Chromosoma (2004) 113: 103–112

DOI 10.1007/s00412-004-0296-2

REVIEW

Frédéric Gachon . Emi Nagoshi . Steven A. Brown .

The mammalian circadian timing system: from gene expression

Received: 5 May 2004 / Revised: 14 May 2004 / Accepted: 16 May 2004 / Published online: 3 August 2004
# Springer-Verlag 2004

Abstract Many physiological processes in organisms (meaning about a day). Rhythms with substantially shorter
from bacteria to man are rhythmic, and some of these and longer period lengths are called ultradian and
are controlled by self-sustained oscillators that persist in infradian, respectively. This classification is somewhat
the absence of external time cues. Circadian clocks are arbitrary, in that the period length (τ) of ultradian rhythms
perhaps the best characterized biological oscillators and can range from fractions of seconds to about 20 h, and that
they exist in virtually all light-sensitive organisms. In of infradian rhythms from about 30 h to decades.
mammals, they influence nearly all aspects of physiology Examples of ultradian rhythms are heartbeat frequencies
and behavior, including sleep-wake cycles, cardiovascular (τ≈1 s in man), respiratory oscillations in yeast (τ≈40 min,
activity, endocrinology, body temperature, renal activity, Murray et al. 2001), somite deposition during vertebrate
physiology of the gastro-intestinal tract, and hepatic embryogenesis (τ≈90 min, Pourquie 2003), and foraging
metabolism. The master pacemaker is located in the rhythms of the common vole Microtus arvalis (τ≈2.5 h,
suprachiasmatic nuclei, two small groups of neurons in the Gerkema and van der Leest 1991). Infradian rhythms
ventral part of the hypothalamus. However, most periph- include female estrus cycles (τ = days to months,
eral body cells contain self-sustained circadian oscillators depending on the species), circannual mating cycles (τ≈1
with a molecular makeup similar to that of SCN year), and emergence cycles of some cicada (τ≈13–17
(suprachiasmatic nucleus) neurons. This organization years (Hoppensteadt and Keller 1976).
implies that the SCN must synchronize countless sub- The sole observation of rhythmic behavior or physiol-
sidiary oscillators in peripheral tissues, in order to ogy does not necessarily indicate that a biological clock is
coordinate cyclic physiology. In this review, we will involved. For example, the circannual sexual cycle of
discuss some recent studies on the structure and putative hamsters, which in males manifests itself by a 20-fold
functions of the mammalian circadian timing system, but oscillation of testicle weight, is controlled by seasonal
we will also point out some apparent inconsistencies in the changes in melatonin secretion (Bartness et al. 1993). In
currently publicized model for rhythm generation. turn, the variations in melatonin secretion are brought
about by seasonal changes in day length (photoperiodism).
Only processes that continue to oscillate in the absence of
What are biological clocks? external time cues (e.g., temporal changes in light intensity
and temperature), are considered to be outputs of biolog-
Most biochemical and physiological processes fluctuate in ical timekeepers. The same cyclic processes can be
a temporal fashion, and some of these do so with a controlled by environmental cues in one species and by
relatively constant period length. Cycles with a period biological clocks in another species. For example, yearly
length (τ) of approximately 24 h are considered to be reproduction cycles are governed by photoperiodism in the
circadian, derived from the Latin words circa diem Siberian hamster and by a true circannual timekeeper in
certain ground squirrels and fruit bats (see Lincoln et al.
2003).
Communicated by E.A. Nigg
A further complication in studying biological clocks is
F. Gachon . E. Nagoshi . S. A. Brown . J. Ripperger . that they are only detectable if synchronized within an
U. Schibler (*) organism or within populations of cells. Thus, if cultures
Department of Molecular Biology, Sciences III, University of of yeast cells (Saccharomyces cerevisiae) are grown in a
Geneva,
30, Quai Ernest Ansermet, chemostat that assures a constant supply of oxygen and
1211 Geneva-4, Switzerland nutrients, oxygen consumption oscillates with a constant
e-mail: [email protected] period length of about 40 min (Murray et al. 2001).
104

