We then coexpressed BMAL1 with CLOCK and confirmed that the anti-AcBMAL1 antibody readily recognizes CLOCK-induced acetylation at Lys537 (Figure 6B). In the same assay, coexpression of SIRT1, but not SIRT2 and SIRT3, induced specific deacetylation of BMAL1 (Figure 6B), indicating that SIRT1 specifically deacetylates BMAL1. We also confirmed that SIRT1 readily deacetylates BMAL1 in an in vitro deacetylation assay (Figure S4). Importantly, the SIRT1(H363Y) enzymatically deficient mutant did not affect the acetylation state of BMAL1 (Figure 6C). Furthermore, BMAL1 deacetylation by SIRT1 is responsive to NAD+ and significantly attenuated by NAM (Figure 6D), suggesting that the acetylation of BMAL1 is an event regulated by cellular metabolism.

SIRT1 Contributes to Circadian Control In Vivo

To determine the effect of SIRT1 on the cyclic acetylation of BMAL1 we first compared MEFs from WT mice and Sirt1−/− animals. Upon serum shock, acetylation at Lys537 is cyclic in WT cells, whereas it is sustained and mostly constant in Sirt1−/− MEFs (Figure 7A). Interestingly, lack of acetylation appears to also influence BMAL1 phosphorylation levels (Figure 7A). As phosphorylation has been linked to BMAL1 stability (Cardone et al., 2005; Kondratov et al., 2003), it is interesting to observe that BMAL1 appears indeed to be expressed at higher levels in the absence of SIRT1. These results suggest that SIRT1-controlled acetylation could constitute a critical regulatory step in the control of BMAL1 protein stability.
Next we sought to demonstrate the role of SIRT1 in vivo. To do so, we used tissue-specific Sirt1−/− mice in which the loxed gene was selectively deleted by albumin promoter-driven Cre recombinase in the liver. The original mutant mice have the unique deletion of exon 4 of the Sirt1 gene, which encodes the conserved SIRT1 catalytic domain (Cheng et al., 2003). Livers were collected from mice entrained at different times of the circadian cycle and used to analyze BMAL1 acetylation and gene expression levels. Paralleling the results obtained with the Sirt1−/− MEFs (Figure 7A), BMAL1 acetylation is significantly increased and only mildly rhythmic in the livers from the mutant mice (Figure 7B). BMAL1 phosphorylation is also slightly increased with respect to WT mice, although not to the extent observed in the Sirt1−/− MEFs (see also Figure S5). Finally, expression of the circadian genes Cry1 and Per2 is also significantly altered (Figure 7C), reminiscent of the results obtained with the Sirt1−/− MEFs (Figure 2).

Discussion

A large array of metabolic processes follows the rhythmicity of the circadian cycle. The presence of molecular links that reveal functional wiring between the clock machinery and metabolic pathways has been invoked (Rutter et al., 2002; Schibler and Sassone-Corsi, 2002), and much compelling evidence has accumulated (Turek et al., 2005; Wijnen and Young, 2006). We have proposed that the HAT function of CLOCK may be controlled by changing cell energy levels or, conversely, could regulate them (Doi et al., 2006; Grimaldi et al., 2007). The finding that a core element of the clock machinery directly elicits histone modifications underscored the link between circadian physiology and chromatin remodeling. These notions suggested that NAD(H)-dependent energy pathways in the cell could influence the HAT function of CLOCK:BMAL1. We reasoned that CLOCK-mediated acetylation, and thereby transcriptional activation, could be counterbalanced by transcriptional silencing induced by NAD+-dependent HDACs (Imai et al., 2000; Landry et al., 2000). Intriguingly Sir2, a NAD+-dependent HDAC, had been functionally linked to Sas2 (Kimura et al., 2002; Suka et al., 2002), a protein of the MYST family of HATs to which CLOCK belongs (Doi et al., 2006; Nakahata et al., 2007).
Our results indicate that SIRT1 could function as a molecular rheostat of CLOCK-mediated HAT function, by modulating the timing of histone lysine acetylation (Figure 7D). SIRT1 also modulates the circadian machinery by controlling the acetylation levels of BMAL1 (Figure 6), a core circadian element whose CLOCK-induced acetylation is important for circadian physiology (Hirayama et al., 2007). BMAL1 is acetylated at a key, conserved lysine at position Lys537. We have shown that Lys537 acetylation increases the efficacy of the repressor CRY to silence CLOCK: BMAL1-mediated transcription, an event essential to obtaining proper circadian oscillations (Hirayama et al., 2007). Importantly, the oscillatory acetylation patterns of H3 and BMAL1 differ in their timing: BMAL1 acetylation is sustained at a circadian time when H3 acetylation is at minimal levels (at 24 hr post-serum shock; compare Figures 4A and and7A).7A). This difference nicely fits the scenario of a dual role for CLOCK-mediated acetylation, implicated both in transcriptional activation of circadian promoters (acetylation of H3; Doi et al., 2006) and in their subsequent down-regulation following acetylation of BMAL1- and CRY-mediated repression (Hirayama et al., 2007). These findings raise the fascinating possibility that CLOCK and SIRT1 enzymatic activities may be temporally regulated by additional posttranslational modifications. Interestingly, H3 Lys14 acetylation was shown to be significantly modulated by the phosphorylation of the nearby Ser10 residue (Cheung et al., 2000b; Lo et al., 2000). Of relevance to circadian control, phosphorylation at Ser10 has been linked to light-induced activation of clock gene expression in the supra-chiasmatic nucleus (SCN) (Crosio et al., 2000).