Identification of Hub Genes, Single-Nucleotide Polymorphisms, and Potential Drug Targets In Breast Cancer Using Transcriptomic Analysis

Summary

This study investigated mitochondrial oxidative stress–related genes associated with therapeutic resistance in HER2-positive breast cancer. Using transcriptomic datasets, integrated bioinformatics, and clinical data (n = 4,929), the authors identified MTHFD2 and PRDX3 as key differentially expressed genes with potential prognostic value. In silico. analyses suggest functional effects of genetic variants.

Abstract

HER2-positive (HER2⁺) breast cancer often develops resistance to therapies like Lapatinib, potentially involving mitochondrial metabolic and redox reprogramming. This study aimed to identify mitochondrial oxidative stress–related genes (MOS-DEGs) as biomarkers of resistance and to characterize their functional nsSNPs and structural impacts. RNA-seq datasets (GSE231524, GSE231525) were analyzed using DESeq2 to identify DEGs, which were intersected with mitochondrial oxidative stress genes to obtain MOS-DEGs. Functional enrichment, PPI analysis, ROC analysis, expression profiling, and survival analysis were performed. nsSNPs were evaluated using multiple predictive tools, and structural impacts were assessed through secondary and 3D modeling. Integrated transcriptomic and clinical analysis identified MTHFD2 and PRDX3 as central MOS-DEGs displaying opposing regulatory roles in mitochondrial redox metabolism. MTHFD2 was significantly upregulated and associated with poor prognosis (HR = 1.53, p = 1.1 × 10⁻¹⁶), while PRDX3 exhibited protective expression patterns (HR = 0.73, p = 7.7×10⁻¹⁰). ROC analysis suggested their potential as predictors of therapeutic response. nsSNP analysis revealed five deleterious variants in MTHFD2 and four deleterious variants in PRDX3, with rs1471336772 (MTHFD2) and rs747786383 (PRDX3.) identified as the most pathogenic. These variants were predicted to be damaging based on multiple computational scoring systems, including SIFT ≤ 0.05, PolyPhen-2 ≥ 0.85, deleterious CADD scores, and high REVEL scores, and were located within catalytic or redox-active domains. Structural modeling suggested that these substitutions may destabilize conformation, disrupt metal- and cofactor-binding sites, and affect NADPH regeneration or thioredoxin-dependent peroxidase activity. Molecular dynamics predictions indicated potential loss of structural stability and altered flexibility, suggesting possible functional impairment of mitochondrial redox control. This study identifies MTHFD2 and PRDX3 as regulators of mitochondrial oxidative stress in HER2⁺ breast cancer. Deleterious nsSNPs in these genes may contribute to altered redox balance, potentially influencing metabolic adaptation (MTHFD2) and antioxidant defense (PRDX3).

Introduction

Breast cancer is the most commonly diagnosed cancer in women worldwide and a major cause of cancer-related death despite substantial progress in early detection and targeted therapeutics^1,2,3. Molecular heterogeneity within breast tumors underlies diverse clinical outcomes and treatment responses, posing persistent challenges for precision oncology^4,5. Among the recognized molecular subtypes, human epidermal growth factor receptor 2 (HER2)-positive breast cancer accounts for approximately 15–20% of all breast cancers ....

Protocol

Data retrieval and preprocessing

This study was conducted using publicly available transcriptomic and genetic data; no human or animal subjects were directly involved. Transcriptomic datasets relevant to HER2⁺ breast cancer therapy resistance were retrieved from the NCBI Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/)¹¹. Two RNA-seq datasets, GSE231524 and GSE231525, were selected because of their specific focus on HER3-driven resistance and DUSP6 inhibition in HER2+ breast cancer cell lines (BT474 and MDA-MB-453). These datasets included parental, drug-tolerant,....

Results

Differential gene expression analysis

Differential expression profiling was performed to investigate transcriptional alterations associated with HER2⁺ breast cancer progression and therapy resistance using two independent RNA-seq datasets, GSE231524 and GSE231525. In the first dataset (GSE231524), a total of 19,727 genes were initially quantified. Following normalization, filtering, and application of the adjusted regression model (ARM) for differential expression, .......

Discussion

Mitochondrial oxidative stress is a defining hallmark of tumor adaptation, governing cellular survival, metabolic plasticity, and therapy resistance. The present study systematically explored mitochondrial oxidative stress–related differentially expressed genes (MOS-DEGs) in HER2⁺ breast cancer by integrating transcriptomic data, functional enrichment, clinical validation, and detailed in silico mutational and structural analyses. The integration of differential expression with mutation profilin.......