Moreover, under such growth conditions the transcripts of insufficient for auto-repression, a new cycle of Per and
most yeast genes fluctuate in concentration with the same Cry transcription can start (Albrecht and Eichele 2003;
40 min period length (Klevecz et al. 2004). This ultradian Reppert and Weaver 2002). Many additional components
respiratory clock probably also functions in yeast cells contribute to the robustness of this molecular clockwork
grown in normal liquid cultures or on agar plates, but it circuitry. For example, the orphan nuclear receptor and
escapes detection under these conditions because the repressor REV-ERBα interconnects circadian transcription
phases of individual cells do not become synchronized. of the positive and negative “limbs” of the oscillator. Rev-
Likewise, cultured mammalian cells harbor robust circa- Erbα transcription is regulated by the same components
dian oscillators (Balsalobre et al. 1998). However, unless that control Per and Cry transcription, and the resulting
the daily oscillations in gene expression generated by circadian accumulation of REV-ERBα leads to periodic
these cell-autonomous timekeepers are recorded in repression of Bmal1 and Clock transcription. In turn, this
individual cells, they are only noticeable after synchroni- leads to a rhythmic expression of Bmal1 and (to a lesser
zation by chemical signals (see below) (Balsalobre et al. extent) Clock mRNA that is antiphasic to Rev-Erbα
1998, 2000b; Nagoshi et al. manuscript in preparation). expression (Preitner et al. 2002).
Posttranslational mechanisms such as protein phosphor-
ylation also play important roles in generating oscillations
Molecular circadian oscillator: model and open of approximately 24 h. For example, casein kinase 1ε
questions (CK1ε), initially identified as an essential Drosophila
clock component (Price et al. 1998), phosphorylates PER,
Circadian oscillators have been genetically and biochemi- CRY, and BMAL1 proteins (Eide et al. 2002; Eide and
cally dissected in several model organisms. In 1990, Virshup 2001; Lee et al. 2004), and hypomorphic Ck1ε
Hardin et al. discovered that the Drosophila period protein mutant alleles (dubbed Tau) cause a dramatic shortening of
is required for the cyclic accumulation of its own mRNA the period length in hamsters (Lowrey et al. 2000). In
and therefore proposed that the period gene is engaged in keeping with this observation, an autosomal dominant
an autoregulatory feedback loop (Hardin et al. 1990). mutation in the human Per2 gene that inactivates a CK1ε
Since then negative feedback loops of clock gene expres- phosphoacceptor site results in familial advanced sleep
sion have been uncovered in all genetic model systems for phase syndrome (FASPS) (Toh et al. 2001). CK1δ, a close
circadian oscillators, including cyanobacteria, neurospora, paralog of CK1ε, has also been found to be associated
plants, and mammals. Excellent and detailed reviews are with PER–CRY complexes and may therefore perform a
available on all of these systems (Albrecht and Eichele similar function as CK1ε (Lee et al. 2001). Hypopho-
2003; Froehlich et al. 2003; Golden 2003; Reppert and sphorylated PER proteins have a higher metabolic stability
Weaver 2002; Roenneberg and Merrow 2003; Staiger than their hyperphosphorylated counterparts, and this may
2002; Stanewsky 2003); here we will concentrate on a
brief discussion of some recent findings about the
mammalian circadian timing system.
The identification of mammalian clock genes has
greatly profited from genetic studies in the fruit fly
Drosophila melanogaster, as most essential Drosophila
clock genes have orthologs in mammals. Remarkably,
however, one essential transcription factor of the circadian
oscillator, dubbed CLOCK (for circadian locomotor output
cycles kaput) was first identified by a forward genetic
approach in the mouse (King et al. 1997).
The cartoon in Fig. 1 displays a simplified version of
the mammalian circadian molecular oscillator. The key
negative components of this genetic circuitry are the four
genes encoding the repressors cryptochrome 1 (Cry1),
cryptochrome 2 (Cry2), period 1 (Per1), and period 2
(Per2). These genes are activated by the two PAS domain
basic helix-loop-helix transcription factors CLOCK and
BMAL1, the key positive components of the circadian
oscillator. PER and CRY proteins form heteropolymeric
Fig. 1 Simplified model of mammalian circadian oscillator. This
complexes of unknown stoichiometry, and once these model explains several biochemical findings and observations made
complexes have reached a critical concentration in the cell in mice or hamsters carrying mutations in clock genes. However, as
nucleus, they interact with the CLOCK-BMAL1 hetero- shown in Fig. 2, it does not explain how PER and CRY proteins
dimer and thereby annul the activation potential of these determine the phase of cyclic Per and Cry mRNA expression.
Phosphorylation by CKIε is believed to render PER proteins less
transcription factors. As a consequence Cry and Per stable (indicated by a repression bar). However, it is also
mRNAs and proteins decrease in concentration, and once conceivable that phosphorylation by CKIε and CKII augments the
the nuclear levels of the CRY–PER complexes are activity of PER–CRY complexes (indicated by arrows, see text)
 105

lead to an increased accumulation of PER proteins. Hence, repressing activity of this protein (Lin et al. 2002;
in both Tau hamsters and human FASPS subjects, Nawathean and Rosbash 2004). Given the sequence
threshold levels of PER complexes required for auto- similarities of mammalian and insect PER and CKII
repression are expected to be reached faster than in the proteins, we consider it likely that mammalian CKII also
corresponding wild-type individuals. As a consequence participates in the modulation of PER activity in the
the period length of the oscillator shortens and its phase mammalian system. Likewise, glycogen synthase kinase 3
becomes advanced. In Drosophila, casein kinase II (CKII) (GSK3) appears to play a role in both the Drosophila and
also phosphorylates PER and thereby enhances the the mammalian circadian oscillator (Harms et al. 2003),

Fig. 2a–c How do CRY and PER proteins determine different by PER–CRY repressor complexes. c Hypothetical model for
phases of circadian gene expression? a Circadian clock gene circadian Per2 transcription. Circadian Per2 transcription must be
expression in mouse liver. The three top panels show an immuno- regulated by a different mechanism than that described for Rev-Erbα
blot of liver nuclear proteins harvested at 4 h intervals around the mRNA expression in b. As shown in a, repression starts only toward
clock with antibodies raised against mCRY1, mCRY2, and mPER2 the end of nuclear PER–CRY accumulation and persists for several
proteins. The bottom panel displays the temporal accumulation of hours after PER–CRY has reached trough levels. This delay could
Rev-Erbα mRNA (solid line), Per2 mRNA (broken line), and Cry1 be explained if hyper-phosphorylation of PER proteins were
mRNA (dotted line) in liver, as measured by TaqMan reverse required for the initiation of repression of Per2 transcription, and
transcribed polymerase chain reaction (RT-PCR). It can be assumed if the repressed chromatin state outlasted the presence of PER–CRY
that the phase of circadian transcription precedes that of circadian complexes. In this purely speculative model it is assumed that
mRNA accumulation by about 2 h. The data are adapted from repression involves the formation of relatively stable heterochro-
Preitner et al. (2002). b Hypothetical model for circadian Rev-Erbα matin domains around the Per2 gene. Histone modifications, such as
transcription. Circadian Rev-Erbα mRNA expression is nearly methylations and phosphorylations, as well as the recruitment of
antiphasic with the circadian accumulation of PER and CRY heterochromatin components, such as HP1, may participate in this
proteins (see panela). Hence, the extent of repression of Rev-Erbα process. The heterochromatin-specific signatures are represented by
mRNA transcription correlates closely with the nuclear concentra- red symbols on the nucleosomes of compacted chromatin. After
tions of PER and CRY proteins, and a mechanism allowing rapidly PER–CRY complexes disappear, the heterochromatin-specific his-
reversible transitions between active and inactive chromatin must be tone modifications and/or macromolecular components are gradu-
operative. Conceivably histone acetylation and deacetylation may be ally removed, and a new round of transcription can ensue. The
involved (Etchegaray et al. 2003), although this has not yet been active state of the chromatin may be associated with histone
demonstrated experimentally for the Rev-Erbα locus. Blue stars acetylation (and other histone modifications), as in the model
represent putative acetyl groups on nucleosomes of the extended displayed in b. Rhythmic Cry1 transcription may be governed by a
chromatin strings. In the repressed chromatin state, these acetyl similar mechanism
groups are assumed to be removed by histone deacetylases recruited
106

