Insulin resistance is a main determinant in the development of type 2 diabetes mellitus and a major cause of morbidity and mortality. The circadian timing system consists of a central brain clock in the hypothalamic suprachiasmatic nucleus and various peripheral tissue clocks. The circadian timing system is responsible for the coordination of many daily processes, including the daily rhythm in human glucose metabolism. The central clock regulates food intake, energy expenditure and whole-body insulin sensitivity, and these actions are further fine-tuned by local peripheral clocks. For instance, the peripheral clock in the gut regulates glucose absorption, peripheral clocks in muscle, adipose tissue and liver regulate local insulin sensitivity, and the peripheral clock in the pancreas regulates insulin secretion. Misalignment between different components of the circadian timing system and daily rhythms of sleep–wake behaviour or food intake as a result of genetic, environmental or behavioural factors might be an important contributor to the development of insulin resistance. Specifically, clock gene mutations, exposure to artificial light–dark cycles, disturbed sleep, shift work and social jet lag are factors that might contribute to circadian disruption. Here, we review the physiological links between circadian clocks, glucose metabolism and insulin sensitivity, and present current evidence for a relationship between circadian disruption and insulin resistance. We conclude by proposing several strategies that aim to use chronobiological knowledge to improve human metabolic health.
The circadian timing system consists of a central brain clock in the hypothalamic suprachiasmatic nucleus and peripheral clocks in tissues, including the liver, muscle, adipose tissue and pancreas.
Misalignment between different components of the circadian timing system and daily rhythms of sleep–wake behaviour and food intake might contribute to the development of insulin resistance.
Strategies to improve metabolic health by circadian synchrony include modulating light exposure, modulating rhythmic behaviour and chronotherapy.
Circadian molecules are a promising new treatment option for insulin resistance.
Insulin resistance in liver, muscle and adipose tissue is a pivotal pathophysiological process in the development of type 2 diabetes mellitus (T2DM), which has been designated as one of the four priority noncommunicable diseases by the WHO1. Major complications of T2DM are retinopathy, chronic kidney disease, neuropathy, peripheral vascular disease, myocardial infarction and stroke. The central approach in the prevention and treatment of insulin resistance is lifestyle modification. The second step is medication, with metformin being the cornerstone of the oral glucose-lowering drugs. If necessary, glycaemic control can be further improved with additional medication, including oral sulphonylureas, oral sodium–glucose cotransporter 2 (SGLT2) inhibitors, injectable glucagon-like peptide 1 (GLP1) receptor agonists or injectable insulin2. Traditionally, research has focused on the quantity and quality of physical activity, food intake and medication; however, circadian factors including the timing of light exposure, physical activity, food intake, medication and sleep–wake behaviour might also prove important for the prevention and treatment of insulin resistance.
The mammalian circadian timing system consists of a central brain clock in the hypothalamic suprachiasmatic nucleus (SCN), and peripheral clocks in other brain regions and tissues throughout the body, including muscle, adipose tissue and liver. The SCN receives a direct projection from the retina, via which environmental light synchronizes the approximately 24 h rhythm of the SCN with the exact 24 h rhythm of the environment (Fig. 1). The entrained timing signal from the SCN is forwarded via neural and hormonal signals and body temperature to the peripheral clocks. The molecular mechanism of the central and peripheral clocks is based on transcriptional-translational feedback loops, which are present in almost every cell of the human body.
In this Review, we describe the physiological links between circadian clocks, glucose metabolism and insulin sensitivity. We also present current evidence for the relationship between circadian disruption and insulin resistance, with a focus on human studies. Finally, we propose several strategies to implement chronobiological knowledge with the aim to improve human metabolic health. The chronobiology terms and metabolic terms we use are defined in Box 1 and Box 2, respectively.
Circadian control of insulin sensitivity
The circadian timing system
The mammalian circadian timing system is composed of a central pacemaker in the bilateral SCN of the anterior hypothalamus and a multitude of peripheral clocks in other brain areas and peripheral tissues (Fig. 2). The circadian timing system serves to prepare an organism for the alternating opportunities and challenges that go along with the rhythmic changes of the daily light–dark cycle. This concept is illustrated by the evolutionary advantage of an internal clock that matches the period duration of the environment, as demonstrated in cyanobacteria3,4.
The discovery of the molecular mechanism that keeps these clocks functioning was awarded the 2017 Nobel Prize in Physiology or Medicine. The central feature of this molecular mechanism is the transcriptional–translational feedback loop (TTFL) involving the core clock genes: the period genes (PER1, PER2 and PER3), cryptochrome genes (CRY1 and CRY2), ARNTL (also known as BMAL1), CLOCK (or its orthologue NPAS2) and the genes encoding the nuclear receptors REV-ERB (NR1D1 and NR1D2) and ROR (RORA, RORB and RORC). The rhythmic signal produced by this molecular clock has a period of approximately 24 h, which is a circadian period.
