Circadian Rhythms of the Hypothalamus: From Function to Physiology

Hypothalamic regulation of feeding begins with the ARC. Positioned lateral to the third ventricle, the ARC forms a complex with the median eminence that is unguarded by the blood–brain barrier (BBB). This unique positioning allows the ARC to sense metabolite, nutrient, lipid and hormone fluctuations in the bloodstream. From this vantage point, the ARC is highly attuned to metabolic and hunger signals to stimulate its first order neuronal subtypes, either the appetite inhibiting pro-opiomelanocortin (POMC)/cocaine amphetamine-regulated transcript (CART) neurons, or the appetite-inducing neuropeptide Y (NPY)/agouti-related protein (AgRP) neurons which function antagonistically to each other. Though the ARC in mice loses the rhythmic expression of clock and GABAergic-related genes when the timing of feeding is restricted to the rest period [89], the ARC still maintains a weak but rhythmic dopamine release when the SCN clock is disrupted [90]. In fact, PER2::LUC recordings reveal endogenous circadian rhythms in an explanted ARC, which can be sustained up to eight days, with higher amplitude peaks in the dorsal ARC versus the lateral ARC [91]. Thus, the ARC appears to have its own autonomous clock. Moreover, it is clear that the intrinsic ARC clock is used to mediate the timing and quantity of food intake. Mice placed on RF display rhythmic changes in AgRP/NPY neuronal activity [92]. When the function of AgRP neurons is ablated in neonatal mice, there is diminished FAA, suggesting that AgRP neurons are involved in FAA [92]. Cre-mediated deletion of Bmal1 expression in AgRP neurons (ABKO) in adult mice results in increased feeding, daytime hepatic gluconeogenesis, and respiratory exchange ratio in the ABKO mice, although it is unknown whether these mice express FAA [93]. Lesioning of NPY-ARC neurons causes the uncoupling of feeding from sleep–wake cycles in rats; however, it does not result in loss of food predictability under RF [94]. Similarly, using saportoxin conjugated to leptin to eliminate ARC leptin neurons disrupts the rhythmic integration of activity and body temperature while leaving FAA unaltered in rats [95]. These studies indicate that the FEO is not solely dependent on the ARC.

Using transcriptomics to analyze appetite-regulating pathways in AgRP and POMC neurons of the hypothalamus has revealed an enrichment of circadian signaling factors during food deprivation [96]. Other studies have also revealed changes in the oscillation patterns of clock genes in ARC neurons in response to meal timing [97]. In addition, the neurons of the melanocortin system, predominately localized to the ARC, are well-established contributors to the hypothalamic regulation of metabolism and potential regulators of glucose homeostasis [98].

There are a few crucial molecular signals and receptors in the ARC that are essential to mediating feeding and appear to be involved in FAA. A recent study identified a circadian role for the neurotrophin receptor p75NTR activity in AgRP neurons in driving FAA [99]. p75NTR is a brain-derived neurotrophic-factor (BDNF) receptor that rhythmically mediates oscillation of certain metabolic liver genes [100]. Mice lacking p75NTR in either the AgRP neurons or the entire brain express FAA only during the active phase, but not the rest phase [99]. Additionally, the mammalian target of rapamycin (mTOR) protein, highly expressed in the ARC and PVN, is implicated in rhythmic feeding [101,102]. mTOR is a nutrient-activated serine-theronine kinase which regulates cell growth and metabolism. Activation of hypothalamic mTOR signaling in mice results in anorexia and substantial weight loss [102]. Interestingly, the knockout of mTOR complex 1 (mTORC1) dampens the rhythmic expression of AgRP and NPY, but is negligible for the regulation of feeding and energy homeostasis [103]. However, in POMC neurons, the mTORC1 signaling pathway is necessary to carry out leptin-mediated suppression of food intake [104]. Together, these findings suggest that molecular signals within the ARC mediate the timing of food consumption, but may also potentially function as part of the FEO.

The function of the LH is best summed by the question “to rest or ingest?” The LH tightly controls energy expenditure and food intake to maintain energy homeostasis [105]. In particular, the abundant orexin (OX) neurons of the LH are crucial for feeding and wakefulness and are sensitized to promote arousal, feeding, locomotion and drinking in the face of energy deficits [106,107,108]. Widespread projections to the autonomic nervous system (ANS) and the SCN as well as interactions with the neuroendocrine system makes the LH an ideal candidate for involvement in FAA and the FEO. In a RF experiment with a two-hour feeding window, lesions of the LH attenuated FAA behavior as measured by locomotion; however, LH-lesioned rats still show some FAA [109]. However, mice under RF show increased activity of OX-LH neurons in anticipation of feeding [110], suggesting some contribution of the LH in FAA.

Relatively recent research has begun to more thoroughly examine the PVN from a circadian perspective. Inhibitory GABAergic projections from the ARC innervate the PVN, a hypothalamic nuclei that functions as an integrator of metabolic, neuroendocrine, and satiety signals with direct inputs to the ANS. Owing to its diverse functions, the PVN contains a variety of neural subtypes which both receive and send out projections from neighboring hypothalamic nuclei including the DMH, VMH, LH, SCN, MOPA, and SPVZ [111,112]. Ablation of single-minded homolog 1 (SIM1)-positive neurons, a marker of PVN neurons, leads to hyperphagia (over-eating) and obesity [113]. Abe et al. 2002 demonstrated that the PVN maintains 92% rhythmicity following excision from brain tissue suggesting that a semi-autonomous clock is present [71]. However, other studies centered on the examination of SCN grafts reveal that the PVN is still heavily dependent on the SCN for rhythmic activity [114]. A prime example of this includes a recent optogenetic study, which identifies a SCN–PVN–LH neurocircuit involved in the regulation of wakefulness [115]. These data suggest that the cortisol-releasing hormone (CRH) neurons in the PVN may act as a “pulse generator”, whose excitability is regulated by SCN innervations [115].