and the same may hold true for protein phosphatase 2A In contrast to most biochemical processes, circadian
(Sathyanarayanan et al. 2004). rhythms are temperature-compensated. That is, within the
Biochemical work by McKnight and coworkers on physiological temperature range the period length changes
NPAS2 (a CLOCK paralog in the forebrain) and CLOCK very little if the temperature is increased or decreased.
suggests that circadian rhythms could be influenced Thus Q10—the ratio between the rates measured for (bio)
directly by cellular metabolism (Rutter et al. 2001). At chemical reactions at two temperatures differing by 10°C
least in vitro, the dimerization of NPAS2 or CLOCK with within the physiological range—is nearly 1 for the
BMAL1 and/or the binding of the resulting heterodimers circadian period length (Dunlap 1999). This temperature
to their DNA recognition sequences (E-boxes) are compensation may be important for poikilothermic organ-
dramatically modulated by the ratio of reduced to oxidized isms, such as fish, amphibians and reptiles, in order to
nicotinamide adenine dinucleotides. High [NAD(P)H/ anticipate daytime irrespective of ambient temperature.
NAD(P)+] ratios facilitate the occupancy of E-boxes by Surprisingly, however, two recent reports have demon-
NPAS2–BMAL1 or CLOCK–BMAL1 heterodimers while strated that even mammalian circadian oscillators are
low [NAD(P)H/NAD(P)+] ratios inhibit this process. This temperature-compensated (Izumo et al. 2003; Tsuchiya et
opens the exciting possibility that circadian oscillators can al. 2003). The molecular basis for temperature compen-
adapt their phase to the nutrient state of the cell (see sation remains elusive, but it is conceivable that synthesis
below). It will be interesting to examine whether and degradation rates of clock gene products are
mutations in the NAD binding sites of NPAS2 or modulated in the same direction by increasing or
CLOCK will affect the function of these transcription decreasing temperature. Moreover, temperature-dependent
factors in circadian rhythm generation. changes in the accumulation of clock components may be
Although impressive progress has been made in the compensated by temperature-dependent interactions be-
genetic and biochemical dissection of mammalian circa- tween them. For example, at low temperatures the
dian oscillators, the widely advertised oscillator circuitry synthesis of Cry and Per mRNAs and proteins may be
presented in Fig. 1 should be regarded as a working reduced, but due to a higher affinity between CRY and
hypothesis. In fact, many important issues remain PER proteins under these conditions, fewer CRY–PER
unanswered or are even at odds with this oversimplified complexes might be required for repression.
scheme. For example, the model posits that the cyclic
transcription of Rev-Erbα, Cry, and Per is governed by the
same mechanism. Yet the phase of circadian Rev-Erbα Master and slave circadian oscillators
mRNA accumulation differs from that of circadian Per2
and Cry1 mRNA accumulation by about 9 h and 11 h, The suprachiasmatic nucleus (SCN) was positively
respectively (Fig. 2a). The accumulation of PER and CRY identified as the master circadian pacemaker in mammals
proteins is antiphasic with that of Rev-Erbα transcription more than a decade ago by an elegant lesion and
but almost in phase with Per and Cry transcription. Thus, transplantation experiment. Surgical SCN lesions render
while the trough in Rev-Erbα expression can readily be laboratory animals completely arrhythmic with regard to
explained by the abundance of PER–CRY complexes, wheel-running activity. Remarkably the rhythms can be
more complicated mechanisms must be operative in CRY- restored by the transplantation of fetal SCN tissue, and the
mediated and PER-mediated auto-repression. Because period length of the rescued rhythmicity is determined by
PER phosphorylation is considerably retarded with regard the SCN implant of the donor (Ralph et al. 1990; Silver et
to PER protein accumulation, this apparent conflict could al. 1996). Recently, circadian expression of a luciferase
be resolved if only CRY–PER complexes containing reporter gene under the control of the mouse Per1
hyperphosphorylated PER proteins were capable of auto- promoter has been recorded in real time in individual
repression, and if auto-repression were associated with neurons of SCN explants kept in tissue culture. This
changes in chromatin structure that outlast the presence of elegant experiment has shown that most neurons contain
PER and CRY proteins. This hypothesis is supported to circadian oscillators (Yamaguchi et al. 2003). A similar
some extent by the observation that histone acetylation conclusion has been reached by recording circadian firing
around the Per1 and Per2 promoters is rhythmic and frequencies in individual hamster SCN neurons obtained
parallels the transcription rates of these genes in mouse by the enzymatic dissociation of SCN tissue. While all
liver nuclei (Etchegaray et al. 2003). A hypothetical model neurons show circadian fluctuations in electrical activity,
of how PER–CRY may generate different phases of target the periods vary dramatically between different cells. Yet
gene expression is displayed in Fig. 2b,c. If PER and CRY the average period length corresponds to the period length
regulate the cyclic transcription of Rev-Erbα and Per2 (or of the locomotor activity of the donor, a clock output
Cry1) indeed by different mechanisms, a signature directly regulated by the SCN (Liu et al. 1997). Hence, in
determining which mechanism is used must exist in each vivo the SCN neurons must be coupled to synchronize
of the PER/CRY target genes. These signatures may be their molecular oscillators. This is probably accomplished
binding sites of yet unknown transcription factors or by both synaptic communications and paracrine mechan-
epigenetic marks on DNA (e.g., CpG methylations) or isms. Gene knockout studies in mice suggest that the cell
histones (posttranslational modifications). adhesion molecule NCAM-180 and the VIP/PACAP
receptor VPAC2 may be involved in the paracrine
 107

