A circadian sleep disorder reveals complexity in the clock

Abstract

Circadian rhythms are established by transcription of clock genes and autoregulatory transcriptional feedback loops. In this issue, Xu et al. (2007), characterize mice expressing a human Per2 mutation identified in patients with Familial Advanced Sleep Phase Syndrome. Their results reveal that PER2 phosphorylation— by CK1δ and other kinases—is surprisingly complex and has opposite effects on PER2 levels and period length.

The ability to predict sunset, sunrise and other time cues by circadian clocks exists in most organisms. In vertebrates, these clocks are based on cell-autonomous transcriptional feedback loops of clock proteins, such as the basic helix-loop-helix-PAS domain transcription factors BMAL1, CLOCK, PER1, 2, 3 and CRY1, 2. These proteins establish a stable oscillation period of ~ 24 hr. CLOCK:BMAL1 heterodimers bind to DNA sequences (E-Boxes) and stimulate transcription of output genes that regulate behavior, metabolism and physiology. In addition, CLOCK:BMAL1 activate Per and Cry transcription. PER:CRY complexes form in the cytoplasm and translocate to the nucleus where they repress transcriptional activity of CLOCK:BMAL1. A second feedback loop has also been identified, involving Rev-erbα transcription and repression of BMAL1 transcription. Built-in biochemical delays in the progress of these loops results in 24 hr rhythmicity.

Mutations in clock components or their regulators can affect rhythmicity by altering the free-running period (τ, the period in which behavior continues to oscillate in the absence of external cues) and can result in circadian sleep disorders such as Familial Advanced Sleep Phase Syndrome (FASP) in humans (Lowrey et al., 2000; Xu et al., 2005). In contrast with delayed sleep phase syndrome (extreme night owls), which commonly affects adolescents and improves with age, advanced sleep phase (extreme early birds) is a more specific phenotype Toh et al. (2001) found a large FASPS pedigree with a serine to glycine substitution (S662G) in the (h)PER2 gene, with in vitro data suggesting a potential phosphorylation site by CK1ε (Toh et al., 2001). Another FASPS pedigree was later identified with a CK1δ threonine to alanine (T44A) mutation, an alteration shown to reduce activity of this enzyme in vivo (Xu et al., 2005). Interestingly, although introduction of the mutant T44A CK1δ transgene did shorten τ as expected in mice, it actually lengthens τ in Drosophila, suggesting different regulatory mechanisms across species.

In this issue of Cell, Xu et al. (2007) characterize transgenic mice expressing hPER2 with the S662G mutation. When the hPER2 S662G transgene was introduced in wild type and mPer2 knockout animals, mice had a shorter τ, thus mimicking human FASPS. Furthermore, mice with a serine to aspartate (S662D) mutation in hPER2, which should mimic phosphorylation, had a longer τ. Together with prior studies of FASPS CK1δ T44A transgenic mice, this study validates mice models to functionally test human circadian mutations.

Using the transgenic mice expressing phosphorylation site mutants of hPER2, the authors determined how PER2 phosphorylation at Ser 662 may affect the working of the inner clock. Interestingly, Ser 662 is the first of a series of five closely-spaced serines and phosphorylation of Ser 662 facilitates phosphorylation of the other four residues by CK1. Previous studies had suggested the involvement of both CK1ε and CK1δ in the phosphorylation of PER2. Because of its role in FASPS, the authors therefore explored the role of CK1δ. The authors showed that modulating the levels of wild-type CK1δ did not affect τ in wild type mice. Surprisingly, however, they found that decreased CK1δ amounts increased τ in both hPER2 S662G and S662D transgenic animals. Increased CK1δ amounts in short τ hPER2 S662G transgenic mice made τ even shorter. These experiments indicate that CK1δ does not phosphorylate PER2 at Ser 662 and probably influences the clock by phosphorylation of PER2 at other sites, or by phosphorylation of other circadian proteins. Additionally, because increasing CK1δ activity reduces τ, whereas phosphorylation at S662 prolonged τ, it is suggested that phosphorylation at other PER2 sites leads to opposite effects on PER2 overall activity. This model is supported by in vitro and Drosophila studies suggesting that PER phosphorylation increases its degradation, an effect expected to reduce τ and thus opposite to what would be expected in the hPER2 S662G mutation where the site is lost.

