Identification of novel human glioblastoma-specific transcripts by serial analysis of gene expression data mining

Abstract

Glioblastoma multiforme (GBM) remains the most common and aggressive primary brain tumor in adults with a poor median survival, and molecular biomarkers for GBM pathogenesis are in need. The objective of this study is to identify potential novel genes for GBM pathogenesis by gene expression data mining. Available SAGE libraries of GBM, astrocytoma, and normal brain tissues were collected from the Cancer Genome Anatomy Project (CGAP). Significance analysis for microarray (SAM) and CGAP-SAGE-Genie-DGED were used to identify differentially expressed tags, and specific tags that were differentially expressed only in GBM were further selected. Tags to genes association was performed by CGAP-SAGE-Genie-SAV. Immunohistochemistry was used to investigate distribution and validate expression of the interested gene. Three genes were significantly differentially expressed just in brain. up-regulated expression of STAB1 and down-regulated expression of SH3GL2 and DNM3. Immunohistochemistry assay indicated that STAB1 mainly expressed in vascular endothelial cells and over-expressed in GBM samples compared to normal samples. Our study shows that data mining of public sources of gene expression is an effective way to identify novel tumor-associated genes, and this work may contribute to the identification of candidate genes for GBM angiogenesis.