The present goal is to explore how the SCN influences rhythms of circulating steroid secretion, and how synchronization and de-synchronization of the CTS impact and are impacted by steroid hormone actions on cellular/molecular clocks located throughout the body, including those of the SCN. We explore plasma levels, inter-individual variability, differences among species and between the sexes. Rhythmic response patterns are locally unique. A great number of studies have examined oscillation in one brain nucleus or one bodily tissue at a time and the challenge is to understand the system as a whole.

The CTS presents an empirically accessible system for studying coordination of signals from the environment and from within the body. The molecular, cell-based clock mechanism can be impacted by external and internal stimuli in a number of ways, thereby producing alignment or misalignment of temporal changes in the body. To understand how coordination of bodily clocks is achieved, we try to capture simultaneously ongoing events at several temporal scales, from ultradian to circadian to menstrual/ovulatory to seasonal, and at different levels of analysis from cells to organs to body-wide events. The cartoon in Figure 1 is a visual representation at a glance of all the elements at play in the CTS. The visual summary allows us to highlight the idea that the CTS is not fundamentally hierarchically organized, and that one could begin a discussion at any starting point in the circle. The image also conveys the idea that everything depicted happens simultaneously. Finally, it describes the sequence of topics covered in this review, highlighting the key elements of the storyline which begin and end with the SCN.

Circulating steroid hormones of both gonadal and adrenal origin fluctuate daily, and act on tissues throughout the body. Daily hormone rhythms, however, are not fixed characteristics but are modulated by various behavioral, physiological and environmental conditions (Figure 2B). As an example, steroids affect level of physical activity (reviewed in Vitale et al., 2018) and physical activity in turn can improve muscle function, modulate body weight, and reduce feelings of depression. In contrast, physical inactivity is associated with increased incidence of cancer, obesity, diabetes, etc. Examination of the entire system impacted by circulating steroid hormones, reveals how the timing systems of the body regulate local and global responses and are themselves regulated.

On the output side, the SCN sends temporal information to the rest of the brain and body via both neural efferents (Morin, 2013) and by diffusible signals (Lesauter and Silver, 1998; Li et al., 2012). The timing signal from the SCN to hypothalamic neuroendocrine cells is apparently not continuously available, but appears to be communicated at a specific time of day, namely at the transition from light to dark measured by the onset of FOS expression (Butler et al., 2012). Interestingly, transplant studies of the SCN indicate that the circadian signal for hormone secretion is different from that regulating circadian locomotor behavior (Meyer-Bernstein et al., 1999). SCN transplants restore rhythms of locomotor activity, but not rhythmic endocrine secretion, indicating that they are based on different output signals. One possible explanation is that the latter require neural efferents while the former do not. Another possibility is that signals of SCN origin travel to their target sites via a portal system that courses between the SCN and the organum vasculosum of the lamina terminalis, a circumventricular organ lying in the 3rd ventricle (Yao et al., 2021). From the cerebral spinal fluid humoral signals can reach large volumes of parenchyma.

Suprachiasmatic nucleus control of steroid rhythms is achieved through actions on tropic hormones, on pineal melatonin, and via the autonomic nervous system (ANS). The major neural efferents of the SCN to neuroendocrine systems include both mono- and multi-synaptic connections. Efferents of SCN VIP neurons that lie in the core region of the SCN synapse monosynaptically onto hypothalamic gonadotropin releasing hormone (GnRH) neurons (van der Beek et al., 1993; Van Der Beek et al., 1997). The density of innervation differs between the sexes and increases during puberty, associated with maturation of the hypothalamic-pituitary-gonadal axis (HPG) (Horvath et al., 1998; Kriegsfeld et al., 2002). VIP₂ receptors are localized in the GnRH neurons (Smith et al., 2000). In brain slices, GnRH neurons from both female and male mice are excited by VIP through its receptor-mediated signaling, measured by increased firing rate and intracellular calcium (Piet et al., 2016). This effect was independent of stage of the estrous cycle and the time of day that the slices had been collected. In a multisynaptic pathway to releasing hormones, efferents of AVP neurons of the SCN shell synapse directly onto hypothalamic kisspeptin (Kp) neurons which in turn contact GnRH neurons (reviewed in Simonneaux, 2020). The SCN also sends monosynaptic efferents to neurons in the paraventricular nucleus. These neurons are thought to participate in balancing the sympathetic and parasympathetic systems to many organs including the pineal, gonads and adrenal glands (Buijs et al., 2019). Both routes regulate the glandular cellular clocks modulating steroid biosynthesis, discussed below.

