The molecular clock has been conserved from cyanobacteria to mammals and is believed to align behavioral and biochemical processes with the diurnal cycle. This cellular mechanism has been an advantage to increase the fitness of organisms through the ability to anticipate food availability or predator presence. Accumulating evidence has revealed that the circadian clock is intimately interconnected with the metabolic cycle. But, the nature of these interconnections is yet not clear at the transcriptional level, and the contribution of cis-regulatory modules has not been elucidated. Accessible chromatin regions of the genome are implicated in diverse processes, such as gene regulation through enhancers and promoters, insulation of genomic domains or alternative splicing. They allow cell-type specific programs that are controlled by tissue-specific transcription factors and chromatin modifiers, although tissue-specific regulation of the circadian clock remains unclear. Therefore, we compared genome-wide BMAL1 binding and chromatin accessibility in the liver and in NIH3T3 fibroblasts. Moreover, for the first time, our study explores the dynamics of accessible regions, histone 3 lysine 27 acetylation (H3K27ac) and RNA polymerase II (Pol II) every 4 hours over one day in mouse liver. In this study, we show that a substantial fraction of these accessible sites are oscillating during a diurnal cycle with a circadian period. We observed that these accessible regions are enriched in the proximity of actively transcribed genes, and that they are dynamically affected by the binding of transcription factors such as BMAL1. We investigated wild-type (WT) and Bmal1-/- genotypes, in night restricted feeding regimen and in Light-Dark (LD) cycle, to study the circadian clock regulatory network underlying diurnal transcription. Therefore, we applied a penalized generalized linear model to infer the activity of transcription factor binding motifs in oscillating accessible sites using Pol II loadings at the transcription start sites of nearby genes. We were able to recapitulate the known regulatory elements of the circadian clock, notably E-box, D-Box, and ROR-responsive elements (RRE) in the WT genotype. On the other hand, we found that Forkhead box (FOX), glucocorticoids responsive elements (GRE), and C-AMP Response Element (CRE) were the main contributors of the regulation of oscillating genes in Bmal1-/-. Finally, using a mixture model to detect footprints with a base pair resolution, we studied the dynamics of the accessibility overlapping E-box. Our last analysis suggested that BMAL1/CLOCK is binding on double E-boxes with a spacer of 6 or 7 bp in a hetero-tetramer configuration. A 3D structure model further supported this binding mode. In Summary, we used DNase I-seq and ChIP-seq of Pol II and H3K27ac to study the circadian chromatin landscape in a 4h time-resolved experimental design. We uncovered the underlying circadian transcriptional regulatory network and, we dissected the chromatin accessibility around BMAL1 binding sites at a base pair resolution, which led to an unappreciated mode of binding of BMAL1/CLOCK in a hetero-tetramer conformation on double E-boxes. Lastly, we found tissue-specific factors that might contribute to tissue-specific binding of BMAL1/CLOCK.
EPFL2016
