Circadian rhythms in mammals are regulated by a system of endogenous circadian oscillators (clock cells) in the brain and in most peripheral organs and tissues. One group of clock cells in the hypothalamic SCN (suprachiasmatic nuclei) functions as a pacemaker for co-ordinating the timing of oscillators elsewhere in the brain and body. This master clock can be reset and entrained by daily LD (light–dark) cycles and thereby also serves to interface internal with external time, ensuring an appropriate alignment of behavioural and physiological rhythms with the solar day. Two features of the mammalian circadian system provide flexibility in circadian programming to exploit temporal regularities of social stimuli or food availability. One feature is the sensitivity of the SCN pacemaker to behavioural arousal stimulated during the usual sleep period, which can reset its phase and modulate its response to LD stimuli. Neural pathways from the brainstem and thalamus mediate these effects by releasing neurochemicals that inhibit retinal inputs to the SCN clock or that alter clock-gene expression in SCN clock cells. A second feature is the sensitivity of circadian oscillators outside of the SCN to stimuli associated with food intake, which enables animals to uncouple rhythms of behaviour and physiology from LD cycles and align these with predictable daily mealtimes. The location of oscillators necessary for food-entrained behavioural rhythms is not yet certain. Persistence of these rhythms in mice with clock-gene mutations that disable the SCN pacemaker suggests diversity in the molecular basis of light- and food-entrainable clocks.
Circadian rhythms in mammals are regulated by a system of circadian clocks in the brain and in most peripheral organs and tissues. Circadian clocks provide a temporal framework for the internal biochemical processes underlying physiology and behaviour (internal synchrony) and co-ordinate these processes with environment cycles associated with the solar day (external synchrony). External synchrony requires that circadian clocks be adjustable by local environmental time cues, so that the right behaviour (e.g. wake-up) occurs at the right time of day (e.g. morning or evening). Two clock parameters that can be adjusted are clock frequency, or period (the duration of one complete cycle, denoted by the Greek letter ‘τ’), and clock phase (the position within a cycle, denoted by ‘f’). A stimulus that can control f and τ, and thereby entrain the clock, is designated a ‘Zeitgeber’ (‘time giver’).
The daily cycle of LD (light–dark) is almost universally effective as a circadian Zeitgeber. Light exposure early in the night (simulating an abnormally late sunset) induces an acute phase delay of the circadian cycle, setting it back to an earlier phase. Light exposure late in the night (simulating an abnormally early sunrise) induces an acute phase advance, setting the clock forward (Figure 1B). Light during the middle of the day has little or no effect. This circadian rhythm of sensitivity to light can be represented as a PRC (phase–response curve) by plotting the magnitude and direction of a phase shift as a function of the time within the circadian cycle when light exposure occurred (Figure 1A). The shape of the curve reveals how entrainment to LD cycles can be achieved: if the clock runs fast, evening light delays its progress; and if the clock runs slow, morning light advances its progress, and the clock is trapped at a characteristic phase relative to the LD cycle .
Entrainment of circadian clocks by daily LD cycles ensures that animals specialized to be active in the day or night maintain an appropriate chronotype (e.g. diurnal or nocturnal). Genetic variations in clock τ or sensitivity to light (the shape of the PRC) can support adaptive variations in the LD-entrained phase of the clock, producing, for example, early birds that catch the worm. There are, of course, cats that hunt the birds, and the cat that can entrain its foraging rhythm to that of the bird would probably also forage more efficiently. This raises the question as to whether circadian rhythms are rigidly controlled by LD cycles, or whether phase can be modified by exposure to daily cycles of food availability, or the activity patterns of predators, competitors or potential mates, which also bear directly on survival and reproductive success. This essay is concerned with specializations of the mammalian circadian system that confer sensitivity to non-photic stimuli and synchronize behaviour and physiology more directly to daily rhythms of other animals or sources of food.