The period of the endogenous circadian timing system does not match the exact 24 h rhythm of the outside world and, therefore, has to be reset every day. Environmental light is the most important Zeitgeber for resetting the central pacemaker, reaching the SCN through a direct connection from intrinsically photosensitive retinal ganglion cells through the retinohypothalamic tract. The remaining clocks in the circadian timing system therefore depend predominantly on the SCN for entrainment to the light–dark cycle of the outside world.
The SCN sends its entrained timing signal to the peripheral tissue clocks through the autonomic nervous system, hormonal signals (including melatonin and cortisol), modulation of body temperature, and behavioural signals, such as physical activity and food intake. As most peripheral clocks do not receive direct light information, they are also sensitive to these other Zeitgebers. This scenario is especially true for peripheral clocks in metabolic tissues such as liver, white adipose tissue (WAT), brown adipose tissue (BAT), pancreas and muscle, which use the metabolic signals resulting from food intake for their entrainment5,6,7,8 (Fig. 1).
Circadian rhythm in glucose metabolism
In healthy humans, plasma glucose tolerance depends on the time of day of glucose ingestion, with glucose tolerance being higher in the morning than in the evening9,10. This diurnal rhythm in glucose tolerance is partly mediated by the diurnal rhythm in whole-body insulin sensitivity11. Moreover, the time-dependent glucose tolerance in healthy individuals is strongly mediated by the rhythm in pancreatic β-cell glucose sensitivity (that is, pancreatic glucose-induced insulin secretion), as demonstrated by studies that use hyperglycaemic clamping12 and a triple-tracer mixed-meal technique13. However, because participants in these studies were sleeping at night and were awake during the day, it remains unclear to what extent these morning–evening differences are the result of behavioural and environmental differences or are caused by a direct influence of the circadian system. A 2015 study using a circadian misalignment protocol demonstrated that the diurnal rhythm in glucose tolerance is robustly regulated by the circadian timing system, separate from influences of behavioural and environmental changes14. Consistent with the results from diurnal studies, the endogenous circadian influence on glucose tolerance results from a stronger β-cell response in the circadian morning14.
Clock control of insulin sensitivity
In this section, we discuss the clocks in the different tissues and organs that are involved in the control of glucose metabolism and explain their role in the regulation of insulin sensitivity and insulin secretion.
The central clock
The central clock in the SCN not only synchronizes the peripheral clocks described in the sections below, but also affects multiple processes that influence the diurnal rhythm in glucose metabolism, including the physiological daily rhythms in sleep–wake behaviour, food intake, hormone secretion, insulin sensitivity and energy expenditure (Fig. 3).
The SCN controls the daily rhythm in sleep–wake behaviour15,16 via its connections with hypothalamic areas such as the subparaventricular zone, the ventrolateral preoptic area and the dorsomedial hypothalamus17. The SCN presumably also has a direct role in the control of food intake18,19, which is enhanced via the regulation of the sleep–wake cycle since food intake requires a waking state. The circadian control of food intake might be mediated via direct neuroanatomical connections between the SCN and the hypothalamic arcuate nucleus, which is involved in the regulation of food intake20, but indirect connections between the SCN and areas involved in the rewarding aspects of food could also have a role21.
In addition to the aforementioned roles, the SCN controls the daily rhythm of release of several hormones affecting glucose tolerance. Firstly, the activity of the hypothalamic–pituitary–adrenal axis is regulated via connections from the SCN to the paraventricular nucleus, resulting in a diurnal rhythm of cortisol secretion, with a peak before the onset of the active period22. The glucocorticoid cortisol affects insulin signalling and reduces insulin secretion23. Secondly, the circadian rhythm of melatonin (also known as the hormone of darkness, as it is exclusively released during the dark phase in diurnal and nocturnal species alike), which affects insulin secretion24,25, is orchestrated by output from the SCN, via the paraventricular nucleus and the intermediolateral column to the pineal gland26. Thirdly, the diurnal rhythm in growth hormone, which antagonizes insulin action in liver and muscle27, is partly regulated by the SCN via its control of the sleep–wake cycle27,28,29.
Furthermore, SCN lesion studies in rodents demonstrated that the SCN controls the diurnal rhythm in whole-body insulin sensitivity30 and reported that within 8 weeks of an SCN lesion being created, rodents are insulin resistant31. In humans, a role for the SCN in the control of insulin sensitivity is suggested by misalignment protocols demonstrating the endogenous circadian control of glucose tolerance, independent of behavioural rhythms14,32.