coordination of circadian SCN gene expression and important timing cues for peripheral clocks (see below),
outputs (Harmar et al. 2002; Shen et al. 1997). they probably contribute to the synchronization of these
Although many unicellular organisms have circadian timekeepers.
timekeepers, it came as a surprise that vertebrate tissue Until a few months ago, it was thought that peripheral
explants and even immortalized cell lines harbor func- clocks dampen after a few cycles, and that only oscillators
tional circadian clocks (Balsalobre et al. 1998; Yamazaki in SCN neurons were self-sustained. However, in an
et al. 2000). The oscillators in cultured cells can readily be elegant study Takahashi and his colleagues have recently
observed after a short treatment with high concentrations shown that circadian timekeepers in liver and lung
of serum or a wide variety of chemicals that induce known explants can generate up to 20 (or more) daily cycles of
signaling pathways, including those involving the activa- Per2-luciferase expression (Yoo et al. 2004). A similar
tion of protein kinase A (PKA), protein kinase C (PKC), number of daily cycles of luciferase activity could also be
mitogen activated protein kinase (MAPK), glucocorticoid recorded in serum-induced NIH3T3 fibroblasts kept in
receptor (GR), endothelin, glucose, and retinoic acid tissue culture when the Bmal1 promoter drives circadian
receptor (RAR) (Akashi and Nishida 2000; Balsalobre et luciferase expression (Nagoshi et al. manuscript in prep-
al. 2000a,b; Hirota et al. 2002; McNamara et al. 2001; aration). Hence, there is little reason to believe that SCN
Yagita and Okamura 2000; Yagita et al. 2001). Circadian neurons possess more robust clockwork than peripheral
gene expression in tissue culture cells can also be induced cell types or fibroblasts in tissue culture. In spite of this
by square-wave temperature cycles (e.g., by 12 h resemblance between central and peripheral clocks, the
33°C/12 h 37°C). Moreover, body temperature oscilla- terms of master and slave oscillators are still justified.
tions, recorded by telemetry in the intraperitoneal cavity of Thus, while circadian gene expression persists in periph-
mice and simulated in the cell culture medium by a eral organs of SCN-lesioned mice, their phases are no
computer-directed incubator, can sustain previously in- longer coordinated, neither between the same organs in
duced cyclic gene expression (Brown et al. 2002). Thus, different individuals nor between different organs in the
although body temperature rhythms are not the most same individual (Yoo et al. 2004). Therefore, although

Fig. 3a, b Synchronization of the mammalian timing system. a The windows during the day. Feeding and/or fasting are the most
mammalian circadian timing system is composed of countless self- dominant Zeitgebers (timing cues) for subsidiary clocks in periph-
sustained oscillators that are operative in both suprachiasmatic eral tissues. However, the SCN also uses more direct timing cues,
nucleus (SCN) neurons and peripheral cell types. The central such as humoral and neuronal signals, to entrain the phases of
pacemaker in the SCN is synchronized every day by the photoperiod peripheral clocks (see text). b Peripheral clocks continue to oscillate
via visual and non-visual photoreceptor cells in the retina (see text). in SCN-lesioned mice, but their phases are no longer coordinated in
Rhythmic neuronal and hormonal SCN outputs drive circadian rest– these behaviorally arrhythmic animals (Yoo et al. 2004)
activity cycles, which in turn limit feeding time to certain time
108

circadian clocks in the periphery keep ticking in the is one of these pathways (Le Minh et al. 2001). For
absence of an SCN, they must be synchronized by the example, circadian corticosterone production/secretion is
SCN pacemaker (Fig. 3). controlled by the SCN, and the GR agonist dexamethasone
is a potent phase shifting agent for peripheral clocks in
intact animals and tissue culture cells, but not in the SCN
Phase entrainment of central and peripheral oscillators itself. Moreover, daytime feeding inverts the phase of
circadian liver gene expression much more rapidly in mice
As implied by their name, circadian clocks can measure with a liver-specific glucocorticoid receptor gene knock-
the day length only approximately. For example, in out, while the oscillators in other, GR-proficient tissues of
constant darkness the period length of most laboratory the same animals show slow phase inversion kinetics.
mice is slightly below 24 h. Therefore, the circadian Finally, all tissues of adrenalectomized mice display rapid
timing system must be corrected every day by a few phase inversion kinetics upon daytime feeding. Given the
minutes, in order to stay in harmony with geophysical multitude of additional signaling pathways that can affect
time. The photoperiod is the key synchronizer of the circadian oscillators in tissue culture cells (see above), it is
circadian clock in all well-studied systems. In mammals, likely that the phase entrainment of peripheral clocks by
photic inputs are perceived by the retina and transmitted in hormones, paracrine/autocrine factors, and body temper-
the form of electrical signals directly to SCN neurons via ature rhythms is complex and redundant. Hence the
the retino-hypothalamic tract. Neuronal signaling at the detailed genetic and biochemical dissection of the
synapses between optic nerve endings and SCN neurons mechanisms involved in the synchronization of peripheral
involves primarily the neurotransmitters glutamate and clocks will be a major challenge in the field.
PACAP (Hannibal 2002). Recent studies have added an
unexpected facet to the photic synchronization of
mammalian circadian clocks. Even visually blind mice, The benefits of circadian clocks
whose retinas lack all classic cone and rod photoreceptor
cells of the outer layer, can still efficiently phase entrain If cycles in physiology and behavior can be controlled by
their circadian pacemaker through light-sensitive ganglion environmental time cues, what benefits do biological
cells in the inner layer (Berson 2003; Freedman et al. clocks offer to their owners? A key difference between
1999). These neurons contain the photopigment melanop- cycles that are driven by external time cues and biological
sin, and their axons project directly to SCN neurons. Only timekeepers is that only the latter can anticipate environ-
mice that are deficient in both classic photoreceptors and mental changes. For example, the circadian clock of plants
melanopsin are incapable of adjusting their clock to light– allows them to produce photo-system I and II components
dark cycles (Hattar et al. 2003; Panda et al. 2003). These already before sunrise, so that photosynthesis can
mice are still rhythmic, but are incapable of phase-shifting commence as soon as light energy is available (Harmer
their clock in response to light. et al. 2000). On a similar note, a nocturnal rodent
Because the phases of peripheral circadian clocks possessing a circadian timekeeper can anticipate dusk in
depend on output signals of the SCN, it is sufficient to his underground habitat and does not have to forage
entrain the phase of the latter to adapt the entire timing periodically to examine whether sunset is approaching.
system to the physiological needs of the organism. Such anticipation may considerably reduce exposure to
Feeding time is the major Zeitgeber (timing cue) that day-active predators and thus provide a selective advan-
entrains the phase of peripheral oscillators. Thus, feeding tage to an animal possessing a circadian timing system.
nocturnal rodents such as mice or rats exclusively during The temporal separation of chemically incompatible
the day for a week or longer gradually inverses the phase reactions is another important function of biological
of circadian gene expression in liver, pancreas, heart, timing systems. Cyanobacteria (Synechococcus elonga-
skeletal muscle, kidney, and lung (Damiola et al. 2000; tus), in which nearly all genes are transcribed in a
Stokkan et al. 2001). Importantly, feeding time has little circadian fashion, can perform both nitrogen fixation and
influence on the phase of circadian SCN gene expression, photosynthesis (Berman-Frank et al. 2001). If these
and daytime feeding thus completely uncouples the phases processes were conducted simultaneously, the oxygen
of central and peripheral clocks (Damiola et al. 2000; generated by photosynthesis would poison the nitroge-
Stokkan et al. 2001). However, as soon as food is offered nase, and nitrogen fixation would be inefficient. The
ad libitum—even after an extended period of daytime circadian clock solves this problem by separating the
feeding—the SCN rapidly reestablishes its dominance and phases of photosynthesis and nitrogen fixation. Likewise,
resynchronizes peripheral clocks to the normal, nocturnal circadian oscillators in mammalian liver cells may help to
mode (Le Minh et al. 2001). Since the feeding time-driven separate glycogen synthesis and degradation in time,
uncoupling of peripheral oscillators from the SCN pace- limiting the former to the absorptive phase and the latter to
maker takes much longer (7–10 days) than their re- the postabsorptive phase (Ishikawa and Shimazu 1976).
synchronization to the normal mode, one has to postulate Some important biochemical reactions can produce
that the SCN can also synchronize peripheral clocks via harmful side products, and it would be advantageous to
more direct routes than imposing feeding time. Several limit such activities to a time window during which they
observations strongly suggest that glucocorticoid signaling are needed. For example, cytochrome p450 enzymes
 109

