The Mammalian Circadian Timing System: The daily rhythms of genes, cells, and organs
In mammals, most vital processes are subject to circadian variations. Thus sleep-wake cycles, locomotor activity, heartbeat, blood pressure, renal plasma flow, body temperature, sensorial perception, and the secretion of many hormones fluctuate during the day in an orderly fashion. The daily timing of these physiological parameters persists under constant conditions and must therefore be controlled by one or more circadian pacemakers.
The molecular circadian oscillator
Owing to genomic, genetic, and biochemical approaches, several essential mammalian clock genes could be isolated and studied during the past few years. The protein products of these genes can be grouped into feedback loops consisting of positive and a negative limb (for review see S. M. Reppert, D. R. Weaver, Nature 418, 935-941, 2002 ). The PAS domain helix-loop-helix transcription factors BMAL1 and CLOCK are the principal constituents of the positive limb, whereas the proteins PER1, PER2, CRY1 and CRY2 are the major components of the negative limb. In addition, protein kinases, including CK1ε and CK1γ, appear to play important roles in modulating the activities and/or stabilities of positive and negative limb components. Briefly, BMAL1 and CLOCK activate the transcription of Per and Cry genes. PER and CRY protein for multi-subunit complexes, and once these complexes have reached a critical threshold concentration, they annul the activity of BMAL1 and CLOCK. As a consequence, the levels of PERs and CRYs falls below the threshold concentration required for autorepression, and a new 24-hour cycle of Per and Cry expression can initiate. The accumulation of Bmal1 mRNA and, to a lesser extent, of Clock mRNA is also circadian, but the phase of cyclic Clock and Bmal1 expression is nearly opposite to that of Per and Cry mRNA accumulation. In fact, PER and CRY proteins activate rather than repress Bmal1 and Clock transcription, and thereby generate the antiphasic expression of the positive and negative limb members. We have identified the orphan nuclear receptor REV-ERBα as the molecular link that couples the two interconnected feedback loops ( Preitner et al., 2002 ).
Circadian clocks everywhere....
In mammals, the master circadian clock resides in the suprachiasmatic nucleus (SCN) at the base of the hypothalamus. As indicated by the word "circadian" (from Latin circa diem), this clock can tell time only approximately, and it must be readjusted every day by light-dark cycles. This resetting is accomplished via the retino-hypothalamic tract, which transmits light information from the retina directly to SCN neurons.
Circadian pacemakers were originally believed to exist only in a few specialized cell types, such as SCN neurons. However, in recent years, circadian clocks have been found in most cell types and even in immortalized cell lines (see Figure 1).
Time-lapse microscopy of individual fibroblasts expressing a yellow fluorescent protein under the control of the Rev-erbα locus (Fig. 1 B) and mathematical modelling of real-time luminescence recordings of cell populations expressing fire fly luciferase in a circadian fashion clearly indicate that fibroblast clocks are self-sustained and cell-autonomous. In fact, these oscillators even keep ticking during cell division ( Nagoshi et al., 2004 ). Hence, the circadian oscillators of cultured fibroblasts appear to be as robust as the ones operative in SCN neurones.
Synchronization of peripheral clocks
Transcriptome profiling approaches by several laboratories, including ours, have indicated that a large fraction of circadian liver transcripts encode proteins involved in food processing and detoxification ( Kornmann et al., 2001; R. A. Akhtar et al., Curr Biol 12, 540-550, 2002; S. Panda et al., Cell 109, 307-320, 2002; K. F. Storch et al., Nature 417, 78-83, 2002; Gachon, F and Schibler, U. unpublished data). We thus wondered whether the adaptation to and the anticipation of feeding might be a major purpose of circadian liver gene expression. Indeed, our restricted feeding experiment show that feeding time is the dominant Zeitgeber for peripheral oscillators ( Damiola et al., 2000 ). Thus, the phase of the molecular circadian oscillators in many peripheral tissues, including liver, kidney, heart, and pancreas, is completely inverted in mice and rats fed exclusively during the light phase for several consecutive days. When fed ad libitum, these nocturnal rodents ingest most of their food during the night. Interestingly, feeding time does not affect the central pacemaker in the SCN. Hence, feeding time can completely uncouple the peripheral clocks from the central SCN timekeeper. Conceivably, the SCN entrains peripheral oscillators mostly through imposing rest/activity cycles and thus feeding time. More direct signaling pathways controlled by the SCN (e.g. the cyclic secretion of hormones, Balsalobre et al., 2000 ) and body temperature cycles ( Brown et al., 2002 ) may also contribute to the phase setting of peripheral oscillators (see Figure 2).