To determine a potential role for the PVN in FAA, mice with PVN lesions were tested for FAA activity [109,116,117]. In spite of its known role in energy intake, lesion of the PVN did not disrupt FAA [118]. In spite of these findings the PVN still presents an attractive possibility for being part of the FEO. For example, oxytocin (OT) has recently been postulated as the main hormonal cue for FAA due to findings linking food entrainment in rabbits to OT-PVN neurons [119]. Although the SCN has direct projections to the PVN, the PVN neuroendocrine neurons are also indirectly modulated by the SCN via its innervation of the DMH, ARC and MPOA [120,121].

SIM1-PVN neurons can be divided into two satiety-sensitive subtypes, melanocortin receptor 4 (MC4R) and prodynorphin (PDYN) neurons [122]. Ablation of these neuronal subtypes leads to hyperphagia, obesity, and reduced energy expenditure [123,124]. PVN neurons act as secondary order neurons receiving direct projections from the primary order neurons of the ARC. Therefore, ARC-PVN projections are vital to regulating food intake. Rats under a narrow four-hour daily RF window showed elevated NPY levels in the PVN prior to feeding time [125]. ARC-mediated NPY release in the PVN is therefore likely an important orexigenic signal for FAA. From another perspective, the SCN’s projections to the PVN are also important to feeding through a light-coupled pathway. During the daytime, light exposure suppresses food intake and increases cFos expression in AVP-SCN neurons and OT-PVN neurons in mice [126]. Retrograde tracing has revealed that AVP-SCN neurons terminate on OT-PVN neurons, while blocking OT neurons eliminates the light induced suppression of food intake [126]. These findings highlight a potential neurocircuit explaining light induced suppression of food intake.

One of the main initial contenders for the FEO was the DMH [127]. Gooley et al. 2006 reported that DMH lesions in rats resulted in loss of FAA as well as disrupted body temperature and wakefulness, and that the degree of food entrainment corresponded to the number of DMH cells remaining post-lesion [128]. Subsequent studies showed that expressing Bmal1 in the SCN of Bmal1^−/− mice resulted in a restoration of light-entrainable but not food-entrainable circadian rhythms [129]. Injection of the Bmal1 vector to the DMH restored food-entrainable but not light-entrainable circadian rhythms, suggestive of a possible role of the DMH as the site of the FEO [129]. However, follow-up studies contested these findings, rendering the DMH’s contribution to FAA inconclusive at best [130,131]. Specifically, robust FAA has been found in rats with lesioned DMH [132], though original results obtained by Gooley et al. 2006 differed in several key areas: food type, lesion method, food locations, and cage configuration, among others. Cage configuration has been ruled out as a major factor [130].

Interestingly, the DMH has reciprocal feedback to the SCN whereby it can suppress SCN activity, leading to increased locomotor activity during the rest phase in mice [133]. One potential mechanism for DMH suppression of homeostatic feeding is through BDNF tropomyosin receptor kinase B (TrkB) receptor-expressing DMH neurons [134]. Mutation of the Bdnf or TrkB receptor genes leads to obesity both in humans and mice [134]. TrkB-expressing DMH (DMH^TrkB) neurons are also activated post-refeeding or during overnight fasting [134]. Chemogenetic activation of DMH^TrkB neurons elicits the suppression of food intake during the active phase, whereas chemogenetic inhibition of the same neurons leads to food intake only during the rest phase [134]. However, one study revealed that optogenetic stimulation of inhibitory GABAergic DMH neurons projecting to PVN in particular actually promote food intake [135]. Collectively, these data suggest that the DMH is still an integral player in the circadian regulation of energy intake, though its contribution to FAA is less clear.

Finally, the VMH has also been postulated to contribute to FAA. The VMH is innervated by multiple hypothalamic regions including the ARC, DMH and LH [136,137], and is characterized by steroidogenic factor 1 (SF1) neurons that are highly sensitive to glucose and leptin [138,139,140]. Lesions of the VMH do not completely inhibit FAA; although under TRF, mice with VMH lesions display increased arousal-associated with FAA [141]. Five to nine weeks following the VMH lesion surgery, Mistlberger et al. 1984 reported FAA was dampened or absent in VMH lesions rats; however, recovery of FAA was seen during a second phase of RF 14–21 weeks after the surgery [142]. These findings indicated the VMH is not ultimately essential for FAA.

