Humans and other mammals exhibit a remarkable array of cyclical changes in physiology and behaviour. These are often synchronized to the changing environmental light–dark cycle and persist in constant conditions. Such circadian rhythms are controlled by an endogenous clock, located in the suprachiasmatic nuclei of the hypothalamus. This structure and its cells have unique properties, and some of these are reviewed to highlight how this central clock controls and sculpts our daily activities.
Early researchers in the field of biological rhythms recognized that the daily behavioural and hormonal rhythms of laboratory rodents were readily influenced by environmental lighting regimes. Since many of these rhythms were sustained in the absence of cyclical changes in lighting, but could be altered by brief exposure to light, it was reasoned that the clock or pacemaker responsible for these rhythms should be found within the visual system of the brain (for a review, see ). The development in the 1970s of new ways to visualize neural connections enabled researchers to map novel pathways from sense organs to the brain. Injection of such anatomical tracers into the eye revealed that the tracer was transported from the retina to an area of the hypothalamus called the SCN (suprachiasmatic nuclei, meaning ‘nuclei above the optic chiasm') . Studies were then initiated to experimentally destroy (‘lesion') this part of the brain to determine whether and how circadian rhythms in behaviour and physiology were affected. Lesions of the SCN, but not other hypothalamic structures, abolished rhythms in plasma corticosterone  in addition to drinking and general locomotor activity , thereby implicating the SCN as a possible site of a circadian pacemaker in mammals. Further studies revealed that grafts of fetal SCN tissue into the hypothalamus of SCN-lesioned arrhythmic adult rodents restored rhythms in behaviour (but not endocrine function). Grafts of other brain structures do not have these rhythm-promoting actions . Hence, a combination of lesion and transplant approaches has led to the widely accepted view that the SCN function as the master circadian pacemaker in mammals.
The SCN are composed of bilateral lobes adjacent to the ventral floor of the third ventricle, one of the fluid-filled structures of the brain (Figure 1). Each rugby-ball-shaped lobe contains ~8000 neurons in addition to ~1000–2000 non-neuronal glial cells. SCN neurons are very small (~8–12 μm in diameter, approx. half the size of neurons found in other hypothalamic sites, such as the paraventricular nuclei) and are so densely packed that the SCN lobes are readily distinguishable in brain sections stained for standard histological markers, such as Nissl substance. SCN cells have a number of remarkable properties. (i) Metabolic measurements in vivo demonstrate a pronounced day–night rhythm in SCN metabolic activity, with 2-deoxyglucose uptake being much higher during the day than during the night . (ii) In vivo, rodent SCN neurons show a pronounced rhythm in spontaneous electrical activity, with high levels of AP (action potential) discharge during the day [peak at ~CT6 (circadian time 6)] and lower frequencies at night . This in vivo rhythm is sustained when the SCN is isolated from the surrounding brain by fine knife cuts that sever its neural connections. Remarkably, in vitro, neurons in brain slices containing the SCN also express a similarly phased circadian rhythm in spontaneous AP production [8,9]. The peak in this rhythm is used as a marker of the mid-phase (~CT6) of the SCN circadian pacemaker in vitro. (iii) Neonatal and fetal SCN neurons that are dissociated and cultured on glass plates embedded with MEAs (multi-electrode arrays) can sustain spontaneous rhythms in electrical discharge . The period of these rhythms is determined genetically by the intracellular molecular clock. These and other studies (see the chapter by Herzog [10a] and below) establish that individual SCN neurons can function as cell-autonomous circadian pacemakers.