Regulation of circadian rhythms by behavioural state
Circadian rhythms can be shifted by arousal during the rest period
In chronobiology research, laboratory rodents are typically housed in cages with litter that needs to be changed every few days. Cage cleaning stimulates arousal, and in Syrian hamsters maintained in DD (constant dark) or LL (constant light), this can shift the circadian clock, depending on when the hamster is disturbed. Arousal stimulated in the mid-to-late ‘subjective day’ (the sleep phase of the nocturnal animal's rest–activity cycle, when light would normally be on) induces a phase-advance shift, while arousal stimulated late in the ‘subjective night’ (the active phase) or early in the subjective day can induce a phase-delay shift . The PRC, like the response to light, is thus bidirectional, but its shape is very different, indicating a distinct non-photic neurochemical pathway capable of resetting the clock machinery (Figures 1A and 1C).
Phase shifts induced by cage cleaning are small, of the order of 30–60 min, but some other procedures for stimulating arousal can produce much larger shifts. Most Syrian hamsters confined to a novel wheel during the mid-subjective day run for an extended time, and runners exhibit phase-advance shifts in the 2–3 h range. Running itself is not necessary, as phase shifts of similar magnitude can also be induced in hamsters kept awake by ‘gentle handling’, an arousal procedure that minimizes stress and activity . More stressful methods of stimulating behavioural arousal during the day are surprisingly ineffective. Physical restraint, loud noise, confinement to a platform over water, and stimulant drugs (caffeine, modafinil, yohimbine and amphetamine) all potently suppress sleep but do not induce phase shifts when administered during the subjective day [4–6].
In rats and mice, single episodes of activity stimulated in the subjective day generally do not produce large phase shifts, but running stimulated at the same time each day can entrain free-running rhythms in DD. The phase of entrainment in these species indicates that scheduled activity can advance the clock late in the subjective day or early in the subjective night and can delay the clock late in the subjective night [7,8]. Consistent with this phase sensitivity, the τ of rodents in DD shortens when they are provided a wheel and concentrate on running early in the subjective night.
Sensitivity of the circadian clock to neural or endocrine correlates of activity and arousal may represent an adaptation for adjusting circadian rhythms to promote synchrony with prey and mates or segregation from predators and competitors. If so, then single or daily episodes of arousal should be capable of altering the phase of entrainment to LD cycles. Such effects have been demonstrated, although the interactions are complex and not fully understood . For example, contrary to simple additivity of the PRCs to arousal and light, running stimulated every day in the middle of the light period delays rather than advances the onset of LD-entrained nocturnal activity in hamsters. By contrast, a single episode of running stimulated at the same circadian phase greatly potentiates the phase advance of nocturnal activity that occurs in response to an 8 h advance of the LD cycle. In DD, arousal can inhibit phase shifts to light at night, at times when arousal alone has minimal or no effect, and brief light exposure can inhibit phase shifts to arousal in the subjective day, when light alone has no resetting effect . Exposure to bright light for 24–48 h has the opposite effect, greatly potentiating the size of phase shifts to a bout of arousal stimulated in the subjective day . These complex interactions may reflect three clock properties: (i) the distinct PRCs to light and activity; (ii) transient deformation of PRCs following exposure to a resetting stimulus or to overnight bright light; and (iii) mutual inhibition between light and arousal, possibly due to convergence on the same clock substrates but with opposite effects (discussed further below).
Although the relevance of non-photic clock resetting to the survival of animals in natural habitats remains to be demonstrated, it is nonetheless clear that, in at least some species, the circadian clock can be modulated by correlates of behavioural state. These responses define a non-photic gateway to the circadian clock that can potentially be exploited in the search for ‘chronobiotics’ (clock-resetting compounds). This search requires analysis of neural input pathways that convey non-photic signals to the clock.
Neural pathways for non-photic regulation of the light-entrained master pacemaker
The circadian timekeeping system in mammals consists of a bona fide master clock or pacemaker located in the hypothalamic SCN (suprachiasmatic nuclei) that co-ordinates secondary or subordinate circadian oscillators located in other brain regions and in most peripheral organs and tissues. Neural ablation studies reveal that the SCN is essential both for LD entrainment and for maintenance of rhythmicity at the tissue level in peripheral organs in the absence of Zeitgebers .
