Aging and Circadian Rhythms

Circadian Timing System Regulates Sleep Timing and Consolidation

The circadian timing system is one of the two major sleep regulatory systems [43, 44] (the other being a homeostatic sleep-wake process). The circadian timing system is a major determinant of the timing of sleep and sleep structure in humans, and many aspects of sleep vary markedly with circadian phase in both young and older adults [45-48]. A proper alignment between the timing of sleep and the circadian phase of sleep is important for sleep duration and quality, as demonstrated in both healthy subjects [49-51] and in some clinical conditions [52, 53]. The circadian timing system has a major influence on the timing and duration REM sleep [44], and has a smaller but still significant impact on many aspects of nonREM sleep. The circadian drive for wakefulness increases across the biological day, reaching its maximum in the evening hours when homeostatic sleep pressure is high, the so-called “wake-maintenance zone” [54, 55]. The circadian drive for sleep reaches its maximum during the early morning hours just prior to habitual awakening time, when homeostatic sleep pressure is low [29, 56]. Under ideal conditions, the circadian rhythm of sleep-wake interacts with the homeostatic sleep-wake process to allow for consolidated sleep (and wake) in humans [57-63]. Studies in young adults have demonstrated that even a small change in the circadian time of sleep can have large impact on the ability to consolidate sleep throughout the night. Thus, age-related changes in circadian rhythms or circadian sleep regulation may underlie the sleep timing and consolidation changes seen in aging, and if so may be a target for therapeutics to improve sleep.

As outlined in earlier chapters, circadian rhythms are endogenously generated oscillations in physiology and behavior with a near-24-hour period. Human circadian period averages slightly longer than 24 hours, with a range of about 23.5 – 24.5 hours in sighted adults [64-70]. The circadian system is synchronized to the 24-hour day by signals from the environment, a process called entrainment. In humans, as in most mammals, entrainment typically occurs via light-dark exposure. Light has a phase-dependent effect on the circadian system, meaning that the effect of a given light stimulus is dependent on the phase (or biological time of day) at which the light exposure occurs. Light exposure in the late evening and early night shifts the timing of rhythms later (phase delay shifts), light exposure in the late night and early morning shifts the timing of rhythms earlier (phase advance shifts), while light exposure in the middle of the biological day produces very small changes in rhythm timing [71, 72]. Plots of the magnitude of the phase shift with respect to the phase at which the light exposure was given are called phase response curves (PRCs). The phase relationship between the circadian system and the entraining signal is referred to as the phase angle (or phase angle of entrainment; see Figure 1). Circadian period interacts with the PRC in the entrainment process, and individuals with different periods (and/or different magnitude PRCs) will have different phase angles of entrainment [70, 73].

Open in a separate window

Figure 1

Schematic illustrating some of the key features of circadian rhythms. A. Phase (down arrows) refers to a reference point in the ~24-hour rhythm, in this case the peak of the rhythm. The duration from the phase on one cycle to the same phase on the next cycle (dashed line) is the period (cycle length) of the rhythm. Period can only be assessed under controlled experimental conditions. B. The near-24-hour circadian rhythms are entrained (synchronized) to the environment through periodic signals from the environment, typically light-dark exposure (indicated by the bar across the top of Panel B). The relationship between the entraining signal (here, lights on, represented by the right dashed vertical line) and the phase of the rhythm (here the peak of the rhythm, represented by the left dashed vertical line) is referred to as the phase angle of entrainment (horizontal arrow). This phase angle is dependent on the period of the rhythm, the strength of the entraining signal, and the phase-dependent response to that entraining signal.

Age-related changes in any of the structures involved in generating or entraining circadian rhythms, and/or age-related changes in any of the critical features or processes involved in entrainment may therefore contribute to altered circadian rhythm timing with advancing age. We outline below the evidence for alterations in circadian rhythms with age, and how these might contribute to age-related changes in sleep timing and consolidation.

Phase of Entrainment

The phase relationship between a circadian rhythm of interest and the signal from the environment that entrains the rhythm (typically the light-dark cycle) is referred to as the phase angle of entrainment [91]. There is evidence from animals that the timing of the rhythm of locomotor activity with respect to the timing of the light-dark cycle is altered in aging. Studies in hamsters found that activity onset is earlier with respect to lights out in older animals, and reentrainment after the light-dark cycle is shifted is faster [92, 93]. However, a study in mice reported delayed activity onset and slower reentrainment [94]. Studies of phase angle in humans have either reported no difference with age [83, 95-97], or have found that older people show an altered phase angle, such that the timing of the phase of their rhythms of core body temperature and melatonin occur later with respect to sleep (and lights out) [1, 86, 87]. This latter finding means that older adults are not only sleeping at an earlier clock time, they are sleeping at an earlier biological time.