Finally, the central clock is responsible for the circadian regulation of multiple components of energy expenditure, such as the sleep–wake cycle15,16, diet-induced thermogenesis33, resting energy expenditure34 and (at least in rodents) BAT activity35,36,37.
The gut clock
Glucose enters the body via the gastrointestinal tract. Intestinal cells throughout the intestinal tract contain a molecular clock38,39 and this gut clock is synchronized by signals resulting from food intake38. The gut clock regulates intestinal motility40 and nutrient absorption (Fig. 4). ARNTL regulates the expression of membrane glucose transporters and thus matches the timing of maximal monosaccharide uptake to the habitual feeding period41. Brush border disaccharidases, including sucrase, display a circadian rhythm in activity42,43, but the mechanism regulating this circadian activity remains to be elucidated (Fig. 4).
The muscle clock
Human skeletal muscle has an autonomous molecular clock44,45 (Fig. 5). Rodent data showed that the SCN synchronizes the skeletal muscle clock46,47, but signals resulting from physical exercise48,49 and food intake have also been shown to be involved in synchronization49,50,51.
Cultured rodent myotubes express circadian rhythmicity in insulin sensitivity52. CLOCK and ARNTL regulate muscle insulin sensitivity via changes in protein levels and membrane translocation of the insulin-sensitive glucose transporter GLUT4 (ref.53), as well as through the modulation of the insulin signalling pathway via expression of the deacetylase SIRT1 (ref.54). Furthermore, a 2017 study showed that the muscle clock regulates muscle insulin sensitivity via histone deacetylation of metabolic genes by HDAC3 (ref.55) (Fig. 5). Consistently, human muscle tissue shows a diurnal rhythm in insulin sensitivity with higher insulin sensitivity in the morning than in the evening56, as well as a diurnal rhythm in mitochondrial oxidative capacity, which peaks in the evening57.
The adipose tissue clock
WAT contains an autonomous circadian clock as shown in both rodent58,59 and human60,61 in vitro models (Fig. 6). Similar to the muscle clock, the WAT clock is synchronized by the SCN62 and by signals resulting from food intake63,64.
Adipocytes from rodents have circadian rhythmicity in glucose uptake52. In line with this observation, in human WAT ~25% of the transcriptome shows diurnal variation, including pathways involved in the regulation of glucose uptake65. Subcutaneous WAT explants from humans who are obese show an intrinsic diurnal rhythm in insulin signalling as determined by AKT phosphorylation, with peak insulin sensitivity at noon66. Rodent data indicated that this diurnal rhythm in adipose tissue insulin sensitivity could be the result of circadian regulation of the retinol-binding protein receptor STRA6 (ref.67). In addition, CLOCK and ARNTL regulate the expression of key enzymes in the regulation of lipolysis such as adipose triglyceride lipase (ATGL), lipoprotein lipase (LPL) and hormone-sensitive lipase (HSL)68,69 (Fig. 6).
The liver clock
The liver contains an autonomous clock that is synchronized by the SCN72,73 via a combination of autonomic signals and endocrine signals64 (Fig. 7). The liver clock also responds strongly to the timing of food intake, as the liver clock can be uncoupled from the SCN clock by inverting the daily feeding rhythm74. The liver clock regulates several pathways involved in the control of glucose and lipid metabolism, as indicated by microarray73,75,76, proteomic77,78 and metabolomic79,80,81 studies.
By synchronizing the diurnal rhythms in gluconeogenesis and glucose export to the habitual fasting period, the liver clock in rodents is essential to maintain euglycaemia82,83. The repression of gluconeogenesis during the feeding period is mediated by the interaction of CRY (which has its diurnal peak of expression during the feeding period) with the glucocorticoid receptor84 and with G protein-coupled receptor signalling85. The overall result of these interactions is the repression of the expression of rate-limiting gluconeogenetic genes. In addition, insulin-mediated suppression of gluconeogenesis is partly dependent upon CRY-mediated FOXO1 degradation86,87 (Fig. 7). In view of this information, it is tempting to speculate that the liver clock contributes to the diurnal rhythms in hepatic glycogen content88 and in hepatic insulin sensitivity13 that are observed in healthy individuals.
In addition to the regulation of gluconeogenesis, the liver clock regulates the diurnal rhythm in mitochondrial dynamics (Fig. 7). Therefore, the liver clock is involved in regulating mitochondrial glucose oxidation and fatty acid oxidation89,90, which protects the liver against oxidative stress during fasting91.