Collectively, studies involving numerous nuclei of the hypothalamus suggest that the FEO may not be localized to one region, but rather dispersed among CNS and peripheral components, with one or more components residing in the hypothalamus to coordinate or respond to peripheral energy status [143,144,145]. One notable study found that in mice with liver-specific deletion of Per2, FAA was almost completely abolished [146]. Interestingly, using RNA sequencing, researchers identified the liver-derived ketone body, β-hydroxybutyrate (βOHB) as essential for driving FAA in mice [146]. The restoration of FAA from these liver-specific mutant mice by the administration of exogenous βOHB suggests that the liver may also be a fundamental part of the FEO. The influence of peripheral energy states on hypothalamic regulation of rhythmic feeding has also proved to be important reciprocal information. For example, the adipokine adiponectin (ADIPOQ) is released from adipose tissue, and regulates diurnal feeding rhythms through hypothalamic clocks [147]. Adipoq-deficient mice display augmented daytime food intake which is further reinforced by the disrupted expression of core clock gene and appetite-associated hypothalamic genes in the ARC and DMH [147]. For a rough timeline of mechanistic insights, see Figure 3.

The ARC plays a crucial role in hypothalamic metabolic regulation. A majority of the AgRP neuron population reside in the ARC, which communicates directly and indirectly with the SCN [112,157]. SCN projections to the ARC contribute to the ARC’s sensitivity to glycemia [158]. Lesion or electrolytic ablation of the SCN alters glucose metabolism and impairs the robust circadian firing of POMC α-melanocyte-stimulating hormone (α-MSH) ARC neurons, suggesting that some of ARC rhythmicity is derived in part from SCN communication [159,160]. The reciprocal projections are also important; when the ARC to SCN projections are severed in rats, rhythmicity in locomotion, corticosterone (CORT) and body temperature is lost under constant darkness [161]. This indicates that ARC rhythmicity is important in synchronizing body wide functions. Multiple studies using either brain slice cultures from transgenic mice or Per1/Per2 in situ hybridization have shown the ARC has its own diurnal rhythmicity expressed by clock genes [71,162]. A recent study using the forebrain-specific BMAL1 KO mice revealed a cell autonomous clock in hypothalamic AgRP neurons, which specifically controls glucose production during fasting and energy intake during the active phase [163]. ARC regulation of energy expenditure occurs via the ANS which indirectly projects through the PVN, VMH and LH [157,164]. Leptin fluctuations stimulate AgRP and POMC ARC neurons whose firing through hypothalamic intermediaries stimulate changes in heart rate, blood pressure and hepatic insulin sensitivity [165,166]. A crucial ARC neural subtype is the Kisspeptin-expressing neurons (Kiss1) which upon silencing leads to bodyweight gain in mice with dysregulated rhythms in feeding, sleep and body temperature without affecting the circadian expression of the SCN [57]. Dense projections of the Kiss1-ARC neurons to the SPVZ and DMH regions along with wiring to AgRP and POMC neurons in the ARC suggest that the Kiss1-ARC neurons can also converge on these SCN targets to rhythmically regulate their output [57]. An intriguing ARC-neural subtype from a circadian standpoint are pituitary adenylate cyclase-activating polypeptide (PACAP) neurons, PACAP is well known to modulate light-induced phase resetting in the SCN, and PACAP expression neurons of the ARC can inhibit food intake [167] through a POMC-dependent mechanism [168].

At the molecular level, AMPK activity results in the activation of Agrp expression and the inhibition of Pomc expression through regulation of mTORC1 pathway. AMPK inhibits mTOR signaling preventing mTORC1-mediated oxidative metabolism in POMC neurons which inhibit food intake [104] while also promoting Agrp expression which is otherwise inhibited by the mTOR pathway [169]. Various factors such as insulin, feeding, glucagon-like peptide-1 (GLP-1), and leptin inhibit AMPK activity in the ARC. However, other signals such as ghrelin, hypoglycemia and fasting trigger AMPK-ARC activity (Figure 4). The deletion of SIRT1 from POMC neurons results in increased propensity to develop obesity, specifically in female mice under nutrient stress [170]. Researchers further show decreased energy expenditure due to impaired sympathetic nerve activity in WAT [170]. SIRT1 overexpression in POMC neurons, protects against age-associated weight gain, adipose expansion, and reduced energy expenditure [170]. It is clear from these studies that the ARC has an essential role in rhythmic regulation of body wide metabolism.

The PVN lies downstream of the ARC as an integrator of metabolic, neuroendocrine, and satiety signals. Innervated by inhibitory GABAergic projections from the ARC, the excitatory glutamatergic neurons of the PVN terminate on several regions, including the nucleus tractus solarius (NTS) and the parabrachial nucleus (PB) [171,172,173]. Though its role in energy intake is well accepted, only recently has circadian rhythmicity in the PVN been shown to be a critical regulator of rhythmic energy intake and metabolism. Recently, diurnal variations in the firing of PVN neurons has been observed. Deletion of Bmal1 expression from the PVN of mice inhibits this diurnal firing, leading to reduced diurnal rhythmicity in metabolism [174]. Interestingly, BMAL1 controls the rhythmic expression of GABA-A receptor γ2 subunit, and the absence of diurnal γ2 rhythms results in loss of diurnal metabolism. Thus, the dynamic responsiveness of BMAL1-driven PVN neurons to GABAergic neurotransmission is essential for mediating diurnal metabolism [174].

Upon activation, the MC4R-PVN neurons signal satiety to decrease food intake. The recent study by Mazier et al. 2019 reports a potential molecular pathway for PVN-mediated satiety [175]. mTORC1 blocks the synthesis of endocannabinoids (eCB) by the PVN neurons; however, AMPK-mediated inhibition of mTORC1 enables eCB production to resume in the PVN. Binding of eCB to the cannabinoid receptor 1 (CB1R) on target POMC-ARC neurons causes increased glutamate release and thus increased glutamatergic input to the PVN eliciting reduced food intake [175]. Identifying involvement of mTORC1, which is closely associated with TTFLs in the core clock mechanisms suggests a molecular basis for rhythmic activity of PVN neurons and, therefore, PVN-mediated circadian regulation of metabolism.

