Keywords

1 Markers of Circadian Rhythms in Humans In Vivo

The master clock is located in the suprachiasmatic nucleus (SCN) of the hypothalamus [1, 2]. Peripheral cell-autonomous clocks have also been observed in tissues outside the brain, such as fibroblasts, red blood and mononuclear cells, adipose tissue, pancreatic islet cells, skeletal myotubes, hepatocytes, and cardiomyocytes, and in immortalized cancer cell lines [3–14]. Unlike nonhuman models, scientists do not have direct access to the SCN in humans and instead use marker rhythms driven by the SCN to indicate phase, amplitude, and period of the circadian clock. Estimations of circadian timing of the internal biological clock (i.e., phase) and the strength or robustness of the observed oscillation (i.e., amplitude) are informative when determining changes in response to environmental or physiological perturbations. Phase represents the time within the circadian cycle at which a particular event occurs (e.g., minimum, maximum, onset, offset, midpoint), whereas amplitude is commonly defined as the magnitude between the mesor and the maximum of a rhythm. The mesor is defined as the value midway between the maximum and minimum of a fitted rhythm or time series, and thus amplitude is approximately half of the minimum to maximum range of the rhythm.

1.1 Melatonin

The most commonly used circadian marker rhythm in humans is the melatonin rhythm. Melatonin is easily measured in saliva, blood, and urine and is the most precise marker of circadian phase and period [15–19] when exposure to light is controlled. High melatonin levels are considered to be a marker of the biological night (Fig. 13.1). During entrainment when the circadian clock is synchronized to the 24 h day, melatonin levels rise on average ~2 h prior to habitual bedtime [20], peak during the nighttime, and return to low levels shortly after habitual wake time [21]. Note that there are large individual differences in the timing of the melatonin rhythm, which are larger in the modern environment versus after exposure to the natural light-dark cycle [21]. Further, there are large individual differences in peak melatonin levels [22]. A common misperception is that exposure to darkness increases melatonin levels. Rather, the SCN controls the melatonin circadian rhythm via a multisynaptic pathway. This pathway includes an efferent projection to the paraventricular nucleus of the hypothalamus (PVN) and a descending projection to sympathetic preganglionic neurons in the upper thoracic spinal cord which, in turn, project to the superior cervical ganglion (SCG). Postganglionic nerve fibers from the SCG then release norepinephrine to stimulate beta- and alpha-adrenergic receptors on the pineal gland [23]. Beta-adrenergic receptor activation signals the pineal gland to synthesize melatonin from tryptophan via several enzymatic steps. If maintained in constant conditions, the melatonin rhythm continues to rise and fall independent of light exposure. If exposure to light occurs during the biological night, melatonin levels will be acutely reduced as photic input from the retina to the SCN results in inhibition of the multisynaptic SCN-pineal circuit, removing the rate-limiting step of melatonin synthesis. Thus, to accurately assess melatonin levels, samples must be collected every 30–60 min under dim light conditions (e.g., <8 lx maximum).

Fig. 13.1
figure 1

Phase relationships between the primary circadian phase marker rhythms driven by the SCN in humans. High melatonin and low core temperature levels represent the biological night, whereas low melatonin and high core temperature levels represent the biological day. Cortisol levels are lowest in the first part of the biological night, peak near the end of the biological night, and decrease across the biological day

Full size image

A variety of markers have been developed to quantify the timing of melatonin rhythm. Changes over and under arbitrarily defined thresholds are often used. For example, melatonin onset is most commonly defined as when melatonin levels rise above 10 pg/ml in plasma, or 3–4 pg/ml in saliva, as salivary levels are 30–40 % of plasma levels [19, 24]. Other thresholds for melatonin onset are when melatonin levels rise two standard deviations above a stable low daytime baseline [19] and the time of a threshold change is calculated by linear interpolation. If the entire melatonin rhythm is assessed, individualized thresholds can be computed to determine the thresholds of when melatonin levels reach 25 % of the three-harmonic fitted peak-to-trough melatonin amplitude [18] and the 50 % mean crossing [15], for example. Melatonin offset is considered the time at which melatonin levels return to low daytime levels, falling below these thresholds, and the melatonin midpoint is the time midway between the onset and offset. The fitted melatonin peak has also been used as a phase marker, and comparisons of the various markers show similar variability in their estimates of circadian phase [15].