The pancreatic clock
The presence of an autonomous circadian pancreatic clock92 has been demonstrated not only in rodents93,94,95, but also in human islets and dispersed human islet cells (that is, the cells were cultured as seperate or single cells, not as an intact islet)96,97. The pancreatic clock is synchronized to the light–dark cycle95 via signals derived from the central brain clock in the SCN that include autonomic neuronal signals98, melatonin release93, glucocorticoid release96 and changes in body temperature96. The amplitude of oscillations in the expression of clock genes in cultured rat islets is dependent on the glucose concentration in the culture medium95.
Pancreatic islets isolated from rats show a circadian rhythm in insulin secretion93. CLOCK and BMAL1 activate the transcription of genes involved in insulin biosynthesis, insulin transport and glucose-stimulated insulin secretion99 (Fig. 8). In line with this observation, disruption of the pancreatic clock causes defective insulin secretion94,100,101. Similarly, in human pancreatic islets one group has confirmed that the pancreatic clock controls insulin secretion97.
Insulin resistance of liver, muscle and adipose tissue, which is initially compensated for by increased insulin secretion, is an early characteristic in the development of T2DM. Of note, in addition to insulin resistance, β-cell failure contributes to the development of T2DM102.
Insulin resistance in skeletal muscle is characterized by a reduced insulin-stimulated glucose uptake as a result of reduced insulin signalling and GLUT4 translocation103. As skeletal muscle is responsible for the majority of glucose uptake in the postprandial state104, skeletal muscle insulin resistance contributes to elevated postprandial glucose levels and reduced glucose tolerance.
The main role of the liver in the maintenance of glucose homeostasis is to release glucose — that is, endogenous glucose production — when plasma glucose and insulin levels are low. Under normal conditions, endogenous glucose production is strongly suppressed by insulin. Following hepatic insulin resistance, endogenous glucose production remains unsuppressed despite high plasma insulin levels, thereby contributing to elevated glucose levels via enhanced gluconeogenesis and reduced glucose uptake105. Insulin also suppresses de novo lipogenesis and VLDL production in the liver, and therefore hepatic insulin resistance is also characterized by elevated VLDL secretion106.
The role of adipose tissue insulin resistance in the development of T2DM is more indirect than that of muscle and hepatic insulin resistance. Insulin suppresses adipose tissue lipolysis; therefore, patients with T2DM are characterized by having elevated levels of plasma free fatty acids107. Tissues such as liver and muscle take up the additional circulating plasma free fatty acids, which contributes to ectopic lipid accumulation. This ectopic fat accumulation in itself strongly contributes to the development of liver and muscle insulin resistance108.
Circadian disruption and insulin resistance
The first clue that the circadian timing system might be involved in the pathophysiology of insulin resistance was the observation in the 1960s of an altered daily rhythm in glucose tolerance in patients with T2DM109. Later, observations including the development of metabolic syndrome in the Clock mutant mouse110, the discovery that food intake at the wrong circadian phase (the habitual sleeping phase) causes obesity in mice111 and the observation that circadian misalignment results in decreased glucose tolerance in humans112 led to the proposal of the circadian disruption hypothesis113. Sophisticated tissue-specific pancreatic94,99,100,114, hepatic82,91, muscle53,115 and adipose116 transgenic and knockout models gave further support for this hypothesis. On the other hand, several studies, including transgenic mouse models117,118,119 and studies with desynchronized food intake120,121,122, have not been able to confirm the circadian disruption hypothesis, as they reported no negative metabolic effects of circadian disruption. An overview of the metabolic phenotypes of published transgenic animal models is outside the scope of the present Review, but can be found in several previous papers123,124,125.
According to the circadian disruption hypothesis, metabolic health is optimal when the different daily rhythms, including the behavioural fasting–feeding and sleep–wake rhythms, hormonal and autonomic nervous system rhythms and central and peripheral clock rhythms, oscillate in synchrony with each other. By contrast, misalignment between certain components of this system, such as between behavioural and tissue clock rhythms, can result in circadian disruption and the development of insulin resistance and T2DM. In the sections below, we discuss epidemiological and experimental human studies that investigated the association between insulin resistance and several forms of circadian disruption (Box 3).
Clock genes in humans
In line with the rodent clock gene mutation studies, human mutations in several clock genes were shown to contribute to the genetic susceptibility to obesity, insulin resistance and T2DM. Observational studies have shown associations between single nucleotide polymorphisms in ARNTL126 and T2DM, between specific haplotypes of CLOCK and obesity127,128, between polymorphisms in CRY2 and elevated fasting glucose129,130 and between polymorphisms in the circadian clock gene NR1D1 and obesity131. Inspired by these findings, several investigators explored gene–behaviour interactions and showed that interactions between diet and clock gene mutations affect fasting glucose132, insulin resistance133,134, body weight135,136 and T2DM137.
