Circadian Regulation of Brain and Behavior: A Neuroendocrine Perspective

Abstract

Neuroendocrine systems are key regulators of brain and body functions, providing an important nexus between internal states and the external world, which then modulates appropriate behavioral outputs. Circadian (daily) rhythms are endogenously generated rhythms of approximately 24h that help to synchronize internal physiological processes and behavioral states to the external environmental light-dark cycle. Given the importance of timing (hours, days, annual) in many different neuroendocrine axes, understanding how the circadian timing system regulates neuroendocrine function is particularly critical. Similarly, neuroendocrine signals can significantly affect circadian timing, and understanding these mechanisms can provide insights into general concepts of neuroendocrine regulation of brain circuits and behavior. This chapter will review the circadian timing system and its control of two key neuroendocrine systems: the hypothalamic-pituitary-gonadal (HPG) axis, and the hypothalamic-pituitary-adrenal (HPA) axis. It will also discuss how outputs from these axes feedback to affect the circadian clock. Given that disruption of circadian timing is a central component of many mental and physical health conditions, and that neuroendocrine function is similarly implicated in many of the same conditions, understanding these links will help illuminate potentially shared causality, and perhaps lead to a better understanding of how to manipulate these systems when they begin to malfunction.

1. General Introduction

Neuroendocrine systems are key regulators of brain and body functions, providing an important nexus between internal states and the external world, which then modulates appropriate behavioral outputs. For instance, as nicely demonstrated in this volume, reproductive hormones impact several brain areas to prime them to respond to signs and signals in the environment that indicate it is the appropriate time to engage in reproductive behaviors. This is important because the tightly orchestrated set of responses ensures that energy is not wasted at times when reproduction would at best be unfruitful, or at worst negatively impact survival of mother and/or offspring. Thus, a common theme in neuroendocrine regulation of behavior is the tight coordination of physiological/body function with environmental context. In this regard, the circadian (daily) timing system is a key regulator not only of neuroendocrine function, but of general neurobehavioral and physiological systems. This chapter will explore a few key aspects of what is known about the generation, regulation, and implementation of circadian rhythms in the brain and body. It will also spend some time investigating direct regulation of the hypothalamic-pituitary-gonadal (HPG) and hypothalamic-pituitary-adrenal (HPA) axes by the circadian clock, and their bidirectional impacts on circadian rhythms.

1.2. Circadian Rhythms: A definition

Circadian rhythms have been found in nearly all organisms that have life spans longer than a day. These rhythms are endogenously generated (i.e. they persist even in the absence of external cues) and allow organisms to anticipate the daily change in the environment that are a result of the approximately 24h period defined by the rotation of the Earth about its axis. It is important to note that circadian rhythms are but one of a set of “biological rhythms” that have varying periods and are also important for neuroendocrine function. For instance, the circannual timing system is influenced by photoperiod and helps seasonal animals predict the onset of changes in the climate that are important for breeding seasons. However, the circadian system is perhaps one of the most well-characterized of these biological timing systems, with their molecular underpinnings, neurobiological basis, and system-wide organization particularly well-understood.

1.3. The Suprachiasmatic Nucleus and the Molecular Clock

The focus of this chapter will be on the vertebrate circadian system, and more specifically on the mammalian circadian timing system. From many experiments undertaken in the early part of the twentieth century, it was clear that mammals had a circadian system and that the seat of this clock must be somewhere in the brain. However, it was unclear where in the brain the circadian clock was located, or indeed, if the circadian clock was actual localized to a single brain nucleus. Early work attempted to find the circadian clock using a series of tract tracing studies from the retina. The hypothesis was that since it was known that circadian rhythms are influenced by the light-dark cycle, then there should be a direct pathway from the retina to the locus of the clock. Work by Moore and Klein (Klein and Moore 1979; Moore and Klein 1974) finally determined that circadian rhythms are entrained (or synchronized) to external environmental time by photic (light) information conveyed by the retinohypothalamic tract (RHT) to a small collection of cells above the optic chiasm in the anterior hypothalamus, known as the suprachiasmatic nucleus (SCN) (Klein and Moore 1979; Moore and Klein 1974). Importantly, this pathway is non-image forming, but is necessary and sufficient for entrainment to light. If the primary visual pathway is transected at the level of the optic tract beyond the optic chiasm (i.e., thalamic relays caudal to the SCN), then animals are visually blind, but the circadian system is able to respond to photic cues and the entrainment is still possible (Klein and Moore 1979; Johnson, Moore, and Morin 1988). This pathway was further elaborated with the discovery of dedicated pathway of non-image forming retinal cells that convey photic information to the SCN. While both rods and cones contribute to entrainment, a novel subset of intrinsically photosensitive retinal ganglion cells (ipRGCs) containing melanopsin were also discovered to be central to entrainment (Panda, Sato, et al. 2002; Ruby et al. 2002; Hattar et al. 2003; Lucas et al. 2003; Panda et al. 2003). All three photoreceptors must be knocked out to abolish entrainment (Hattar et al. 2003), however, specific ablation of these cells prevents entrainment even while preserving vision (Guler et al. 2008). That the SCN receives photic input is but one of the myriad pieces of evidence that have been collected demonstrating its position as the master clock in mammals. The SCN shows rhythms of electrical activity and gene expression both in vivo and in vitro (Inouye and Kawamura 1979; Green and Gillette 1982; Groos and Hendriks 1982; Shibata et al. 1982; Yamazaki et al. 2000; Ono, Honma, and Honma 2015), and SCN rhythms persist for many cycles in vitro, demonstrating that input from other brain areas is not necessary. In addition, lesioning the SCN abolishes circadian rhythms (Moore and Lenn 1972; Stephan and Zucker 1972), and transplants of SCN tissue restore rhythmicity in hosts whose own rhythms have been eliminated by SCN lesions (Lehman et al. 1987; Silver et al. 1990). While restoring host rhythms indicates SCN tissue may is to sustain rhythms, it does not fully demonstrate that the SCN is the clock. Follow-up experiments using SCNs from hamsters with a naturally occurring mutation that affected period length (“tau” mutant, 20h) demonstrated that transplanted SCNs restore the period of the activity rhythm to that of the donor, rather than of the host — a remarkable demonstration of the ability of a brain area to determine behavioral outputs (Ralph et al. 1990; Silver et al. 1996).