Another major neural pathway through which SCN efferents regulate hormone secretions is by way of actions on the pineal gland through a multisynaptic innervation via the superior cervical ganglion. There is a vast literature on pineal melatonin, developed ever since it was determined that the duration of pineal melatonin secretion encodes the duration of the night, which of course becomes progressively longer during fall and winter and shorter during the spring and summer. Systemically circulating melatonin acts on multiple sites including steroidogenic cells in testes, ovaries, and adrenal glands that ultimately affect plasma steroid levels. The literature on the functions and mechanisms of action of melatonin is beyond the scope of this review and the reader is directed to Pfeffer et al. (2018), Cipolla-Neto et al. (2021), and Reiter and Sharma (2021).

The occurrence of a daily rhythm underlying the ovulatory luteinizing hormone (LH) surge and more broadly, the ovulatory cycle, has been known for decades (Everett and Sawyer, 1949). Unraveling the mechanisms underlying this daily event began in the early 2000s, with the discovery of ultradian “pulses” and circadian “surges” (Clarkson et al., 2017; Simonneaux, 2020; Nagae et al., 2021; Uenoyama et al., 2021). Pulsatility of GnRH is necessary for the release of pituitary gonadotropic hormones (Belchetz et al., 1978; Knobil, 1980; Urbanski et al., 1988) and involves a mechanism termed the “GnRH pulse generator.” Pulsatile GnRH secretion is produced by a population of hypothalamic Kp neurons of the arcuate nucleus (ARC). These cells generate synchronized GnRH release about every 90 min, driving pulsatile pituitary gonadotropin secretion. Pulsatility in plasma gonadal steroid levels is temporally linked to pulsatile pituitary gonadotropin.

The frequency of GnRH pulsatility determines the proportion of LH versus follicle stimulating hormone (FSH) released such that low frequency pulsatility of GnRH favors FSH synthesis whereas high frequency pulsatility facilitates LH (Wildt et al., 1981). In both sexes, testosterone (T) and progesterone (P) pulses occur ∼10–40 min after the peak of the LH pulse. Estrogens follow closely or are concurrent with T and P pulses. In women, the episodic secretion of LH and estradiol (E2) is correlated with systematic changes across the menstrual cycle. The inter-pulse intervals of LH and E2 steadily decrease over the follicular phase, and significantly increase during the luteal phase (Bäckström et al., 1982). The frequency of GnRH pulses, however, do not increase during ovulation, indicating that a mechanism other than the pulse generator underlies the ovulatory LH surge (Adams et al., 1994).

The aforementioned pulsatility of gonadal trophic hormones and ovarian hormones is regulated by the SCN. By integrating information about hormonal pulses, the SCN becomes the key to timing the ovulatory surge in LH. Environmental light cues synchronize the timing of AVP and VIP release from the SCN. In the late afternoon, under the condition of high circulating E2, VIP directly activates GnRH neurons. Vasopressin produces positive feedback on Kp neurons located in the POA/AVPV in the rostral hypothalamic regions (reviewed in Matsuda et al., 2019). In addition, the ARC Kp neurons in the preoptic area (POA)/anteroventral periventricular nucleus (AVPV) produce GnRH pulsatility in both males and females while only in females Kp neurons of the POA/AVPV region mediate the preovulatory GnRH/LH surge. Thus, anatomically distinct hypothalamic regions and neuropeptide projection pathways are critical to the timing of the LH surge and GnRH pulse generator (Figure 3).

As in the HPG axis, pulsatility is required for rhythmic corticosteroid secretion. Specifically, adrenocorticotropic hormone (ACTH) pulses are indispensable for generating pulsatile glucocorticoids (Spiga et al., 2011). In contrast to the HPG axis however, the pulse generator of the corticosteroids seems to be downstream of the hypothalamus. This conclusion derives from evidence that in the HPA axis, at an optimal concentration, a constant level of corticotropin releasing hormone (CRH) can generate pulsatile secretion of the ACTH and CORT (Walker et al., 2010, 2012). This suggests that the pituitary and adrenal glands form a closed feedback loop capable of fine tuning the ultradian oscillations of ACTH and glucocorticoids.

