A Novel Framework for the Identification and Characterization of Putative Positive and Negative cis-Acting Regulatory Elements in Arabidopsis thaliana

Environmental stress causes the transcriptional reprogramming and adaptation of cell metabolism to new, often non-optimal conditions. Gene expression is induced when specific transcription factor (TF) binds to cis-acting regulatory element (CRE) in the promoter. Although substantial knowledge has been collected over last decades, it is still unclear how CRE combinations control gene expression in plants. Simple presence/absence models or combinatory rules are not sufficient for the prediction of gene expression patterns in the promoter sequence. Thus, we hypothesize that known CREs are influenced by other positive and/or negative motifs (pCREs and nCREs, respectively) in promoters. The following paper presents a novel framework for the identification and characterization of these motifs, based on the expression patterns under multiple stresses in Arabidopsis thaliana. The putative pCREs and nCREs that might have impact on the activity of well-known ABA-responsive element (ABRE) element were identified. In total 73 pCREs and 23 nCREs were identified. In the group of pCREs known ABRE-coupling elements (e.g. evening element and dehydration responsive element) were found. The nCREs included known repressors, like BOXBPSAS1, suggesting that the used methodology is correct. Moreover, found motifs were significantly overrepresented in conserved non-coding sequences and DNase I hypersensitive sites. A number of machine learning algorithms were used build a predictive models for ABA_3h responses on the basis of ABRE and found pCREs. All of them showed significant improvement in ABA_3h response prediction when compared to the information in ABRE. Additionally, inclusion of nCREs improved the performance of predictive models. Our results can lead to better understanding of transcriptional regulation in plants.