We only identified three studies that investigated tissue clock gene expression rhythms in patients with T2DM. One study described a reduced amplitude of the daily rhythm in leukocyte clock gene expression in patients with T2DM138. Another study investigated the diurnal rhythm in clock gene expression in gluteal subcutaneous adipose tissue and surprisingly found no differences between lean participants, participants with obesity and patients with T2DM139. A study in circadian myotube explants described unaltered clock gene expression rhythms, but a decreased amplitude in NR1D1 expression in patients with T2DM44. In sum, indications of altered tissue clock rhythms in patients with T2DM are very limited.
Effects of light
Daylight is the main synchronizer of the central clock. Our modern lifestyle, however, is characterized by reduced light exposure during the day and increased light exposure during the night. These lifestyle changes have a substantial effect on the alignment of our circadian timing system to the solar day, as illustrated by an elegant study that investigated the effects of camping in natural light–dark conditions on human daily sleep–wake behaviour140.
In several animal models, investigators have shown that dim light at night disturbs diurnal rhythms of food intake and locomotor behaviour120,125,141, causing obesity and reduced glucose tolerance in mice125,141 but not in rats120. In line with these findings, observational studies in humans have shown correlations of exposure to light at night with obesity142,143 and T2DM144.
Under conditions of controlled food intake and physical activity, bright ambient light directly reduces insulin sensitivity in a time-dependent manner in healthy individuals145. When healthy participants are kept awake during the night, bright light causes increased levels of glucose in plasma146. In patients with T2DM, bright morning light increases fasting and postprandial levels of glucose147. A 2017 study in rats reported wavelength-dependent effects of ambient light on glucose tolerance, with white and green light but not blue and red light reducing glucose tolerance148, but whether these observations translate to humans remains to be determined.
Melatonin is secreted by the pineal gland and shows a pronounced diurnal rhythm. During the dark period, plasma levels of melatonin are high149, and melatonin secretion is acutely suppressed by light exposure26. Melatonin acutely increases insulin secretion in cultured human islets24. By contrast, melatonin administration in healthy women acutely decreases glucose tolerance150,151, an effect that is dependent on a common gain-of-function variant of the melatonin receptor gene MTNR1B152.
The role of melatonin signalling in the pathophysiology of T2DM remains a topic of lively debate153. On the one hand an association exists between reduced melatonin levels and the incidence of T2DM154, and rare loss-of-function mutations in MTNR1B are associated with an increase in the risk of T2DM155. On the other hand, one publication suggests that increased pancreatic β-cell melatonin signalling might reduce insulin secretion in humans25.
Accumulating evidence from both epidemiological and experimental studies shows that behavioural sleep–wake rhythms affect the risk of developing insulin resistance. A 2015 meta-analysis of prospective studies showed that both individuals who sleep for short periods and those who sleep for long periods are at increased risk of developing T2DM, with a proposed ‘optimal’ sleep duration of 7–8 h per night156.
The interpretation of epidemiological studies, however, should be made with caution. It has been suggested that the relationship between long sleep duration and adverse health could be the result of reversed causality, undiagnosed disease, residual confounding and the subjective reporting on sleep duration possibly representing time in bed157. On the other hand, investigators are in general agreement that poor sleep quality increases the risk of obesity and T2DM. A meta-analysis showed that people with reduced subjective sleep quality are at increased risk of developing T2DM158. In line with this finding, patients with obstructive sleep apnoea are at increased risk of developing T2DM, which could be mediated by increased food intake and/or decreased physical activity159,160, among other mechanisms, owing to disturbed sleep161.
Several well-controlled human experimental studies shed further light on the relationship between sleep deprivation and insulin sensitivity. The seminal experimental study showed reduced glucose tolerance after five nights of chronic partial sleep loss (4 h per night) compared with five well-rested nights (12 h per night) in healthy human participants, under conditions of controlled food intake and physical activity162. Subsequent experimental studies under controlled conditions confirmed reduced liver163, adipose164 and whole-body163,164,165,166,167,168 insulin sensitivity as a result of sleep restriction to 4–6 h per night for 1–14 nights in healthy individuals. By contrast, other studies under controlled conditions found only short-term effects169 or no effect170 of sleep restriction on insulin sensitivity, which could be the result of milder sleep restriction169 or a mitigating effect of the negative energy balance owing to the experimental design170. Further to these observations, studies have shown that experimental slow-wave sleep suppression resulted in reduced whole-body insulin sensitivity in healthy individuals171,172,173.
The proposed mechanisms for the effects of sleep restriction and sleep disturbance on insulin sensitivity include an altered sympatho–vagal balance162,171,172 and increased circulating levels of catecholamines167 or cortisol167,172. In conditions of ad libitum food intake, increased food intake as a result of sleep restriction or disturbance probably contributes to decreased insulin sensitivity159,160.