Photic entrainment of the SCN is mediated by the retinohypothalamic tract, a bundle of nerve fibres originating from intrinsically photoreceptive retinal ganglion cells that make synapses with SCN neurons, where they release glutamate and PACAP (pituitary adenylate cyclase-activating polypeptide) as signalling molecules . A secondary source of photic input is provided by the IGL (intergeniculate leaflet), a retinorecipient component of the thalamus that mediates effects of LL on clock τ . The SCN is also innervated by a number of brain regions known to regulate behavioural states, which are of special interest in the search for clock input pathways that may mediate non-photic entrainment. Neurotransmitters utilized by some of these inputs can induce phase shifts, but evidence suggesting a role in non-photic regulation of circadian rhythms is available only for the monoamine 5-HT (5-hydroxytryptamine, also known as serotonin) [12,13]. 5-HT pharmacology is complicated by the existence of at least 14 receptor subtypes, several of which are found in the SCN, either presynaptically (e.g. 5-HT1b receptors on retinohypothalamic axon terminals) or postsynaptically (e.g. 5-HT1a/2a/2c/5a/7), and others (5-HT1a autoreceptors) that regulate the activity of serotonergic cells of origin outside of the SCN, within the midbrain median and dorsal raphe nuclei. Consequently, systemically applied 5-HT receptor ligands have multiple sites of action, producing effects that may be difficult to interpret. Nonetheless, the application of multiple experimental approaches, including pharmacology, electrophysiology, microdialysis, lesions and receptor knockouts, has provided convergent evidence consistent with a role for 5-HT in non-photic modulation of the SCN pacemaker, although puzzles remain.
SCN tissue slices maintained in vitro exhibit circadian rhythms of neural activity that persist for several days or more. Application of 5-HT1a/5a /7 receptor agonists during the mid-subjective day resets the phase of peak neural activity by ~3 h [14,15]. In behaving rodents, systemic injections of the 5-HT1a/7 agonist 8-OHPAT may induce small phase advances during the subjective day, but injections directly into the SCN have surprisingly little or no effect , unless 5-HT receptors are supersensitized by prior treatment with a 5-HT synthesis inhibitor . Neurotoxic lesions of 5-HT afferents in hamsters do not block phase-advance shifts to daytime arousal in hamsters , but such lesions in mice do impair entrainment to scheduled treadmill running in the subjective night, and eliminate the τ-shortening effect of nocturnal wheel running [7,19].
Another set of observations indicates that 5-HT inhibits phase-resetting responses to light at night . Nonspecific 5-HT agonists such as fluoxetine, and agonists selective for presynaptic 5-HT1b receptors or postsynaptic 5-HT1a/7 receptors, attenuate activation of SCN neurons by light or glutamate and suppress light-induced phase advances and delays. Conversely, 5-HT lesions or genetic deletion of the 5-HT1a receptor enhances the magnitude of phase shifts to light [21,22], whereas pharmacological blockade of 5-HT receptors blocks the inhibitory effect of behavioural arousal on light-induced shifts . Given these effects, it is surprising that in rats, activation of 5-HT2c or 5-HT3 receptors mimics rather than inhibits phase-resetting effects of nocturnal light . Whether selective activation of these receptors occurs under physiological conditions is unclear, and it remains possible that the net effect of endogenous 5-HT activity is to promote non-photic clock responses .