Circadian Amplitude

There are numerous reports of reduced circadian rhythm amplitude with aging. In animals, reduced amplitude of the rest-activity cycle [94, 98-100], as well as the amplitude of multi-unit electrical activity in the SCN have been reported [101-104]. In studies of human circadian rhythms, most reports find a reduced temperature amplitude with age [31, 78, 82, 105], and many but not all find reduced amplitude of the rhythms of melatonin and other hormones [7, 96, 106]. While changes in the electrical activity of the SCN likely lead to alterations in output rhythm amplitude, the functional consequences of alterations in output rhythm amplitude are not well understood.

Circadian-Sleep Interaction

As outlined above, the circadian system interacts with a sleep-wake homeostatic process to regulate the timing and consolidation of sleep in humans. Studies using protocols such as the forced desynchrony protocol have been used to separate circadian from sleep-wake-dependent influences on sleep and waking performance, and to compare those influences between young and older adults. Those studies have demonstrated that the sleep of older adults is much more vulnerable to circadian misalignment than the sleep of young adults [12, 29, 31, 32, 39, 48, 107]. There is a much narrower range of circadian times when the end of sleep can remain consolidated in older adults compared with young adults, and a corresponding reduction in the range of circadian phases at which alertness and performance is impaired in older subjects [1, 62, 108, 109], suggesting an age-related reduction in the circadian drive for sleep in the early morning [12, 29, 31, 32, 39, 107]. Studies using ultra-short sleep-wake cycles have also reported a reduction in the circadian drive for wakefulness in the evening (the “wake-maintenance zone”) [39, 96]. Together, these findings suggest that there may be a reduction in the circadian rhythm of sleep-wake propensity that occurs in aging.

Circadian Period

It was hypothesized that a shortening of circadian period with age could explain the shift in sleep timing with age, and there was evidence from some animal studies that period was shorter in older animals [110-112]. An initial series of forced desynchrony studies in which circadian period was assessed in healthy older adults and compared with young adults found no difference in period with age [64], and in a follow-up study of a larger group of young and older adults we found the same result [70]. A study of six blind men who each had their period estimated twice over a ~10-year interval found no evidence for a shortening of period with age within an individual [113]. Together, these findings suggest that an age-related shortening of circadian period does not underlie the advance in circadian rhythms and sleep timing with age in humans.

Response to Light

Light is the primary environmental signal influencing circadian rhythms, and serves to synchronize the near-24-h circadian system to the 24-h environmental day [64-67, 91, 114-116]. Most humans spend relatively little time in outdoor levels of light [117-119] and therefore indoor light plays a dominant role in synchronizing circadian rhythms for most people. Use of artificial illumination in the evening has been shown to partially suppress and alter the timing of the melatonin rhythm and sleep in young adults [120-123]. Thus, the pattern of exposure to light in the evening is a likely mechanism contributing to circadian timing and sleep in older adults, and there is evidence to support age-related differences in light exposure patterns in older adults living in the community [124, 125].

Whether the response to light differs between young and older adults is also relevant to sleep and circadian rhythm timing, and a number of studies of the circadian response to light in older vs. young adults have been carried out over the past two decades, with mixed findings. Klerman and colleagues [126] used a bright (10,000 lux) light stimulus of 5 hours/day over 3 days in young and older adults, and delivered the light stimuli across the phase delay and the phase advance regions of the PRC. They found no evidence for an age difference in phase-shifting response when light was presented in the phase delay region, and a suggestion that there might be reduced responses in the older participants in the phase advance region. Benloucif and colleagues [127] used a 4-hour 3,500 lux stimulus delivered in the phase delay region and also found no evidence for an age difference in phase-shifting response. Kim and colleagues [85] tested a 2-hour light stimulus of 2,000 or 8,000 lux delivered at a variety of different phases and did not find significant differences in phase-shifting responses between young and older participants. Duffy and colleagues [128] used a 6.5-hour light stimulus in the phase delay region and tested a wide range of stimulus intensities. They found no difference in phase delay response for low (<100 lux) or high (>1,000 lux) light levels, but did find evidence for a reduced responsiveness among the older subjects in the intermediate range, with a half-maximal response shifted to 263 lux compared with 119 lux in the young adults.

