In silico data filtering to identify new angiogenesis targets from a large in vitro gene profiling data set

Abstract

The objective of this study was to use gene expression data from well-defined cell culture models, in combination with expression data from diagnostic samples of human diseased tissues, to identify potential therapeutic targets and markers of disease. Using Affymetrix oligonucleotide array technology, we identified a common profile of genes upregulated during endothelial morphogenesis into tubelike structures in three in vitro models of angiogenesis. Rigorous data selection criteria were used to identify a list of over 1,000 genes whose expression was increased more than twofold over baseline at either 4, 8, 24, 40 or 50 h. To further refine and prioritize this list, we used standard bioinformatic algorithms to identify potential transmembrane and secreted proteins. We then overlapped this gene set with genes upregulated in colon tumors vs. normal colon, resulting in a subset of 128 genes in common with our endothelial list. We removed from this list those genes expressed in 6 different colon tumor lines, resulting in a list of 24 putative, vascular-specific angiogenesis-associated genes. Three genes, gp34, stanniocalcin-1 (STC-1), and GA733-1, were expressed at levels 10-fold or more in colon tumors compared with normal mucosa. We validated the vascular-specific expression of one of these genes, STC-1, by in situ hybridization. The ability to combine in vitro and in vivo data sets should permit one to identify putative angiogenesis target genes in various tumors, chronic inflammation, and other disorders where therapeutic manipulation of angiogenesis is a desirable treatment modality.