1.2 Body Temperature and Cortisol

Two other commonly used circadian marker rhythms in humans are body temperature and cortisol (Fig. 13.1). The SCN controls rhythms in body temperature via multisynaptic projections to the preoptic temperature control center of the hypothalamus, and through direct effects of melatonin on peripheral vasodilation. The SCN also controls rhythms in cortisol via input into the endocrine hypothalamic pituitary adrenal (HPA) axis, as well as through a multisynaptic neural pathway to the adrenal glands that bypasses the HPA axis [25]. Core and distal skin (e.g., hands and feet) body temperatures show approximate inverse rhythms with high core and low distal skin temperatures during the daytime and low core and high distal skin temperatures at night. The fitted minimum of the core body temperature rhythm is the most common circadian temperature phase marker. Accurate assessment of the circadian rhythms in body temperatures requires control of posture, activity, and sleep, as changes in these factors alter the observed temperature rhythm (e.g., physical activity acutely increases core temperature and sleep reduces core temperature). Cortisol levels peak in the morning, decrease across the day, are low near habitual bedtime, and rise throughout the night (Fig. 13.1). Given the pulsatile nature of cortisol, accurate assessment of the cortisol rhythm requires frequent sampling (e.g., every 20–30 min).

2 Protocols to Evaluate Circadian Phase and Amplitude in Humans

2.1 Constant Routines

The constant routine protocol, a modification by Czeisler and colleagues (discussed in [26]) of the Mills test [27], can be used to assess the phase and amplitude of the clock immediately upon release from entrainment into “constant conditions” (Table 13.2). The constant routine protocol controls for factors that influence circadian variables of interest by making constant or equally distributing factors such as ambient light and temperature, physical activity and posture, nutrition intake, and sleep-wake state across the circadian cycle (Fig. 13.2, Table 13.1). Constant environmental conditions include dim-ambient light commonly maintained at ~1.5 lx in the angle of gaze (~0.6 W/m2), ambient temperature in the thermoneutral range (e.g., 22–24 °C), and a light bed sheet pulled up to the waist to maintain a constant temperature microclimate. Constant behavioral conditions include continuous wakefulness to control for sleep-induced changes in physiology and constant posture (i.e., bed rest with the head raised to a 35–45° angle) to control for posture or activity-induced changes (e.g., participants remain in bed and use bedpans/urinals). Continuous monitoring by research staff is required to ensure constant wakefulness and consistent posture. Concurrent brain wave recording is also recommended to maximize continuous wakefulness, as microsleeps will occur and unattended subjects will fall asleep. Nutrition intake is often in the form of miniature snacks distributed equally across the circadian cycle (e.g., hourly) or via less commonly used continuous enteral feeding or venous glucose infusion (Table 13.1). Meals are typically identical across the constant routine (e.g., one-fourth of a ham sandwich, room-temperature juice and water), isocaloric, and total daily calories consumed are increased by ~5 % to account for the higher energy needs associated with extended wakefulness [28]. At a minimum, ~24 h is required to assess a full circadian cycle if melatonin is used as a circadian phase marker, and more time (i.e., 1.5–2 days) is needed if body temperature is used, given residual effects of prior sleep on temperature. Constant routines are often ~40 h in duration and thus recovery sleep occurs at habitual bedtime.