5-HT neurons of the dorsal raphe innervate the thalamic IGL, which in turn innervates the SCN, releasing NPY (neuropeptide Y), enkephalin, neurotensin and GABA (γ-aminobutyric acid) as signalling molecules. Although the IGL is innervated by the retina, intra-SCN administration of receptor agonists for IGL neurotransmitters mimics the phase-resetting effects of behavioural arousal in the day and can attenuate rather than mimic phase shifts to light at night [9,25–27]. A critical role for NPY in clock resetting by behavioural arousal is suggested by the evidence that IGL ablation, or NPY antibody injected into the SCN, can block phase shifts to arousal or entrainment to scheduled running [7,9,28,29]. It therefore seems likely that the actions of behavioural arousal on the SCN pacemaker, including resetting of phase and modulation of photic responses, are regulated by both 5-HT and NPY transmission, organized in serial and parallel pathways. The details of this neural circuitry remain to be fully clarified.
Biochemical and molecular basis of non-photic entrainment
The circadian clock in mammals is based on autoregulatory transcriptional–translational feedback loops . At the core of the clock are two Per (Period) and two Cry (Cryptochrome) genes, which are transcriptionally activated by CLOCK (circadian locomotor output cycles kaput)–BMAL1 (brain and muscle ARNT-like 1) protein complexes bound to E-box promotors. PER and CRY proteins form dimers that translocate to the nucleus where they inhibit the activity of CLOCK–BMAL1, thereby suppressing their own transcription. PER–CRY are gradually degraded by the proteasome, freeing CLOCK–BMAL1 to initiate a new cycle of transcription. Secondary loops involving the nuclear receptor proteins REV-ERBα and RORα (RAR-related orphan receptor α) enhance the amplitude, precision and stability of the core loops. Within this system, any stimulus that can acutely alter clock-gene transcription or clock-protein levels has the potential to reset the timing loops.
In SCN neurons, endogenous expression of Per1 is high during the mid-subjective day and declines to a nadir in the subjective night, in antiphase with Bmal1 expression. Light, via glutamate release from retinohypothalamic terminals, activates calcium-dependent second-messenger cascades in SCN neurons, culminating in the binding of phosphorylated cAMP-response element binding protein to cAMP-responsive elements on the Per1 gene promoter, thereby activating Per1 expression. During the day, light has little or no effect on PER1 protein levels, as Per1 expression is already maximal. At night, when Per1 expression falls, photic stimuli can drive it up, acutely delaying the decline early in the night or advancing the rise late at night (Figures 2A–2C). This action of light on Per1 expression is propagated through the interlocking feedback loops and can account for the defining features of the PRC to light.
Non-photic stimuli appear to reset the clock in a similar fashion, but with an opposite sign. Behavioural activity, NPY and 5-HT1a/7 agonists all inhibit SCN neuronal activity and suppress immediate early gene (e.g. cFos) and Per1 gene expression (Figures 2D and 2E). This can explain the maximal sensitivity of the circadian clock to behavioural arousal during the day, as well as the inhibitory effects of these non-photic stimuli on phase shifts to light at night [31,32]. Second-messenger pathways coupling non-photic inputs with clock genes remain to be fully defined, but probably involve changes in the activity of MAPKs (mitogen-activated protein kinases) such as ERK1/2 (extracellular-signal-regulated kinases 1/2). Non-photic manipulations suppress the activation of ERK1/2 in most of the SCN , but activate ERK1/2 in some dorsolateral SCN cells . Phosphorylation of Elk-1, a transcriptional regulator that is a downstream target of MAPK, is not affected by some non-photic manipulations . Dexras1 (dexamethasone-induced Ras-related protein 1), which can regulate G-protein activity and attenuate MAPK activation, has been reported to inhibit non-photic phase shifting, as mice deficient in Dexras1 exhibit larger phase shifts to exercise . However, this observation may be due to altered photic responses in these mice .
