In the SCN, as well as in most peripheral cells, a transcriptional-translational feedback loop (TTFL) occurs in the expression of a handful of clock genes, which lasts ~24 hours [17]. The TTFL consists of a positive and negative feedback loop (Figure 2). In the negative feedback loop, a heterodimer composed of BMAL1 and CLOCK or NPAS2 [18] binds to the E-box of three different period (Per) genes and two different cryptochrome (Cry) genes. After translation and post-translational modification, PER and CRY proteins translocate back into the nucleus, where CRY interacts with the CLOCK-BMAL1 heterodimer to inhibit its own transcription. As a result of this negative feedback loop, Bmal1 mRNA reaches peak levels twelve hours in antiphase with Per and Cry mRNA levels [2].

Besides Per and Cry, the CLOCK-BMAL1 heterodimers activate Rev-Erbα expression, starting the positive feedback loop. REV-ERBα represses the transcription of Bmal1. Therefore, when CRY inhibits the CLOCK-BMAL1 heterodimer, it also inhibits the expression of Rev-Erbα. This inhibition leads to the de-repression—thus activation—of Clock and Bmal1 transcription and resets the TTFL [2].

The TTFL does not complete in exactly 24 hours. Therefore, the SCN receives afferent information from external zeitgebers to synchronize its clock gene transcription to the time of day [19]. This entrainment allows the SCN to convey accurate temporal cues, in synchrony with the environment, to peripheral clocks.

Light is one of the most potent zeitgebers for the SCN [20]. A subset of intrinsically photosensitive retinal ganglion cells (ipRGCs) expresses melanopsin, a photosensitive pigment. When these ipRGCs detect light or receive photic information from photoreceptor cells in the retina, they process and relay this information to the SCN through the retinohypothalamic tract [21–23]. Glutamate, which is the main neurotransmitter of this tract, stimulates NMDA receptors and consequently L-type voltage-dependent Ca²⁺ channels on the SCN neurons, leading to an increase in intracellular calcium concentration ([Ca²⁺]_i) [2, 24–26]. This increased [Ca²⁺]_i can adjust transcription of the clock genes in the TTFL by activating the cAMP response element-binding protein (CREB)/CRE transcriptional pathway [27]. Besides photic input, non-photic input like behavioural activity is also known to be able to modulate the SCN phase [28, 29].

The SCN has historically been considered as the ‘master oscillator’ that forces circadian timing on ‘slave’ peripheral clocks [2]. However, recent perspectives, like that of Starnes and Jones [30], are more nuanced, proposing that interactions between the SCN and peripheral clocks are often bidirectional. Moreover, while some peripheral clocks need constant SCN input, others only require periodic cues to align their rhythm to the environment.

The two-process model of sleep regulation [31, 32] is often used to explain the timing of sleep within a circadian framework. It describes the interaction between two regulatory processes of sleep: a circadian mechanism (process C) and a sleep-dependent homeostatic mechanism (process S). Process S, which represents the sleep pressure, accumulates during wakefulness and decreases during sleep. Process S is thought to be reflected in the activity of the slow waves in the non-rapid eye movement (NREM) sleep electroencephalogram (EEG). Process C is responsible for sleep/wake initiation and is controlled by the SCN [31]. Sleep onset occurs when the rising process S passes the upper sinusoidal sleep-initiating threshold of process C. During sleep, process S decreases until it intersects with the lower wake-inducing threshold of process C, and wakefulness is initiated. While processes S and C work independently, sleep is optimal when the homeostatic drive for sleep is aligned with the circadian timing of sleep [33, 34].

This entanglement of sleep homeostatic and sleep-regulating circadian processes makes that sleep can be changed by factors that influence the clock. Exposure to light during the night can delay or advance the phase of the circadian clock depending on the timing of light exposure. Light at the beginning of the subjective night will cause a phase delay, while light at the beginning of the subjective day causes a phase advance [20]. Consequently, looking at the model, the sleep homeostatic process S might intersect sooner or later during the day with the thresholds of process C, thus resulting in advanced or delayed sleep onset, respectively.

