Timing - Understanding Central and Peripheral Clocks

The autonomic nervous system provides a direct pathway between the SCN and target organs. Identified targets include the submandibular salivary glands (31, 32), liver (33-35), pancreas (31), lung (36) and adrenal glands (31, 37, 38). The regulation of these tissues, both sympathetic and parasympathetic, have been directly traced from the SCN to target peripheral tissues (39).

Melatonin is a hormone produced and released into the circulation by the pineal gland. This release is controlled by the SCN via the sympathetic nervous system with minimal levels circulating during the day and maximal levels reached during the night. Melatonin synchronises peripheral clocks with the SCN, acting as a global zeitgeber that controls the transcription/translation cycle of peripheral clock genes. This has been demonstrated in both cell culture (40) and in animal models (41). The mechanism by which melatonin synchronises peripheral clocks has not yet been elucidated but it has been suggested that melatonin may mediate its effects on peripheral clocks via post-transcriptional processes (42) and this in turn, may explain why phase shifts occur more slowly in peripheral clocks than in the SCN.

Glucocorticoids (corticosterone in rodents, cortisol in humans) are hormones produced in the adrenal gland and this release is controlled by the SCN via the sympathetic nervous system. Glucocorticoid response elements have been found to be located in the regulatory regions of mouse core clock genes Bmal1, Cry1, Per1 (43, 44), and Per2 (45) suggesting that glucocorticoid binding may lead to the transcriptional activation of these and other clock genes (46). While glucocorticoid receptors are expressed throughout the brain and periphery, the SCN does not express these receptors (47). On their own, circulating glucocorticoids have been shown in the rat liver to maintain peripheral clock gene rhythmicity in the absence of autonomic nervous system input from the SCN or feeding signals, albeit with lower amplitude of expression (48). Exogenous administration of the glucocorticoid analogue dexamethasone has been shown to phase shift clock gene rhythms in rat peripheral tissues grown in the laboratory (49).

The SCN controls the core body temperature rhythm in mammals (50). It has been shown that rat fibroblasts cells grown in the laboratory can maintain previously induced rhythms when exposed to the natural temperature rhythms measured in mice for a number of days compared to fibroblasts maintained at a constant temperature, but not establish a new rhythm in these cells (51). Temperature effects are most likely mediated by heat shock factor 1 (HSF1) which has been shown to oscillate in liver cells and modify its rhythm in line with temperature changes (52). The use of HSF1 inhibitors have been demonstrated to stop the shift in transcription of clock genes due to changes in temperature rhythms (53).

Behavioural processes such as locomotor activity and feeding are also influenced by the SCN in mammals (54, 55). These in turn can influence endocrine activity and body temperature. Under unrestricted feeding conditions, activity and feeding follow a regular rhythm in line with the SCN (and light/dark cycle).

As shown above, the SCN employs a number of ways to communicate environmental light information to the peripheral clocks to ensure physiological processes are appropriately coordinated during sleep and wake.

Activity, in the form of exercise, has been examined as a possible entrainer. In the SCN, previous experiments have shown that scheduled exercise can enhance the shift of behavioural rhythms to a change in light schedule in mice (a nocturnal animal). Adding three hours of wheel running at the start of the dark period (their normal active time) for four days had no additional phase shifting effect on SCN Per1 expression when exposed to either an eight hour phase advance or delay in light schedule (56). When the exercise is administered at the end of the dark period the results are the same (104). When the light schedule remains constant but exercise is scheduled eight hours before lights off (either voluntary or forced) Per2 expression in the SCN remains entrained to the light schedule (105).