involved in the hepatic detoxification of xenobiotic by a transcriptional cascade (Schibler et al. 2003). A recent
substances can generate reactive oxygen species (Bondy report suggests that posttranscriptional mechanisms may
and Naderi 1994). This offers perhaps a rational explana- also be used in establishing cyclic protein accumulation
tion for the circadian expression of many cytochrome (Baggs and Green 2003). Nocturnin, a mRNA dead-
p450 genes, with peak expression during the absorptive enylase, accumulates with a robust circadian rhythm in the
phase (Furukawa et al. 1999; Lavery et al. 1999). retina and many peripheral tissues (Wang et al. 2001).
Considerable insight into the possible functions of the Although nocturnin targets have not yet been identified,
circadian timing system has been gained from transcrip- the cyclic poly(A) tail shortening of such mRNAs could
tome profiling studies. For example, in liver the affect both their stability and translation efficiency in a
accumulation of about 5–10% of all mRNAs undergoes circadian manner.
a daily oscillation (Akhtar et al. 2002; Duffield et al. 2002;
Panda et al. 2002; Storch et al. 2002). The identification of
these cyclic transcripts has led to the conclusion that a Conclusions and perspectives
large fraction of circadian genes are involved in the
processing of food components. Thus, many enzymes Since the identification of the first mammalian clock gene
playing important roles in the metabolism of proteins, in 1997 (King et al. 1997), we have witnessed an
lipids, carbohydrates, and xenobiotic substances appear to explosion of genetic and biochemical studies shedding
be synthesized in a rhythmic fashion. If the anticipation of light on the mammalian circadian timing system. Never-
the absorptive phase is indeed a major task of circadian theless, many enigmas persist, and we are far from
gene expression, the finding that feeding time is the understanding the rhythm-generating clockwork circuitry.
dominant Zeitgeber for peripheral clocks makes perfect On the bright side, mathematical modeling has provided
physiological sense, at least for organs such as liver, proof-of-principle that the auto-regulatory feedback loops
pancreas, and the gastrointestinal tract. proposed for the mammalian clock can generate stable
Transcriptome profiling studies have unveiled two other oscillations (Leloup and Goldbeter 2003). However, the
important aspects of the circadian timing system. First, kinetic parameters used in such simulations have not been
different circadian transcripts within one tissue can determined experimentally. While genetics has made and
accumulate with many different phases, and secondly, will make invaluable contributions to the identification of
most circadian genes are expressed in a tissue-specific clock components, we are in urgent need of quantitative
manner. Again, both of these properties are physiologi- biochemical approaches to make further progress. Thus,
cally relevant. As mentioned above, different phases are we will have to purify protein complexes containing the
required for the temporal separation of biochemically core clock components, identify their subunits, determine
incompatible processes (e.g., glycogen synthesis and their stoichiometry in the complex, estimate the synthesis
degradation), and cell type-specific circadian gene expres- and decay rates of the central clock components, and
sion is in keeping with the concept that different tissues determine the equilibrium and rate constants with which
must control the timing of different functions. For they interact. None of these will be easy tasks, but the
example, while our kidney tubules probably play a determination of the rate and equilibrium constants driving
minor role in circadian food metabolism, they adjust the macromolecular interactions in living cells will be
efficiency of filtration and water-reabsorption to our rest– particularly challenging. Cells are not test tubes, and the
activity cycles (Ballauff et al. 1991). Many overseas effective concentration of a macromolecule cannot simply
travelers must have noticed that jet lag not only affects the be obtained by dividing its cellular amount by the cell
timing of our vigilance state, but also that of urine volume. Indeed, the effective cellular concentrations of
production. macromolecules are influenced dramatically by parameters
How do molecular oscillators of the type presented in that are difficult to measure, such as macromolecular
Fig. 1 govern overt rhythms in physiology and behavior? crowding (Ellis 2001) and subcellular compartmentaliza-
The transcription of some genes, like the gene specifying tion.
the neuropeptide arginine vasopressin, is regulated directly The identification of signaling pathways through which
by core components of the oscillator: it is activated by peripheral oscillators are synchronized in the intact
CLOCK–BMAL1 and repressed by CRY–PER complexes organism will be another enticing research area. Feeding
(Jin et al. 1999). However, the expression of other clock time has been shown to have a strong influence on the
output genes, such as the ones encoding pyridoxal kinase phase of subsidiary clocks. However, restricted feeding is
and some cytochrome p450 enzymes, is controlled by the always associated with restricted fasting, and we do not
PAR basic leucine zipper (PAR bZip) transcription factors even know whether it is the former or the latter that
DBP, TEF, and HLF (Lavery et al. 1999; Gachon et al. synchronizes peripheral clocks. Recent observations made
2004). At least Dbp is a direct target gene of clock core in our laboratory are actually more compatible with a
components, similar to arginine vasopressin (Ripperger et scenario in which periodic starvation is the dominant
al. 2000). This more indirect pathway, in which clock- Zeitgeber. Hence, systemic chemical signals released
controlled transcription factors govern the rhythmic during the postabsorptive phase might be better candidates
expression of downstream genes, permits the generation for timing cues than signals released during the absorptive
of a large number of different phases within the same cell phase. Again, the unequivocal identification of such
110