The eventual transition from wakefulness to sleep and vice versa depends on a bistable mutually inhibitory system. During wakefulness, arousal-promoting regions in the reticular formation and posterior hypothalamus, like the Raphe nucleus, Locus Coeruleus, and the lateral hypothalamus (LH), are active [35–38]. These centres, collectively known as the ascending reticular activation system, inhibit the sleep-promoting neurons in the ventrolateral preoptic nucleus (VLPO) of the hypothalamus [39]. However, when the VLPO is more active, it inhibits the arousal-promoting regions using GABA and rapidly induces sleep [40]. This ‘flip-flop switch’, as it was named by Saper et al. [41], efficiently promotes fast switching between sleep and wakefulness and prevents intermediary states.

The flip-flop switch is thought to be stabilized by orexin, a neuropeptide produced in the LH. Orexin stimulates the arousal-promoting regions, thus reinforcing wakefulness and biasing the switch against sleep. This wake-promoting action is counteracted by the VLPO, which, when active, inhibits the orexin neurons in the LH [42].

The SCN plays an important role in sleep-wake transitions, but has only a few direct neuronal projections to the VLPO or orexin neurons [43–45]. Instead, the SCN influences these areas through an indirect, multi-synaptic pathway. The SCN projects to the subparaventricular zone [45, 46], which subsequently relays to the dorsomedial nucleus of the hypothalamus (DMH) [47]. The DMH has extensive connections to the VLPO and orexin neurons [43, 44] and is essential in relaying SCN information to the sleep-wake regulating network [48].

During the day, when SCN neuronal activity is high, the human SCN is thought to indirectly inhibit the VLPO and stimulate orexin production [49, 50] (the opposite then is thought to occur in nocturnal rodents). Both mechanisms drive the switch toward the waking state. By the end of the day, when SCN activity decreases and sleep pressure is highest, the VLPO is activated and sleep is initiated.

Sleep consists of two distinct states: rapid eye movement (REM) sleep and NREM sleep. 75% of sleep is spent in NREM sleep, during which the parasympathetic system is most active, leading to decreased heart and respiration rates [51, 52]. During REM sleep, the activity of the sympathetic nervous system dominates, and respiration and heart rate are enhanced [51, 52]. The duration of this ‘dreaming’ state increases through the night and occupies approximately 25% of total sleep [53]. Sleep onset typically begins with a period of NREM sleep, which ends in wakefulness or REM sleep. Humans usually experience 4 to 5 cycles of alternating NREM and REM states during the night [53, 54].

Previously, we have demonstrated that REM sleep significantly increases SCN neuronal activity, and that this enhancement decreases at the end of REM sleep. This effect was observed independent of the circadian phase, and SCN neuronal activity was shown to correlate with EEG slow wave activity [55]. Therefore, the interactions between the SCN and sleep homeostatic mechanisms appear to be bidirectional.

GCs are produced from cholesterol in the zona fasciculata of the adrenal gland cortex through a series of enzymatic conversions [62]. The regulation of GC synthesis and secretion starts in the paraventricular nucleus (PVN), which releases corticotropin-releasing hormone (CRH) into the median eminence either in response to stress or the circadian clock. In the anterior pituitary, CRH binds to its G-protein coupled receptor (GPCR) on corticotropic cells, activating Gα_s and raising cAMP concentration, which ultimately causes exocytosis of adrenocorticotropic hormone (ACTH) into the bloodstream [63, 64]. When ACTH binds to its GPCR in the zona fasciculata of the adrenal cortex, intracellular movement of steroid precursors between cytosolic stores and mitochondria is accelerated, and the production of GCs is increased [62]. In humans, the most important GC hormone is cortisol, while corticosterone is the main GC in rodents [65].

To prevent excessive GC production and optimize secretion in unstressed situations, cortisol/corticosterone exerts negative feedback on the PVN and the anterior pituitary [66]. Research in rodents has revealed two types of corticosterone-related negative feedback: a fast nongenomic mechanism and a slower genomic mechanism. The nongenomic mechanism involves GC-mediated suppression of CRH and ACTH release by the PVN and anterior pituitary, respectively [66]. Studies have demonstrated that bilateral adrenalectomy in rats rapidly increases circulating ACTH levels and CRH immunoreactivity in the PVN [67, 68]. Levin et al. [69] add to this observation by demonstrating that exogenously administered corticosterone acts at the PVN to decrease ACTH levels in adrenalectomized rats. Lesioning the PVN before administering corticosterone, in accordance with the proposed mechanism, increases plasma ACTH levels [70].