A systematic review and meta-analysis showed that in patients with established T2DM, individuals who sleep for a short duration or a long duration and individuals with lower sleep quality have reduced glycaemic control compared with individuals who get adequate sleep174. Although several of the studies included in the systematic review and meta-analysis corrected for physical activity, the meta-analysis did not correct for food intake or physical activity, so it is possible that these correlations are partly confounded by increased food intake or decreased physical activity174.
The incidence of obstructive sleep apnoea is high in patients with T2DM. Furthermore, in patients with comorbid obstructive sleep apnoea and T2DM, poor glycaemic control correlates with the severity of obstructive sleep apnoea, a finding that could again be partly confounded by increased food intake and/or decreased physical activity161.
Chronotype and social jet lag
An individual’s chronotype could also be a risk factor for insulin resistance. Observational studies show that evening chronotypes are at increased risk of developing T2DM compared with morning chronotypes175, even when results are corrected for sleep duration and physical activity (food intake was not corrected for in this study)176. Some evidence suggests that this increased risk could be the result of increased social jet lag — the discrepancy between the social (behavioural) and endogenous (circadian) time. Individuals with an evening chronotype who are working regular daytime hours are at increased risk of social jet lag.
Social jet lag is associated with the development of T2DM, independently of sleep duration177,178,179, even when results are corrected for food intake and physical activity177. Patients with T2DM who are evening chronotypes show worse glycaemic control compared with patients who are morning chronotypes, a finding that might, in part, be mediated by an increase in evening food intake180; however, an association between poor glycaemic control and chronotype, independent of sleep duration, total food intake and physical activity, does also exist180,181.
Shift work and jet lag
Shift workers are at increased risk of developing T2DM, as shown by a 2015 meta-analysis of observational studies182; the degree of increased risk relates to the number of night shifts per month183. This increased risk of T2DM might be mediated by a combination of acute and chronic effects.
Experimental circadian misalignment under strictly controlled conditions acutely decreases glucose tolerance and insulin sensitivity both in non-shift workers and in chronic shift workers14,32,112,184,185,186. To our knowledge, the chronic effects of repeated phase shifts on food intake, physical activity and insulin sensitivity have not been studied experimentally in humans, but several animal studies show that repeated phase shifts cause increased food intake, increased body weight and disturbed glucose metabolism187. A 2014 translational study showed that repeated jet lag in mice causes reduced glucose tolerance via disturbance of the intestinal microbiome. Fascinatingly, faecal transfer from jet-lagged humans into germ-free mice also reduced glucose tolerance in mice, suggesting that the microbiome clock might have an important role in the development of insulin resistance owing to repeated phase shifts188.
Does circadian disruption contribute to insulin resistance?
Taken together, the hypothesis that circadian disruption contributes to the development of insulin resistance in humans is supported by the following findings: decreased glucose tolerance caused by experimental circadian misalignment in humans; the association between human clock gene polymorphisms and insulin resistance; the experimentally observed effects of night-time light exposure and sleep disturbance on glucose metabolism; and the association of short sleep duration, long sleep duration, low sleep quality, late chronotype, social jet lag and shift work with insulin resistance. Therefore, it seems probable that disturbance of the central and/or tissue clock rhythms (Fig. 9) contributes to the pathophysiology of insulin resistance at the tissue level. Furthermore, circadian disruption might cause misalignment of nutrient fluxes. For instance, a mismatch between hepatic glucose production, muscle glucose uptake and carbohydrate intake could contribute to elevated levels of glucose and an imbalance between lipid storage in WAT, lipid oxidation in BAT and hepatic lipid production might contribute to ectopic lipid accumulation.
Circadian synchrony and metabolic health
Modulating light exposure
Light provides the main input for the SCN, and optimization of daily light exposure can therefore increase circadian synchrony140. To our knowledge, however, no published randomized controlled trials (RCTs) have investigated the effects of long-term natural light exposure on insulin sensitivity or glycaemia.
One potential strategy is the adaptation of architecture and/or indoor lighting conditions. For example, one RCT showed that supplementing daytime indoor light conditions with bright artificial light in homes for the elderly improves cognitive functioning, sleep quality and the diurnal rhythm of locomotor activity189,190. Another study found that increasing blue light intensity in the morning with a system of wavelength-controlled light bulbs and LEDs at home improved subjective sleep quality in elderly women compared with low morning blue light intensity191.
A second potential strategy is to limit the use of screens from computers, tablets and smartphones in the evening, or to use blue light filters with these devices. An experimental study showed that reading a paper book in the evening reduced sleep-onset latency and improved daytime alertness the next day compared with reading a book on a light-emitting tablet192.
