The Epigenetic Layer in Bladder Cancer. Bladder cancer (BC) is one of the most common urological malignancies worldwide. Beyond genetic mutations, aberrant DNA methylation -- the addition or removal of methyl groups at CpG sites -- plays a central role in regulating gene expression in cancer. Hypomethylation can activate oncogenes, while hypermethylation silences tumor suppressor genes. Both processes contribute to bladder cancer initiation and progression.
Gene expression microarray studies and DNA methylation microarray studies have each been performed independently for bladder cancer, but results from single datasets are often inconsistent and subject to sample-size limitations. Integrating both data types from multiple public datasets can produce a more reliable and comprehensive picture of the epigenetic and transcriptomic landscape of bladder cancer.
Study Objective. This study aimed to identify genes that are simultaneously aberrantly methylated and differentially expressed in bladder cancer -- so-called aberrantly methylated-differentially expressed genes (AMDEGs) -- and to determine which core genes and pathways these AMDEGs implicate in bladder cancer development. The approach combined four GEO public microarray datasets, gene ontology analysis, KEGG pathway analysis, protein-protein interaction (PPI) network construction, and TCGA database validation.
Four GEO Datasets Combined. Two gene expression datasets were downloaded from the Gene Expression Omnibus (GEO): GSE3167 (40 BC samples and 20 normal urothelial samples) and GSE65635 (27 BC samples and 7 normal samples). Two DNA methylation datasets were also obtained: GSE37816 (8 BC samples and 8 normal samples) and GSE33510 (113 BC samples and 18 normal samples). Differentially expressed genes (DEGs) and differentially methylated positions (DMPs) were identified using the limma package in R, with FDR less than 0.05 and absolute log2 fold change greater than 0.5 as cutoffs for DEGs, and FDR less than 0.05 for DMPs.
Aberrantly methylated-differentially expressed genes were defined as genes meeting both criteria simultaneously: a gene had to be differentially expressed between tumor and normal tissue AND differentially methylated, with the methylation and expression changes in opposite directions (hypomethylation paired with upregulation, or hypermethylation paired with downregulation). Using this filter, 71 hypomethylated/upregulated genes and 89 hypermethylated/downregulated genes were identified as AMDEGs.
PPI Network Construction and Module Analysis. The STRING database was used to build protein-protein interaction networks for both AMDEG sets. Cytoscape software with the MCODE plugin identified network modules (clusters of highly interconnected genes), and the top 5 hub genes (highest connectivity within the network) were designated as core genes for each group. Gene ontology and KEGG pathway enrichment analysis was performed using DAVID bioinformatics tools. All core genes were subsequently validated for methylation and expression status in the TCGA bladder cancer database.
Hypomethylated/Upregulated Gene Enrichment. Gene ontology analysis of the 71 hypomethylated/upregulated AMDEGs found significant enrichment in biological processes including cell-cell adhesion, positive regulation of cell growth, and blood vessel development. Molecular function enrichment included poly(A) RNA binding, KDEL sequence binding, and cadherin binding involved in cell-cell adhesion. KEGG pathway analysis showed enrichment in the p53 signaling pathway (hsa04115, p = 0.00029, genes: SFN, ATP1B1, SERPINB5), metabolic pathways (hsa01100, p = 0.00171), and fructose and mannose metabolism (hsa00051, p = 0.00179).
Hypermethylated/Downregulated Gene Enrichment. The 89 hypermethylated/downregulated AMDEGs were enriched in biological processes including cell adhesion, positive regulation of cell proliferation, signal transduction, and extracellular matrix organization. KEGG pathway analysis identified insulin secretion (hsa04911, p = 1.88 x 10^-6, genes: ABCC8, ADCY9, KCNMA1, RYR2, VAMP2), calcium signaling pathway (hsa04020, p = 4.02 x 10^-6), cAMP signaling pathway (hsa04024, p = 9.80 x 10^-5), and focal adhesion (hsa04510, p = 0.00011) as the most significantly enriched pathways.
The finding that hypomethylated/upregulated genes are enriched in p53 signaling and metabolic pathways while hypermethylated/downregulated genes cluster around calcium signaling, cAMP signaling, and focal adhesion -- all pathways with established roles in bladder cancer -- provides biological coherence to the methylation-driven transcriptional changes identified in this study.