It is estimated that most, if not all, SCN neurons synthesize the inhibitory neurotransmitter GABA (γ-aminobutyric acid), and functional GABAA receptors are present throughout the SCN. This indicates that the SCN is composed of a network of inhibitory neurons and it would be tempting to speculate that the SCN neurons are homogeneous. However, within this GABAergic network, the location of SCN neurons can be partly distinguished and differentiated by their neuropeptide content. The neuropeptide AVP (arginine vasopressin) is produced by neurons in the medial aspect of the SCN, whereas VIP (vasoactive intestinal polypeptide) is mostly found in neurons in the ventral part of the SCN (Figure 1). Other neuropeptides such as GRP (gastrin-releasing peptide) and SP (substance P) are predominantly found in neurons of the central part of the SCN. In the rat and hamster, the RHT (retinohypothalamic tract; see below), which conveys light information from the retina to the SCN, densely innervates VIP and GRP neurons. This led some researchers to schematize the SCN as having a ‘core' area where non-rhythmic cells integrate environmental light input and a ‘shell' area where the actual master pacemaker cells are contained . Both GABAergic and peptidergic neurotransmission is then used to link activity between and within these functional subregions of the SCN. However, this model does not appear universally applicable between different species (for a detailed review, see ). What is apparent is that within the rodent SCN, VIP signalling is very important for synchronizing SCN neurons (see below). Further research is necessary to determine whether the other neurochemical phenotypes of SCN neurons have distinct roles in circadian timekeeping.
Photic and non-photic resetting
There are two basic types of stimuli that reset the SCN circadian pacemaker and hence behavioural rhythms. These stimuli alter the SCN at different phases of the SCN circadian cycle and communicate with the SCN via different pathways. Many nocturnal laboratory rodents, such as rats, hamsters and mice, exercise voluntarily in running-wheels, and this rhythm in locomotor activity shows a very readily measurable day–night profile under LD (light–dark) conditions that is sustained with near-24h periodicity in DD (constant dark) or LL (constant light; see Figure 2). Typically, the onset of this behavioural rhythm is used as a marker of the beginning of subjective night (CT12). Steady-state changes in the phase of these onsets are used to interpret how a stimulus resets the behavioural rhythm and hence the phase of the underlying SCN circadian pacemaker. Exposure to light or photic stimuli characteristically shifts the nocturnal rodent SCN only during the subjective night or active phase of the circadian cycle; exposure to light during the day has minimal phase-resetting actions. This pattern of resetting [PRC (phase–response curve)] is called a photic PRC and is mediated via the RHT (Figures 2 and 3). In contrast, stimuli that promote arousal phase-shift the SCN pacemaker during the middle of the day or inactive phase and have minimal resetting actions during the active subjective night (Figure 2; see also the chapter by Mistlberger and Antle [12a]). This temporal pattern of sensitivity is very different to that of photic stimuli and hence has been named as the ‘non-photic' PRC. Examples of the non-photic stimuli include sleep deprivation, forced exercise, exposure to a novel environment etc. There are two main pathways that communicate non-photic information to the SCN. The first is the GHT (geniculohypothalamic tract), which conveys neural signals from the IGL (intergeniculate leaflet) of the thalamus. The second pathway originates in the MR (median raphe) of the brain stem (Figure 3). It is the daily resetting actions of photic and non-photic stimuli that are responsible for sculpting and shaping our daily rhythms.
The neurochemical signals of the photic and non-photic input pathways are also different. Glutamate is the principal neurochemical of the RHT, whereas NPY (neuropeptide Y) and 5-HT (5-hydroxytryptamine; also known as serotonin) are key neurochemicals of the GHT and MR pathways respectively. Glutamate is the brain's main excitatory transmitter, whereas NPY and 5-HT are known to be predominantly inhibitory in the mammalian central nervous system. In situ hybridization and immunohistochemical studies show that both ionotropic and metabotropic glutamate receptors are expressed in the SCN, and a variety of NPY and serotonergic receptors are present in this structure. Consistent with this, electrophysiological studies confirm that retinal illumination increases the electrical activity of rodent SCN neurons in vivo, and electrical stimulation of the optic nerve in SCN brain slices excites SCN neurons in vitro. These actions are blocked by pre-treatment with glutamate receptor antagonists . In particular, activation of the ionotropic glutamate receptors are implicated in photic entrainment. Complementary studies of NPY and 5-HT indicate that these neurochemicals predominantly suppress/inhibit SCN neuronal activity. The precise roles of the receptors mediating these inhibitory actions are difficult to elucidate, as a number of NPY and 5-HT receptors are located both pre- and post-synaptically in the SCN. Overall, at the level of the electrical activity of an SCN neuron, photic and non-photic stimuli differ by their excitatory and inhibitory actions respectively.