The SCN is a compact structure of only about 10,000 cells, but this seemingly small collection of cells belies an incredibly complex and elegant neuroanatomical and functional organization. SCN cells are heterogeneous in neurotransmitter and peptidergic content, as well as in their cellular morphologies, and show distinct differences in their afferent and efferent connections. This heterogenous organization of the SCN has been shown to be central to its special abilities to act as a clock. At the tissue level, in vivo the SCN the shows a seemingly coordinated pattern of activity, with a period of about 24h. Puzzlingly, dissociated individual SCN neurons exhibit different free-running periods (Welsh et al. 1995), which has led to the hypothesis that SCN cells must communicate amongst themselves in an organized circuit that is functionally important. Generally speaking, the SCN has a ventromedial subregion that has come to be known as the “core”, and dorsolateral region known as the “shell”. This organizational scheme is based on anatomical, neurochemical, and functional studies (reviewed in Antle and Silver (2005)). Functionally, the shell shows high-amplitude oscillations of many core “clock genes” including the Period1 and Period2 genes (Per1, Per2; discussed further below in the context of molecular clocks). Whilst in the core, these rhythms are very low amplitude, or in some studies undetectable. Conversely, when the animal is exposed to a light pulse there is a rapid and transient induction of Per1 and Per2, while in the shell compartment, Per1 and Per2 are not induced by light (Hamada et al. 2001). Intense study of the SCN over the past decade, using both advanced analytical tools and cutting-edge imaging techniques has made this brain structure a powerful model to understand how small networks of neurons interact in order to have a cohesive output. A full discussion of how this phenomenon is manifest in the SCN is beyond the scope of this chapter, but current models suggest that the spatiotemporal pattern of SCN activity is a dynamic interaction between the various SCN components, but can also be influenced by the environmental light-dark cycle (Hamada et al. 2001; Herzog 2007; Yan et al. 2007; Pauls et al. 2014; Evans et al. 2013). However, the specific mechanism that underlies this unique ability of the SCN remains unknown, though multiple mechanisms have been proposed for coupling among SCN neurons (Antle et al. 2003; Antle et al. 2007).

While there has been tremendous work undertaken to unravel the complex organization of the SCN at a circuit level, even more work has been undertaken at the level of the molecular clock. The incredible findings at the molecular level is based largely upon work conducted in Drosophila (fruit-fly), a fact that was recently acknowledged with the awarding of a Nobel Prize in Physiology or Medicine to Rosbash, Young, and Hall in 2017.

In the SCN, rhythms are generated by interactions between cells. But at the level of individual cells, circadian rhythms are generated by a clock-like molecular timing mechanism (Figure 1). This is what underlies the “cell-autonomous” nature of circadian rhythms: while networks of neurons are needed to time behavior and physiology, individual cells can show clear circadian rhythms. Circadian clocks in animals have a highly conserved molecular mechanism based on self-regulating transcriptional–translational feedback loops that underlie rhythmic expression of core “clock” genes/proteins. In mammals, this consists of a set of transcriptional activators and transcriptional repressors. The transcriptional activator proteins BMAL1 and CLOCK, bind enhancer elements (E-boxes) on DNA, and promote the transcription of Period and Cryptochrome genes. These genes encode for repressor proteins, which feed back to inhibit BMAL1/CLOCK function. Inhibiting the enhancers thereby reduces their own expression: a classic “negative feedback” loop. In addition to these core clock genes and proteins, several kinases are important to help set the “pace” of the reactions, including casein kinase 1 /ε (CK1/ε). These factors provide fine control of variables such as period length and amplitude of the core clock machinery. Remarkably, this biochemical cascade, takes approximately 24h to complete — an elegant example evolution at work (Koike et al. 2012; Takahashi 2017).

Figure 1. Diagram of the Molecular Circadian Clock.

A simplified depiction of the molecular circadian clock mechanism showing the transcriptional-translational feedback loop (TTFL) in mammals. In the nucleus, BMAL1 – CLOCK protein complex binds to the promoter region of genes of the Per and Cry family, as well as to clock-controlled genes (ccg). This increased transcriptional activity results in production of their protein products, which then shuttle to the cytoplasm, where PER/CRY protein dimers translocate back to the nucleus. Once nuclear transport has occurred, they act to inhibit BMAL1/CLOCK, thus reducing their own transcription. In addition to the TTFL, kinases pathways, including casein kinase 1 epsilon (CK1e) phosphorylate PER in the cytosol, which targets the protein for degradation. This provides a time delay that enables the molecular clock to cycle with a period of approximately 24 hours.

In addition to core molecular elements of the clock, the circadian system can influence many cellular pathways by direct interaction which “clock-controlled genes” (CCGs). These downstream genes can be rhythmically regulated by the core clock and encode proteins involved in tissue specific effects. This schema allows for a cellular clock that is common amongst nearly all somatic cells, but then allows for tissue specific changes. Specifically, transcription analysis suggests only ~10% overlap of CCGs between tissues (Akhtar et al. 2002; Panda, Antoch, et al. 2002). Moreover, even when the same CCGs are found to be regulated rhythmically in different tissues, in many cases they have different distributions of peak phase and amplitudes in different tissues. This level of tissue specificity provides both a challenge but also a potential benefit. The challenge is that the wide variability of patterns of gene and protein expression throughout the body hamper our abilities to determine what the basic operating principles of the molecular clocks and CCGs may be. However, a benefit is that tissue specific regulation of CCGs may be possible, providing for the potential of tissue specific temporal drug targeting. This idea has already been implemented in the concept of “chronotherapy” for cancers. By understanding how tumors may present different circadian activity than surrounding healthy tissue, timing chemotherapy or radiotherapy to have maximal effect on the cancer with less effects on healthy tissues (Levi 2006).