While all the previous studies used polychromatic light sources, additional studies using monochromatic light stimuli have also been conducted. Herljevic and colleagues [129] used a 30-minute light stimulus of short (456nm) wavelength light delivered in the phase delay region and found significant differences in melatonin suppression between young and older women, but no age difference when a longer (548nm) wavelength light stimulus was used. In a study of 2 hours of intermittent short or long wavelength monochromatic light delivered in the phase advance region, Sletten and colleagues [130] reported that phase shifting responses were slightly larger in the young participants, although the difference was not statistically significant. Najjar et al. [131] studied a series of non-visual responses to monochromatic light in young and older adults and found a shift in peak sensitivity to longer wavelengths in the older participants, but no change in melatonin suppression. Thus, while there are some suggestions of changes in light sensitivity with aging in humans, the differences in response to light in healthy older adults are not strong, and additional research in this area is needed to better understand whether changes in light sensitivity contribute to sleep and circadian rhythm timing changes with age.

Light Transmission

The changes in circadian responses to light that have been observed in some studies may be due to age-related changes in the pathway through the eye, along the RHT, and/or within the SCN [132]. There is extensive evidence for changes in the transmission of light through the crystalline lens with age [132-134]. The aging lens accumulates yellow pigmentation which selectively reduces transmission of short wavelength light [131, 132, 134]. While the exact relevance of this for humans living freely in environments where they can control ambient polychromatic lighting is not yet clear, a study of nearly 1,000 Danish adults found that the age-related increase in yellowing of the lens was associated with greater reported sleep disturbances [135].

There are also changes in the pupil with aging, with older adults having a smaller pupil than young adults. Daneault and colleagues tested whether this impacts response to monochromatic light exposure. They found that while older adults had smaller pupils at dark-adapted baseline and at all light levels tested, the reduction in pupil size in response to light was not different between young and older subjects [136]. Thus, available evidence suggests that light transmission through the lens is altered with age, specifically reducing transmission of short wavelength light. Age-related changes in retinal function have also been reported in humans [137, 138], and there is a report that the number of ipRGCs declines with age in rodless-coneless mice [139], although that same group reported no change in responsiveness to light in the same type of older mice [140].

SCN

While studies of human SCN function cannot be carried out, there is a general consensus based on animal studies that there are age-related changes in the SCN (reviewed e.g. in [141]). Studies carried out more than two decades ago demonstrated that the locomotor activity pattern of aged animals was much more consolidated after transplantation of fetal SCN, suggesting that some unknown factor(s) that had declined with age had been re-introduced or improved [142-145]. There is strong evidence of altered patterns of electrical activity in the SCN of aged animals [101-103]. This is likely due to altered synchrony amongst SCN neurons, which leads to a reduced rhythm of multi-unit activity [104]. Within individual SCN cells, changes in cell membrane properties that alter electrical activity of the cells have been demonstrated in older animals [146]. There is conflicting evidence about whether the size or cell number within the SCN is altered with age [147-150], but there is general consensus that the aged SCN shows reductions in the number of cells expressing two major peptides, vasoactive intestinal polypeptide (VIP) and arginine-vasopressin (AVP) in both animals and humans [148-158].

Clock Gene Expression

There is some evidence from animal studies that clock gene expression is altered in aging, although not all studies are in agreement. One study found that expression of Per1 in response to an entraining light stimulus was reduced in aged hamsters, and this was associated with a significantly longer time to resynchronization [159]. That same study also found that the amplitude of Per1 and Per2 were not altered in older hamsters studied in constant darkness, but Bmal1 and Clock were altered in older hamsters. In a study of young and older mice, the amplitude of Per2 expression (but not expression of Per1, Clock, or Cry) was found to be reduced in the SCN of older mice [160]. A more comprehensive study of clock gene expression in young and older mice found age-related differences in expression of Per2, Bmal1, Rev-erbα, Dbp, and Dec1 expression in the SCN of the older mice [161]. Thus, even with the limited number of studies thus far, there is evidence that the molecular clockwork itself may be altered in aging, although much more research in this area remains to be done.

The authors wish to thank Ms. J. Hong for assistance with the manuscript. This work was supported in part by NIH grants P01 AG09975, R01 AG044416, and R01 HL094654. KMZ is supported by a fellowship from the Finnish Cultural Foundation and a grant from the Gyllenberg Foundation; EDC is supported by a fellowship from institutional training grant T32 HL007901.