A hyper-dynamic nature of bivalent promoter states underlies coordinated developmental gene expression modules

作者：Shah Akshay; Oldenburg Anja; Collas Philippe^*

来源：BMC Genomics, 2014, 15(1): 1186.

DOI：10.1186/1471-2164-15-1186

• Chromatin state
• Differentiation
• Hidden Markov Model
• Promoter
• Transcriptome

Background: Chromatin remodeling is crucial for proper programing of developmental gene expression. Recent work provides a dynamic view of post-translational histone modifications during differentiation; however there is little insight on the evolution of combinatorial genome-wide patterns of chromatin marks, excluding an essential aspect of developmental gene regulation.
Results: We report here a 15-chromatin state Hidden Markov Model which describes changes in chromatin signatures in relation to transcription profiles during differentiation of human pre-adipocytes into adipocytes. We identify nineteen modules of gene expression reflecting multiple waves of transcriptional up-and down-regulation which characterize adipogenic differentiation. From our model, we developed chromatin state matrices fitting each of these transcription modules to show how the complexity and dynamic nature of chromatin signatures relate to expression patterns. Spatial relationships between chromatin states underlie a high-order chromatin organization in differentiating adipocytes. We show the importance of gene expression level in generating diversity in chromatin signatures, and show that the hyper-dynamic nature of H3K4me2/H3K27me3-marked 'bivalent' promoter states underlies many of the gene expression patterns associated with adipogenic differentiation.
Conclusions: Our results reveal the highly dynamic nature of bivalent promoter states within the adipogenic lineage. The data constitute a valuable resource enabling the assessment of possibilities to alter the adipogenic program.

• 出版日期2014-12-30