Our knowledge of the neural substrates of photic and non-photic resetting is constantly changing and expanding. In the 1980s and 1990s, the prevailing view was that circadian light information was captured by rods and cones, but over the past decade, there has been a revolution in our understanding of photoreception in vertebrates. One of the main classes of neurons in this structure, the RGCs (retinal ganglion cells), were thought to be light-insensitive such that they relied on rods and cones to sense and capture photic information. However, ipRGCs (intrinsically photosensitive RGCs) were discovered and subsequently determined to express a novel photopigment, melanopsin [14,15]. Although this photopigment is expressed by a comparatively small proportion of RGCs, these ipRGCs are intrinsically photosensitive, and electrophysiological recordings in vitro show that they depolarize in response to exposure to light . The temporal profile of this depolarization is a slow onset and offset, and resembles the temporal profile of SCN neuron responses to retinal illumination. Studies in rodless coneless mice establish that ipRGCs alone are sufficient for entrainment of circadian rhythms to environmental light . IpRGCs are particularly sensitive to blue light wavelengths. More recently, ipRGCs have been subdivided into a number of different classes, and it is now apparent that the projections of ipRGCs are much more extensive than previously thought, raising the possibility that ipRGCs convey particular kinds of photic information to many other brain areas . Future studies are necessary to determine whether and how neuronal activity in these areas is influenced by the activation of ipRGCs. A key implication here is that the RHT is unlikely to be a specific projection to the hypothalamus and that the name, RHT, is a misnomer. Furthermore, many ipRGCs also contain the neuropeptide PACAP (pituitary adenylate cyclase-activating polypeptide) , and PACAP modulates the actions of glutamate in the SCN [20,21]. Thus we are indeed in our infancy in our understanding of the communication of light information to the neural circadian system.
Similarly, knowledge of arousal and non-photic neural mechanisms continues to expand. In the late 1990s, a new group of neuropeptides, called the orexins (also named the hypocretins) were identified and found to be synthesized exclusively by neurons in the lateral hypothalamus [22,23]. Orexin neurons are crucial for sustaining wakefulness; mice lacking orexin neurons or orexin receptors cannot maintain wakefulness and show rapid transitions in brain state. Orexin neurons are activated by arousal-promoting stimuli, including those that reset the SCN pacemaker . Orexin neurons innervate many neural structures of the circadian system, including the IGL and MR in addition to the SCN region . Consistent with the observation that orexin receptors are present in the SCN region, the electrical activity of SCN neurons is influenced by orexin  and can be reset by exogenous applications of orexins given during the subjective day . Hence, orexin neurons may transpire to be major players in relaying non-photic information throughout the neural circadian system.
Molecular basis of the SCN clock
The molecular basis of circadian timekeeping in mammals is remarkably similar to that of insects (see the chapter by Glossop [27a]), with many of the key molecular components evolutionarily conserved. This intracellular oscillator is composed of both positive and negative feedback/feedforward loops. To summarize briefly, the transcription factors CLOCK or NPAS2 dimerize through their PAS domains to BMAL1 and positively drive the rhythmic transcription of the per1-2 (period) and cry1-2 (cryptochrome) genes. Following translation, PER and CRY proteins accumulate in the cytoplasm, where they are phosphorylated by CK1ε (casein kinase 1ε) and CK1δ, and translocate back into the nucleus. Here, they exert their negative drive to the system by inhibiting CLOCK–BMAL1 transcriptional activity, essentially inhibiting their own transcription. Over several hours, the phosphorylated PER–CRY complexes are broken down and the negative influence on their own transcription is removed, restarting the cycle. Several other auxiliary loops (e.g. Dec1-2 and Reverb α) stabilize the system. The speed at which PER–CRY complexes are degraded sets the speed of the molecular clock, which usually cycles with a periodicity of ~24 h. Mutations that affect the interactions between PER proteins and CK1 accelerate the clock  and underpin familial advanced sleep-phase syndrome in humans , whereas the after-hours and overtime mutations, which slow the degradation of CRY, decelerate the clock [30,31]. More recent results have indicated an increased complexity to this model of the molecular clock. Biochemical (e.g. changes in intracellular Ca2 + and cAMP) and epigenetic factors (e.g. alteration of chromatin structure) are both under control of the clock and regulate the transcription of core clock genes [32,33].