The VMH has been well known for nearly a century for its essential regulation of metabolism [176]. In 1980, Ishikawa and Shimazu found that bilateral electrolytic lesions of the VMH, but not the SCN, substantially reduced the hepatic rhythms of glycogen metabolism in rats fed during the active phase [177]. This implicated a greater role of VMH in rhythmic metabolic homeostasis. Although no overt PER2::LUC rhythms have been observed in explanted VMH organotypic slices [91], hypothalamic lesions of the VMH are known to dramatically increase food consumption, reduce energy expenditure, and induce hyperglycemia and obesity [178]. Knockout of Vglut2 in SF1-VMH neurons results in hypoglycemia in mice [179]. Using mice with Bmal1 ablation in the SF1-VMH neurons, Orozco-Solis et al. 2016 found that these mice displayed augmented BAT activity during their active phase, leading to increased body temperature, increased energy expenditure, and reduced body weight [180]. Like the ARC, the VMH contains PACAP-expressing neurons, and PACAP-VMH neurons are sensitive to energy status and involved in glucose homeostasis [181]. Blocking PACAP signaling leads to attenuated leptin-induced hypophagia and hyperthermia [182] indicating that these PACAP-VMH neurons are key mediators of leptin-regulated energy homeostasis.

Interestingly, the liver-derived endocrine hormone fibroblast growth factor 21 (FGF21), which signals starvation to modulate fuel partitioning and metabolism, has substantial effects on the peripheral tissues as well as the CNS to regulate circadian behavior and metabolism [183]. FGF21 can transverse the BBB to directly activate excitatory glutamaterigic VMH neurons eliciting suppressed sugar and carbohydrate intake and increased energy expenditure [184]. FGF21 also affects calcium signaling in glucose-sensitive VMH neurons, resulting in increased glucose sensitivity [184]. Thus, the diurnal expression of this hepatic-derived hormone regulated by lipid metabolism has the ability to diurnally affect specific VMH neurons. Another circadian pathway linked to BAT thermogenesis and energy expenditure begins with the rhythmically oscillating thyroid hormone triiodothyronine (T3) which was identified as an activator of the AMPK pathway in SF1-VMH neurons [185]. AMPK activity in SF1-VMH neurons simultaneously decreases ceramide-induced endoplasmic reticulum stress to promote BAT thermogenesis, while also increasing c-Jun N-terminal kinase (JNK) pathway activation to promote hepatic lipid metabolism through PNS and SNS innervations from the VMH [185]. While AMPK evidently plays a central role in VMH function in the periphery, the presence of the deacetylase SIRT1, stimulated by insulin, is also crucial for normal VMH function. For example, mice lacking SIRT1 are hypersensitive to dietary obesity [186]. Contrastingly, overexpression of SIRT1 in the SF1-VHM neurons render mice more resistant to diet induced obesity due to increased energy expenditure and improved insulin sensitivity in skeletal muscle [186]. These intriguing findings suggest that the non-autonomous clock of the VMH has an important role in rhythmic energy expenditure through regulation of BAT activity.

Sitting dorsal to the VMH and populated by NPY and α-MSH terminals, the DMH is another primary target of ARC, SCN and SPVZ innervation. Dense DMH projections to the VLPO, the locus coeruleus, and the OX-LH neurons are thought to mediate sleep–wake cycles [91,187,188]. Furthermore, a plethora of afferent and efferent connections between hypothalamic sites involved in blood pressure, feeding, metabolism, and thermoregulation, renders the DMH essential to the integration of hypothalamic circadian rhythms [66,91,187]. Though Per2::LUC rhythms support the presence of a semi-autonomous oscillator in the DMH, the oscillator dampens more rapidly than that of the ARC [91]. Excitotoxic lesions of the DMH have been shown to attenuate locomotion, feeding, wakefulness and most substantially serum CORT levels [66]. It is clear that energy status strongly controls DMH activity, as it receives visceral afferent signals from the PB and responds to feeding-related peripheral hormones such as leptin, ghrelin and CCK [98]. Selective activation of the DMH leptin-sensitive (DMH^LepRb) neurons by pharmacosynthethic receptors increases BAT thermogenesis and locomotion, resulting in weight reduction without affecting food intake [189]. Moreover, CRE-mediated deletion of DMH^LepRb neurons produces an opposite effect, highlighting a role for the DMH^LepRb neurons in controlling energy expenditure [189]. A similarly powerful diurnal hormone is GLP-1 produced both by the intestinal cells and the NTS neurons, of which the GLP-1-NTS neurons synapse onto the glutamatergic DMH neurons [190]. Injections of GLP-1 into the DMH increases BAT thermogenesis, while knockdown of the GLP-1 receptor in the DMH increases adiposity, attenuates energy expenditure and BAT thermogenesis, and increases insulin resistance [190]. Both leptin and GLP-1 are secreted in response to nutrient intake, thus these hormones connect the zeitgeber of food intake to the metabolic clock in part through their interactions with the DMH.