The genomic mechanism suppresses CRH gene expression and pro-opiomelanocortin—the precursor to ACTH—transcription in pituitary corticotroph cells [66]. The exact mechanisms behind genomic suppression are largely unknown [66], but several rodent studies have shown that GCs inhibit pro-opiomelanocortin gene expression [71–73].

CRH is secreted in an ultradian rhythm. The amplitude and frequency of these ultradian peaks are determined by the phase of the circadian rhythm in GC secretion [74]. GC levels are highest just before an animal’s active period, which is early morning for diurnal species and early evening for nocturnal species [75–78]. This latter GC peak may permit the body to respond and prepare effectively for physical stressors during the active period [79]. On top of the ultradian rhythm, stressors like hypovolemia or fear can cause acute (~1 hour) peaks in GC release by enhancing the PVN’s CRH secretion [80].

The earlier presented evidence shows that the SCN and the adrenal molecular clock can regulate GC secretion. GCs themselves appear to act as a mediator to align the rhythms of peripheral clocks to that of the SCN. For instance, when rats are given dexamethasone—a potent GC receptor (GR) agonist—in constant darkness, this injection acts as a zeitgeber for the regulation of intraperitoneal temperature [107]. Furthermore, GC rhythms play a role in coordinating day-night cycles of bladder capacity [108], heart function [109], and metabolism [110].

GCs exert their effects through binding to the GR, a nuclear receptor known as NR3C1. When bound, the complex migrates from the cytoplasm to the nucleus, where it functions as a stimulatory transcription factor by binding to GC response elements (GREs) in the promotor of certain genes. Clock genes Per1 and Per2 were found to contain GREs [111], suggesting a direct impact on the phase of the peripheral molecular clock by GCs. Therefore, it is not surprising that the genes integral to circadian rhythm regulation are particularly sensitive to the loss of GR function, as found in zebrafish models [112].

Interestingly, studies have shown that clock proteins can interact with the GR to alter its function. For example, REV-ERBα affects the subcellular localization of the GR [113], and CLOCK-BMAL1 can acetylate the GR, thereby inhibiting its activity [114]. CRY1 and 2 also seem to oppose the actions of the GR [115]. These mechanisms might help with adjusting the phase of peripheral clocks to that of the external environment. For instance, when GC levels are acutely elevated due to stress, they can disrupt the normal synchronization of peripheral clocks to the SCN [116]. By inhibiting the GR through e.g., CLOCK-BMAL1, unwanted desynchronisation can be mitigated [114].

The GR is expressed in many peripheral cells and in the brain, but has not been found in adult SCN neurons [61, 117]. This finding suggests that GC levels do not influence the rhythmic activity of SCN neurons. However, GRs have been identified in the foetal/neonatal SCN, gradually disappearing as gestation progresses [118]. As maternal GCs can pass the placenta [119], they can entrain the foetal SCN to the maternal circadian rhythm by acting at the GRs [120].

In contrast to adult SCN neurons, GRs were found in the glial cells associated with the SCN and PVN [121, 122]. These glial cells actively control SCN activity by regulating extracellular GABA and glutamate concentrations [123]. GCs have been shown to stimulate the expression of glial fibrillary acidic protein (GFAP), an integral component of glial filaments, in SCN and PVN-associated glial cells [122]. This upregulation suggests a role for GCs in modulating glial plasticity, possibly impacting the SCN network.

The distribution of temporal information from the central clock has been described above using mainly neuronal pathways. However, the SCN also produces and releases a number of neuropeptides, which help to regulate circadian rhythms in physiology and behaviour [126]. The released signals from the SCN can reach peripheral clocks through a specialized vascular circulation pathway to the lamina terminalis, which has recently been described to function as a portal vein [127]. Additionally, the CSF may take up and distribute signalling molecules from the SCN through a system of perivascular channels known as the glymphatic system (Figure 4) [128]. The main function of the glymphatic system was suggested to be the clearance of waste from the brain, the failure of which supposedly contributes to neurological diseases like AD [129]. The SCN, sleep, and GC signalling all seem to influence the glymphatic system, which immediately raises the question about the role of temporal alignment for the proper functioning of the glymphatic system. This part will focus on the function of the glymphatic system and the circadian rhythmicity of an important component, the ChP. Subsequently, the influence of GCs, circadian rhythms and sleep on the glymphatic system will be discussed.