signals will require both biochemical and genetic strate- Akhtar RA, Reddy AB, Maywood ES, Clayton JD, King VM, Smith
gies. AG, Gant TW, Hastings MH, Kyriacou CP (2002) Circadian
cycling of the mouse liver transcriptome, as revealed by cDNA
Last but not least, we wish to emphasize that very little microarray, is driven by the suprachiasmatic nucleus. Curr Biol
is known about the perhaps most important question, 12:540–550
namely whether and how the circadian clock influences Albrecht U, Eichele G (2003) The mammalian circadian clock. Curr
the fitness and longevity of an organism. To the best of our Opin Genet Dev 13:271–277
Baggs JE, Green CB (2003) Nocturnin, a deadenylase in Xenopus
knowledge, there are no published studies comparing the laevis retina: a mechanism for posttranscriptional control of
life expectancy of intact and SCN-lesioned laboratory circadian-related mRNA. Curr Biol 13:189–198
animals in a rigorous fashion. Per2 knockout mice display Ballauff A, Rascher W, Tolle HG, Wember T, Manz F (1991)
a high susceptibility to hyperplasias and gamma-ray- Circadian rhythms of urine osmolality and renal excretion rates
of solutes influencing water metabolism in 21 healthy children.
induced malignant tumors (Fu et al. 2002). However, it Miner Electrolyte Metab 17:377–382
cannot be excluded that PER2 also performs functions that Balsalobre A, Damiola F, Schibler U (1998) A serum shock induces
are not related to circadian rhythm generation, and it circadian gene expression in mammalian tissue culture cells.
remains thus unclear whether the tumor proneness is the Cell 93:929–937
consequence of a defect circadian oscillator. Serious health Balsalobre A, Brown SA, Marcacci L, Tronche F, Kellendonk C,
Reichardt HM, Schutz G, Schibler U (2000a) Resetting of
problems have also been associated with clock perturba- circadian time in peripheral tissues by glucocorticoid signaling.
tions in man. For example, nurses performing rotating Science 289:2344–2347
night shift work have been reported to develop breast Balsalobre A, Marcacci L, Schibler U (2000b) Multiple signaling
cancer with an abnormally high frequency (Schernhammer pathways elicit circadian gene expression in cultured Rat-1
fibroblasts. Curr Biol 10:1291–1294
et al. 2001). However, these studies will have to be Bartness TJ, Powers JB, Hastings MH, Bittman EL, Goldman BD
confirmed by examining larger cohorts. For the moment, (1993) The timed infusion paradigm for melatonin delivery:
jet lag, advanced or delayed sleep phase syndromes, sleep what has it taught us about the melatonin signal, its reception,
disturbances, and depressions are still the best-known and the photoperiodic control of seasonal responses? J Pineal
Res 15:161–190
medical consequences of circadian clock perturbations. Berman-Frank I, Lundgren P, Chen YB, Kupper H, Kolber Z,
All in all, we must admit that the arguments developed Bergman B, Falkowski P (2001) Segregation of nitrogen
above about how circadian clocks may improve the fitness fixation and oxygenic photosynthesis in the marine cyanobac-
of a mammalian organism are still based on educated terium Trichodesmium. Science 294:1534–1537
guesses. Therefore, we are in urgent need of experimental Berson DM (2003) Strange vision: ganglion cells as circadian
photoreceptors. Trends Neurosci 26:314–320
approaches that rigorously scrutinize the benefits of Bondy SC, Naderi S (1994) Contribution of hepatic cytochrome
circadian clocks. Elegant studies with cyanobacteria have P450 systems to the generation of reactive oxygen species.
shown that such approaches exist, and that the discrim- Biochem Pharmacol 48:155–159
ination between the effects of clock gene mutations on Brown SA, Zumbrunn G, Fleury-Olela F, Preitner N, Schibler U
(2002) Rhythms of mammalian body temperature can sustain
clock-related and non-clock-related functions is possible peripheral circadian clocks. Curr Biol 12:1574–1583
(Ouyang et al. 1998). In these experiments, long and short Damiola F, Le Minh N, Preitner N, Kornmann B, Fleury-Olela F,
period mutations in the essential circadian clock gene kaiC Schibler U (2000) Restricted feeding uncouples circadian
provide a strong selective advantage to cyanobacteria oscillators in peripheral tissues from the central pacemaker in
the suprachiasmatic nucleus. Genes Dev 14:2950–2961
when these are co-cultured in long and short light–dark Duffield GE, Best JD, Meurers BH, Bittner A, Loros JJ, Dunlap JC
cycles, respectively. As the same mutations can have (2002) Circadian programs of transcriptional activation,
opposite effects under different circadian growth condi- signaling, and protein turnover revealed by microarray analysis
tions, it must be the resonance of the clock with the of mammalian cells. Curr Biol 12:551–557
photoperiod that improves fitness. As mutations affecting Dunlap JC (1999) Molecular bases for circadian clocks. Cell
96:271–290
circadian period length are also available for hamsters and Eide EJ, Virshup DM (2001) Casein kinase I: another cog in the
mice, such experiments should, in principle, also be circadian clockworks. Chronobiol Int 18:389–398
feasible for mammalian organisms. Eide EJ, Vielhaber EL, Hinz WA, Virshup DM (2002) The circadian
regulatory proteins BMAL1 and cryptochromes are substrates
of casein kinase Iepsilon. J Biol Chem 277:17248–17254
Acknowledgements We would like to thank Nicolas Roggli for Ellis RJ (2001) Macromolecular crowding: an important but
the artwork. Research from our own laboratory discussed in this neglected aspect of the intracellular environment. Curr Opin