The LH is an extensively interconnected region which consists of numerous distinct neural subtypes such as glucose-sensitive OX neurons which are activated in response to insulin-induced hypoglycemia [191]. Though known for its role in diurnal feeding anticipation and wakefulness, the LH’s contribution to metabolic regulation from a circadian standpoint has not been fully elucidated. Notably, unlike the PVN whose destruction produces obesity, the destruction of the LH instead results in anorexia [72,192]. Viral tracing experiments reveal bidirectional projections between the ARC, VMH and PVN, indicating interconnectedness in crucial metabolic regions [193,194]. Moreover, the OX-LH neurons can regulate BAT thermogenesis to effect overall energy expenditure [195,196]. GLP-1 receptors are also present in the LH and are critical for controlling food intake and bodyweight [197]. In all, the LH’s diverse inputs and neural subtypes suggests further roles in rhythmic metabolic homeostasis that are yet to be discovered.

A range of neuroendocrine hormones are produced by several hypothalamic nuclei and circulated by secretion into the third ventricle. This includes the well-known oscillatory glucocorticoid hormone, CORT, which is produced by PVN corticotrophin-releasing hormone (CRH) neurons and is critical in the hypothalamic–pituitary–adrenal (HPA) stress axis. Although typically associated with stress, CORT, has been shown to reflect variations in metabolic rate independent of psychological stress [230]. CORT circulation is highly rhythmic, returning to the brain through the ARC. ARC detection of CORT also follows a circadian pattern, dependent on target glucocorticoid and mineralocorticoid receptors in the ARC [231]. Prolonged elevation of glucocorticoid levels in mice results in overconsumption of food via inhibition of CRH-expressing neurons, lowering CORT levels and stimulating NPY expression [232,233]. Additional studies have revealed that both NPY and AgRP expression are differentially responsive to stress via direct innervation from PVN-CRH neurons to the ARC [234,235,236]. NPY neurons of the ARC innervate the PVN, resulting in NPY-mediated CORT production, food intake, and increased PVN activity [237,238]. NPY injection in the PVN of mice results in BAT thermogenesis and increased WAT lipoprotein lipase enzymatic activity [55]. Microarray analyses reveal that rhythmic NPY neurons in part control circadian transcriptional activity in the mouse liver [239]. These findings underscore a bidirectional relationship between the HPA axis and rhythmic ARC and PVN neurons, suggesting an important feedback loop whereby stress-induced chronodisruption influences metabolism and food intake. The recent identification of a neurocircuit whereby GABAergic SCN neurons project to CRH-PVN neurons which excite the OX-LH neurons stimulating wakefulness in mice [115] further highlights the diverse involvement of the CRH neurons and their CORT production in balancing homeostatic processes.

The hypothalamic-regulated hormone melatonin is synthesized in the pineal gland and is integral to sleep/wakefulness. Melatonin peaks during the dark phase in both diurnal and nocturnal organisms, whereas most hormones are expressed at opposite phases between nocturnal and diurnal organisms. This suggests that melatonin release is regulated by mechanisms upstream of the unknown biological diurnal/nocturnal switch. Control of melatonin release occurs via direct excitatory glutamatergic inputs from the SCN to the pineal gland. During the dark phase, direct GABAergic inhibitory signals from the SCN to the PVN inhibit the PVN to pineal gland projections [118,240]. Rhythmicity in melatonin provides essential regulatory control in metabolism; when melatonin synthesis is abolished by pinealectomy, glucose tolerance is impaired and blood glucose rhythmicity is completely lost under ad libitum feeding conditions [241,242]. Thus, the melatonin-mediated regulation of plasma glucose levels is controlled by a delicate balance of glutamatergic and GABAergic pre-autonomic hypothalamic inputs. The daily rise in plasma glucose is produced by inhibition of GABAergic activity at sympathetic pre-autonomic neurons of the PVN, which also increases hepatic glucose output [243]. Altogether, melatonin acts as a key integrator of circadian rhythms and energy metabolism through its effects on the hypothalamus and the periphery.

Gut-derived polypeptides have also been implicated in circadian rhythmicity. Apart from the previously discussed GLP-1 and FGF21 hormones, other peripherally-derived hormones also modulate metabolism in a hypothalamus-dependent manner. The gastrin-releasing peptide (GRP) is a mediator of food intake and locomotor activity, and can induce light-like resetting of the SCN [244]. Recently, microbes residing in the gut have been shown to be highly relevant circadian factors in metabolism. Studies have demonstrated that certain microbial taxa and their secretions exhibit diurnal oscillations [245,246,247,248]. Additionally, the timing of food intake and chronodisruptions such as jet lag and shift work, can alter the abundance and functions of gut microbes [248,249]. Most interestingly, germ free mice which lack gut microbiota, actually display altered SCN and hepatic transcriptional rhythms, particularly in core clock genes and metabolic pathways [248]. While the gut microbe interaction with the clock machinery is still under investigation, these studies suggest an important relationship that may influence overall energy metabolism. Collectively, rhythmic cross-talk between the extra-SCN hypothalamus and the periphery is critical for metabolic homeostasis.