Human studies looking at central clock timing have shown that depending on the time of day, exercise can have an effect on the following melatonin onset. Phase shifts have been seen in both directions with evening exercise (106-108) but this did not carry through to the following night where the melatonin onset was no longer different to the no exercise group (106). When combined with a change in light schedule, exercise may assist with the re-entrainment of the SCN. When placing subjects onto a 23.67 hour day, twice daily two hour bouts of exercise scheduled three and seven hours after waking were able to significantly advance the melatonin peak when compared to subjects who did not exercise (107). Another set of experiments in which subjects had a nine hours phase delay in their sleep/wake cycle under dim light conditions (< 5 lux), three 45 minute bouts of exercise performed in the evening over seven days produced a significant phase delay of melatonin onset of 3.17 hours when compared to a no exercise group (1.67h phase delay) (109). These findings suggest that while exercise could be used to help the SCN to entrain to a new light schedule, the amount required to make a significant impact is most likely impractical for individuals moving to a different light schedule, such as travellers moving to a different time zone.

When examining the effect of exercise on phase shifting peripheral clocks in animal models a different picture emerges, with variation in responses between tissues. When the light schedule remains unchanged but exercise is scheduled during the normally inactive period for mice, eight hours before lights off (either voluntary or forced) for four weeks, mouse Per2 expression in skeletal muscle is phase advanced up to 3 hours but a non-significant phase advance is seen in the lung (105). These results demonstrate that altering the timing of muscle activity has a direct effect on the muscle clocks and the associated increase in oxygen demand may have an influence in lung clocks though the results from this study were not conclusive. As expected, no change in Per2 expression was seen in the SCN. In humans, changes in gene expression in skeletal muscle after a single bout of resistance exercise showed that at six hours post exercise, there was a modest yet statistically significant upregulation of expression of Cry1, Per2, and Bmal1. As only two time points were examined, it was not possible to determine if gene expression had phase shifted for these clock genes.

When exercise accompanies a change in light schedule, the changes in the rhythms of clock gene expression depend on whether the light schedule is advanced or delayed from baseline. After four days of an eight hours advance in light schedule Per1 expression in the mouse liver did not entrain to the new light schedule in either the control group or the group that exercised at the start of the dark period (56). However, in the same set of experiments Per1 expression did entrain to the new light schedule in the lung and skeletal muscle in the exercise group but not in the control group. When light exposure was phase delayed by eight hours both groups showed phase delays in Per1 expression in all tissues examined but exercise did not significantly affect the rate of phase shift over control. If the exercise is given at the end of the dark period, phase advanced mice no longer show re-entrainment in lung and skeletal muscle after 4 days (104). To date there have been no studies that have studied the effect of activity on peripheral clock timing in humans.

It has been shown that exercise can effect circulating levels of melatonin but results have been mixed with some studies showing an increase (106, 110), a decrease (111), or no change (106, 108, 110) and this appears to be dependent on exercise duration, intensity, and time of day. Therefore, it appears that any possible effects of exercise on peripheral gene expression are unlikely due to melatonin. If exercise can directly affect peripheral clock gene expression a possible candidate to mediate this effect are glucocorticoids. It has been demonstrated in mice and rats that bouts of increased activity increases corticosterone levels (112).

Mammals display an internally generated circadian temperature cycle controlled by the SCN, both directly by melatonin (50) and indirectly by sleep/wake and activity cycles. The effects of environmental temperature cycles have been examined to see if they also influence circadian rhythmicity. Mice that were exposed to an inverted environmental temperature profile with normal 12L:12D cycle for six days (37°C at night, 24°C during the day) showed no change in SCN rhythmicity, but expression rhythms in the liver were shifted by 9.5 hours. The core body temperature rhythm was also shifted with the normal temperature peak shifted from late afternoon to late evening and activity peak shifted from early evening to early morning. Observations showed that feeding rhythms in these mice also shifted by 6.5 hours as they consumed the majority of their food during the day (51). In addition to the studies that invert the environmental temperature cycle, the effect of a two hour temperature increase during the light period on peripheral clock timing has been examined in mice (113). Per2 peak expression immediately after the heat increase was phase advanced in the kidney and submandibular gland but not in the liver. Exposing mice to the two hour temperature increase on two consecutive days saw Per2 peak expression advance in the liver as well as the kidney and submandibular gland immediately after the exposure to the second temperature increase. When measured 24 hours later, Per2 peak expression had returned to the same phase as seen before the exposure to the two hour temperature increase. To our knowledge, research on the effects of environmental temperature on central and peripheral clocks has not been conducted in humans.