To understand the molecular basis for resetting of the clock, the effects of photic/non-photic stimuli and/or their neurochemical correlates on the expression of clock genes in the SCN have been studied. Consistent with the observation that NPY acts to suppress SCN neuronal activity, giving NPY or non-photic stimuli in vivo at the time that the SCN clock is reset by non-photic stimuli suppresses SCN expression of per1/per2 . In vivo, peripheral injections of a serotonin agonist suppresses per gene expression at the same phase (CT9–CT10) that this compound resets behavioural rhythms . Similarly for photic stimuli given during the subjective night, exposure to light in vivo or glutamate agonists in vitro increases per1 expression in the SCN [36,37]. Hence, similar to their actions on SCN neuronal activity, non-photic and photic stimuli have very different actions on the core molecular clock in the SCN.
To further understand intra- and extra-SCN influences on clock-gene regulation, transgenic mice in which bioluminescent [luc (luciferase)] or fluorescent [EGFP (enhanced destabilized green fluorescent protein] reporters are driven by clock-gene promoters have been generated (Figure 1A). Transgenic animals in which the promoter of the gene is fused to the luc reporter gene have been generated to monitor circadian rhythms in transcription. These animals include c-fos::luc mice, per1::luc rats and per1::luc mice . Here, when the promoter of the gene is activated, luc is transcribed. The resulting luc enzyme acts on its substrate, luciferin, which is included in the culture medium, and this reaction produces photons of light that can be measured with photosensitive devices. These devices include photomultiplier tube assemblies or highly sensitive luminescence-imaging microscopy systems. More recently, a transgenic animal has been generated with two different luc genes expressed under the promoters of per2 and bmal1 . The luc proteins induce differently coloured emission spectra that can be imaged simultaneously using optic filters to separate the spectra, thus allowing concurrent measurement of two different clock-reporter constructs. To assess the clock protein as opposed to clock mRNA activity, an mPer2Luc knockin mouse was produced that accurately reports the level of PER2 protein .
Per1 promoter activity has been tracked in mice using EGFP . This construct, unlike normal GFP constructs, which are relatively stable, has a half-life of ~2.1 h and thus dynamic changes in promoter level can be assessed. Visualization of per1-promoter-driven GFP activity requires fluorescent excitation of the GFP protein. Fluorescence microscopy produces high-quality images; however, over time the reporter is bleached, and cells and tissues suffer phototoxic damage, so this approach is not suitable for long-term imaging. Conversely, bioluminescence imaging requires no external energy source for visualization and SCN explants have been maintained in culture for over 600 days .
Electrical and chemical communication
At one time it was thought that the maintenance of timekeeping in the SCN did not depend on conventional forms of intercellular communication. Schwartz [41a] showed that impairment of AP-dependent communication through infusion of the sodium-channel blocker TTX (tetrodotoxin) into the SCN in vivo suppressed the expression of circadian rhythms in behaviour. However, on termination of the TTX infusion, behavioural rhythms returned at a time of day that was consistent and predictable from their phasing pre-TTX treatment. This indicates that the SCN had maintained its phase during the blockade of AP-mediated synaptic communication. Hence, it was the circadian control of behaviour and not SCN clock phase that had been impaired by the TTX treatment. However, in vitro studies of the actions of TTX on per1::luc expression in SCN brain slices are not consistent with this. Yamaguchi and colleagues [41b] found that TTX greatly suppresses per1::luc expression and seems to disrupt the synchrony between SCN neurons. This indicates that the in vitro SCN is much more sensitive to TTX than the SCN in vivo or that TTX infusions in vivo may have been acting outside the SCN rather than on the SCN itself. Other TTX-independent forms of neural communication are possible via gap junctions, and connexin proteins, key constituents of gap junctions, are present in the SCN . Mice lacking connexin32 have imprecise behavioural rhythms and their SCN neurons show diminished electrical coupling [43,44]. Perhaps the in vitro SCN slice culture has reduced levels of connexins, thereby making it more sensitive to perturbations in AP production.