While the timing of light exposure has robust control over physiological functions and behavior, the timing of food intake also plays a crucial role as well. Time-restricted feeding (TRF) offers a novel non-pharmacological treatment that shortens the window of energy intake to 8–12 h without inadvertently restricting total caloric intake. Preclinical and animal studies clearly demonstrate that TRF can prevent metabolic disease even in mice fed a HFD [260]. Though some studies suggest that TRF activates OX-PVN neurons [261] at the molecular level, TRF strengthens the otherwise dampened rhythms of clock genes, improves mTOR, AMPK and CREB signaling, reduces adiposity and alters the liver metabolome [260,262]. When TRF is applied during the rest phase, the phase of clock gene expression in peripheral tissues reverses, as does the phase of satiety hormones such as leptin, ghrelin and insulin, leading to weight gain and hepatosteatosis [263]. Using a rotating light cycle to mimic shift work, chow-fed mice show altered phases of insulin and CORT secretion, while HFD mice under rotating light have complete loss of rhythmicity in hepatic lipogenic gene expression [264]. Simply delaying feeding time by 4 h elicits a shift in peripheral clocks and increases weight gain in mice fed HFD [265]. Thus, TRF can actually reverse the deleterious side effects of obesity [266,267]. Even when only adhering to TRF on the weekday and following ad libitum feeding on the weekends in HFD fed mice, the beneficial effects of TRF are still seen [266,268]. At the hypothalamic level, TRF has been shown to restore disrupted locomotor activity in mice with lesioned DMH or jointly lesioned DMH and SCN [269]. Interestingly, hypothalamic leptin sensitivity is improved when TRF is aligned to the daily rhythms of leptin [270]. MC4R neurons in the ARC are highly involved in the regulation of the melanocortin system, which has been demonstrated to mediate the benefits of TRF on energy balance [271]. When the timing of food intake is misaligned with the individuals’ intrinsic clock, as is seen in shift workers or chronically jet-lagged individuals, long-term health ramifications exist.

Diet is also capable of remodeling circadian rhythms behaviorally, transcriptionally, and metabolically [22,260,272,273]. For example, feeding mice a ketogenic diet or overexpressing the FGF21 (whose production is stimulated by a ketogenic diet) augments locomotion during the daytime and results in a phase advance in mice [183]. Mice placed on standard HFD have a profound reorganization of their circadian gene expression and metabolite production, in several metabolic tissues, including liver, adipose tissue, and the CNS [22,272,273]. However, hypothalamic PER2::LUC rhythms in the SCN, DMH or ARC appear unaltered under HFD [91]. Is it possible that other hypothalamic regions are affected, or that the Per2 locus itself is protected from nutrient challenge in the hypothalamus? The so called “Western diet” feeding across 16 weeks in mice results in attenuated neuropeptide expression of Npy, Pomc and AgRP [274]. Interestingly, these ARC neuropeptides are also altered in hypothalamic cell lines derived from global Bmal1-KO mice [274]. The prevalent saturated fatty acid in Western diet known as palmitate is capable of stimulating Npy expression [274]. However, palmitate exposure disrupts Bmal1 and Nampt expression, the latter an integral enzyme of the NAD⁺ salvage pathway, which interacts with the core circadian clock, to induce neuroinflammation in hypothalamic neurons [275]. These findings highlight a mechanistic link between Bmal1 expression and the effects of Western diet on rhythmic expression of energy related hypothalamic neuropeptides. The sensitivity of circadian timekeeping to unhealthy diets or stress is further supported by studies showing that HFD feeding blunts CRH-PVN neuron responsiveness to nutritional challenges and stress [276]. Typically stimulation of these neurons elicits their rapid activation; however, clamping CRH-PVN neurons at either high or low levels of activation appears to mimic the same blunting of responsiveness induced by HFD [276]. In both these models, blunting of CRH-PVN neurons results in reduced diurnal rhythmicity in feeding and energy expenditure [276]. AgRP-ARC neurons also display blunted inhibitory activity while under HFD [277]. The PVN, DMH, VMH and ARC have direct innervations to the ANS, some of which terminate in adipose tissue. These extensive projections allow hypothalamic control of WAT and BAT activity [278]. An important leptin-mediated feedback loop between the brain and adipose tissue has been shown to be important for long-term energy balance in mice. Specifically, the disruption of circadian rhythms through chronic jet lag is sufficient to induce leptin resistance independent of obesity risk factors [229].

Interestingly, the PVN has also been uncovered as an influential mediator of food selection. An intriguing study by Okamoto et al. 2018 demonstrates that activation of AMPK-regulated CRH-PVN neurons drives the selection of carbohydrates over fat in mice under a restricted feeding paradigm [279]. Altogether, this discovery that fasting may simulate AMPK to activate a subset of CRH-PVN neurons to induce preference for carbohydrates lays the foundation for further elucidation of the relationship between obesity, diet and the circadian clock.

Circadian differences exist between sexes, which may extend to circadian regulation within specific hypothalamic nuclei. For example, recent research revealed that men and women placed in either circadian-aligned or circadian-misaligned conditions across 3 days displayed different physiological changes [280]. Specifically, women expressed circadian misalignment by changes in their energy homeostasis through decreased leptin and increased ghrelin levels [280]. On the other hand, men show increases in hedonic appetite, specifically for carbohydrates [280]. Comparing adolescent girls and boys under similar social jet lag, researchers found that a late chronotype is associated with increased adiposity in girls, but not boys [281]. These studies are two of several that underscore sex-specific differences in the expression of circadian misalignment, suggesting that studies may need to take into account sex when attempting to understand the mechanisms by which chronodisruption alters energy balance and circadian alignment.