It has been shown in mice and rat models that changes in feeding schedule does not have an effect on SCN rhythmicity in light:dark conditions (114-117) or constant darkness (114). There is, however, a suggestion that timed caloric restriction may affect SCN rhythmicity. Mice that had received 66% of average daily caloric intake during their normal sleep period at ZT6 (six hours after lights on) demonstrated a decrease in amplitude of Per1 expression in the SCN and a phase advance of 1.4 hours when compared to ad libitum fed mice or mice that were given a normocaloric intake at ZT 6 (118). To date there is a lack of human studies looking at the effect of meal timing on SCN rhythmicity.

Animal models examining the effect of food timing on peripheral clocks has been shown to be dependent on the tissue examined. In the rat lung, timed restricted feeding to 4 hours during lights on (when they would normally be asleep) was shown to result in a six hour phase advance in Per1 expression after seven days with no further shift seen after 19 days (115). When the time food was available was extended to eight hours there was no longer any difference in Per1 expression when compared to ad libitum fed rats. Restricting feeding in mice to when the lights were on did not alter gene expression in the kidney after four days while in the pancreas the results were not clear (114). In the heart, one study in mice was unable to demonstrate a clear result (114), but others have shown phase advances in a number of clock genes in both mice and rats (117, 119). The most commonly examined tissue in animal studies is the liver. When feeding is restricted to lights on, mouse (114, 120) and rat (115, 120) models show an almost complete inversion of expression in a number of genes after seven to nine days, while others found this inversion occurs even faster (119). Further restricting feeding to just four hours during lights on gave similar results (115, 116).

When kept under constant light conditions, Per2 expression in mouse liver, kidney, and submandibular gland showed a decrease in amplitude and a broadening in the distribution in peak phases. While remaining in constant light, timed restricting feeding to what would have been lights off (20:00 – 08:00) was unable to increase the amplitude of expression of Per2 in these tissues but it did reverse the distribution of peak phases to what is seen in animals kept in light/dark conditions (121).

Whether food timing has an effect on a particular peripheral clock may depend on the types of signals it receives from the SCN. The regular timing of meals in a rhythmic pattern has been demonstrated to maintain clock gene expression rhythmicity and amplitude in the liver of rats when autonomic nervous system and corticosterone inputs from the SCN are abolished (48). Glucose production in the rat liver has been shown to be controlled by sympathetic nerves by the SCN regardless of feeding schedule (33, 34). It has also been shown that when rats are restricted to daytime feeding, Per1 expression in rat submandibular salivary glands remains in sync with the SCN or becomes arrhythmic (32). The submandibular salivary glands receive direct signalling from sympathetic nerves from the SCN, destruction of this neural pathway in rats causes Per1 expression shift 180 degrees and entrain to the daytime feeding schedule. In the same rats, liver Per1 expression was shifted by feeding schedule (32), and this shift has also been shown by others (114, 115, 119, 122-124). Feeding induced increases in temperature could also have an entraining effect on peripheral clocks (46). Further evidence of differential signalling to peripheral clocks has been shown using parabiosis – two mice joined together in order to share their blood supply. Parabiotic pairs of mice consisting of one intact and one with the SCN removed were entrained to a 12L:12D cycle before being released into constant darkness for three days. On examination it was found that the mice with the SCN removed were able to maintain circadian rhythms of clock gene expression in the liver and kidney (suggesting a blood borne signal was responsible) but not in heart, spleen, and skeletal muscle (suggesting possible alternate signalling) (125).

Again, at this point in time there are few data available on the effect of meal timing on peripheral clocks in humans. This, in part, has to do with the ability of researchers to directly access suitable tissues over time to determine if time peripheral clocks are likely to be influenced by the timing of eating. The identification of surrogate markers of peripheral clock activity would assist in observing the possible effects of meal timing in humans.

