How circadian clocks keep time: the discovery of slowness

Circadian clocks are biological timing systems that synchronize physiology and behavior with the daily light/dark cycle. Circadian timekeeping is based on cell-autonomous feedback loops that oscillate with a specific endogenous period under constant conditions. Recurrent environmental stimuli such as light and temperature synchronize (entrain) cell-based circadian oscillators under physiological conditions. In mammals, light signals from the retina are transduced to the suprachiasmatic nucleus in the hypothalamus, which consists of a network of tightly coupled robustly oscillating clock neurons. These signals are in turn relayed to cellular clocks throughout the body. Most of the components of cellular circadian clocks are known, and the major interactions driving the feedback loops have been identified. However, the principles by which the clock components accurately measure time at the molecular level are only partially understood. In this review, we focus on the molecular mechanisms identified thus far that contribute to slow biochemical processes underlying the measurement of time on a 24-hour scale in the circadian clocks of the cyanobacterium Synechococcus elongatus , the filamentous fungus Neurospora crassa , and the metazoans Drosophila melanogaster , Mus musculus and humans.


Circadian inhibitors
Intrinsically disordered with multiple phosphorylation sites, dimerization via coiled-coils or PAS domains Circadian clocks are biological timing systems that synchronize physiology and behavior with the daily light/dark cycle. Circadian timekeeping is based on cell-autonomous feedback loops that oscillate with a specific endogenous period under constant conditions. Recurrent environmental stimuli such as light and temperature synchronize (entrain) cell-based circadian oscillators under physiological conditions. In mammals, light signals from the retina are transduced to the suprachiasmatic nucleus in the hypothalamus, which consists of a network of tightly coupled robustly oscillating clock neurons. These signals are in turn relayed to cellular clocks throughout the body. Most of the components of cellular circadian clocks are known, and the major interactions driving the feedback loops have been identified. However, the principles by which the clock components accurately measure time at the molecular level are only partially understood. In this review, we focus on the molecular mechanisms identified thus far that contribute to slow biochemical processes underlying the measurement of time on a 24-hour scale in the circadian clocks of the cyanobacterium Synechococcus elongatus, the filamentous fungus Neurospora crassa, and the metazoans Drosophila melanogaster, Mus musculus and humans.

Prokaryotic circadian clocks
Synechococcus elongatus has the most well studied prokaryotic circadian clock. It is built on a transcriptional-translational feedback loop (TTFL) that controls the cyclical expression of its core clock proteins KaiA, KaiB, and KaiC (left column). The AAA+ hexamer KaiC is the central enzymatic driver of the cyanobacterial clock, executing a cycle of slow autophosphorylation and dephosphorylation at two sites in its CII domain over the course of the day; KaiC phosphorylation is stimulated by KaiA during the day and inhibited by KaiB at night [1]. Formation of the nighttime KaiABC repressive complex is restricted by the phosphorylation state of the CII domain [2], extraordinarily slow ATP hydrolysis at the KaiC CI domain (only ~15 ATP/day due to a non-ideal position of a water molecule in the active site [3]) and the slow metamorphosis of KaiB from its inactive ground-state fold to a rare fold-switch form (fsKaiB) that is competent to bind KaiC [4]. The rate of CI ATP hydrolysis controls pacemaker timing through an ordered series of conformational changes in KaiC [5] that allow binding of fsKaiB and sequestration of KaiA [4,6], allowing the slow autodephosphorylation of KaiC to begin the cycle anew. Remarkably, this pacemaker can function in the absence of transcription in vivo [7] and can be fully reconstituted with purified proteins in vitro [8,9], facilitating a deep mechanistic understanding of this circadian clock.

Eukaryotic circadian clocks
The core circadian clocks of fungi and animals rely on interlocked TTFLs. The heterodimeric circadian transcriptional factors (TFs) WC1-WC2 (WCC), CLK-CYC and CLOCK-BMAL1 rhythmically control the expression of their inhibitors FRQ, TIM and PER, as well as PER1/2/3 and CRY1/2 in Neurospora, Drosophila and mammals, respectively. The association of FRQ with FRH, Drosophila PER with TIM, and mammalian PER1/2 with CRY1/2 stabilizes the complexes [10][11][12]. Casein kinase 1 (CK1) is conserved in eukaryotic circadian clocks and anchored to FRQ and PERs. These large molecular complexes accumulate in the nucleus and inhibit their TFs through a phosphorylation-induced release from DNA [13][14][15]. In mammals, CRY association can also directly inhibit CLOCK-BMAL1 [16,17]. Circadian pacemaking appears to be associated with highly tuned protein interactions [18], and the slow, progressive hyperphosphorylation of FRQ and PERs [19], followed by inactivation of repressive complexes and their degradation to start a new circadian cycle. Additional interlocked feedback loops associated with the core oscillator link the circadian pacemaker to cellular metabolism [20]. These loops are critical for robust circadian function under physiological conditions and are not discussed here.

Activators of circadian transcription
CLK-CYC from Drosophila and CLOCK-BMAL1 from mouse are conserved orthologs that share a DNA-binding domain coupled to tandem PAS domains [21]. This structural architecture is conserved in the evolutionarily unrelated WC1-WC2 complex (WCC) from Neurospora [22]. PAS domains are sensory and ligand binding domains found in all kingdoms of life [23], so it is tempting to speculate that these PAS domain-containing TFs may represent relics of an ancient signal transduction machinery for diurnally recurring ligands.

Inhibitors of circadian transcription
Measuring time on a circadian scale requires slow kinetics in crucial biochemical steps. Phosphorylation of clock proteins typically occurs at a slow rate over a large number of phosphorylation sites. FRQ or PER both contain long intrinsically disordered regions (IDRs) with many potential phosphorylation sites that are targets for progressive phosphorylation [19,24]. Therefore, FRQ and PER, together with their major kinase, CK1, appear to play a major role in circadian timekeeping. Other biochemical steps control the pacemaker, such as how tightly CRY1/2 bind to CLOCK-BMAL1 [18].

Casein kinase 1 and the slowness of circadian phosphorylation
The kinase domain of CK1 has several conserved anion binding sites that facilitate its activity on pre-phosphorylated (primed) substrates [25]. However, the main feature that makes CK1 suitable for circadian function is its ability to slowly phosphorylate unprimed sites with low affinity [19,26]. Phosphorylation of such low-affinity sites in FRQ and PER relies on sitespecific anchoring of CK1 to a structurally conserved binding domain (CK1BD), increasing its local concentration and allowing the kinase to come into contact with sites throughout the protein via dynamic looping of the IDRs [19]. The regulation of progressive phosphorylation by substrate-looping may underlie some of the slowness and temporal precision of the circadian pacemaker. In mammalian PER1/2, phosphorylation of a series of sites in the FASP (Familial Advanced Sleep Phase) region [27], located within the CK1BD, constitutes a phosphoswitch that further slows CK1 phosphorylation, including at a degron for ß-TrCP-mediated proteasomal degradation [25,28]. A similar mechanism has been described for Drosophila PER [29] and proposed for Neurospora FRQ [19].
The IDRs of FRQ and PER contain several motifs whose function may be sensitive to phosphorylation, including NLSs, NESs, degrons, and intra-or intermolecular interaction motifs. Therefore, phosphorylation of a large number of sites in different regions of the protein may be functionally redundant, progressively weakening and eventually inactivating and degrading circadian repressor complexes through various mechanisms and pathways.