Fig. 13.2
figure 2

The constant routine protocol is typically preceded by subjects maintaining their habitual and consistent wake-sleep schedule in the home environment. The first one to two nights in the laboratory may consist of the habitual wake and sleep schedules with typical meals. The constant routine protocol on days 3–4 is composed of miniature hourly snacks (s), constant posture, wakefulness, dim light, and ambient temperature. Participants are continuously monitored during the constant routine to ensure wakefulness and compliance with procedures. A 40 h constant routine permits assessment of multiple measure of circadian phase such as body temperature, melatonin, and cortisol and recovery sleep at the habitual sleep time. Constant routines also commonly precede and follow exposure to a phase-shifting stimulus to determine the change in circadian phase. B breakfast, L lunch, D dinner, S snack, orange boxes habitual meals, black boxes scheduled sleep, underline scheduled wakefulness

Full size image

Comparisons between the constant routine and typical sleep-wake conditions demonstrate that the circadian rhythm in melatonin is minimally influenced by sleep-wake state [29], assuming that ambient light is maintained at dim levels. Circadian rhythms in body temperature and cortisol are, however, influenced by sleep-wake state [21, 29, 30]. The sleep-induced decrease in core temperature [31] (increase in distal skin temperature) and posture-/activity-induced increase in core temperature (decrease in distal skin temperature) can mask the circadian temperature rhythms, resulting in imprecise estimates of circadian phase and amplitude. Similarly, there is a sleep-induced decrease in cortisol levels shortly after the beginning of the sleep episode and a wake-induced increase [32, 33].

Limitations of the constant routine procedure include constant wakefulness, which requires a degree of sleep deprivation that may influence outcomes of interest (i.e., circadian time is not the only factor changing); time (i.e., minimum of 24 h); protocol costs; and a tightly controlled environment which cannot be easily performed outside of the laboratory. Additionally, the constant routine cannot be used to assess circadian period nor interactions between circadian phase and wake-sleep influences on physiology and behavior (Table 13.2). Furthermore, like most laboratory protocols, findings may not translate directly to real-world conditions.

Variations on the constant routine include constant posture protocols that either permit sleep and typical meals [34] or modified constant routines that permit sleep and some changes in posture (e.g., bathroom breaks). Constant posture and modified constant routine protocols still include control for dim-ambient light and thermoneutral ambient temperature and can therefore still be used to determine circadian timing of the melatonin rhythm. The constant posture protocol has been used to measure the melatonin rhythm during wakefulness and sleep from blood samples taken via an indwelling catheter with an extension tubing that exits a room porthole to allow blood to be assessed with minimal disruption of sleep [34]. Individuals can also be awakened from sleep to obtain saliva samples from which to assess melatonin levels. When assessing the melatonin rhythm using saliva instead of blood sampling, food and fluid intake are typically proscribed ~30 min prior to collecting saliva samples, and the mouth is rinsed 30 min prior to sample collection to reduce the risk of food contamination of saliva samples. Furthermore, if posture is not constant, seated posture is typically maintained for ~15 min prior to sample collection. If sleep or changes in posture are permitted, circadian body temperature rhythm phase cannot be precisely determined [35, 36].

Limitations of constant posture and modified constant routines include time (i.e., minimum of 24 h) and protocol costs, and require a tightly controlled laboratory environment. These routines cannot be easily performed outside of the laboratory. Additionally, when daytime sleep opportunities are allowed, total sleep time is reduced relative to nighttime sleep opportunities [37], and such differences in total sleep time could influence outcomes of interest.

2.2 Assessment of the Dim-Light Melatonin Onset

The phase of the melatonin rhythm can also be assessed without performing a constant routine. As noted, melatonin, as opposed to temperature and cortisol, is less impacted by posture and meals. As long as light is maintained at dim levels and food and posture are controlled prior to sample collection (e.g., food proscribed 30 min and posture consistent 15 min immediately prior to the sample), saliva and blood samples can be used to accurately assess melatonin levels. Generally, the dim-light melatonin onset (DLMO) can be determined from samples obtained starting ~7 h prior to and ending ~1 to 2 h after habitual bedtime, assuming that the subject is stably entrained. Saliva sampling can also be easily performed outside of the laboratory when measuring and controlling light levels [38–40].

Limitations of the DLMO assessment include the inability to assess melatonin amplitude or other melatonin markers such as the midpoint and offset and melatonin duration. If the subject is not entrained, 24 h sampling may be required to obtain the melatonin onset, which would permit assessment of melatonin offset, midpoint, and amplitude.