Modulating rhythmic behaviour
In view of the strong association between disturbed sleep and impaired insulin sensitivity, sleep improvement could be a sensible approach in the prevention and treatment of insulin resistance193,194, but high-quality intervention studies are currently lacking (Fig. 9). One study in chronically sleep-restricted healthy individuals showed a correlation between improved indices of insulin sensitivity and increased sleep duration after 40 days of sleep extension (~45 min extra each night)195. A study on the metabolic effects of sleep extension in sleep-restricted individuals who are obese is ongoing196.
Evidence-based strategies to treat insomnia include cognitive behavioural therapy (high-quality evidence according to the GRADE criteria), light therapy (low-quality evidence) and exercise (low-quality evidence)197. A short course (<4 weeks) of benzodiazepines or benzodiazepine receptor agonists can be considered if these strategies are not successful197, but concerns exist of negative effects of these hypnotic drugs on glucose tolerance, possibly owing to a reduction in slow-wave sleep after hypnotic drug use193.
Individuals with obstructive sleep apnoea represent a unique population regarding sleep–wake behaviour. A 2017 meta-analysis of four RCTs assessed the effect of continuous positive airway pressure (CPAP) treatment on insulin resistance in patients with obstructive sleep apnoea who had either normal glucose values or prediabetes. The authors found no effect on HOMA-IR, but reported a small reduction of fasting insulin198. A 2017 meta-analysis showed that treatment of obstructive sleep apnoea with CPAP does not improve levels of haemoglobin A1c (HbA1c) or fasting levels of glucose in patients with obstructive sleep apnoea who have established T2DM, despite reduced daytime sleepiness199. The surprising lack of effect of CPAP treatment on glycaemia in patients with obstructive sleep apnoea could be related to treatment adherence199, as two studies with good adherence did show a decrease in HbA1c (ref.200) or mean 24 h glucose concentrations201.
Regular physical activity is one of the cornerstones of the lifestyle changes prescribed to patients with T2DM (Fig. 9). Regular physical exercise decreases insulin resistance and reduces HbA1c (ref.202). As physical activity also shifts the central circadian pacemaker in humans203, improves sleep duration and quality204 and affects the muscle clock48,49, it is possible that some of the beneficial metabolic effects of (daytime) physical activity are mediated through the circadian timing system. To our knowledge, however, no studies have proved this idea. We are also not aware of any studies investigating the optimal timing of physical exercise for the reduction of body weight and insulin resistance.
Individualized nutrition therapy is another core intervention in the prevention and treatment of insulin resistance and T2DM2. Classically, the main focus points for feeding behaviour are calorie reduction and healthy macronutrient distribution. An approach with a focus on the timing of food intake can also be of great value for people with insulin resistance205. A 2017 systematic review on meal timing and frequency in the prevention of cardiovascular disease proposed an approach to eating that included the recommendations to eat a greater share of calories early in the day and to use consistent overnight fast periods206 (Fig. 9).
The advice to consume a greater share of calories early in the day mainly results from RCTs showing that breakfast consumption (compared with breakfast skipping) improves insulin sensitivity207,208,209,210, although not all studies agree211,212. A hallmark study on the overnight fast in 156 North American individuals showed that most people do not consume the ‘normal’ three meals per day within 12 h, but instead showed an irregular eating pattern spread over a >15 hr period. In that same study it was reported that a small group of eight people classed as obese who were treated with ‘time-restricted feeding’ (that is, they were asked to restrict eating to a 10 h period for 16 weeks) lost 3 kg in weight, which persisted over 1 year213. In line with this finding, a 2018 randomized crossover trial in eight individuals with prediabetes showed that isocaloric early time-restricted feeding (that is, a 6 h feeding period with dinner before 15:00 h) reduces insulin resistance compared with a 12 h feeding period214.
Chronotherapy in patients with T2DM
In patients with T2DM, glycaemic control with oral glucose-lowering drugs and/or insulin reduces microvascular and cardiovascular complications2,215. To our knowledge and despite preclinical observations on the effects of metformin on the molecular clock216,217, the time-dependent effects of metformin on blood levels of glucose218 and the time-dependent effects of the sulphonylurea tolbutamide on insulin secretion219, no trials have assessed the chronotherapeutic effects of these frequently prescribed glucose-lowering drugs on clinically relevant outcomes.
The only potential example of evidence-based chronotherapy in T2DM is the dopamine agonist bromocriptine (Fig. 9). Dopaminergic activity shows a diurnal rhythm220 and dopamine signalling increases insulin sensitvity221. When administered in the morning, a quick-release bromocriptine formulation reduces HbA1c and fasting levels of glucose in patients with T2DM2,222. To our knowledge, however, no human trials have compared the effects of different administration times, which would be the ultimate proof that bromocriptine treatment is actually a chronotherapy.