To support the weight of the brain and to nourish it with nutrients, growth factors, ions, and peptides, the brain is bathed in CSF [130]. CSF is produced by the ChPs, which are tufts of highly vascularized, villous structures covered by epithelia that line the brain’s ventricles [131, 132]. The epithelial cells in the ChP are connected by tight junctions, thereby creating a blood-CSF barrier and restricting unwanted movement of water-soluble molecules from the systemic blood circulation into the CSF [133]. Due to the tight junctions and adherence junctions, ChP epithelial cells are polarized and can create osmotic gradients [134]. The extensive system of fenestrated capillary loops in the ChP and the osmotic gradient govern the passive ultrafiltration of plasma into the interstitial fluid of the ChP [135, 136]. To secrete the produced fluid into the ventricles, the ChP epithelium contains aquaporins (AQPs), which are water channels [137]. Ions are actively transported into the CSF by ion transporters [138, 139].

However, increasing buoyancy and nourishing the brain are not the only functions of CSF. A study from 2012 by Iliff et al. [128] demonstrated that CSF from the subarachnoid space can enter the brain parenchyma, mix with interstitial fluid, and drain into the peripheral lymphatic system through perivascular spaces and meningeal lymphatic vessels [140]. This CSF efflux depends on astrocytic AQP4 and is likely modulated by circadian rhythms and sleep/wake cycles [12, 141]. As the system seems to have similar functions as the lymphatic system—clearing waste—and depends on glial cells, it was termed the glymphatic system [128]. The glymphatic system has been shown to clear tau, α-synuclein, and amyloid-β, which are proteins associated with neurodegenerative diseases [128, 142, 143].

A waste-clearing system is most efficient if it has directionality. In the glymphatic system, this direction is from the brain parenchyma toward the peripheral lymphatic system. Several different factors contribute to the directionality of CSF movement. For instance, the rhythmic expansion and contraction of the cerebral arteries propels the fluid through the perivascular space [144, 145], while spontaneous changes in the muscle tone of vessels (vasomotion) support steady CSF movement [146, 147]. At the same time, respiratory cycles create pressure changes that further facilitate CSF flow [148].

The glymphatic system can be compared to a closed hydraulic system, as it relies on fluid pressure and directed flow to clear waste efficiently. Just as poking a hole in a hydraulic system would disrupt its function and pressure, any disruptions in the glymphatic system can compromise its ability to clear waste. Uncareful introduction of tracers or cannulas into the glymphatic system can create leaks that disrupt the closed nature of the system and thereby inactivate it [141, 149]. Likewise, perivascular structures collapse and fill with other fluids postmortem, rendering most histological samples of this system useless [145]. These difficulties in maintaining the glymphatic system’s integrity have caused significant discussion about its existence [150]. The most recent research on the glymphatic system therefore increasingly relies on in vivo imaging techniques like diffusion tensor imaging and positron emission tomography, as this can maintain the system’s integrity and function [151, 152].

Besides the glymphatic system, other pathways have been proposed for clearing brain waste. The intramural periarterial drainage theory suggests that perivascular spaces do not exist. Instead, the CSF enters the brain via arterial pial-glial basement membranes, mixes with interstitial fluid, and leaves along the basement membrane of the smooth muscle cells in the artery [153]. Another theory proposes that convective mixing mechanisms drive the efflux of solutes from the brain parenchyma [154]. As the glymphatic pathway is the most widely accepted waste-clearing theory, this review will exclusively focus on this pathway.

The ChP exhibits many functional daily rhythms. CSF production, for instance, was observed in a diffusion MRI study to peak during the night in humans [155]. Moreover, the composition of CSF varies across the day, with Na⁺ concentrations in human subjects reaching peak values throughout the day and reaching lowest levels during the night [156]. These daily fluctuations can be linked to rhythmic gene expression within the ChP. For example, SNAT3 and NCX4, both sodium transporters, show upregulation during the light phase [157], aligning with the diurnal changes in CSF composition. Additional genes showing high circadian expression are related to the endoplasmic reticulum stress response. Transcriptomic data shows that ER stress response markers peak during the late subjective night in the ChP, optimizing the preparation for stress due to the rapid processing and accumulation of secreted proteins, such as transthyretin [11]. This anticipatory upregulation may prevent ER-stress-related cytotoxicity [158] and fine-tune circadian secretion.