review has been supported by Swiss National Science Foundation Struct Biol 11:114–119
(grant to U.S.), the State of Geneva, the NCCR program “Frontiers Etchegaray JP, Lee C, Wade PA, Reppert SM (2003) Rhythmic
in Genetics,” the Bonnizzi Theler Stiftung, and the Louis Jeantet histone acetylation underlies transcription in the mammalian
Foundation of Medicine. F.G. received a postdoctoral fellowship circadian clock. Nature 421:177–182
from the Fondation pour la Recherche Médicale (France). Freedman MS, Lucas RJ, Soni B, von Schantz M, Munoz M, David-
Gray Z, Foster R (1999) Regulation of mammalian circadian
behavior by non-rod, non-cone, ocular photoreceptors. Science
284:502–504
References Froehlich AC, Pregueiro A, Lee K, Denault D, Colot H, Nowrousian
M, Loros JJ, Dunlap JC (2003) The molecular workings of the
Akashi M, Nishida E (2000) Involvement of the MAP kinase Neurospora biological clock. Novartis Found Symp 253:184–
cascade in resetting of the mammalian circadian clock. Genes 198
Dev 14:645–649
 111
Fu L, Pelicano H, Liu J, Huang P, Lee C (2002) The circadian gene Lee C, Etchegaray JP, Cagampang FR, Loudon AS, Reppert SM
Period2 plays an important role in tumor suppression and DNA (2001) Posttranslational mechanisms regulate the mammalian
damage response in vivo. Cell 111:41–50 circadian clock. Cell 107:855–867
Furukawa T, Manabe S, Watanabe T, Sehata S, Sharyo S, Okada T, Lee C, Weaver DR, Reppert SM (2004) Direct association between
Mori Y (1999) Daily fluctuation of hepatic P450 monooxyge- mouse PERIOD and CKIepsilon is critical for a functioning
nase activities in male rats is controlled by the suprachiasmatic circadian clock. Mol Cell Biol 24:584–594
nucleus but remains unaffected by adrenal hormones. Arch Leloup JC, Goldbeter A (2003) Toward a detailed computational
Toxicol 73:367–372 model for the mammalian circadian clock. Proc Natl Acad Sci
Gachon F, Fonjallaz P, Damiola F, Gos P, Kodama T, Zakany J, USA 100:7051–7056
Duboule D, Petit B, Tafti M, Schibler U (2004) The loss of Lin JM, Kilman VL, Keegan K, Paddock B, Emery-Le M, Rosbash
circadian PAR bZip transcription factors results in epilepsy. M, Allada R (2002) A role for casein kinase 2alpha in the
Genes Dev 18:1397–1412 Drosophila circadian clock. Nature 420:816–820
Gerkema MP, van der Leest F (1991) Ongoing ultradian activity Lincoln GA, Andersson H, Loudon A (2003) Clock genes in
rhythms in the common vole, Microtus arvalis, during calendar cells as the basis of annual timekeeping in mammals—
deprivations of food, water and rest. J Comp Physiol [A] a unifying hypothesis. J Endocrinol 179:1–13
168:591–597 Liu C, Weaver DR, Strogatz SH, Reppert SM (1997) Cellular
Golden SS (2003) Timekeeping in bacteria: the cyanobacterial construction of a circadian clock: period determination in the
circadian clock. Curr Opin Microbiol 6:535–540 suprachiasmatic nuclei. Cell 91:855–860
Hannibal J (2002) Neurotransmitters of the retino-hypothalamic Lowrey PL, Shimomura K, Antoch MP, Yamazaki S, Zemenides
tract. Cell Tissue Res 309:73–88 PD, Ralph MR, Menaker M, Takahashi JS (2000) Positional
Hardin PE, Hall JC, Rosbash M (1990) Feedback of the Drosophila syntenic cloning and functional characterization of the
period gene product on circadian cycling of its messenger RNA mammalian circadian mutation tau. Science 288:483–492
levels. Nature 343:536–540 McNamara P, Seo SP, Rudic RD, Sehgal A, Chakravarti D,
Harmar AJ, Marston HM, Shen S, Spratt C, West KM, Sheward WJ, FitzGerald GA (2001) Regulation of CLOCK and MOP4 by
Morrison CF, Dorin JR, Piggins HD, Reubi JC et al (2002) The nuclear hormone receptors in the vasculature: a humoral
VPAC(2) receptor is essential for circadian function in the mechanism to reset a peripheral clock. Cell 105:877–889
mouse suprachiasmatic nuclei. Cell 109:497–508 Murray DB, Roller S, Kuriyama H, Lloyd D (2001) Clock control of
Harmer SL, Hogenesch JB, Straume M, Chang HS, Han B, Zhu T, ultradian respiratory oscillation found during yeast continuous
Wang X, Kreps JA, Kay SA (2000) Orchestrated transcription culture. J Bacteriol 183:7253–7259
of key pathways in Arabidopsis by the circadian clock. Science Nawathean P, Rosbash M (2004) The doubletime and CKII kinases
290:2110–2113 collaborate to potentiate Drosophila PER transcriptional re-
Harms E, Young MW, Saez L (2003) CK1 and GSK3 in the pressor activity. Mol Cell 13:213–223
Drosophila and mammalian circadian clock. Novartis Found Ouyang Y, Andersson CR, Kondo T, Golden SS, Johnson CH
Symp 253:267–277 (1998) Resonating circadian clocks enhance fitness in cyano-
Hattar S, Lucas RJ, Mrosovsky N, Thompson S, Douglas RH, bacteria. Proc Natl Acad Sci USA 95:8660–8664
Hankins MW, Lem J, Biel M, Hofmann F, Foster RG, Yau KW Panda S, Antoch MP, Miller BH, Su AI, Schook AB, Straume M,
(2003) Melanopsin and rod-cone photoreceptive systems Schultz PG, Kay SA, Takahashi JS, Hogenesch JB (2002)
account for all major accessory visual functions in mice. Coordinated transcription of key pathways in the mouse by the
Nature 424:75–81 circadian clock. Cell 109:307–320
Hirota T, Okano T, Kokame K, Shirotani-Ikejima H, Miyata T, Panda S, Provencio I, Tu DC, Pires SS, Rollag MD, Castrucci AM,
Fukada Y (2002) Glucose down-regulates Per1 and Per2 Pletcher MT, Sato TK, Wiltshire T, Andahazy M et al (2003)
mRNA levels and induces circadian gene expression in cultured Melanopsin is required for non-image-forming photic responses
rat-1 fibroblasts. J Biol Chem 277:44244–44251 in blind mice. Science 301:525–527