The effects of timed melatonin administration in patients with T2DM have been investigated in one small RCT with 36 participants, which compared 3 weeks of melatonin administration with placebo223. The investigators reported no convincing evidence of beneficial metabolic effects of melatonin.
The insulin requirements of patients on insulin therapy vary over the diurnal cycle owing to the diurnal rhythms of sleep–wake behaviour, physical activity, food intake and insulin sensitivity. The research community has made tremendous efforts to improve insulin pharmacokinetics with the aim of matching them to the individual patient’s diurnal pattern in insulin requirements224,225. The question of whether the beneficial effects of insulin are (partly) mediated through central or peripheral clock modulation, however, remains to be resolved.
The artificial pancreas — which consists of an insulin pump controlled by a control algorithm coupled to a continuous glucose sensor226 — was in 2016 shown to increase the length of time spent in target glucose ranges in patients with T2DM who had been admitted to hospital227. One possible approach to further improve the algorithm controlling the artificial pancreas for patients with T2DM would be to incorporate information on the diurnal rhythm of insulin sensitivity.
The administration of exogenous glucocorticoids causes insulin resistance23. Data from a subgroup analysis of a small open-label randomized trial228 and two prospective cohort studies229,230 suggest once-daily modified-release hydrocortisone formulations might be beneficial for metabolic health compared with thrice-daily immediate-release hydrocortisone. Analysis of the small subgroup of patients with comorbid adrenal insufficiency and diabetes mellitus showed that replacement therapy with once-daily modified-release hydrocortisone formulations (which mimic the physiological diurnal rhythm of cortisol levels) might lead to improvements in body weight and HbA1c compared with thrice-daily immediate-release hydrocortisone.
New circadian therapies might arise from large-scale chemical screens looking for clock-improving molecules231,232,233 (Fig. 9). Promising candidates include the REV-ERBα agonist SR9011 and the REV-ERBβ agonist SR9009, both of which directly target the molecular clock and were shown to decrease obesity and hyperglycaemia in diet-induced obese mice. Timed twice-daily administration of these REV-ERB agonists alters metabolic gene expression patterns in muscle and WAT, leading to increased muscle glucose and fatty acid oxidation (increased energy expenditure), in combination with decreased WAT triglyceride synthesis234.
Another promising candidate is the natural citrus compound nobiletin, which has been shown to reduce body weight and improve insulin sensitivity in two different mouse models of the metabolic syndrome (diet-induced obese mice and db/db mice). Nobiletin directly targets the molecular clock by activating RORα and RORγ, thus increasing the amplitude of circadian locomotor behaviour, the rhythm of tissue clock gene expression and the rhythm of hepatic metabolic gene expression. As a result, energy expenditure increases, adiposity decreases and hepatic steatosis decreases235. Finally, in preliminary reports, two different CRY stabilizers were shown to improve glucose tolerance in diet-induced obese mice236 and db/db mice237. The exact mechanism responsible for these metabolic benefits, however, remains to be elucidated.
In conclusion, REV-ERB agonists, ROR agonists and CRY stabilizers are promising circadian molecules for the treatment of T2DM, and human phase I studies of these compounds are to be expected.
Despite the large body of evidence from animal studies, the exact mechanisms mediating the metabolic derangements resulting from circadian disruption remain to be resolved. For example, does circadian misalignment cause a mismatch of glucose and lipid fluxes between the various organs or do disrupted tissue clocks cause insulin resistance at the tissue level, or are both mechanisms involved?
Currently, the clinical utility of the knowledge on circadian clock regulation of insulin sensitivity is only beginning to be explored. A clear need exists for RCTs that investigate the metabolic effects of natural light–dark exposure, sleep improvement, time-restricted feeding and the daily timing of exercise. Clinical trials are needed that investigate methods to prevent metabolic complications in shift workers. With regard to biomarkers, evidence suggests that circadian phase biomarkers can help to optimally synchronize the circadian timing of behavioural or pharmacological interventions238. We are in no doubt that new circadian molecules targeting the molecular clock will be identified within the next 10 years. Furthermore, mathematical models could be an important aid to predict the effects of timed administration of clock agonists239. We expect the further development of promising candidate circadian molecules, including nobiletin, in phase I human trials in the coming years.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
The time difference between two consecutive peaks or troughs, or any other fixed point in the rhythm. In the case of daily or circadian rhythms, this period is exactly or approximately 24 h, respectively. The period of the rhythm in constant conditions is called the free-running period and is denoted by the Greek letter τ.
On a line graph, the amplitude is half the distance between the peak and trough of a daily or circadian rhythm.