1.4. A network of central and peripheral oscillators

Though it is undeniable that the SCN is the master circadian clock, as in nearly all other neuroendocrine pathways, the circadian system is comprised of a “pseudo hierarchical” network of control nodes throughout the brain and body. Originally, the system was conceived as purely hierarchical, with the SCN sitting atop the hierarchy. However, it has become clear that while the SCN is certainly the master circadian clock it controls a network of circadian oscillators throughout the body, which also interact with each other. But how does the SCN control these myriad oscillators? The current models involve two non-mutually exclusive methods: direct neural control and humoral regulation, with indirect regulation (which may be a combination of the previous two) also playing a role.

1.4.1. Direct Neural Control

Tract tracing has been used to carefully delineate neural outputs from the SCN to other brain areas (Morin 2013). As discussed earlier, the SCN has several component parts, each showing different peptidergic contents, and anatomical projections. However, neural efferents originate in both the SCN core and shell. What has been discovered is that many SCN outputs are monosynaptic projections targeting key neuroendocrine cell populations that are involved in the production of essential releasing hormones. Direct projections from the SCN have been traced from the SCN to the medial preoptic area (MPOA), supraoptic nucleus (SON), anteroventral periventricular nucleus (AVPV), the paraventricular nucleus (PVN), dorsomedial nucleus of the hypothalamus (DMH), lateral septum, and the arcuate nucleus (Arc) (van der Beek et al. 1993; Van der Beek et al. 1997; Gerhold, Horvath, and Freeman 2001; Horvath, Cela, and van der Beek 1998; Buijs, Hermes, and Kalsbeek 1998; Kalsbeek et al. 1996; Vrang, Larsen, and Mikkelsen 1995; Kriegsfeld et al. 2002; Egli et al. 2004; Buijs et al. 2003; Kalsbeek and Buijs 2002; Kalsbeek et al. 2000). Given these brain nuclei regulate secretion of factors into the cerebrospinal fluid, pituitary portal system, and general circulation, SCN-derived signals can control widespread systems in the brain and body (Skinner and Malpaux 1999; Skinner and Caraty 2002; Reiter and Tan 2002; Tricoire et al. 2003).

An elegant example of how the SCN can control downstream targets via neural projections comes from work done in the regulation of reproductive hormones. Initial indications that endocrine function may be under direct neural control of the SCN is that hormonal rhythms are eliminated after the severing SCN efferents (Hakim, DeBernardo, and Silver 1991; Nunez and Stephan 1977). In addition, while behavioral rhythms are restored in SCN-lesioned animals following SCN transplants, hormonal rhythms are not (Nunez and Stephan 1977; Meyer-Bernstein et al. 1999; Silver et al. 1996). Making use of the phenomenon known as “splitting” which occurs in hamsters housed in constant light (LL), several groups have specifically probed how the SCN regulates rhythms of reproductive hormones. In “splitting”, hamsters begin to show two separate bouts of activity within each 24h period. These split rhythms seem like in each 24h day, the split hamster experiences two 12h days. Supporting this theory, it was demonstrated that split female hamsters show two daily preovulatory surge of luteinizing hormone (LH), and that the plasma concentration of LH in each of these surges was only about ½ of the plasma concentration of a control female (Swann and Turek 1985). The discovery of rhythmic clock gene expression in the SCN allowed the further probing of this phenomenon, and it was found that in split hamsters the left and right SCNs showed activity that was 180 degrees out of phase with each other (de la Iglesia et al. 2000). The LH surge is regulated by activity of gonadotropin releasing hormone (GnRH) neurons, and split hamsters were found to have activity in GnRH neurons only in one side of the brain at a time — and in the ipsilateral side to the active SCN (De la Iglesia, Meyer, and Schwartz 2003). These findings support the conclusion that the timing of the LH surge must be from a neural signal in the SCN, which is communicated to ipsilateral GnRH neurons. This has to be a neural signal since a diffusible signal from the SCN would impact both hemispheres of the brain indiscriminately.

1.4.2. Humoral Regulation

While the neural regulation of physiology and behavior by the SCN is clear, a remarkable set of studies demonstrate that the SCN can also exert its influence via a set of humoral or diffusible signals. Lesions of the SCN had definitively shown that it is necessary for normal circadian rhythms to be expressed within an animal, however, SCN transplants into SCN lesioned hamsters showed more mixed results.

It is well known that plasma melatonin rises at the start of the dark phase in humans and non-human animals and can be considered a key vertebrate timing hormone (reviewed in Simonneaux and Ribelayga 2003; Johnston and Skene 2015). The pineal gland is the main driver of this timed melatonin release, via a projection from the SCN via the superior cervical ganglion. Thus, the dusk rise in plasma melatonin has been shown to be driven by an interaction between the light-dark cycle and the endogenous clock and serves as an excellent phase marker (Lewy and Sack 1989; Lewy 2007), and importantly can serve as a critical driver of photoperiodic changes (Simonneaux and Ribelayga 2003; Hazlerigg 2012; Hazlerigg and Wagner 2006; Nishiwaki-Ohkawa and Yoshimura 2016). Many studies have demonstrated that melatonin can serve as a phase resetting agent acting through melatonin receptors (MT1 and MT2) found throughout the brain and the periphery of mammals (Dubocovich and Markowska 2005). As such, there is significant interest in the use of melatonin as pharmacological tool to reset the circadian clock after an acute circadian disruption (e.g. transmeridian flight, aka “jetlag”). However, the precise mechanism of action (if there are any effects) remains elusive. Melatonin can alter core body temperature (Johnston and Skene 2015), and it is possible that these changes in core body temperature can also serve as synchronizing cues to peripheral clocks (Schibler et al. 2015), providing yet another pathway by which melatonin can act as a timing hormone. As a caveat, in humans the links between melatonin, body temperature and sleepiness is less clear (Lok et al. 2019). A few caveats are certainly required in this brief discussion of melatonin. The first is that it is likely that nearly 99% of melatonin in the body is not of pineal origin and never released, instead being contained within mitochondria where the molecule can act as a free-radical scavenger independent of its activity at MT1 or MT2 (Zhao et al. 2019). Remarkably, and perhaps unknown to many researchers, many strains of laboratory mice (e.g. C57BL/6) have negligible pineal melatonin production due to a point mutation in arylalkylamine N-acetyltransferase (AANAT) a critical enzyme in the melatonin synthesis (Ebihara et al. 1986; Roseboom et al. 1998; von Gall et al. 2000). These animals have stable and precise circadian rhythms, but these rhythms can indeed be shifted by exogenous melatonin (Dubocovich et al. 2005), demonstrating that melatonin is not necessary for normal circadian rhythms to be expressed, but can still influence circadian timing, largely through the MT1 and/or MT2 receptors.

