Computational detection of stage-specific transcription factor clusters during heart development

Frontiers in Genetics

Farah S. eRaad¹  Wolfram Hubertus eZimmermann¹  Angelica eRoa¹  Edgar eWingender¹  Sebastian eZeidler¹  Shizuka eUchida²  Cornelia eMeckbach⁴  Rebecca eTacke⁴  Mehmet eGültas⁴ 

[1] DZHK (German Center for Cardiovascular Research), Partner site Göttingen;DZHK (German Center for Cardiovascular Research), Partner site Rhein-Main, Frankfurt am Main;Goethe University Frankfurt;University Medical Center Göttingen;

关键词: cardiomyogenesis;    Markov clustering;    MatrixCatch;    TF interaction;    Engineered heart muscle;   

DOI  :  10.3389/fgene.2016.00033

来源: DOAJ

【 摘 要 】

Transcription factors (TFs) regulate gene expression in living organisms. In higher organisms, TFs often interact in non-random combinations with each other to control gene transcription. Understanding the interactions is key to decipher mechanisms underlying tissue development. The aim of this study was to analyze co-occurring transcription factor binding sites (TFBSs) in a time series dataset from a new cell-culture model of human heart muscle development in order to identify common as well as specific co-occurring TFBS pairs in the promoter regions of regulated genes which can be essential to enhancecardiac tissue developmental processes.To this end, we separated available RNAseq dataset into five temporally defined groups: i) mesoderm induction stage;ii) early cardiac specification stage; iii) late cardiac specification stage; iv) early cardiac maturation stage; v) late cardiac maturation stage,whereeach of these stages is characterized by unique differentially expressed genes (DEGs). To identify TFBS pairs for each stage, we applied the MatrixCatch algorithm, which is a successful method to deduce experimentally described TFBS pairs in the promoters of the DEGs. Although DEGs in each stage are distinct, our results show that the TFBS pair networks predicted by MatrixCatch for all stages are quite similar. Thus, we extend the results of MatrixCatch utilizing a Markov clustering algorithm (MCL) to perform network analysis. Using our extended approach, we are able to separate the TFBS pair networks in several clusters to highlight stage-specific co-occurences between TFBSs. Our approach has revealed clusters that are either common (NFAT or HMGIY clusters) or specific (SMAD or AP-1 clusters) for the individual stages. Several of these clusters are likely to play an important role during the cardiomyogenesis. Further, we have shown that the related TFs of TFBSs in the clusters indicate potentialsynergistic or antagonistic interactions to switch between different stages. Additionally, our results suggest that cardiomyogenesis follows the hourglass model which was already proven for Arabidopsis and some vertebrates. This investigation helps us to get a better understanding of how each stage of cardiomyogenesis is affected by different combination of TFs.

【 授权许可】

Unknown