3 Protocols to Evaluate Circadian Period and Amplitude in Humans

Accurate assessment of circadian period in sighted humans requires assessment in the absence of external synchronizers, or under tightly controlled exposure to synchronizers, which ensures their even distribution with respect to circadian phase.

3.1 “Free-Running” Temporal Isolation Protocol

In the 1960s Aschoff and Wever began performing their classic bunker studies in an isolation unit free from natural time cues [41]. Subjects lived in isolation, typically self-selecting their own meal, sleep-wake and light-dark schedules. These and related studies provided information on fundamental concepts of circadian physiology. For example, findings showed circadian rhythms in body temperature and associations between body temperature and sleep, e.g., subjects most commonly chose to initiate sleep on the downward curve of the body temperature rhythm [42, 43] and slept the longest when sleep was initiated near the body temperature minimum. Observed periods for urine corticosteroids and rectal temperature were on average ~25 h [44], and shorter periods were observed in subjects who napped versus those who maintained one primary sleep episode per “day” [45]. Limitations of these “free-running” temporal isolation protocols include subjects self-selected environmental (e.g., light) and behavioral (sleep and meals) events at a limited number of circadian phases. Thus, the nonuniform distribution of zeitgebers, or synchronizers, with respect to circadian phase likely had an impact on the outcomes observed [17, 46]. Experiments by Middleton et al. [47, 48] performed in constant dim light with self-selected sleep-wake schedules, but with knowledge of clock time, revealed circadian periods closer to 24 h (e.g., 24.3 h).

3.2 Forced Desynchrony in Laboratory Conditions

In 1938, Nathaniel Kleitman and Bruce Richardson performed a month-long study ~30 m underground in Mammoth Cave, Kentucky. During their study, they pioneered the forced desynchrony protocol. Kleitman and Richardson lived on a 28 h day and compared circadian temperature rhythms on 28 h versus 24 h day for 1 week each [49]. Regardless of the day length, ~24 h body temperature rhythms were observed, indicating that the ~24 h circadian rhythm in humans is not dependent upon the environmental light-dark cycle. Modern versions of the forced desynchrony protocol established by Czeisler and colleagues [17, 18, 31, 34, 44, 50–55] scheduled individuals to live in the laboratory in dim light-dark, wake-sleep cycles that are outside the range of entrainment of the human circadian clock (e.g., 20 h, 28 h, or 42.85 h days; Fig. 13.3, Tables 13.1 and 13.2). These laboratory studies last for at least 1 week on the non-24 h day length and often 2–4 weeks in order to obtain precise estimates of circadian period upon release from entrainment and revealed circadian period estimates close to 24 h (i.e., ~24.15 h) [17, 18, 53]. In these studies, participants live in an environment free of time cues (i.e., no access to timepieces, sunlight, or electronic devices), and light-dark exposure, physical activity, and food intake are all controlled. The specific day lengths noted above were chosen as their harmonics do not overlap with 24 h (i.e., the sought-after period) nor do their harmonics share any harmonics with a ~24 h period [17].

Fig. 13.3
figure 3

Double raster plot of the 28 h forced desynchrony protocol during which subjects initially start off on a 24 h day (T = 24 h). Typically, participants first undergo a 40 h constant routine (not shown here; but see Fig. 13.2) and then transition to a non-24 h day length that is outside the range of entrainment of the near 24 h period of the human circadian clock under controlled dim light conditions. Common forced desynchrony day lengths include T = 28 h (shown), 20 h, and 42.85 h. The example 28 h day length shown includes a 2:1 wake to sleep ratio of scheduled 18 h and 40 min wake and 9 h and 20 min sleep. The forced desynchrony is also often followed by a second constant routine protocol to assess circadian phase. The blue line illustrates the trajectory of a core body temperature minimum with a period longer than 24 h