As discussed above, though SCN transplants restored many behavioral rhythms (e.g. locomotion, feeding, drinking), they did not restore hormonal rhythms, or rhythms in neuroendocrine axis activity. Specifically, Meyer-Bernstein et al. (1999) showed that in SCN-lesioned hamsters that had behavioral rhythms restored, did not have their estrous cycle rhythms rescued, nor could estradiol induce LH surges at the appropriate circadian time. Rhythms in adrenal hormones were absent in SCN-transplanted hamsters, nor did plasma melatonin rise at the expected time in the late subjective night. This led to the conclusion that it while the neural integration of the graft was enough to restore locomotor rhythms, more extensive (or complex) connections where needed to restore endocrine function. However, in a heroic set of experiments, the necessity of neural integration for restoration of behavioral rhythms was tested by transplanting encapsulating donor SCN tissue in a membrane that prevented neural outgrowth while allowing signals to diffuse between graft and host (Silver et al. 1996). These experiments demonstrated that even with an encapsulated graft, behavioral rhythms could be restored, showing that neural connections from graft to host were not necessary. To further probe the role of diffusible signals in regulation of SCN function, an in vitro coculture technique was developed by Maywood and colleagues. In these experiments, a “control” wild-type SCN was placed upon a membrane that enabled communication via shared culture media with a target host SCN that was rendered arrhythmic via genetic deletion of different neuropeptides or molecular components of the core circadian clockwork (Maywood et al. 2011). These paracrine signaling experiments demonstrated that a control SCN could rescue circadian pace-making ability in cultured SCN slices which were deficient in vasoactive intestinal peptide (VIP), which is key in communicating between cellular oscillators in the SCN. Moreover, they also demonstrated that arrhythmic SCN tissue carrying a Cry-null mutation (thus missing a core clock element), could be induced to become rhythmic when co-cultured with the wild-type SCN.

Thus, not only are diffusible humoral signals sufficient to regulate rhythms in behavior, they are also able to act at the level of the SCN to synchronize SCN oscillators. Conversely, neural outputs seem necessary to generate rhythms in neuroendocrine signals and endocrine axes more generally.

2. Circadian control of reproduction

2.1. Rhythms of the HPG axis.

The HPG is a primary regulator of reproductive function. It is also evident that circadian rhythms are present at many levels of the HPG, albeit with significant species differences (Khan and Kauffman 2012). In many species, including rodents, these rhythms are under control of the SCN clock, as both SCN lesions and genetic manipulations of core clock disrupt HPG function (Chu et al. 2013; Gray et al. 1978; Miller et al. 2004). In mice, mutations of the core clock genes Clock or Bmal1 lead to altered LH surge rhythmicity, and abnormal estrous cyclicity (Chu et al. 2013; Gray et al. 1978; van der Horst et al. 1999). Remarkably, human females who have a single-nucleotide polymorphisms in ARNTL (Bmal1) have more miscarriages (Kovanen et al. 2010), and though this effect is likely mediated by a non-CNS site of action, it further substantiate a role for the clock in regulation of the basic function of the HPG. The circadian contribution to HPG regulation has been well characterized in rodent models, with the Syrian hamster being a gold-standard model system. In entrained (i.e. 24h light-dark cycle) the estrous cycle is nearly exactly 4 days long. Early studies clearly demonstrated the link between the endogenous circadian period and estrous cycle length, showing that in free-running (i.e. in constant conditions) hamsters, the estrous cycle is a multiple of four of the free-running period (Fitzgerald and Zucker 1976). If hamsters are entrained to a non-24hr light-dark cycle (e.g. 22h), or if the circadian period is lengthened (e.g. using heavy water) the duration of the estrous cycle is still, reliably, a quadruple multiple of the circadian day.

2.2. From SCN to HPG: Neural Control of HPG Function by the Circadian Clock.

As described above, the SCN is a compact yet particularly heterogenous structure in terms of neuropeptidergic content. Of the many types of cells within the SCN, two seem particularly important in the regulation of HPG function: vasoactive intestinal peptide (VIP), and arginine vasopressin (AVP) containing neurons. In both cases, these cell types seem to exercise their regulation of the HPG via actions, directly or indirectly, on gonadotrophin releasing hormone (GnRH) cells (Figure 2).

Figure 2. Circadian Control of the Hypothalamic-Pituitary-Gonadal (HPG) Axis.

A depiction of the main nodes of the circadian control of the HPG axis. While vasoactive intestinal polypeptide (VIP) neurons of the suprachiasmatic nucleus (SCN) project directly to gonadotropin releasing hormone (GnRH) cells in the preoptic area (POA), arginine vasopressin (AVP) cells project to Kisspeptin (KISS) neurons in the anteroventral periventricular area (AVPV), which then synapse onto GnRH neurons. Gonadotropin inhibitory hormone/RFRP-3 (RFRP-3) neurons in the dorsomedial hypothalamus (DMH) play an important role in the circadian gating of responses of both GnRH and KISS neurons. Another population of gonadal hormone sensitive KISS neurons (aka Kisspeptin, Neurokinin B, Dynorphin; KNDy neurons) in the arcuate nucleus (ARC) are also inhibited by GnIH and project onto terminals of GnRH neurons in the median eminence, and act as important regulators of GnRH secretion. Outputs of the gonads, particularly androgens and estrogens, can impact circadian timing. While androgen receptors (AR) have been found within the core region of the SCN, suggesting a direct effect, estrogen type one receptor (ESR1) is only sparsely expressed in the SCN, with most effects likely being mediated by extra-SCN hypothalamic relays. Beyond the scope of this review, gonadal hormones also directly act upon many of the neuron types described in this pathway. Oscillations in clock genes (depicted by the small oscillator) have been found in many levels of the HPG-axis, suggesting local timing control is a key mechanism by which the SCN can regulate the overall activity of the axis.