Another aspect of intercellular communication is revealed through investigations of neuropeptidergic transmission. Studies with mice lacking VIP or its cognate receptor, VPAC2, (VIP−/− and Vipr2−/− mice respectively) establish that intercellular VIP–VPAC2 signalling is important for normal circadian timekeeping. Both VIP−/− and Vipr2−/− mice show disrupted behavioural rhythms and diminished levels of core clock-gene expression and immediate-early gene expression in the SCN [45–47]. Electrophysiological studies indicate that SCN neurons in brain slices from VIP−/− and Vipr2−/− mice have diminished synchrony, with a reduced proportion of cells showing detectable rhythms in electrical activity [47–50]. Patch-clamp studies reveal that Vipr2−/− SCN neurons tend to hyperpolarized (i.e. less electrically excited) than their wild-type counterparts and this is consistent with the reduced frequency of AP production observed in extracellular recordings from these neurons in vitro. In Vipr2−/− mice crossed with per1::luc or per1::EGFP mice, some Vipr2−/− SCN neurons sustain a degree of molecular rhythmicity, but many do not and there is diminished synchrony and co-ordination of molecular activities between these neurons (Figure 1) [51,52]. In contrast, mice lacking GRP receptors, which are normally present in the SCN, have a diminished response to light, but sustain behavioural rhythms, indicating that loss of neuropeptide signalling does not generally drastically affect the SCN timekeeping function. Hence, intercellular communication via VIP–VPAC2 signalling is very important for normal SCN timekeeping and the circadian control of behaviour.
VIP also resets the wild-type rodent SCN circadian pacemaker. Microinjection of VIP into the SCN of hamsters free-running in constant conditions resets their behavioural rhythms in a pattern that partly resembles the phase-shifting effects of light . Similarly, in vitro, exogenous VIP resets the firing-rate rhythm of rat SCN neurons, with a temporal pattern of sensitivity mimicking that of the RHT transmitter, glutamate . These resetting actions of VIP seem to recruit cAMP-dependent mechanisms, as blockade of adenylate cyclase alters the resetting actions of VIP . Since VIP shifts the SCN in a light-like manner, it is perhaps not surprising that mice lacking VIP–VPAC2 signalling do not synchronize normally to the LD cycle. Indeed, recent evidence indicates that a non-photic stimulus is more effective at organizing behavioural rhythms in these mice .
SCN control of behaviour
Lesion and transplant studies have firmly established that the SCN controls the circadian profile of behaviour and physiology . Precisely how it does this is less clear. Encasing a fetal SCN graft in material that prevents the graft forming synaptic connections with the host brain does not prevent the restoration of behavioural rhythms . This implicates the rhythmic secretion of neurochemicals as a paracrine signal from the graft in the circadian control of behaviour. Whether such mechanisms are also the norm in the SCN-intact adult animal is unknown.
The identity of this neurochemical signal has thus far eluded researchers. There are at least six candidates: VIP, PK2 (prokineticin 2), TGFα (transforming growth factor α), cardiotropin-like cytokine, GABA and glutamate (Figure 3). The central concept here is that these SCN molecules are rhythmically synthesized and released. They could be communicated locally from SCN efferents, secretion into the cerebrospinal fluid of the ventricles or via passive diffusion in the extracellular space between neurons (‘volume transmission'). Whatever the release mechanism, these signals are postulated to alter the activity of brain centres controlling behaviour and physiology. For example, based mostly on anatomical studies demonstrating that PK2-containing SCN neurons project throughout the hypothalamus and thalamus to areas where expression of PK2 receptor is expressed , it is suggested that PK2 signals from the SCN are very important for conveying circadian information throughout the brain .
Unfortunately, the demonstration of rhythmic release of such neurochemicals is technically difficult, and hence precise knowledge in this domain is limited. One exception is the assessment of glutamate and GABA. Using the implantation of microdialysis probes into the PVN (paraventricular nuclei), Kalsbeek, Buijs and others have shown that glutamate and GABA release from SCN efferents modulates PVN neuronal activity and influence the timing of the autonomic nervous system (see the chapter by Kalsbeek [58a]).