Regulation of circadian rhythms by daily cycles of food availability
Food is the dominant Zeitgeber for circadian clocks in most peripheral organs
When food is freely available, nocturnal laboratory rodents eat primarily at night, a circadian rhythm controlled by the SCN and entrained by light. Peripheral organs regulating metabolism and the digestion, absorption, utilization and storage of nutrients also exhibit circadian rhythms, appropriately synchronized with the feeding rhythm. If food availability is restricted to the day, the activities of these organs shift to track the new daily rhythm of food intake. Clock-gene rhythms in peripheral organ cells are also rapidly reset, but the phase of clock-gene rhythms in the SCN pacemaker is not, as long as the LD cycle remains unchanged . These results imply that SCN control of peripheral physiology may be exerted primarily by its control over feeding behaviour (Figure 3). There are humoral and neural outputs from the SCN to hypothalamic autonomic centres that innervate peripheral organs , and these may play some role in modulating peripheral rhythms when food is freely available , but evidently these signals are weak in competition with timing signals directly associated with food intake.
Feeding-related signals that entrain peripheral clocks in behaving animals may include nutrients (e.g. glucose), metabolic hormones (e.g. insulin) or autonomic outputs from brain regions that sense food intake. Entrainment signals may also be provided by biochemical pathways directly affected by energy metabolism and nutrients  (Figure 4). Key among these is the intracellular ratio of reduced to oxidized NAD cofactors, a ratio that is affected by cellular energy metabolism . The NAD ratio has been shown to alter the DNA-binding activity of CLOCK–BMAL1 in vitro. The NAD ratio has also been shown to alter chromatin remodelling by regulating the activity of SIRT1 (sirtuin 1), thereby altering gene expression. A variety of nutrient-responsive pathways could also interact with the core molecular components of the circadian clock. The nutrient-responsive kinase AMPK (AMP-activated protein kinase) can affect the stability of the clock protein CRY1 . Another nutrient-responsive factor, PGC-1α [PPAR (peroxisome proliferator-activated receptor)γ co-activator 1α], can positively regulate transcription of Bmal1 and Rev-Erbα through co-activation of RORα . PGC-1α also regulates the PPAR family, members of which modify activity and expression of BMAL1 and REV-ERBα. The extent to which interactions between cell metabolic factors and clock genes regulate entrainment of peripheral tissues to feeding schedules remains to be clarified, but it seems apparent that sensitivity to daily cycles of nutrient intake may be a fundamental property of peripheral circadian clock cells.
Restricted feeding schedules can also entrain clock-gene rhythms in many brain regions, readily decoupling these oscillators from the SCN pacemaker, which as noted is not shifted by feeding schedules in competition with LD cycles . In the absence of LD cycles, daily feeding schedules can sometimes entrain the SCN (Figure 5A) or can restore SCN rhythmicity disrupted by long-term exposure to LL [45,46]. When this occurs, the phase of SCN entrainment is such that the daily active period begins prior to mealtime. The feeding-related signals that entrain the SCN have not been determined, and may include neural correlates of behavioural arousal associated with food acquisition. The fact that the SCN, particularly in rats, usually does not entrain to daily feeding schedules even in DD or LL (Figures 5B and 5C) implies that SCN clock cells are either less sensitive to intracellular metabolic cues or are shielded from extracellular correlates of daily feeding cycles that may directly affect cell metabolism in peripheral clock cells.
A separate circadian oscillator generates food-anticipatory activity rhythms
When nocturnal rodents entrained to LD are restricted to one daily meal in the middle of their sleep phase (the light period), not only do peripheral organs shift, but the animals become awake and active 1–3 h prior to mealtime [46,47]. This is interpreted as food-anticipatory activity, because behaviour is directed at sources of food, such as a food trough or an operant lever that delivers food pellets at mealtime. Also, if a daily feeding is skipped, activity continues through the expected duration of food availability, declines and then rises the next day prior to the usual mealtime. This meal-associated rhythm persists for as long as the animal can be safely food deprived, for example at least 96 h in adult rats (Figure 5C), and is not dependent on the presence of an LD cycle. In rats maintained in DD, food-anticipatory activity emerges even in cases where the SCN fails to entrain to scheduled meals and free-runs with a different (usually longer) τ. In these cases, two daily rhythms of activity are apparent, one expressing a periodicity different from 24 h, corresponding to the SCN phase, and the other expressing a period that matches the feeding schedule (Figure 5B). This implies that behaviour is controlled by at least two circadian clocks, an inference confirmed by analysis of the properties of food-anticipatory activity rhythms in rats following surgical ablation of the SCN. These animals are behaviourally arrhythmic when food is freely available, but exhibit a robust food-anticipation rhythm when limited to one daily mealtime (Figure 5D). The rhythm persists during several cycles of total food deprivation, resets gradually when mealtime is shifted, and fails to emerge if the feeding intervals are greatly different from 24 h. These properties are consistent with regulation of foraging activity by an FEO (food-entrainable circadian oscillator) located outside of the SCN.