VIP and AVP neurons are spatially distinct within the SCN, with the former being located more ventrally and the latter being located more dorsally. Efferents of SCN VIP neurons contact GnRH neurons (Van der Beek et al. 1997), with female rats showing higher GnRH innervation by VIP than do males (Horvath, Cela, and van der Beek 1998). Developmentally, it has been demonstrated that in hamster, VIP contacts increase following puberty (Kriegsfeld et al. 2002). Ex vivo, VIP can affect firing rates of GnRH neurons (Piet et al. 2016), and importantly, VIP has a time-of-day sensitive effect on GnRH neuron firing, driving an increase in firing rate only at around the time of the LH surge, but much less of an effect at other times (Christian and Moenter 2008). Similarly, in female mice, pharmacological blockade of the VPAC2 receptor attenuates neuronal firing of GnRH cells only during the afternoon proestrus surge.

AVP in the plasma peaks in a coincident manner with the preovulatory LH surge (Schwartz, Coleman, and Reppert 1983). In rat, AVP injected into the medial preoptic area (MPOA) induces an LH surge in SCN-lesioned animals. In Clock mutant mice, LH surges are not detectable, but can be induced by central injections of AVP (Miller et al. 2006). Even in in vitro models, AVP has clear effects on GnRH rhythms. Preoptic area and SCN co-cultures show that GnRH release is in phase with AVP release, (Funabashi et al. 2000). However, what is interesting about AVP cells in their regulation of HPG function is that, unlike VIP, there is as yet no direct evidence of AVP innervation of GnRH neurons, and, perhaps more importantly, GnRH neurons only express the AVP receptor (V1a) sparsely, if at all (Kalamatianos et al. 2004). How does AVP have such a pronounced impact on the LH surge without communication directly to GnRH cells? An intermediate player had to be at work in this circuitry.

One potential way this mystery could be solved is that the GnRH cells themselves are capable of generating part of this rhythmic sensitivity. There is evidence of time-of-day dependent sensitivity of GnRH cells to VIP and AVP (Williams et al. 2011). However, there is not strong evidence for functional consequences of a cell-autonomous GnRH circadian clock. While GnRH neurons in vivo do not have endogenous rhythms in clock gene expression, GnRH mRNA content does show a diurnal pattern (Kriegsfeld 2006; Schirman-Hildesheim, Ben-Aroya, and Koch 2006). The data from in vitro preparations are clearer. GnRH neurons show circadian rhythms in firing rates in hypothalamic slice cultures, even without an SCN present (Christian, Mobley, and Moenter 2005). In GnRH secreting GT1–7 cell lines, rhythms can be detected in both GnRH mRNA and many clock genes (Gillespie et al. 2003), and dominant negative overexpression of mutant Clock-Δ19 in these cell lines decreases the frequency of GnRH pulses (Chappell, White, and Mellon 2003). Thus, one part of this missing connection between SCN AVP and GnRH/HPG function could be explained by intrinsic rhythms of GnRH cells.

While a role for cell autonomous GnRH clocks may be part of this story, it cannot be the only part. Two other players have now been implicated in the both general regulation of the HPG, but also in the circadian control of ovulation: Kisspeptin and gonadotropin releasing inhibitory hormone (GnIH). Kisspeptin (KISS), and its receptor GPR54 (KISS1r), was first implicated in regulation of gonadal function when a mutation in GPR54 was found in hypogonadal individuals (de Roux et al. 2003; Seminara et al. 2003). The neuroanatomy and function of the KISS signaling system across mammalian species has been extensively documented (Lehman, Hileman, and Goodman 2013). Many studies have demonstrated that the effects of KISS are via actions on GnRH, and are important for the regulation of ovulation. First, KISS1r mRNA is found in GnRH cells of many species (Han et al. 2005; Irwig et al. 2004; Messager et al. 2005), and application of KISS increases GnRH cell firing in vitro (Liu, Lee, and Herbison 2008). Second, KISS cell bodies are concentrated within the anteroventral periventricular (AVPV) and arcuate (ARC) regions of the rodent brain, and express many sex-steroid receptors (Smith 2008). Third, studies with transgenic KISS knockout mice show that KISS is necessary for the expression of LH surges (Dror, Franks, and Kauffman 2013). In the species explored, most ARC KISS cells co-express neurokinin B (NKB) and dynorphin, positive and negative regulators of the reproductive axis, respectively (Goodman et al. 2007). These neurons have been coined “KNDy” (K=KISS, N=NKB, and Dy=dynorphin) (Lehman et al. 2010; Lehman, Coolen, and Goodman 2010). Interestingly, these neurons also co-express receptors for NKB and dynorphin and are reciprocally interconnected, which underpins their unique ability to propagate stimulatory or inhibitory signaling (Lehman et al. 2010; Lehman, Coolen, and Goodman 2010), and also express progesterone and estrogen receptors, which enables steroid negative feedback (Foradori et al. 2002; Smith et al. 2007). As for the circadian control of ovulation and these neurons, KISS mRNA expression is highest at the time of the LH surge in the AVPV, and express the highest levels of FOS expression on the afternoon of proestrus (Williams et al. 2011; Robertson et al. 2009). The SCN has also been shown to project to the AVPV and regulate the function of resident KISS neurons (Kriegsfeld et al. 2004; Leak and Moore 2001), further supporting the notion that the timing of the LH surge may occur via projections to AVPV KISS cells. More recently, Ca²⁺ imaging approaches have elegantly demonstrated both AVP and VIP can directly modulate ARC KISS neurons in a sex-dependent manner, which shows a rostro-caudal anatomical organization in the ARC (Schafer et al. 2018).