Like most cells in the body, the cells in the ChP contain a molecular clock that can coordinate rhythmic gene expression [159]. However, unlike most peripheral clocks, the cells in the ChP can maintain rhythmic expression of clock genes in vitro without input from the SCN. This autonomous rhythmicity has been demonstrated in organotypic explants of the ChP from PER2::LUC mouse models, which showed consistent rhythmic expression over several days [10, 160]. The clock gene expression in the ChP is not only persistent, but can even show stronger oscillations than the SCN itself [10]. This observation is likely explained by the presence of gap junctions between ChP cells, which allow for strong intercellular coupling [161]. This notion is supported by the finding that inhibiting the ChP gap junctions prolongs the circadian period and reduces the PER2::LUC amplitude [10].

Although the SCN appears to be redundant for the rhythmic expression of ChP clock genes in organotypic slices, in vivo studies indicate that the SCN does play a role in sustaining ChP oscillations [11]. Sládek et al. [11] examined transcriptome profiles in control and SCN-lesioned mice, revealing significantly decreased amplitudes in Clock and other rhythmically expressed ChP genes in the SCN-lesioned group. Besides the SCN-to-ChP communication, a coculturing study has found that the ChP also influences the SCN, likely through signalling molecules in the CSF [10]. This reciprocal influence might enhance the synchronization between the SCN and peripheral clocks.

While the glymphatic system is increasingly being accepted as an integral part of central nervous system physiology, the diurnal rhythm of its clearance function is still under debate. To date, the majority of evidence suggests a higher flow of CSF and clearance during sleep [12, 155, 165, 166]. Since sleep regulation and the central circadian clock are closely intertwined, as described above, it remains unclear whether this increased clearance is sleep-driven or circadian-driven, and studies are needed to distinguish the effect of sleep states from the potentially direct effects of the circadian clock on glymphatic flow.

One of the primary signalling pathways the SCN uses to synchronize peripheral clocks to the external environment is rhythmic GC release [81–83]. In the early morning, the SCN stimulates GC production by the HPA-axis through the PVN/DMH, and via sympathetic activation of the splanchnic nerve [88, 97, 98]. Notably, the adrenal gland possesses an intrinsic clock, which is aligned by the SCN, but can also independently release GCs [106]. This morning increase in GCs activates the arousal-promoting areas and inhibits the sleep-promoting areas (VLPO) [124, 125]. Simultaneously, the homeostatic sleep pressure reaches its trough and the circadian process promotes wakefulness [31, 32].

During the day, when its activity is highest, the SCN inhibits the VLPO and stimulates orexin production, stabilizing wakefulness throughout the day [49, 50]. As the evening approaches, the activity of the SCN gradually declines, and the inhibition of the VLPO wanes. In addition, the pineal gland, whose circadian rhythm is entrained by the SCN, starts producing melatonin [186, 187]. Melatonin can further attenuate SCN firing [188–190], amplifying the decreased inhibition of the VLPO. In addition, sleep pressure, which has been accumulating throughout the day, is thought to reach the sleeping threshold of process C. This series of events facilitates the transition from wakefulness to sleep.

During sleep, the ChP, entrained by GCs, increases CSF production [155]. The prevailing hypothesis indicates that the glymphatic system reaches peak efficacy during sleep, clearing the brain of amyloid-β, tau, and α-synuclein [128, 165]. As the night progresses and ends, the sleep pressure gradually declines, and GC release and SCN activity rise, allowing the cycle to restart.

In AD, those processes are disrupted. AD is a progressive neurodegenerative disorder characterized by dementia. The severity of dementia correlates with the amount and distribution of neurofibrillary tangles, which likely cause the clinical symptoms of the disease. The tangles are mainly composed of cytoskeleton-associated tau proteins [191]. The cascade of events that leads to neurofibrillary tangle formation is thought to be initiated by atypical amyloid-β secretion [192].