The SCN is not part of the HPA pulse generator; however, it plays a role in modifying the number of CORT pulses thereby producing the daily variation of this hormone (Waite et al., 2012). Specifically, SCN lesions cause an increase in the number of CORT pulses during the light period and this change abolished the diurnal rhythm of circulating CORT. This result indicated the SCN exert an inhibitory effect on the HPA pulse generator during the (behaviorally) inactive phase of LD cycle.

In summary, in both the HPG and HPA, ultradian pulses occur at multiple levels and contribute to the higher order diurnal and circadian oscillation rhythms. In one attempt to conceptualize the relationship between short and long term oscillations, a coupled oscillator model that assumed variable-strength coupling within and among hormonal responses was proposed (Grant et al., 2018). In this model various temporal rhythms of the brain and periphery could be linked by a deeper understanding of coupling mechanisms.

The ovaries, testes and adrenal glands are differentially modulated by the CTS, thus in order to understand which CTS signals affect particular steroid levels, it is important to know its source(s). Specific circulating steroids may originate from one or more sources (summarized in Table 1 for male–female sex steroids; as circulating corticoids are almost exclusively produced by adrenal cortex, they are not included in the table). The source of a steroid can be determined by assays performed before and after ovariectomy (OVX), orchidectomy or adrenalectomy (ADX), and in some cases, following adrenal hormone suppression by the potent synthetic glucocorticoid, dexamethasone treatment. Most data on daily and circadian variation of circulating steroids are available for humans and rodents, specifically rats and mice. (Though there is lovely data on other species, including Siberian and Syrian hamsters, this work is beyond the scope of the present review). Importantly, there are two major differences in sex steroid metabolism between humans and rodents (specifically rats and house mice). In humans, but not in these rodents, serum sex steroids are bound to high affinity steroid hormone binding globulin (Jänne et al., 1998). Second, in humans but not rats and house mice, the adrenal glands secrete substantial amounts of C-19 androgens, precursors of dehydroepiandrosterone (DHEA) and androstenedione (Δ4) (van Weerden et al., 1992; Boonstra et al., 2008; Soma et al., 2015; Quinn et al., 2016).

Studies of circulating hormones have produced a confusing picture of daily changes. Part of the problem is attributable to methodological constraints. Episodic steroid hormone secretion and diurnal variation in serum levels makes the interpretation of a single value uninterpretable. In animal studies, the problems of drawing blood samples in small rodents, a stressful procedure, is compounded by the difficulties in collecting sufficient material for serial samples within individuals (Yoo and Napoli, 2019). A source of differences among study results is the interval after an experimental manipulation that the assay is performed the pre- and post- menopause or andropause state, and in females, the stage of menstrual/ovulatory cycle. Pre-analytic methodological factors can also be a source of different results among studies (Gholib et al., 2021). In work with humans the importance of accurate steroid level measurements in diagnosis of diseases such as adrenogenital syndrome, precocious and delayed puberty hypercortisolism and adrenal insufficiency stimulated the development of more accurate and sensitive assays and improved sample collection methods. Newer, non-invasive methods involve fecal, urinary or salivary sampling, along with improvements in assay detection methods (discussed in Nilsson et al., 2015; Conklin and Knezevic, 2020) and the possibility of measuring numerous steroids in a single sample (Hill et al., 2019) have improved the quality of information on daily rhythms. Finally, adding to the difficulty of interpreting the results is that a great many different experimental designs have been used in studies of hormone levels with a variety of photic conditions, sampling times, sampling intervals, and sample sizes (summarized in Table 2).