Neural basis of food-anticipatory activity rhythms
In mammals, the SCN is a uniquely light-entrainable master pacemaker, illustrating the principle of localization of function in the central nervous system. Within other vertebrate circadian systems, this function may be distributed among several structures (e.g. the avian hypothalamus, pineal gland and retina), or may be a property of clock cells in every tissue (e.g. in zebrafish). Whether the circadian timing mechanism for food-anticipatory behavioural rhythms in mammals is localized to a single structure, comparable to the SCN, is distributed among several structures or is represented in every brain region and peripheral organ that participates in the control of behaviour is presently uncertain. A distributed organization is suggested by the nearly universal entrainability of non-SCN oscillators to feeding schedules, and by the steady accumulation of evidence against a critical role for specific brain regions and peripheral organs.
Peripheral organs expressing food-entrainable clock-gene rhythms could provide timing signals for behaviour by hormonal or autonomic neural outputs. An example would be the oxyntic gland cells in the stomach, which exhibit food-anticipatory secretion of ghrelin, a hormone that stimulates ingestive behaviour by actions on ghrelin-receptive neurons in the hypothalamic arcuate nucleus. A critical role for this peripheral signal is not supported by observations that food-anticipatory activity rhythms are preserved (in fact, enhanced) following ablation of the arcuate nucleus  and persist in mice lacking ghrelin or ghrelin receptors [49,50].
A potential role for gastric and other peripheral food-entrainable clocks is further called into question by the results of experiments that demonstrate dissociations between peripheral clock-gene rhythms and food-anticipatory behavioural rhythms. In these experiments, a few days of free access to food, following entrainment to a single daytime meal, was sufficient to reset peripheral organs back to a nocturnal phase, but did not reset behavioural food-anticipatory rhythms, as revealed when the rats were subsequently food deprived for several days . During the food-deprivation test, behavioural activity reappeared at the prior mealtime, whereas peripheral organs retained a nocturnal phase. Similarly, when two meals were provided each day, one in the light period and one in the dark period, rats behaviourally anticipated both, but peripheral clock-gene rhythms were not shifted. Not all peripheral sites of food-entrainable clock-gene expression have been evaluated in this way, but the results so far weigh against the idea that peripheral oscillators might directly drive behavioural rhythms or entrain central oscillators that drive behaviour.
The role of central neural oscillators in the induction of food-anticipatory behavioural rhythms has been evaluated primarily using the lesion method. Most of these studies have targeted single structures, and have not succeeded in eliminating food anticipation [52,53]. Hypothalamic areas have been of special interest, as these contain circuits that integrate neural and endocrine stimuli associated with food intake and body weight, and regulate ingestive behaviour and metabolism. Ablation of cells in the dorsomedial hypothalamus has been reported to attenuate food-anticipatory responses in some measures, but complete removal of this region in other studies has revealed little or no deficit . It remains possible that food-entrainable oscillators sufficient to drive food-anticipatory rhythms are distributed throughout the hypothalamus [55,56], endowing this critical circadian mechanism with a high degree of resistance to disruption. Participation of other brain regions, for example circuits that process reward stimuli, is also not ruled out, particularly for anticipation of resources that do not involve caloric restriction (e.g. palatable snacks, water, salt and mating ).
