Identification of putative transcriptional regulatory networks in Entamoeba histolytica using Bayesian inference

Abstract

Few transcriptional regulatory networks have been described in non-model organisms. In Entamoeba histolytica seminal aspects of pathogenesis are transcriptionally controlled, however, little is known about transcriptional regulatory networks that effect gene expression in this parasite. We used expression data from two microarray experiments, cis-regulatory motif elucidation, and a naïve Bayesian classifier to identify genome-wide transcriptional regulatory patterns in E. histolytica. Our algorithm identified promoter motifs that accurately predicted the gene expression level of 68% of genes under trophozoite conditions. We identified a promoter motif (A/TAAACCCT) associated with high gene expression, which is highly enriched in promoters of ribosomal protein genes and tRNA synthetases. Additionally, we identified three promoter motifs (GAATGATG, AACTATTTAAACATC/TC and TGAACTTATAAACATC) associated with low gene expression. The promoters of a large gene family were highly enriched for these motifs, and in these genes the presence of ⩾2 motifs predicted low baseline gene expression and transcriptional activation by heat shock. We demonstrate that amebic nuclear protein(s) bind specifically to four of the motifs identified herein. Our analysis suggests that transcriptional regulatory networks can be identified using limited expression data. Thus, this approach is applicable to the multitude of systems for which microarray and genome sequence data are emerging.