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
AnnotationDbi	Bioconductor	v1.64.0	R annotation package used for HGNC gene symbol standardization and annotation.
apeglm	Bioconductor	method in DESeq2	Method used for log2 fold-change shrinkage in differential expression analysis.
Biobase	Bioconductor	v2.62.0	R package used to access and manage GEO-derived expression and metadata objects.
CADD	University of Washington / Kircher Lab	web tool	Combined Annotation Dependent Depletion score used for nsSNP pathogenicity prediction.
clusterProfiler	Bioconductor	v4.8.1	R package used for GO and KEGG enrichment analysis.
ComplexHeatmap	Bioconductor	v2.18.0	R package used for heatmap generation and expression clustering.
Cytoscape	Cytoscape Consortium	software	Network visualization platform used with STRING-derived PPI networks and CytoHubba prioritization.
dbSNP	NCBI	database	Population variant database used for cross-validation of prioritized nsSNPs.
DESeq2	Bioconductor	v1.42.0	R package used for normalization, variance modeling, and differential expression analysis.
dplyr	CRAN / tidyverse	v1.1.3	R package used for data manipulation and intersection of gene sets.
DynaMut	University of Melbourne / BioSig	web server	Tool used to estimate mutation-associated stability and flexibility changes in proteins.
EnhancedVolcano	Bioconductor	v1.22.0	R package used to visualize differential expression results as volcano plots.
Ensembl Genome Browser	EMBL-EBI / Ensembl	database	Source of canonical transcript sequences for MTHFD2 and PRDX3.
Ensembl Variant Effect Predictor (VEP)	EMBL-EBI / Ensembl	web tool	Tool used to annotate missense variants and integrate predictive scores.
ExAC	Broad Institute	database	Population-level exome database used for nsSNP cross-validation.
Gene Expression Omnibus (GEO)	NCBI	database	Repository used to retrieve transcriptomic datasets for HER2-positive breast cancer analysis.
Gene Ontology	Gene Ontology Consortium	GO:0006979	Ontology resource used to retrieve oxidative stress-related genes and for enrichment analysis.
GEOquery	Bioconductor	v2.70.0	R package used to access GEO count matrices and metadata.
ggplot2	CRAN / tidyverse	v3.5.0	R package used for data visualization, clustering plots, and graphical outputs.
ggraph	CRAN	R package	R package used for graph and network visualization during enrichment and pathway representation.
gnomAD	Broad Institute	database	Population variant database used to check frequency and distribution of prioritized nsSNPs.
GOplot	CRAN	R package	R package used for GO enrichment visualization including chord and summary plots.
GSE231524	NCBI GEO	accession	RNA-seq dataset used for analysis of parental, drug-tolerant, and resistant BT474 phenotypes.
GSE231525	NCBI GEO	accession	RNA-seq dataset used for analysis of DUSP6 knockdown in HER2-positive breast cancer cells.
Human MitoCarta3.0	Broad Institute	database	Curated mitochondrial gene resource used to define mitochondrial gene sets.
Human Oxidative Stress Gene Database (HOSGDB)	HOSGDB	database	Database used to retrieve oxidative stress-related genes.
igraph	CRAN	R package	R package used for network representation and pathway visualization.
iMutant 3.0	University of Bologna / Biofold	web server	Tool used to predict mutation effects on protein stability.
Kaplan–Meier Plotter	KMplot	web tool	Online platform used for recurrence-free survival analysis in breast cancer cohorts.
KEGG	Kyoto University	database	Pathway database used for oxidative phosphorylation and pathway enrichment analysis.
KEGGREST	Bioconductor	R package	Package used to retrieve and annotate KEGG pathway information.
Lapatinib	Not specified in manuscript	1 μM treatment condition	HER2-targeted inhibitor used in the source experimental datasets to derive drug-tolerant/resistant phenotypes.
MetaLR	Integrated via Ensembl VEP	score	Computational pathogenicity metric used for variant prioritization.
MutPred2	MutPred	web server	Tool used to predict functional consequences of amino acid substitutions.
NetSurfP 3.0	Technical University of Denmark	web server	Tool used for secondary structure and solvent accessibility prediction.
org.Hs.eg.db	Bioconductor	v3.18.0	Human genome annotation package used for gene identifier mapping.
pathview	Bioconductor	R package	Package used to map enriched genes onto KEGG pathways.
pheatmap	CRAN	v1.0.12	R package used for heatmap plotting and clustering visualization.
PolyPhen-2	Harvard / Brigham and Women's Hospital	web tool	Computational predictor used to estimate structural/functional impact of amino acid substitutions.
Protein Data Bank (PDB)	RCSB PDB	database	Source of experimentally resolved protein structures for MTHFD2 and PRDX3.
PSIPRED	University College London	v3.2	Protein secondary structure prediction server used for 2D topology analysis.
PyMOL	Schrödinger / PyMOL	v3.1	Molecular graphics software used to visualize protein structures and mutation sites.
REVEL	Integrated via Ensembl VEP	score	Ensemble pathogenicity score used for missense variant prioritization.
ROCplotter	ROCplot.org	web tool	Online platform used for ROC-based validation of gene expression in breast cancer response cohorts.
RStudio	Posit	v4.3.1/v4.3.2 environment	Statistical computing environment used for transcriptomic, enrichment, and visualization workflows.
SIFT	A*STAR	web tool	Predictive tool used to classify amino acid substitutions as tolerated or deleterious.
STRING	STRING Consortium	database	Protein–protein interaction database used for network analysis and hub gene identification.
TNMplot	TNMplot.com	web tool	Platform used to compare gene expression across normal, tumor, and metastatic breast tissues.
VennDiagram	CRAN	v1.7.3	R package used to visualize overlap among DEGs and curated mitochondrial gene sets.