Circadian transcription regulatory networks in mammals

Guillaume Rey

Circadian oscillator networks underlie behavioral and physiological daily rhythms in most living organisms ranging from bacteria to humans. In mammals, the oscillators in peripheral organs are entrained by a master pacemaker residing in the suprachiasmatic nucleus (SCN), which is itself synchronized by daily light–dark cycles.

Figure 1 : Negative limb of the mammalian circadian clock CLOCK/BMAL1 activates Per and Cry gene families whom protein products will repress their own activation.

These molecular clocks are based on interlocked negative feedback mechanisms that rely on several key transcriptional regulators. Among those, the principal activator is called CLK/CYC in insects and CLOCK/BMAL1 in mammals. Using microarray technologies, the rhythmic expression of genes has been studied intensively especially in mouse liver, where a large number of cycling genes have been identified. However, the gene regulatory network underlying this dynamic control of transcription at the genome scale remains at the moment elusive. In a previous study, we used a comparative genomics approach to identify and model CLK/CYC and CLOCK/BMAL1 bound enhancers. The presence of two highly conserved tandem E-box-like (E1-E2) motifs was detected among the prominent CLK/CYC targets genes in flies. A hidden Markov model (HMM) that allows a variable spacer between the two E-boxes was derived from these sequences and validated with functional genomics datasets. Interestingly the fly model was also able to predict known CLOCK/BMAL1 targets in mouse with high enrichment. A phylogenetic analysis showed that this motif is evolutionary conserved among mammals, fishes and insects.


Figure 2 : Probabilistic model of E1-E2 (A) Multiple alignment of the promoter region of Period gene for 12 fly species. The double E-box motif is shown by the E1 and E2 labels. (B) Converged hidden Markov model (HMM) fitted on 5 core clock fly genes. A canonical E-box (E1) is separated by 6 or 7 base pairs from a second more degenerate version (E2).

In order to further characterize and validate our E1-E2 predictions, we asked whether the consensus sequence derived from our model is able to bind BMAL1/CLOCK in vitro. We performed electromobility shift assays (EMSA) on the fly E1-E2 model to test our consensus sites. We showed that our double E-box motif is bound by two BMAL1/CLOCK dimers with a strong cooperativity. We aim to dissect the circadian transcriptional network in mammals by using both experimental and bioinformatics approaches. Combining promoter sequences analysis to infer transcription factor binding sites with available timeseries of mouse liver expression data, we use a linear model of transcriptional regulation to identify cyclic transcription factor (TF) activities. The predicted TF activities for known circadian cis-elements, in particular the E1-E2 element, are in good agreement with published biochemical data. Interestingly, our model also predicts consistent cyclic activities for TFs involved in the entrainment of the circadian clock as well as in the liver metabolism.