Clock genes play a key role in the common steroidogenic pathway. Both StAR and CYP11A1, components of the common steroid biosynthetic pathway, are ccgs. The transcription of StAR while not CYP11A1 oscillates in adrenal and gonadal steroidogenic cells in vivo. However, for both molecules to reach their normal expression level, an intact cellular clock is necessary. In the testes, it is noteworthy that rhythmicity of StAR occurs only in the Leydig cells but is not detected when the testis is studied as a whole organ (Alvarez et al., 2008; Son et al., 2008). The level of StAR and CYP11A1 transcripts in isolated mouse Leydig cells becomes rhythmic after synchronization by dexamethasone treatment (Chen et al., 2017). StAR expression is significantly reduced in cultured Leydig cells of BMAL1 knockout mice and goats but CYP11A1 is not changed (Alvarez et al., 2008; Xiao et al., 2021). In TM3 mouse Leydig cells bmal1 knockdown decreased CYP11A1 mRNA (Ding et al., 2020). In ovaries StAR transcription in the human luteinized granulosa cells can be entrained by T stimulation (Chen et al., 2016). Administration of a Rev-erbα agonist elevated Per2 amplitude and advanced its phase in mouse granulosa cells and significantly increased StAR expression (Chen et al., 2012). In addition, steroidogenic factor 1-driven-bmal1-deletion resulted in significantly decreased StAR transcription in the mouse ovary (Liu et al., 2014). In mouse granulosa cells, the oscillation of CYP11A1 expression is synchronized to bmal1 (Chu et al., 2019). Silencing bmal1 decreased the expression of CYP11A1 in granulosa cells.

In the adrenal gland, the StAR protein is rhythmically expressed (Son et al., 2008). To explore whether adrenal clock proteins are driving rhythms in StAR expression, the investigators used a cell line in which a defective mutation of CLOCK (CLOCKΔ19) was used to disrupt the molecular clock. This manipulation caused decreased StAR transcription. On the other hand, overexpression of CLOCK and BMAL1 in the same cell line increased the level of StAR protein. In contrast to StAR, daily P450scc concentration is constant. In another study, the level of cholesterol associated with P450scc was also not changed over time of day (Brownie et al., 1979). These results show that in the adrenal gland StAR rather than P450scc is a key molecule in the circadian regulation of the common steroidogenic pathway.

The expression of StAR and CYP11A1 are also influenced by gonadotropins and adrenocorticotropins. In the testes, hCG induces a series of clock gene expression in rat Leydig cells, including Per1, Rorb Rev-erba, Rev-erbb, and BMAL1, and also increased StAR mRNA level (Baburski et al., 2019). The same study also showed that low levels of circulating LH and T in male rats altered expression levels of clock genes bmal1, Per2, Cry1, Cry2, Rorα/β, Rev-erbα/β, decreases the level of StAR and CYP11A1 and eliminates rhythmicity in CYP11A1. In rat granulosa cells, LH entrains the oscillation of StAR and CYP11A1 (Chen et al., 2013), and FSH stimulates oscillation of StAR (Chen et al., 2012). In the adrenal cortex, ACTH is a strong phase-setting signal (Yoder et al., 2014) and ACTH pulses induce StAR and CYP11A1 in rat adrenal glands (Spiga et al., 2011).

The expression of StAR is also under the diurnal regulation of melatonin in both gonads and adrenal glands [testes (Frungieri et al., 2017), ovaries (Tamura et al., 2009; Rai and Gosh, 2021) and adrenal glands (Minnetti et al., 2020)].

In the testes, in addition to StAR and CYP11A1 of the common synthetic pathway, 3β-HSD, CYP17A1 and 17β-HSD (encoding 3βHSD2, P450c17 and 17βHSD respectively) are also ccgs. In mouse and goat cultured Leydig cells, dexamethasone entrained the circadian variation of 3β-HSD, CYP17A and 17β-HSD (Chen et al., 2017; Xiao et al., 2021). bmal1 knockout in goat Leydig cells reduced the expression of 3β-HSD whereas bmal1 overexpression increased 17β-HSD mRNA level. Global knockout of bmal1 reduced the expression of 3β-HSD, CYP17A and 17β-HSD in mice (Alvarez et al., 2008). Some testicular steroidogenic genes are also under the regulation of upstream hormonal signals. The de novo synthesis of P450c17 in mouse Leydig cells requires cyclic adenosine monophosphate (cAMP) stimulation, which can be induced by the gonadotropins (Payne and Youngblood, 1995). 3βHSD2 has a relatively high basal level without cAMP. However, to reach its maximum expression, it requires cAMP induction. The expression of CYP17A1 oscillates in the normal rat Leydig cells but its rhythmicity is lost in Leydig cells of rats with insufficient LH and T (Baburski et al., 2019).