An intriguing complexity to this problem of circadian control of behaviour is that the SCN electrical activity in diurnal rodents also peaks during the day and not at night as might be expected . Similarly, per-gene expression in the SCN of such day-active rodents is higher during the day than night and thus does not differ greatly from that seen in nocturnal rodents . This suggests that a combination of SCN output and the interpretation of this efferent signal by downstream substrates determines the phase of an animal's behaviourally active phase.
In addition to direct SCN signals, the SCN can also influence the brain and body by indirectly regulating the release of the pineal hormone, melatonin. Melatonin-binding sites are present in many brain regions and peripheral tissues. The SCN efferents act through the PVN and spinal cord to ultimately regulate the activity of the pineal gland, and it is via this regulation that the SCN can communicate circadian and daylength information to the pineal gland. The SCN itself expresses melatonin-binding sites and its phase can be reset by exogenous melatonin . Hence, the SCN relays circadian-clock-phase information to the brain and body both directly and indirectly, and it is possible that some of these signals feed back on the SCN itself.
Relationship between the intracellular clock and cellular activity
The relationship between clock-gene expression and SCN cellular activity remains a key question in chronobiology. Since the peak firing rate of SCN neuronal activity follows when per1 expression is approaching maximal levels, it is tempting to speculate that high levels of per1 transcript lead to elevated electrical activity. Indeed, studies of mouse per1::EGFP SCN neurons show that these neurons become more excitable and produce more APs, as the level of EGFP fluorescence is at approx. maximal levels . However, a more recent study suggests that per1::EGFP neurons become so excitable during the day that they cannot fire APs and that it is the non-per1::EGFP neurons that are firing APs at maximal rates during the middle of the day [62a]. This naturally occurring highly excited state of SCN per1 neurons during the day may explain why SCN neurons are highly resistant to the effects of excitotoxins that kill most central mammalian neurons. Future studies are needed to clarify the functional relationship between per1 expression and SCN cellular excitability.
Studies of brain and tissues from PER2::LUC/per1::luc rodents demonstrate that, at least in culture, many brain regions can display circadian rhythms in clock gene/protein expression, raising the possibility that circadian oscillators are present in a variety of tissues [63,64]. Among the best investigated are the olfactory bulbs, hippocampus, mediobasal hypothalamus and habenula [65–68]. These studies also show that it is not only neurons that rhythmically express per genes/proteins; non-neuronal glial cells and, in particular, the ependymal cells lining the ventricular walls of the brain can also exhibit rhythms in PER2::LUC [66,67].
The ubiquitous importance of circadian-clock genes in controlling a range of fundamental biological processes is an emerging theme in biology, highlighting the wide-ranging scope of this field [69,70]. However, although we have seen a recent explosion in our understanding of circadian rhythmicity, not least following the discovery of core clock genes, there are a range of questions that still need addressing. How does the molecular clockwork affect the electrical properties of SCN neurons and vice versa? How does the SCN communicate its phase information to control downstream oscillators? What is the role and function of these downstream extra-SCN oscillators in controlling tissue-specific and physiological functions? We predict that the resolution of such questions will not only further basic scientific knowledge, but will also impact on many areas concerning human health and disease.
• Daily circadian rhythms in physiology and behaviour are controlled by a master pacemaker based in the SCN of the hypothalamus.
• The molecular basis of this pacemaker consists of interlocking feedforward and feedback loops that oscillate with a period of approx. 24 hours.
• The phase of this pacemaker can be adjusted by light input from the retina and by a variety of non-photic input pathways.
• Neurochemical signalling and electrical activity are important in maintaining synchronized rhythmicity within the SCN.
• Outputs from the SCN synchronize oscillators in downstream tissues to co-ordinate physiology and behaviour.
H.D.P. and C.G. are supported by grants from the Biotechnology and Biological Sciences Research Council (U.K.) and the Wellcome Trust.
- © The Authors Journal compilation © 2011 Biochemical Society