If feeding patterns are out of sync with central clock timing, these become dominant zeitgeber over SCN in certain peripheral clocks resulting in a slow shift of peripheral clock gene expression to match feeding rhythms, and this has been demonstrated in mouse (114) and rat (115) liver. This slow shift may be SCN-influenced as SCN-lesioned mice have been shown to rapidly invert clock gene expression (one to two days) in response to restricted feeding during the subjective day (120) and may exist in order to prevent a transient shift in feeding from making immediate, lasting changes to the timing of physiological processes (60). In addition, destruction of the SCN or isolating cells from the SCN in mice results in a dampening in amplitude of clock gene expression in the liver (29, 120, 126). This suggests that even though the timing of peripheral clocks in the liver is determined by the timing of food intake, the SCN still has an effect by influencing the amplitude of gene expression in peripheral tissues.

There has also been some interest into the effect of macronutrient composition on central and peripheral clock timing. Mice subjected to a high fat diet showed no change in liver and adipose tissue timing but the amplitude of clock gene expression rhythms was attenuated (15). In rats, parenternal nutrition, specifically glucose, amino acids, and saline administration, has been demonstrated to affect both SCN and liver Per2 expression (127). Human studies showed that switching from a low fat/high carbohydrate diet to a high fat/low carbohydrate diet had no effect on Bmal1, Nr1d1, Per1, Per2, and Per3 rhythmicity but had a significant effect on amplitude of expression for Bmal1, Per1, Per2, and Per3 in monocytes (75). This change in expression may be an uncoupling of central and peripheral clock timing that in turn alters metabolic processes (75).

The effect of caffeine on clock gene expression has been investigated in both animal models and humans. Caffeine is able to elevate cytosolic calcium and therefore likely to influence the timing of clock gene expression through the activation of early transcription factor CREB (Figure 1). In the SCN of mice that have been removed and observed in the lab, short term administration of caffeine given early during the day did not induce phase shifts in Per2 expression but it did induce phase shifts in Per2 expression in mouse liver, lung, and kidney. These shifts were dependent on the time of administration, with caffeine given early in the day producing advances and administration late at night producing phase delays (128). In mice, long-term caffeine administration has been shown to cause phase delays in Per2 expression in livers that have been removed and observed in the lab (128) but phase advances in liver Per1 and Per2 and delays in Bmal1 expression when they remain in the animal (129). Human studies have shown that caffeine can acutely inhibit salivary melatonin levels (130) and when administered three hours prior to habitual bedtime causes a delay in melatonin onset on the following night (131). This shift in SCN timing would have the effect of delaying sleep onset, which in turn could contribute to shortened sleep times. It has yet to be determined what the effect of caffeine administration later into the night might be, whether it might cause a more significant delay or, as is the case with light, a phase advance in SCN timing.

As demonstrated above, there are a number of factors that can alter the timing of central and peripheral clocks, and these factors do not affect timing in all clocks in a uniform manner. As a result, central and peripheral clock timing can become uncoupled and this may lead to undesirable consequences. Observational studies have demonstrated that changes to activity that affect timing such as shiftwork have negative consequences on health such as obesity (132, 133), diabetes (134), metabolic dysfunction (132, 135-137), cardiovascular disease (134), and cancer (134, 138). Metabolic disturbances, such as lowered glucose tolerance, have been observed humans when individuals eat at night (139). This disturbance is presumed to occur because of the misalignment of peripheral clocks as a result of the change in meal timing out of alignment with the timing of the SCN but, as yet, this has not been elucidated in humans. As a result of the change in meal timing, homeostasis is disrupted and results in changes in leptin release, glucose and energy metabolism, and insulin sensitivity (60). A possible way of offsetting this change in metabolic functioning is by altering meal patterns in shiftworkers so they better fit in with circadian rhythms of nutrient metabolism and glucose tolerance (140). By gaining a better understanding of how these different clocks work, we may be able to develop ways of minimising clock desynchrony and the associated negative outcomes that would otherwise occur.