We also followed the SIRT1-CLOCK interaction during the circadian cycle (Figure 5C). To do so, we coimmunoprecipitated SIRT1 from MEFs by using anti-CLOCK-specific antibody at various times after serum shock. While it would appear that the association undergoes some mild variations, after quantification of three different experiments we concluded that the SIRT1-CLOCK interaction is mostly stable during the circadian cycle (Figure 5C). We have also found that the CLOCK-SIRT1 interaction is not significantly modulated by agents known or likely to influence SIRT1 function, including NAD+, pyruvate, resveratrol, splitomicin, desferroxamine, and glucose (Figure S2).
Finally, the CLOCK-SIRT1 interaction does not appear to require the HDAC function. We used a SIRT1 mutant with a single amino acid substitution that diminishes the deacetylase activity (SIRT1-(H363Y); Vaziri et al., 2001). The mutated protein interacts with CLOCK with efficacy equivalent to that of normal SIRT1 (Figure 5D).
To identify the protein regions involved in SIRT1-CLOCK association, we performed GST pulldown assays (Figures 5E and 5F). We found that the central region of CLOCK (aa 450–570) is necessary for interaction with SIRT1. Interestingly, this region contains the serine/threonine-rich domain, which we predicted to be involved in regulated protein interactions (Doi et al., 2006), and exon 19, the domain originally found to be essential for CLOCK function (Antoch et al., 1997). In SIRT1, the N-terminal region (aa 1–231) is necessary and sufficient for eliciting efficient interaction with CLOCK (Figure 5F). This information is of interest because the same SIRT1 domain is involved in the interaction with other regulatory proteins. Specifically, it has been recently found to mediate the interaction with the histone methyltransferase SUV39H1 (Vaquero et al., 2007), a regulatory event that results in increased levels of the H3K9me3 modification and thereby control of heterochromatin formation.
BMAL1 Acetylation at Lys537 Is Regulated by SIRT1
Recently we have reported that BMAL1 is rhythmically acetylated by CLOCK and that this event is essential for control of circadian function (Hirayama et al., 2007). We have generated an antibody that specifically recognizes acetylated BMAL1 at Lys537 (Figure S3). Because of the interplay between CLOCK and SIRT1, we suspected that the deacetylase that could regulate the dynamic levels of BMAL1 acetylation could be SIRT1. To identify which class of HDAC is responsible for deacetylation of BMAL1, we treated cultured cells expressing Myc-CLOCK and Flag-Myc-BMAL1 with class I and II inhibitor, trichostatin A(TSA), and/or class III inhibitor, NAM, for 6 hr and 16 hr, respectively. Acetylation of BMAL1 at Lys537 is significantly increased by NAM treatment but not by TSA treatment (Figure 6A).