One major contributor to the pathogenesis of AD is neuroinflammation although its role in AD is not entirely known. However, accumulating evidence points to microglial and astrocytes whose regulation of CNS response to injury through pro or anti-inflammatory signals have been strongly implicated [285,286,287]. It is postulated the Aβ plaques stimulate astrocytes and microglial to secrete pro-inflammatory cytokines, chemokines, and reactive oxygen species whose chronic presence dysregulate the immune response that furthers neurodegeneration. Both astrocytes and microglia contain intrinsic clocks that control their morphology, internal processes and contribute to regulation of the sleep–wake cycle [288,289,290]. Circadian impairment in microglia and astrocytes has been implicated in the neuroinflammatory responses that progress AD [291,292,293]. One such study found that by disrupting BMAL1 in microglial using a synthetic REV-ERB agonist, an increased inflammatory phenotype appeared in microglial of the amyloid precursor protein knock-in (App-KI) mouse model. Moreover, cognitive tests on these 2-month-old App-KI mice treated with the REV-ERB agonist showed microglial activation and cognitive decline [291]. Implicating circadian disruption in microglial induces chronic neuroinflammation, which exacerbates AD.

A widely acknowledged symptom in patients with AD-associated dementia or PD is the disruption of sleep–wake cycles. A possible disconnect may exist between the rhythmic oscillations in the SCN and the regions involved in sleep–wake cycle homeostasis such as the LH, VMH, and DMH. However, pathological changes in the SCN of individuals with AD further implicate the potential role of the circadian clock in disease progression [294,295,296]. Increasing evidence points to several extra-SCN hypothalamic regions for the early stages of AD development. Notably, plaques and neuron fibrillary tangles have been observed in the PVN and LH of AD patients [295,297]. Although cognitive decline and dementia are the characteristic hallmarks of AD pathology, they are now recognized as manifestations during late stages of AD [298]. Chronic stress can induce tau pathology, neurodegeneration, and learning impairments in a transgenic mouse model of AD [299]. Treatment with the CRH receptor 1 antagonist prevents these deleterious effects of chronic stress, indicating that CRH production in the PVN mediated by the HPA axis may contribute to AD pathology [299]. However, studies examining circulating cortisol levels in relation to AD show mixed findings. Thus, further investigations are needed to understand the interactions among stress, circadian rhythms and AD pathology.

A study in 2015 revealed that Aβ plaques induce post-translational degradation of the circadian clock interactor and co-regulator CREB, as well as BMAL1, resulting in disruption of Per2 expression [300]. Interestingly, amyloid levels are shown to peak during the day and decrease during sleep; however, when sleep is disrupted the amyloid levels measured by either cerebral spinal fluid or positron emission tomography generally show an increase [301,302,303]. In addition, when circadian genes are deleted in a mouse AD model, the mice develop more Aβ plaques than mice without the deleted clock genes while also displaying disrupted amyloid rhythms [301]. At the clinical level, a 2017 study reported a significant dysregulation in Bmal1 expression was associated with early AD [304]. These data suggest a role for circadian genes and circadian rhythmicity in AD pathogenesis.

A very interesting behavioral pattern in AD patients is the manifestation of Sundown Syndrome or “sundowing.” AD patients with sundowning express temporally-dependent behaviors of agitation, aggression, and/or confusion in the afternoon and evening hours [305]. Interestingly, a higher rate of sundowning is reported during the winter months when sunlight exposure is often limited [306]. Melatonin treatment for individuals with sundowning appears to reduce agitated behaviors [307]. The most studied and clinically used therapy is light therapy for sundowning patients, although, in some cases, the SCN may be too degenerated for light therapy to be effective [308]. One study reported that a non-invasive 40 HZ light flicker was sufficient to ameliorate the AD-associated rhythmic disruption in mice [309]. Sundowning severity increases as AD progresses, and light is sometimes recommended during the evening hours to reduce associated agitation. Interestingly, insomnia and hypersomnia are also positively associated with the degree of sundowning symptoms, suggesting a potential corollary factor of sleep fragmentation [310]. The consistent expression of agitation, aggression and/or confusion at a particular time point in the day strongly indicates a disruption of circadian function in the CNS, most likely the hypothalamus. Todd et al. 2018 identified a neurocircuit where the SCN projects to the SPVZ, whose neurons are most active during the early rest phase and innervate the VMH region [311]. Disruption of GABAergic activity in the SPVZ results in a time-dependent increase in aggressiveness in the mice, offering a potential neural circuit involved in sundowning [311]. The VMH region is also associated with the regulation of fear and anxiety, making it a candidate for involvement in sundowning-associated behaviors [312,313]. Further exploration will likely revolve around the rhythmicity of the SCN and its hypothalamic counterparts to elucidate the underlying neurocircuit involved in sundowning.

This link between our intrinsic clocks and emotional state is not unique to AD or sundowning symptoms; other psychological disorders also have circadian-dependent properties. For example, seasonal affective disorder (SAD) is a type of depression that occurs most commonly during the fall/winter seasons [338]. A reduction in light and changes in the photoperiod length during the fall/winter seasons is presumed to trigger the depressive behaviors characteristic of SAD [339]. Treatment for SAD relies predominantly on phototherapy, in which patients are exposed in the morning to either bright white light or infrared light [327,337]. The finding that SAD prevalence varies depending on the latitude, increasing at higher latitudes where the days are shorter during the fall/winter time, further corroborates the relation between light and emotion [340]. While the underlying mechanism behind SAD is unknown, many consider the clock-regulated hypothalamus to be a prime suspect.