Melatonin has an inhibitory effect on testicular steroidogenesis. In the hamster Leydig cells, 3β-HSD and 17β-HSD expression was reduced by this hormone (reviewed by Frungieri et al., 2017). Pinealectomy started the oscillation of 3β-HSD in the rat Leydig cells and this change can be reversed by melatonin supplement (Baburski et al., 2015).

In the ovary, CYP19A1 (encodes P450 aromatase) expression has a diurnal rhythm in rat and mouse granulosa cells (Chen et al., 2013; Chu et al., 2019). Silencing bmal1 in granulosa cells significantly suppressed the expression of CYP19A1. 3β-HSD has large variations over 24 h but is not rhythmic in rat granulosa cells (Chen et al., 2013). Bmal1 siRNA in these cells decreased the level of 3β-HSD.

The variations of LH and FSH across the reproductive cycle influence the homeostasis among steroidogenic enzymes in the ovary (Niswender et al., 2000). In the transition from follicles to corpora lutea, the expression of P-producing 3β-HSD is increased, and the estrogen-producing CYP19A1 and androgen-producing CYP17A1 are decreased. In the rabbit ovary, the activity of promotor on CYP19 responsive to cAMP in the luteal cells decreased by 50% compared to granulosa cells (Andrieu et al., 2009). Human chorionic gonadotropin is stimulatory to the expression of 3β-HSD and CYP19A1 in granulosa cells. However, unlike StAR, the expression pattern of these two genes is not oscillatory in the human granulosa cells (Chen et al., 2017). Low dosage of LH can stimulate CYP17A1 expression in bovine thecal cells in vitro only when they are not pretreated with a high dose of LH. This is a simulation of the pre-ovulatory milieu (Murayama et al., 2012). When they are pretreated with a high dose of LH, low dose of LH no longer stimulates CYP17A1, a condition that simulates the post-ovulatory environment of theca cells.

Melatonin also regulates the mRNA level of the enzymes important in ovarian androgen and estrogen synthesis. In porcine antral follicles, melatonin inhibited expression of CYP17A1 but not CYP19A1 (reviewed by Tamura et al., 2009; Rai and Gosh, 2021). Pinealectomy increased CYP17A1 in rat theca interna and interstitial cells and CYP19A1 in granulosa cells whereas chronic administration of melatonin decreased CYP17A1 (Maganhin et al., 2014; Lima et al., 2015).

Cues related to feeding are thought to be among the most potent signals for peripheral clocks. Important in the present context, glucocorticoids are thought to play a key role in synchronization of the rest of the body to photoperiod and food availability (Balsalobre et al., 2000). For example, restricting food access to the light phase in nocturnal rodents reverses the timing of peak and trough expression of the clock genes per1, per2, and bmal1 in the adrenal gland (Girotti et al., 2009). Restricting feeding to daytime also reverses the diurnal peak and trough of StAR in the adrenal glands (Girotti et al., 2009; Chung et al., 2017). Interestingly, under these conditions CORT secretion has a bimodal pattern with one peak related to the adrenal clock and the other driven by the SCN, suggesting that the food induced peak is independent of the SCN and that the actions of the central clock on the adrenal glands remain undisturbed. In addition to the effects of time of eating, specific food components also influence the expression of glandular clock genes, steroidogenic molecules, and the concentration of circulating steroids.

As for the food components, in mouse testes, a high fat diet significantly reduced the level of bmal1, clock, per2, cry2 at the time of lights on but not at the time of lights off. Also, this diet initiated a difference between the light on and light off time in expression of SF-1, StAR, CYP11A1, 3β-HSD3, 17β-HSD17 and in plasma T, while no difference was seen in ad libitum fed controls (Wang et al., 2018). In another study, a high fat diet abolished the bimodal secretion pattern of T shown in control rats fed a normal diet, and the average level of plasma T was suppressed (Cano et al., 2008). In this experiment, the high fat diet blunted the daily variations of CORT but increased its mean value by 61%.