Molecular basis of food-anticipatory behavioural rhythms
Single or double clock-gene mutations or knockouts can alter or eliminate light-entrainable SCN-dependent rhythms . Clock-gene rhythms are entrainable by restricted feeding schedules, but contrary to expectations, most clock-gene mutations have surprisingly modest or no effect on food-anticipatory behavioural rhythms . An exception is the per2brdm mutation, which eliminates anticipation of a daytime (subjective day) meal , but single or double null mutations of Per1 and Per2 reveal that these genes are dispensable for food-anticipatory rhythms . Bmal1 knockouts, which disable the SCN pacemaker, also do not eliminate food-anticipatory rhythms in either LD or DD, if appropriate restricted feeding protocols are used [60,61]. Clear persistence of food-anticipatory rhythms in constant conditions (i.e. food deprivation) has not been observed in Bmal1−/− mice, but interpretation is complicated by the metabolic fragility of these mice. Rats can be safely food deprived for 4–5 days, but mice, the species of choice for gene-mutation studies, may become hyperactive or torpid by the second day of fasting, especially if they harbour metabolic defects, as do Bmal1−/− mice. Consequently, failure to observe persistence of food-anticipatory rhythms during fasting may reflect either loss of function or masking of function . Pending the results of additional studies to characterize the properties of anticipatory rhythms in Bmal1−/− mice, it remains possible that the molecular mechanism responsible for food-anticipatory behavioural rhythms is either entirely novel, or employs additional elements that confer an enhanced degree of robustness on genetic perturbation. Similar conclusions apply to methamphetamine-induced circadian behavioural rhythms, which may be an expression of the same timing device that drives food-anticipatory rhythms .
Manipulations of behaviour and food availability have yielded important insights into the afferent regulation and formal structure of the mammalian circadian timekeeping system. Many fundamental questions of function and mechanism remain. Despite the striking effects of behavioural arousal on the phase, τ and light-responsivity of the SCN pacemaker, the extent to which these phenomena participate in regulating the circadian phase of animals in complex natural environments is not at all clear. Many of these effects are robust in Syrian hamsters, but more subtle or absent in rats and mice, and results on other species are sparse. A complete neural systems and cellular account of clock resetting by arousal is far from achieved. Such an account will need to address species differences and the differential efficacy of arousal procedures within species.
Food-anticipatory circadian rhythms, by contrast, are obviously adaptive, but the physical location and molecular basis of the timing mechanism remain unclear. There is much interest in how circadian clock genes and metabolic factors might interact to produce food-entrained rhythmicity, but these interactions may prove to be more relevant to peripheral clocks and physiology than to the induction of behavioural anticipatory rhythms, which can be dissociated from metabolic cycles (e.g. by intervals of ad libitum food access between total-food-deprivation tests). This and other formal properties of food-anticipatory behavioural rhythms suggest a timing mechanism that stands outside of the metabolic cycles associated with daily fasting and refeeding. How this timing mechanism might control anticipation of more than one daily meal is not yet understood, and whether there are physically distinct oscillators for anticipation of other resources, such as water, salt or mating opportunity, is another open question.
• The circadian clock in mammals can be phase shifted or entrained by single episodes or daily schedules of stimulated arousal or activity, and these can modify entrainment to LD cycles.
• Neural pathways that mediate non-photic inputs to the master circadian clock (the SCN) probably employ serotonin or NPY as neurotransmitters, which, via second-messenger pathways remaining to be defined, may reset the clock by suppression of per clock genes.
• Daily feeding schedules entrain circadian rhythms of clock-gene expression in peripheral organs and many brain regions, dissociating these from the SCN pacemaker, which remains synchronized to LD.
• Daily feeding schedules induce food-anticipatory behavioural rhythms generated by a circadian mechanism located outside of the SCN. The role of known clock genes in the generation and entrainment of food-anticipatory rhythms remains to be clarified.
We acknowledge research support from the Canadian Institutes of Health Research (CIHR) and the Natural Sciences and Engineering Research Council of Canada (NSERC).
- © The Authors Journal compilation © 2011 Biochemical Society