In addition to KISS and KNDy neurons, GnIH plays an important part of the circadian regulation of ovulation. GnIH was first described in avian species (Tsutsui et al. 2000; Tsutsui et al. 2009), and more recently the mammalian ortholog (Arg)(Phe) related peptide-3 (RFRP-3) has been uncovered (Tsutsui and Osugi 2009; Tsutsui et al. 2018). RFRP-3 cells are largely concentrated in the dorsomedial hypothalamus (DMH) of rodents, with extensive hypothalamic and limbic projections, and project to GnRH cells and to the median eminence (Kriegsfeld 2006). Functionally, a role for RFRP-3 has been defined by demonstrations that RFRP-3 injections cause a rapid LH suppression in several rodent species. Similar to KNDy neurons, RFRP-3 expressing cells also possess estrogen receptors, and are an important target for steroid negative feedback (Kriegsfeld 2006). With regards to circadian control, it has been well-established that the SCN has significant efferent projections to the DMH region that contains the RFRP-3 cells (Kriegsfeld et al. 2004; Leak and Moore 2001). Moreover, a large proportion of RFRP-3 cells (in excess of 60%) are contacted by terminals of SCN origin (Gibson et al. 2008; Russo et al. 2015). This system has been characterized in a functional context as well, with the SCN acting to suppress activity of RFRP-3 neurons during the LH surge, which allows for the time-coordinated release of the GnRH system from negative feedback (Gibson et al. 2008; Russo et al. 2015). It is important to note that the role for RFRP-3 in mammalian reproduction is not without some controversy or caveats (Reviewed in Angelopoulou et al. 2019).

2.3. From HPG to SCN: Reciprocal Feedback to the Clock

It has been observed that gonadal hormones can modulate circadian rhythms, suggesting that gonadal hormones may feedback to the circadian clock. For instance, cycling hamsters show a clear phase advance on the day of estrus when levels of E2 peak. Conversely, clamping E2 by pellet implant shortens free-running rhythms (Morin, Fitzgerald, and Zucker 1977). Since these aspects of circadian rhythms (phase and period) are known to be driven by the SCN, it suggests that there must be E2 effects on the SCN directly. However, estrogen receptors (ESR) are only sparsely expressed in the SCN (Shughrue, Lane, and Merchenthaler 1997; Hileman, Handa, and Jackson 1999; Gundlah et al. 2000; Vida et al. 2008). Thus, this suggests an indirect circuit may be responsible. Several populations of ESR1 positive cells provide input to the SCN, including the ARC, amygdala, bed nucleus of the stria terminalis (BNST), and the POA (De La Iglesia, Blaustein, and Bittman 1999), providing the basis for such a circuit. Indirect E2 effects may still occur in the SCN, perhaps without ESR activity through gap junction effects on glial cells and neurons. Cultured SCN cells show the presence of gap junctions, both by dye and electrical coupling (Colwell 2000; Jiang, Yang, and Allen 1997; Long et al. 2005), and blockade of gap junctions in vitro affect SCN electrical activity (Shinohara et al. 2000; Prosser et al. 1994). In vivo, male connexin-36-knockout mice show reduced amplitude of locomotor activity rhythms (Long et al. 2005). In female rats, E2 leads to increased expression of the inter-neuronal gap junction subunit connexin-36 in the SCN (Shinohara et al. 2001; Rash et al. 2007). While this hypothesis has yet to be completely tested, it remains a plausible way by which E2 can affect circadian function.

Gonadal androgens also show strong influences on circadian locomotor activity. Castration of male mice results in a longer free-running period, decreased precision and significantly reduced consolidation of daily activity (Karatsoreos et al. 2007; Daan et al. 1975). Replacement of either testosterone (T) or the non-aromatizable dihydrotestosterone (DHT) rescue this phenotype, which suggests that the androgen receptor (AR) mediates the effects of androgens on circadian rhythms, rather than conversion to E2 and effects through the ESR. In male mice, androgens also drive an increased sensitivity of the SCN to light (Butler et al. 2012), which seems to be related to the effect on period. Remarkably, in mice, the SCN contains a highly level of AR, which are concentrated in the core region (Karatsoreos et al. 2007), and local administration of T to the SCN restores intact-typical circadian periods, strongly indicating a direct effect of androgens in the SCN (Karatsoreos et al. 2007; Model et al. 2015). In other species, while AR can be found in the SCN, it seems to be more diffuse than observed in mouse (see (Karatsoreos and Silver 2007)for review). It is interesting to note that there seems to be a clear sex difference in the expression of SCN AR in both humans (Fernandez-Guasti et al. 2000) and in rodents (Iwahana et al. 2008). In mice, ARs are more highly expressed in male than in females. This sex difference in SCN AR are paralleled by functional sex differences, as ovariectomy does not have the same effects on period or the onset activity bout as castration does in males. Thus, effects of androgens appear to be mediated largely by direct effects at the level of the SCN, while the effects of estrogens may occur primarily outside the SCN.

3. Hypothalamic-pituitary-adrenal (HPA) axis

3.1. Rhythms of the HPA axis.

The primary output of the HPA axis are glucocorticoids, such as cortisol (corticosterone in most rodents; CORT), and circadian rhythms of CORT in blood have been well described in many species (Dickmeis 2009), including humans, non-human primates, and rodents (Weitzman et al. 1971; Moore and Eichler 1972; Sachar et al. 1973; Dubey et al. 1983; Czeisler and Klerman 1999; Van Cauter and Refetoff 1985). An important aspect of CORT in the plasma is that its phasing does not seem to be linked to the light-dark cycle per se, but instead to the activity phase of the animal. For instance, between nocturnal and diurnal species, CORT levels rise before waking in both, which results in a peak during the day in diurnal animals, and a peak during the night in nocturnal animals (Wong et al. 1983; Albers et al. 1985; Ottenweller et al. 1987). Rhythms of CORT in humans have been a major target of study because they are associated with several neuropsychiatric disorders. As in other diurnal species, cortisol levels in humans peak around the morning wakeup time.