Top Core Genes for Hypomethylated/Upregulated Network. PPI network analysis of the 71 hypomethylated/upregulated genes identified five core hub genes: CDH1, DDOST, CASP8, DHX15, and PTPRF. CDH1 (E-cadherin) methylation status was previously reported to correlate with poor survival in bladder cancer. CASP8, a caspase involved in apoptosis initiation, has been shown to mediate drug-induced apoptosis in bladder cancer. DHX15, an RNA helicase, has been implicated in prostate cancer progression and glioma suppression. PTPRF is a candidate biomarker for prostate cancer and non-small cell lung cancer.
Top Core Genes for Hypermethylated/Downregulated Network. The five core hub genes from the hypermethylated/downregulated network were GNG4, ADCY9, NPY, ADRA2B, and PENK. ADCY9 (adenylate cyclase 9) modulates signal transduction and its expression correlates with colon cancer TNM staging. GNG4 is hypermethylated in glioblastoma, and its overexpression inhibits glioblastoma cell proliferation. NPY (neuropeptide Y) affects vascular smooth muscle contraction and can stimulate angiogenesis. The hypermethylation of PENK has been validated specifically in bladder cancer and it serves as a diagnostic marker in multiple cancer types.
Module Analysis Reveals Functional Clusters. Within the hypomethylated/upregulated network, three functional modules were identified: a Vibrio cholerae infection module (KDELR2, KDELR1, TUSC3), a protein export module (SEC61G, SRP9, TRAM1), and a proteoglycans in cancer module (PTK6, ERBB3, MAPK13). The proteoglycans module is particularly notable because PTK6 is a predictive protein in bladder cancer and ERBB3 is linked to treatment resistance. For the hypermethylated/downregulated network, two modules emerged: GABAergic synapse (ADRA2B, PENK, ADCY9, DRD4, NPY, GNG4) and vascular smooth muscle contraction (RASL12, ACTA2, MYLK, ACTG2).
Core Genes Validated in TCGA Bladder Cancer Data. All ten core hub genes were validated in the TCGA bladder cancer database. For the hypomethylated/upregulated group: CDH1 was confirmed as hypomethylated (p = 9.26 x 10^-8) and upregulated (p = 0.007); CASP8 as hypomethylated (p = 9.13 x 10^-12) and upregulated (p = 0.001); and PTPRF as upregulated (p = 0.005). For the hypermethylated/downregulated group: GNG4 was confirmed as hypermethylated (p = 9.09 x 10^-9); ADCY9 as hypermethylated (p = 3.21 x 10^-6) and downregulated (p = 8.54 x 10^-13); NPY as hypermethylated (p = 9.48 x 10^-11); ADRA2B as hypermethylated (p = 6.15 x 10^-11) and downregulated (p = 0.00036); and PENK as hypermethylated (p = 1.40 x 10^-7) and downregulated (p = 2.98 x 10^-5).
The cross-validation between GEO-derived findings and the independent TCGA cohort substantially increases confidence that these core gene methylation and expression changes are genuine features of bladder cancer biology rather than dataset-specific artifacts. The highly significant p-values across both methylation and expression validation suggest these genes are consistently and strongly dysregulated across different patient populations and platforms.
Significance for Bladder Cancer Research. These ten core genes represent strong candidates for further investigation as diagnostic biomarkers, prognostic markers, or therapeutic targets in bladder cancer. The convergence of methylation-driven silencing or activation with functional roles in known cancer pathways (p53 signaling, focal adhesion, calcium and cAMP signaling) provides biological rationale for future mechanistic studies and clinical translation.
Value of Multi-Dataset Integration. This study demonstrates that combining multiple independent microarray datasets for both gene expression and DNA methylation substantially improves the reliability of candidate gene identification compared to single-dataset analyses. The integrated approach filtered noise from individual studies and converged on 160 AMDEGs (71 hypomethylated/upregulated and 89 hypermethylated/downregulated) that are consistently altered across datasets.
The identification of GABAergic synapse-related hypermethylated/downregulated genes as a functional module is a novel finding. The authors note that the role of GABAergic signaling in bladder cancer and its regulation by aberrant DNA methylation had not previously been characterized, warranting further mechanistic study. Similarly, the enrichment of hypomethylated genes in the KDEL sequence binding pathway and endoplasmic reticulum membrane localization suggests ER stress-related mechanisms may contribute to bladder cancer.
Limitations and Future Directions. The clinical utility of the identified core genes requires further evaluation in prospective studies. Validation was performed only in the TCGA database, which represents a specific patient population. Additional validation in independent clinical cohorts, functional studies to establish causal roles of the core genes, and investigation of whether epigenetic drugs targeting methylation patterns can reverse the observed changes are needed before these findings can be applied clinically.