Full size image

The traditional forced desynchrony protocol maintains the habitual 2:1 wake to sleep ratio (e.g., 18 h and 40 min scheduled wake and 9 h and 20 min scheduled sleep for the 28 h day), although modifications have been made to examine interactions between sleep homeostasis and circadian phase [56–62]. When scheduling wake-sleep for a 2:1 ratio of the imposed day length, sleep duration is not equal at all circadian phases and will be shorter during the biological day [50, 63]. Furthermore, the forced desynchrony protocol assumes a symmetrical phase response curve for any photic and non-photic phase shifts that may persist despite the minimization of these influences. Relative coordination between observed rhythms—for example, between the melatonin rhythm and wake-sleep cycle—may result and thus influence the observed melatonin rhythm. However, this does not appear to affect the estimation of period, providing that the data train is sufficiently long [17]. The traditional forced desynchrony protocol requires a facility with maximal control over environmental conditions and has a high cost in both subject and experimenter time (Table 13.1). Related recruitment of participants for such longer term in-laboratory studies is laborious, and large demands are placed on laboratory resources for such studies.

Variations on the forced desynchrony protocol include ultrashort sleep-wakefulness cycles. Ultrashort sleep cycles schedule sleep and wake in brief segments across one or more circadian cycles. Versions include the 90 min day with scheduled 60 min wake and 30 min sleep opportunities [64] and the 20 min day with scheduled 13 min wake and 7 min sleep opportunities [65] (Fig. 13.4). The ultrashort versions of the forced desynchrony protocol are more cost-effective and provide a similar overall circadian period estimate as the traditional forced desynchrony protocol [66–68]. However, the precision and the error in the estimate for individual participants have not been reported but are likely larger than the traditional forced desynchrony estimates. Limitations of the ultrashort forced desynchrony protocol include an inability to assess circadian phase and amplitude (with the exception of melatonin phase if light is dim during scheduled wake) (Table 13.2). Further, this approach results in an inability to assess interactions between circadian phase and changes in the duration of wakefulness or sleep on outcomes. Like the traditional forced desynchrony protocol, sleep duration is not equal at all circadian phases.

Fig. 13.4
figure 4

The ultrashort wake-sleep schedule may include a typical night of sleep in the laboratory followed by a day (shown) or 24 h or more of wakefulness (not shown), followed by an alternating schedule of wake and sleep (60 min wake and 30 min sleep, in this example) for 1 or more days

Full size image

3.3 Forced Desynchrony in Free-Living Conditions

Another approach to examine circadian period has been to study visually impaired individuals living outside of the laboratory. Studies of blind individuals in the real world provide circadian period estimates that are similar, but longer, to those found in forced desynchrony protocols, but with less precision due in large part to the infrequent sampling of urine (e.g., every 4–8 h) for circadian phase markers in these studies [69]. Other limitations include the long duration of such field studies and uneven distribution of non-photic time cues in the blind (e.g., activity, sleep-wake cycle, meal times, working hours, social interactions, alcohol, caffeine, medication) that may induce daily advances or delays of the circadian system, effectively shortening or lengthening the measured period. Some blind individuals also maintain photic input into the SCN. Another difference from most laboratory studies is that the observed period in the blind is likely assessed long after the “release” from entrainment and therefore may have different properties.

3.4 Analysis of Data from Forced Desynchrony Protocols to Estimate Circadian Period

Frequently sampled melatonin, temperature, and cortisol data from forced desynchrony protocols are often analyzed with harmonic regression models using an exact maximum likelihood fitting procedure [17]. Data can be fitted with periodic components of the imposed sleep-wake cycle and the sought-for circadian period, together with their harmonics. Such techniques utilize the frequently sampled data available in the dataset and are thus robust and provide the most precise estimates of circadian period and amplitude. Alternatives include fitting linear regression through daily circadian phase estimates, which are less precise as they include error in the phase estimate as well as show higher variance in the period estimate. As with all circadian protocols, the derived estimates of phase, amplitude, and period are only as accurate as the robustness of the analytic techniques and degree of experimental control over factors impinging on the outcome of interest. Masking effects may influence the observed rhythms without impinging upon, and thus not representing, the phase, amplitude, and period of the master clock in the SCN.