A discussion about rhythms in CORT and the HPA axis would not be complete without addressing the findings that circulating CORT rhythms are actually generated by changes in the pulsatile secretion of the hormone, that occurs in an ultradian fashion, with about a 1hr period (Walker et al. 2012; Spiga et al. 2014; Russell, Kalafatakis, and Lightman 2015). Remarkably, and differently than the HPG, data show that this ultradian pattern of CORT secretion is driven by feed-forward mechanisms and timed delays in biological processes at the level of both the pituitary and adrenal glands; it does not require a hypothalamic pulse generator (Walker et al. 2012; Walker, Terry, and Lightman 2010). The functional significance of these ultradian pulses is an area of significant research, with findings demonstrating that the adrenal responds optimally to a pulsatile pituitary adrenocorticotropic hormone (ACTH) profile, given that constant ACTH infusion results in significantly reduced CORT levels (Spiga et al. 2011).

3.2. HPA Rhythms: Role of the Suprachiasmatic Nucleus

As noted above, there has been significant work aimed at understanding how rhythms in the HPA are generated, with the pulsatile profile seemingly regulated extra-hypothalamically. For instance, the adrenal gland shows high amplitude rhythms in core clock genes, which form the basis for the rhythmic responsiveness to ACTH, and to both physiological and physical stressors (Ungar and Halberg 1962; Kalsbeek et al. 2003; Bittman et al. 2003; Oster et al. 2006). Thus, at the level of the target glad (in this case the adrenal), circadian clocks gate sensitivity to other signals on a daily basis, much like the elegant control of ovulatory rhythms as discussed above.

While peripheral clocks are clearly involved in regulation of HPA rhythms, the SCN is known to be necessary for overt circadian rhythms in CORT. As noted above, lesions of the SCN completely eliminate CORT rhythms, and these rhythms are not restored by SCN transplants (Meyer-Bernstein et al. 1999; Moore and Eichler 1972). Moreover, a period of circadian “forced desynchrony” to disrupt SCN function has also been shown to drive changes in corticosterone rhythms in rats (Wotus et al. 2013). Though the adrenal gland contains the molecular machinery to express circadian rhythms, adrenal rhythms of clock gene expression are dependent upon the SCN (Guo et al. 2006). Together, these lines of evidence strongly implicate a neural projection from the SCN to the adrenal that can regulate rhythms in this tissue (Figure 3). There are two efferent pathways from the SCN that have been implicated in its regulation of HPA rhythms. The first involves monosynaptic projections from the SCN to the a neighboring hypothalamic region, specifically onto corticotrophin releasing hormone (CRH) neurons in the PVN (Vrang, Larsen, and Mikkelsen 1995; Kalsbeek et al. 1996; Buijs, Hermes, and Kalsbeek 1998). Output from these neurons is clearly important in both the generation of circulating CORT rhythms, as well as for the regulation of response to acute stressors. Complimenting this monosynaptic pathway, a multisynaptic pathway from the SCN to the adrenal cortex itself has also been discovered (Buijs et al. 1999), the specific role of which has not been fully characterized.

Figure 3. Circadian Control of the Hypothalamic-Pituitary-Adrenal Axis.

The SCN projects to corticotrophin releasing hormone (CRH) neurons of the paraventricular nucleus (PVN) which then stimulate release of adrenocorticotropic hormone (ACTH) from the pituitary that then causes release of corticosterone/cortisol (CORT) from the adrenal cortex. CORT can then affect peripheral organs (such as the liver) and other brain regions by causing transcriptional activation of many hundreds of genes, including clock genes. While there is no apparent effect of glucocorticoid receptor (GR) activation in the adult SCN, this indirect action of CORT on peripheral tissues and brain regions can subsequently affect the SCN. There is also neural innervation of the adrenal cortex from the SCN via a multisynaptic projection via the PVN, to the spinal cord and via the Splanchnic nerve to the adrenal cortex. Remarkably, a direct monosynaptic projection from the SCN to the adrenal cortex has also been discovered. It is important to recognize that oscillators (depicted by the small oscillator symbol) exist at nearly every level of this neuroendocrine axis, providing an important local timing cue that can affect how upstream signals are processed, and downstream signals are communicated.

3.3. Effects of Adrenal Rhythms on Brain and Periphery

While the SCN is essential in driving rhythms in the HPA, the SCN itself does not express glucocorticoid receptors (GR) (Balsalobre et al. 2000; Rosenfeld et al. 1988), but much as E2, adrenal hormones have indirect effects on the SCN. For instance, treatment with glucocorticoids affects sleep in humans, and also changes AVP expression within the SCN (Liu et al. 2006). Similarly, rodent studies indicate glial fibrillary acidic protein is upregulated in the SCN following treatment with glucocorticoids (Maurel et al. 2000). However, the mechanisms by which these effects occur, and their overall significance for circadian function, have yet to be fully elucidated.

In describing how the SCN can regulate peripheral clocks, perhaps by way of diffusible signals, it has become increasingly clear the glucocorticoids, and rhythms in glucocorticoids, are an important route for humoral control of peripheral clocks in the brain and periphery. Several studies have shown that PER2 is rhythmic in extra-SCN brain areas, including the oval nucleus of the BNST (BNST-OV), and the central (CEA) and basolateral nuclei of the amygdala (BLA) (Amir et al. 2004; Lamont et al. 2005). Remarkably, these rhythms are not monolithic in nature, in that while rhythms in the BNST-OV and CEA depend on an intact adrenal, PER2 rhythms in the BLA are not affected by adrenalectomy. That CORT is a key player in this rhythm is evidenced by experiments that showed PER2 rhythms could be restored with CORT in the drinking water (i.e. a rhythmic route of administration), but not by pellet implantation (i.e. tonic CORT) (Segall et al. 2006). More recently, a role for diurnal changes in CORT has been uncovered in cortical regions, with diurnal CORT rhythms seemingly important for spine turnover in motor cortex, which seems to drive performance in motor learning tasks (Liston and Gan 2011).

Outside of the brain, adrenal corticosteroids are important in the communication of circadian time to the periphery (Balsalobre et al. 2000). Intraperitoneal injection of dexamethasone (DEX), a potent GR agonist, can cause phase shifts in hepatic clock gene expression, demonstrating that GR signaling can synchronize liver clocks. More broad assays of hepatic gene expression has demonstrated that rhythms of 100 genes in the liver lose their rhythmicity following ADX (Oishi, Shirai, and Ishida 2005). It is interesting to observe that nearly 60% of hepatic transcriptome rhythms that are eliminated by SCN lesions can be restored by DEX treatment (Reddy et al. 2007). The mechanisms by which these effects of GR are translated to changes in transcription are myriad. Since only 2/3 of the genes that had their rhythms restored by DEX contain glucocorticoid response elements (GRE) (Reddy et al. 2007), other indirect pathways must be important, and likely rely on interactions between other clock genes or clock controlled genes.

