCI-H2170 cells viability by Dihydroergotamine. (B) Changes in CSNK2A1 expression in NCI-H2170 cells after Dihydroergotamine treatment. (C) Effect of 40 μM Dihydroergotamine on the migration of NCI-H2170 cells. (D) Effect of 40 μM Dihydroergotamine on the invasion of NCI-H2170 cells.