In female mice, a high fat diet significantly decreased bmal1 expression in the ovary near the end of dark phase (Yokoyama et al., 2020). StAR and CYP11A1 were downregulated at the beginning of light phase and CYP11B1 was downregulated at the beginning of dark phase. These changes eventually led to increased daytime levels of caecal CORT. In comparison, in the same study a high fat, high salt diet group had decreased expression of cry2. There was a trend toward reduction of StAR by this diet, and this was accompanied by a significant, bimodal pattern of plasma CORT. In a study of developmental effects, a high fat diet impacted cellular clocks in hormone production by steroidogenic cells. Thus, both maternal and/or post-weaning high fat diet in pregnant rats increased per1, per2 and per3 mRNA expression in the ovary of the female offspring, and elevated plasma E2 (Lin et al., 2017).

An important consideration here is that while there are numerous examples demonstrating powerful effects of feeding in modulating peripheral clocks, some studies indicate that the SCN is resistant to these cues (Balsalobre et al., 2000; Hara et al., 2001; Wakamatsu et al., 2001). But this generalization is not universally true. Feeding-related cues also act on the central clock. Thus, while rhythms of per1 and per2 mRNA in the SCN were not altered by restricted feeding, bmal1 expression was lower in the morning and increased in the evening compared to control animals – a significant feeding condition by time interaction (Girotti et al., 2009). Restricted feeding entrains PER2 oscillation in the SCN of rats whose rhythm were disrupted by constant light (Lamont et al., 2005). The genetic background of the subjects may be a factor underlying the effects of feeding. Thus, the oscillation of per1, per2 and bmal1 in the SCN can be entrained by restricted feeding in CS mice but not C57BL/6J mice under DD (Abe et al., 2007). In another study suggesting strain differences in SCN sensitivity to food cues, four lines of house mice were examined (Castillo et al., 2004). After long term exposure to scheduled feeding under DD, PER2 oscillation in the SCN was synchronized to food anticipatory activity in 100% of the mice in one line, and only some of the mice in the other three lines. In summary, decoupling time of eating and activity and altering diet components produces phase-shifting in glandular clocks and steroidogenesis. Further exploration of work on how eating and what is eaten effectively change diurnal rhythms can be found in Lewis et al. (2020).

Stressors are powerful factors regulating diurnal variations in steroidogenesis. In humans, a combination of stressors including strenuous exercise, sleep and energy deprivation for 5 days leads to a significant decrease in the mean values of plasma T, Δ4, DHEAS, progesterone and CORT, and blunts their diurnal rhythms (Opstad, 1994). However, the effect of stress on peripheral clocks, on steroidogenesis, and on the SCN differs depending on the type and duration of the stressor, and on the time of day of stress administration. Acute stressors including those lasting minutes to hours and those lasting several days, modulate the expression of peripheral clocks and steroidogenic molecules. In contrast, chronic stressors lasting weeks to months sometimes lead to habituation (reviewed in Koch et al., 2017; Tahara et al., 2017). For example, acute stress has a very rapid impact on adrenal steroidogenesis. After 1 h of restraint, increased StAR mRNA was detected in rat zona fasciculata, accompanied by increased CORT secretion (Liu et al., 2013). Restraint stress applied for 15 min advanced adrenal PER2::LUC oscillation by 2 h (Engeland et al., 2016). In another study, 2 h restraint starting during the mid-day didn’t shift adrenal PER2::LUC expression, but repeating the regimen for 7 days did do so (Stagl et al., 2018). Finally, 14 days of subordination stress also led to advanced PER2::LUC in the adrenal glands (Razzoli et al., 2014).

Acute stress also disrupts the cellular clock and steroidogenesis in gonads and other tissues. For example, acute immobilization stress increases bmal1, per1/2 expression, thereby shifting the peak of these clock mRNA expression toward the end of the stress; this stressor also decreases Rev-erbα; at the same time, it blunts the rhythmicity of Rev-erbβ and cry1 (Medar et al., 2021). In fact, a variety of steroidogenic genes in the Leydig cells, including StAR, CYP11A1, 3β-HSD1/2, CYP17A are decreased by this stress regimen. All these changes lead to decreased mesor and blunted rhythm of T, revealing the powerful effect of short-term stress in regulating daily steroid production. For the ovary, there is much research on the impact of stress on ovarian steroid biosynthesis and reproductive function (reviewed by Whirledge and Cidlowski, 2010; Toufexis et al., 2014) and on how circadian clock disruption influences the ovarian steroidogenesis, ovulation, fertility and fecundity (reviewed by Sellix, 2015). To our knowledge however, there is no data available on how the stressors modulate the ovarian clock thereby impacting ovarian steroidogenesis.