In order to examine molecular signaling pathways involved in the phase resetting of peripheral clocks in intact animals, we recently established a mouse model in which hepatocyte clocks can be switched off an on at will, depending on whether doxycycline (dox) is added to the food or drinking water ( Kornmann et al, 2007 ) The system is based on the inducible expression of REV-ERBα from a doxycycline-dependent transgene. In the absence of dox, REV-ERBα accumulates and represses the essential clock gene Bmal1. As a consequence, hepatocyte clocks are arrested and the temporal expression of clock-controlled downstream genes becomes flat. Genome-wide circadian transcriptome analysis in the presence and absence of dox revealed about 350 oscillator-dependent and about 30 oscillator-independent cyclically expressed genes (see Figure 3). The liver genes that continue to oscillate in the absence of local clockworks are likely to be driven by cyclic SCN output cues. Surprisingly, mPer2 was among these systemically-driven genes. As a component whose rhythmic accumulation is controlled by both systemic cues and local oscillators, mPER2 is a strong candidate for a regulator involved in the synchronization of peripheral clocks. This conjecture has recently gained support from at least two lines of evidence. First, mPER2 may be involved in the phase entrainment by body temperature rhythms (Camille Saini, Hans Reinke, and U.S., unpublished data), and the temperature-dependent transcription factor HSF1 may play a role in this process ( Reinke et al., 2008 ). Second, mPER2 accumulation may also be driven by metabolic cues. Thus, we recently found that SIRT1, a NAD+ dependent protein deacetylase, binds to the CLOCK-BMAL1 heterodimers and deacetylates mPER2 in a cyclic manner ( Asher et al 2008 ). Since acetylated mPER2 is more stable than deacetylated mPER2, rhythmic deacetylation is likely to contribute to the circadian accumulation of mPER2. It is thus conceivable that cyclic NAD+/NADH ratios, which are manifestations of metabolic cycles, are sensed by SIRT1 and translated into rhythmic mPER2 protein expression.
PAR bZIP proteins: a family of circadian output regulators
In 1990, we discovered by serendipity that the expression of DBP, a novel bZip transcription factor, oscillates with a daily amplitude of more than 100-fold in liver and other tissues ( Wuarin and Schibler, 1990 ). DBP has two close relatives, HLF and TEF, and these three proteins form a small family of transcription factors, dubbed the PAR bZip protein family. Similar to DBP, HLF and TEF also accumulate according to a robust circadian cycle in various tissues ( Fonjallaz et al, 1996; Falvey et al, 1995 ). At least in the case of Dbp, the circadian expression is directly hardwired to the molecular circadian oscillator (see Ripperger et al., 2000 ). To examine the function of these proteins, we have performed genetic loss-of-function studies in transgenic mice. These experiments have established that Dbp, Tef, and Hlf are clock-controlled genes, rather than essential components of the circadian clock itself. We used a multitude of molecular, behavioral, and physiological approaches to examine the phenotypes of PAR bZip knockout mice and found that PAR bZip transcription factors play important roles in controlling sleep parameters ( Franken et al., 2000 ), susceptibility to epileptic seizures ( Gachon et al., 2004 ), adaptation to jet lag, male aggressiveness, mother instinct, and liver metabolism ( Gachon et al., 2006 ) (See Figure 4).