4. Interlocking loops: Physiological significance

In closing this chapter, it is important to consider the physiological significance of the interactions between biological rhythms and the HPG/HPA axes. What are the potential advantages that this tight integration would provide for an organism? Biological rhythms that are endogenously generated, such as circadian rhythms, provide a mechanism through which recurring events in physiology and the environment can be predicted. Thus, organisms need not simply reactively responding to these regular occurrences but can instead anticipate them. It is thought that this ability provides more efficient regulation of physiological and behavioral processes, thereby ensuring that resource usage is maximized. This hypothesis is logically appealing, and both the HPG and HPA axes provide excellent models to test its validity.

With regards to the HPG and the ovulatory cycle, biological rhythms are critical both on the seasonal and circadian timescales, since ovulation only happens when several different cyclical factors (e.g. estradiol and luteinizing hormone levels) occur in a particular sequence and interval. If these events occur in a different temporal order, ovulation fails, and fertility is thus reduced (Angelopoulou et al. 2019; Beymer et al. 2016; Simonneaux, Bahougne, and Angelopoulou 2017). In species that have limited windows of fertility (e.g. spontaneous ovulators), it is critical to ensure that the sexual motivation is timed to occur at maximal fertility, and the sensitivity to estradiol levels is important to ensure that the follicle is mature before ovulation. Similarly, in the context of HPA function, it is hypothesized that the daily rise in corticosteroids at the time of waking permits a greater mobilization of glucose when the animal is beginning the active phase. Many circadian “clock genes” have GREs in their promoter regions (Reddy et al. 2007; Reddy et al. 2012), permitting interaction between plasma glucocorticoids and local molecular circadian clocks in tissues, thereby adjusting local timing at the tissue level. For example, the daily rise in corticosteroids may help synchronize peripheral clocks in the gut so they are prepared to receive food. Conversely, the lack of GR in the SCN seems to provide the master circadian clock some protection against acute stressor induced phase shifts.

Finally, there is cross talk between the HPA and HPG, with the aspects of the HPG axis being sensitive to stress mediators related to HPA activity, and metabolism. RFRP-3 has been shown to affect, and be affected by, both stress and metabolic signals (Takayanagi and Onaka 2010; Schneider et al. 2017), and both KISS and RFRP-3 can modulate body mass in several species (Cazarez-Marquez et al. 2019; Talbi et al. 2016), and KISS1 signaling is important in linking fertility and nutritional state (Padilla et al. 2019; Padilla et al. 2017). GnRH neurons are also sensitive to glucose (Roland and Moenter 2011b, 2011a), further linking HPG, HPA, circadian, seasonal, and metabolic systems in interlocking loops of regulation. Many aspects of this model for the adaptive significance of rhythms in the HPG and HPA remain experimentally untested and may well be nearly impossible to dissect. While laboratory experiments that disrupt the temporal organization of these systems do show negative outcomes, as discussed above, naturalistic studies are far more difficult to undertake. Seasonality (with regards to reproduction) is an excellent example of a natural phenomenon that can be modeled in the laboratory. While a detailed discussion of seasonality is beyond the scope of this chapter, the changing duration and intensity of light over the year, coupled with changes in temperature and food availability, provides an important cue to animals about changing environmental circumstances that could affect reproductive success (Revel et al. 2007; Simonneaux et al. 2012; Dardente et al. 2019). For instance, it may not be best to give birth at a time when food will be scarce (e.g. the winter), so fertility is highest at times of year such that parturition occurs at an optimal time to make use of environmental resources. A similar set of processes can happen in more acute situations. For example, it may not be wise to spend energy on reproductive function if the organism is coping with a stressor in the environment that immediately threatens survival (e.g. forest fire, major storm), and hence HPA outputs can inhibit HPG function. More mechanistic work still needs to be done to understand the functional significance of the circadian clocks at the different levels of both the HPG and HPA axes, and how they contribute directly to some of these adaptive processes, but the preponderance of the evidence strongly supports an important physiologic role for this tight temporal regulation.

5. Summary

Seasonal reproductive rhythms in animals have been observed for millennia. It has been more recently that circadian rhythms in neuroendocrine function have emerged as an area of focused research. It has become clear that the circadian clock regulates rhythms in the HPG and HPA axes through a variety of control systems (Figure 4), from a central time keeping mechanism in the SCN, to circadian clocks at the level of important neuroendocrine brain regions, and finally to circadian clocks in the endocrine tissues themselves. These rhythms in endocrine function have significant impacts on the health and wellbeing of animals, impacting everything from reproduction, to metabolism, to stress and immune function. They also feedback to the circadian system, both directly and indirectly. Thus, they form an interconnected loop that intimately links circadian rhythms and neuroendocrine function. In the past decades, it has become apparent that this loop can be disrupted, with consequences for both mental and physical health (Karatsoreos 2012, 2014; Karatsoreos and McEwen 2011; Oliver et al. 2012). It is critical to continue investigating the specific mechanisms that regulate these interacting loops so that we may be able to develop countermeasures or treatments when they become desynchronized with each other, and with the solar day.

Figure 4. The Circadian system and Neuroendocrine Regulation.

This schematic depicts that the SCN brain clock can regulate both the HPG and HPA axes through neural (solid arrows) and diffusible (dotted arrows) signals, which then impact physiology and behavior. Similarly, the outputs of the HPG and HPA can have effects back upon the SCN clock. These inter-related loops ensure that physiology, behavior, and recurring environmental events are synchronized to promote optimal function. Circadian disruption is an insidious factor that can drive changes in these carefully balanced systems, leading to disrupted rhythms in HPG and HPA function that could lead to a variety of mental and physical health issues, from depression to infertility.