Some mechanistic insights into circadian links to SADs have been revealed in preclinical and human studies both. Bmal1-specific knockdown in the SCN results in a depressive anxiety-like phenotype accompanied by disrupted CORT rhythmicity [341], highlighting a circadian hypothalamic mechanism contributing to emotion. Seasonal changes in hypothalamic gene expression and hormonal activity have been observed in Siberian hamsters [342]. fMRI imaging in SAD patients during exposure to blue light has revealed enhanced responsiveness in the posterior hypothalamus to auditory emotional stimuli, suggesting this region’s involvement in SAD [329]. In addition, the dysfunction of hypothalamus-related functions in SAD individuals is reflected by alterations in sleep, feeding and metabolism [338,343]. Appetite is markedly increased in SAD patients, particularly for carbohydrate-rich foods, and weight gain, lethargy, sleepiness and sleep issues such as insomnia also arise more frequently in SAD patients [343,344]. These issues implicate potential roles for the PVN, due to its involvement in carbohydrate preference, and the LH, for its regulation of sleep–wake cycles, in SAD pathology. Moreover, seasonal differences in diurnal cortisol release in SAD patients further implicates a role for the PVN and the HPA axis [336]. The changes in the sleep–wake cycle of SAD patients is also suggestive of hypothalamic OX signaling. Recently, a study highlighted a potential SAD pathway comprised of the SCN, OX-LH neurons and the downstream dorsal raphe nucleus [345]. The quantity of OX hypothalamic neurons was reduced in SAD animals kept in dim light to simulate winter light exposure [346]. Furthermore, exposure to normal light in combination with an OX antagonist led SAD mice to a depressive phenotype [346], suggesting that OX neuron signaling is involved with the depressive behaviors associated with SAD.

Thus, while SAD behavior appear to have a light-dependent origin mediated through a rhythmic hypothalamus, dissimilar neurological phenotypes, including mania, have also been correlated to light-dependent hypothalamic activity. Mania is associated with bipolar disorder (BD) in which individuals commonly vacillate between episodes of depression and mania. BD is often accompanied by disrupted rhythms in the sleep–wake cycle, activity, appetite and hormonal secretions. Moreover, bouts of mania often follow a cyclical, even seasonal pattern, suggesting an underlying circadian mechanism. Clock^Δ19 mice actually express a similar behavioral profile to mania. Treatment with the mood stabilizer lithium, a therapeutic for BD patients, eliminates mania behavior in the Clock^Δ19 mice, revealing a novel role for CLOCK in mood regulation [347]. Intriguingly, the Clock^Δ19 mice have augmented dopaminergic activity in a region downstream of the hypothalamus, the ventral tegmental area (VTA) [347]. The expression of CLOCK in the VTA of Clock^Δ19 mice often eliminates the manic phenotype [347]. Furthermore, the knockdown of Clock expression in the VTA results in greater depressive behaviors although less hyperactivity [348]. This is perhaps not surprising, as cocaine-induced dopaminergic surge results in a profound reprogramming of circadian gene expression in the striatum, a processes partially dependent on the dopamine 2 receptor and pro-inflammatory peroxisome protein activator receptor gamma (PPARγ) activity [349].

The VTA is not the only non-hypothalamic region recently uncovered for its rhythmic role in emotion. In schizophrenia, which is characterized by psychotic behaviors, transcript analysis in post-mortem brain tissues schizophrenia patients reveal altered rhythmic gene expression in the dorsolateral prefrontal cortex [330]. The hypothalamic region is still likely involved in schizophrenia as other studies have found increases in hypothalamic volume [350] and attenuated AVP mRNA expression in the hypothalamus [351]. As studies venture beyond the hypothalamus, it is evident the complexity of emotional regulation involves numerous regions. Changes in brainstem substructure volumes have recently been associated with seasonal changes in photoperiod length suggesting a novel role for the brainstem in SAD [352]. A fascinating neurocircuit study by An et al. 2020 uncovered a circadian gated subcortical pathway for the induction of depressive-like behaviors in response to light at night exposure in mice [353]. Electrophysiological studies have demonstrated that the photosensitive ipRGCs, which project to the dorsal perihabenular nucleus (dpHb), have greater excitability during the nighttime. Projections from the dpHb terminate on the nucleus accumbens to mediate depressive-like behaviors induced by light exposure during the nighttime [353]. This novel circuit potentially underlies the higher propensity for shift workers to develop MDD and anxiety disorders [354].

Mitochondrial metabolic dysfunction may also underlie the behavioral changes associated with mood disorders such as BD and schizophrenia. In mice with forebrain-specific mitochondrial mutations, the behavioral phenotype resembles a manic episode and is accompanied with disrupted circadian expression [355]. The interlocking of intracellular metabolism with circadian machinery via the NAD⁺ salvage pathway suggests a direct link whereby circadian disruption may mediate or worsen metabolic dysfunction, leading to poor cellular health or even cell death. Most likely, however, the metabolic disruptions result in altered neural plasticity and hormone production. For example, CLOCK and NAD⁺ regulation of SIRT1 diurnally antagonize CREB-mediated transcription of tyrosine hydroxylase [356]. In Clock^Δ19 mice, this diurnal regulation is absent, resulting in elevated TH levels during the daytime [356]. TH is the rate-limiting enzyme for the synthesis of dopamine, a hormone very likely involved in emotion. Thus, the interactions between core clock mechanisms, metabolism and emotions are more widespread and intertwined than localized circuitry between the SCN and other hypothalamic nuclei.