Defining the human endothelial transcriptome

Abstract

Advances in microarray technology facilitate the study of biological systems at a genome-wide level. Meaningful analysis of these transcriptional profiling studies, however, demands the concomitant development of novel computational techniques that take into account the size and complexity of the data. We have devised statistical algorithms that use replicate microarrays to define a genome-wide expression profile of a given cell type and to determine a list of genes that are significantly differentially expressed between experimental conditions. Applying these algorithms to the study of cultured human umbilical vein endothelial cells (HUVEC), we have found approximately 54% of all genes to be expressed at a detectable level in HUVEC under basal conditions. The set of highest expressed genes is enriched in nucleic acid binding proteins, cytoskeletal proteins and isomerases as well as certain known markers of endothelium, and the complete list of genes can be found at ... We have also studied the effect of a 4-hour exposure of HUVEC to 10 U/mL of IL-1, and detected 491 upregulated and 259 downregulated statistically significant genes, including several chemokines and cytokines, as well as members of the TNFAIP3 family, the KLFfamily and the Notch pathway. Applying these rigorous statistical techniques to genome-wide expression datasets underscores known patterns of endothelial inflammatory gene regulation and unveils new pathways as well.

(cont.) Finally, we performed a direct comparison of direct-labeled microarrays with amplified RNA microarrays for an initial assessment of the effect of the additional noise of amplification on the outputs of the statistical algorithms. These techniques can be applied to additional genome-wide profiling studies of endothelium and other cell types to refine our understanding of transcriptomes and the gene regulatory network governing cellular function and pathophysiology.