Hoppensteadt FC, Keller JB (1976) Synchronization of periodical Pourquie O (2003) The segmentation clock: converting embryonic
cicada emergences. Science 194:335–337 time into spatial pattern. Science 301:328–330
Ishikawa K, Shimazu T (1976) Daily rhythms of glycogen Preitner N, Damiola F, Luis Lopez M, Zakany J, Duboule D,
synthetase and phosphorylase activities in rat liver: influence Albrecht U, Schibler U (2002) The orphan nuclear receptor
of food and light. Life Sci 19:1873–1878 REV-ERBalpha controls circadian transcription within the
Izumo M, Johnson CH, Yamazaki S (2003) Circadian gene positive limb of the mammalian circadian oscillator. Cell
expression in mammalian fibroblasts revealed by real-time 110:251–260
luminescence reporting: temperature compensation and damp- Price JL, Blau J, Rothenfluh A, Abodeely M, Kloss B, Young MW
ing. Proc Natl Acad Sci USA 100:16089–16094 (1998) Double-time is a novel Drosophila clock gene that
Jin X, Shearman LP, Weaver DR, Zylka MJ, de Vries GJ, Reppert regulates PERIOD protein accumulation. Cell 94:83–95
SM (1999) A molecular mechanism regulating rhythmic output Ralph MR, Foster RG, Davis FC, Menaker M (1990) Transplanted
from the suprachiasmatic circadian clock. Cell 96:57–68 suprachiasmatic nucleus determines circadian period. Science
King DP, Zhao Y, Sangoram AM, Wilsbacher LD, Tanaka M, 247:975–978
Antoch MP, Steeves TD, Vitaterna MH, Kornhauser JM, Reppert SM, Weaver DR (2002) Coordination of circadian timing in
Lowrey PL et al (1997) Positional cloning of the mouse mammals. Nature 418:935–941
circadian clock gene. Cell 89:641–653 Ripperger JA, Shearman LP, Reppert SM, Schibler U (2000)
Klevecz RR, Bolen J, Forrest G, Murray DB (2004) A genomewide CLOCK, an essential pacemaker component, controls expres-
oscillation in transcription gates DNA replication and cell sion of the circadian transcription factor DBP. Genes Dev
cycle. Proc Natl Acad Sci USA 101:1200–1205 14:679–689
Lavery DJ, Lopez-Molina L, Margueron R, Fleury-Olela F, Conquet Roenneberg T, Merrow M (2003) The network of time: under-
F, Schibler U, Bonfils C (1999) Circadian expression of the standing the molecular circadian system. Curr Biol 13:R198–
steroid 15 alpha-hydroxylase (Cyp2a4) and coumarin 7- R207
hydroxylase (Cyp2a5) genes in mouse liver is regulated by Rutter J, Reick M, Wu LC, McKnight SL (2001) Regulation of
the PAR leucine zipper transcription factor DBP. Mol Cell Biol clock and NPAS2 DNA binding by the redox state of NAD
19:6488–6499 cofactors. Science 293:510–514
Le Minh N, Damiola F, Tronche F, Schutz G, Schibler U (2001) Sathyanarayanan S, Zheng X, Xiao R, Sehgal A (2004) Posttrans-
Glucocorticoid hormones inhibit food-induced phase-shifting lational regulation of Drosophila PERIOD protein by protein
of peripheral circadian oscillators. EMBO J 20:7128–7136 phosphatase 2A. Cell 116:603–615
112
Schernhammer ES, Laden F, Speizer FE, Willett WC, Hunter DJ, Toh KL, Jones CR, He Y, Eide EJ, Hinz WA, Virshup DM, Ptacek
Kawachi I, Colditz GA (2001) Rotating night shifts and risk of LJ, Fu YH (2001) An hPer2 phosphorylation site mutation in
breast cancer in women participating in the nurses’ health familial advanced sleep-phase syndrome. Science 291:1040–
study. J Natl Cancer Inst 93:1563–1568 1043
Schibler U, Ripperger J, Brown SA (2003) Peripheral circadian Tsuchiya Y, Akashi M, Nishida E (2003) Temperature compensation
oscillators in mammals: time and food. J Biol Rhythms 18:250– and temperature resetting of circadian rhythms in mammalian
260 cultured fibroblasts. Genes Cells 8:713–720
Shen H, Watanabe M, Tomasiewicz H, Rutishauser U, Magnuson T, Wang Y, Osterbur DL, Megaw PL, Tosini G, Fukuhara C, Green
Glass JD (1997) Role of neural cell adhesion molecule and CB, Besharse JC (2001) Rhythmic expression of Nocturnin
polysialic acid in mouse circadian clock function. J Neurosci mRNA in multiple tissues of the mouse. BMC Dev Biol 1:9
17:5221–5229 Yagita K, Okamura H (2000) Forskolin induces circadian gene
Silver R, LeSauter J, Tresco PA, Lehman MN (1996) A diffusible expression of rPer1, rPer2 and dbp in mammalian rat-1
coupling signal from the transplanted suprachiasmatic nucleus fibroblasts. FEBS Lett 465:79–82
controlling circadian locomotor rhythms. Nature 382:810–813 Yagita K, Tamanini F, van Der Horst GT, Okamura H (2001)
Staiger D (2002) Circadian rhythms in Arabidopsis: time for nuclear Molecular mechanisms of the biological clock in cultured
proteins. Planta 214:334–344 fibroblasts. Science 292:278–281
Stanewsky R (2003) Genetic analysis of the circadian system in Yamaguchi S, Isejima H, Matsuo T, Okura R, Yagita K, Kobayashi
Drosophila melanogaster and mammals. J Neurobiol 54:111– M, Okamura H (2003) Synchronization of cellular clocks in the
147 suprachiasmatic nucleus. Science 302:1408–1412
Stokkan KA, Yamazaki S, Tei H, Sakaki Y, Menaker M (2001) Yamazaki S, Numano R, Abe M, Hida A, Takahashi R, Ueda M,
Entrainment of the circadian clock in the liver by feeding. Block GD, Sakaki Y, Menaker M, Tei H (2000) Resetting
Science 291:490–493 central and peripheral circadian oscillators in transgenic rats.
Storch KF, Lipan O, Leykin I, Viswanathan N, Davis FC, Wong Science 288:682–685
WH, Weitz CJ (2002) Extensive and divergent circadian gene Yoo SH, Yamazaki S, Lowrey PL, Shimomura K, Ko CH, Buhr ED,
expression in liver and heart. Nature 417:78–83 Siepka SM, Hong HK, Oh WJ, Yoo OJ, et al (2004) PERIOD2:
LUCIFERASE real-time reporting of circadian dynamics
reveals persistent circadian oscillations in mouse peripheral
tissues. Proc Natl Acad Sci USA 101:5339–5346