The findings of Xu et al. (2007) are strongly supported by a recent in vitro study, which examined the equivalent serine 662 residue in the mouse PER2 protein (Ser 659) and 20 other phosphorylation sites on the PER protein (Vanselow et al., 2006). Here, increased stability and nuclear retention of PER2 was found when Ser 659 was phosphorylated, whereas phosphorylation at other sites led to opposite effects, i.e. increased degradation. Thus the general concept of a differential effect of various phopshorylation sites on PER activity may explain in vivo versus in vitro discrepancies in the effects of various kinase mutations that regulate the clock.

Taken together, these recent studies favor a model where clock timing is strongly regulated by expression, degradation and/or nuclear entry and retention of PER2, and modulated by multiple states of PER2 phosphorylation, some of which are not dependent on CK1ε/δ. The model is still sketchy, as differing effects of the PER2 FASPS mutation were reported on PER2 nuclear entry and degradation based on in vitro studies. Vanselow et al. (2006) propose a model in which phosphorylation at the PER2 FASPS site is critical in mediating PER2 nuclear retention and repression of CLOCK:BMAL1 activity. PER2 degradation is primarily a consequence of increased cytoplasmic localization of PER2 protein that is phosphorylated at CK1-dependant sites distant from Ser 662, and therefore undergoes proteasomal degradation more quickly. Thus in FASPS, PER2:CRY complexes do not stay in the nucleus sufficiently long, thereby shortening τ. In the model proposed here by Xu et al. (2007), less PER2 is produced through an indirect effect of the FASPS mutation on PER2 expression. The normal rise and fall of PER2 thus reaches critically lower levels earlier in the cycle, therefore shortening τ. The models thus agree upon the effects of distant phosphorylated residues (increased PER2 degradation) whereas they differ on the roles of S662 and proximal residues (higher PER2 levels through increased nuclear retention versus increased mRNA transcription). Why would reduced nuclear retention of PER2 and its shorter half-life in FASPS lead to a decrease in PER2 mRNA expression? There is evidence that CRY can stabilize a CLOCK:BMAL1 complex that is inactive (leading to lower transcription in the face of elevated CLOCK:BMAL1 levels) (Kondratov et al., 2006). Is the S662G PER2 mutation perhaps facilitating this action of CRY to decrease transcription of PER2? Another possibility is that CLOCK:BMAL1 transcription (of PER2) is likely coupled with proteosome degradation (Kwon et al., 2006) as has been shown previously for nuclear receptors. If the S662G PER2 mutation interfered with the proteosome-mediated turnover of the CLOCK:BMAL1 activation complex, this could then lead to a decrease in PER2 transcription in FASPS.

The complex interactions observed here highlight the importance of understanding both transcriptional and post-transcriptional regulation of clock proteins. We expect that additional human circadian mutants will be revealed in the near future, helping us understand the inner working of the mammalian clock through careful studies in animal models.

Figure 1. The primary negative feedback loop of the circadian clock.

Clock/BMAL1 activate Per and Cry transcription. PER and CRY form a complex in the cytoplasm. Phosphorylation of PER by CK1ε/δ mediates PER degradation by the proteosome. Phosphorylation of PER2 on S662 (by an unidentified kinase) stabilizes PER2 potentially by causing its nuclear translocation or increased transcription. CRY and CK1 are also translocated into the nucleus. PER/CRY complexes inhibit the activity of CLOCK:BMAL1 as part of a negative feedback loop.