4 Protocols to Evaluate Circadian Rhythms in Physiology and Behavior

Beyond using the above circadian protocols for assessment of phase, amplitude, and period of the master clock in humans, other physiological and behavioral data can also be collected and aligned to the known circadian phase makers (e.g., melatonin or core body temperature) to provide evidence for circadian variation in such parameters (Table 13.2). For example, constant routine protocols have shown circadian variation in blood pressure and thyroid-stimulating hormone (TSH) [29, 70], whereas daily patterns in prolactin, human growth hormone (hGH), and parathyroid hormone (PTH) are sleep-wake dependent [29]. In addition, many aspects of cognitive function, including reaction time, cognitive processing speed, math processing speed and accuracy, visual search abilities, executive function/decision-making skills, as well as alertness, sleepiness, mood, hunger, and appetite are influenced by time awake and/or circadian time of day [50, 55, 71–75]. When considering sleep-wake versus circadian-dependent variables, it is important to consider whether altered sleep duration is playing a role in the observed changes. For example, it is known that hunger and appetite hormones leptin and ghrelin, as well as free fatty acids, are altered during sleep restriction [76, 77]. Therefore, sleep duration must be taken into account when considering the effects of these circadian protocols on outcome variables.

Another circadian protocol used to examine circadian versus wakefulness-sleep patterns in physiology is to shift sleep to the daytime (Fig. 13.5) [37, 78], reflective of patterns found in night work, to see whether daily patterns remain synchronous with the melatonin circadian rhythm, move with the wake-sleep cycle, or are abolished. Using such protocols, findings provide evidence that hGH and peptide-YY (PYY) are wake-sleep/feeding-fasting driven, whereas diet-induced thermogenesis is primarily influenced by the circadian clock [79, 80]. Other manipulations may then be necessary to determine the specific non-circadian factors driving such patterns that are altered by a shift in wake-sleep (e.g., wake-sleep, feeding-fasting, physical activity-inactivity).

Fig. 13.5
figure 5

The shift of sleep to the daytime is a common simulation of a shift work schedule with 1 to 2 days of sleep at night followed by a transition day with a daytime nap and 1 or more days with daytime sleep and evening/overnight wakefulness

Full size image

Forced desynchrony protocols with a reasonable amount of “daily” wake time have been use to describe interactions between circadian and wake-sleep-driven processes. Findings from forced desynchrony protocols have shown many physiological and behavioral variables which are under circadian control and/or interact with wake-sleep factors. For example, findings from forced desynchrony protocols have shown circadian rhythms in physiology such as blood pressure [70], epinephrine, norepinephrine, heart rate, platelet aggregability [81], glucose tolerance in response to meals [82–84], EEG activity during wakefulness and sleep, susceptibility to presyncope [85], and periodic limb movements [86]. Some of these outcomes, including EEG and performance, show changes that are dependent upon the levels of sleep homeostasis and circadian phase and their interaction. For example, in many of the papers cited above, the amplitude of the circadian variation in performance builds with time awake.

Building on this, recent research has focused on the effects of sleep and circadian manipulations on human metabolomics and transcriptomics [87–91] in an effort to elucidate altered mechanisms and biochemical pathways by which these manipulations confer increased risk for disease states and to identify potential health and disease biomarkers.

5 Summary and Conclusions

The circadian protocols reviewed above have been used to elucidate much information about physiology and behavioral outcomes that are controlled and/or modulated by circadian timing. In the end, when interested in determining mechanisms that contribute to daily patterns in physiology and behavior, it is important to recognize that the integrated timing of circadian clock, wake-sleep, feeding-fasting, physical activity-inactivity, and light-dark cycles likely serve to enhance daily oscillations and promote robustness. A well-tuned system that has robust and coordinated oscillations in these parameters is likely to promote maximal cognitive and physiological health outcomes. Understanding the mechanisms underlying daily patterns in physiology and behavior also permits the development of countermeasure strategies for when these daily patterns are unavoidably disrupted, such as during shift work, jet lag and circadian rhythm, and/or sleep-wake disorders. Additional research is needed to understand the interacting factors that produce robust daily patterns in physiology and behavior, some of which will be circadian and others that will be driven by other factors. For example, manipulation of meal timing, physical activity, and environmental conditions in the above circadian protocols will add to our understanding of biological mechanisms controlling daily patterns in physiology.