The effects of chronic stressors differ from those of short-term stress. CORT secretion is significantly reduced under chronic stress compared to the shorter duration regimens (Tahara et al., 2015). Although potent in reshaping the peripheral clocks and steroid synthesis, it is generally reported that stress does not appear to influence the rhythm of SCN PER::LUC (Takahashi et al., 2001, 2013; Razzoli et al., 2014; Tahara et al., 2015). However, this is not a universal finding. Thus, 19 days of social defeat stress in the dark phase on mice leads to decreased PER2 positive neurons in the SCN, as well as resulting in decreased per2 and cry1 in the adrenal glands (Bartlang et al., 2014). Also, 4 weeks of unpredictable stress (a combination of diverse stressors, including restraint, tilt, forced swimming etc.) caused a significant decrease in the peak value of PER2 expression in the SCN about 3 h before lights are off (Jiang et al., 2011).

Like food and stress, exercise also functions as a powerful non-photic synchronizer, both in the periphery and in the SCN. In the periphery, both voluntary and forced exercise affect the clocks and steroidogenesis and increases plasma CORT (Pieper et al., 1995; Kanaley et al., 2001). In a voluntary exercise paradigm in mice that had free access to a running wheel, the peak of PER2::LUC was delayed in liver and adrenal glands compared to animals without access to a wheel (Schroeder et al., 2012). The same study showed that scheduled exercise, implemented by limiting access to a wheel to either early or late night (active) phase delayed liver PER2::LUC oscillation, whereas the adrenal PER2::LUC was delayed only by the early night access. In contrast, forced exercise in the inactive (light) phase significantly advanced PER2::LUC rhythm and per2 and bmal1 mRNA expression in lung, liver and adrenal gland, and also advanced CORT secretion (Wolff and Esser, 2012; Sasaki et al., 2016). Regular voluntary exercise over weeks steadily increased the average level of CORT (Otawa et al., 2007). This effect was probably caused by increased peak value of StAR expression in the adrenal glands.

In the SCN, in contrast to the periphery, voluntary exercise has a stronger effect than forced exercise. Spontaneous activity suppressed SCN firing rate and increased the amplitude of the electrical activity, whereas forced behavior increased the firing rate (van Oosterhout et al., 2012). In a voluntary running paradigm, limiting the time of access to a wheel influences SCN responses. In mice held in an LD cycle, limiting access to a running wheel in the early night led to reduced PER2::LUC amplitude in the SCN (Schroeder et al., 2012). In contrast, limiting access to a running wheel during the light phase did not affect PER2::LUC oscillation in the SCN (Wolff and Esser, 2012). Manipulating access to wheel running also entrains locomotor rhythms in VIP and/or VIP receptor knockout mice. In these genetically modified animals, scheduled exercise under DD increased the amplitude of PER2::LUC oscillation in the ventral SCN, an effect opposite to that seen in the wild type animals (Power et al., 2010; Hughes et al., 2021). These results indicate the potential of voluntary exercise in shaping the SCN clock is phase dependent.

Forced exercise however, produced by confining subjects to a treadmill, did not affect the SCN (Wolff and Esser, 2012; Sasaki et al., 2016). The reason of the differential effects of voluntary and forced exercise on the SCN may be that forced exercise is a psychological stressor, pointing to different mechanisms underlying the entraining potential of the two types of exercise. Taken together, the examples show that non-photic entraining signals including feeding, stress and exercise are potent in shaping rhythmicity in peripheral clocks and steroid synthesis. Steroid hormones, especially glucocorticoids, modulated by non-photic cues relay circadian signals to extra-SCN brain regions and peripheral tissues (review in Spencer et al., 2018). Given that feeding, chronic stress, and voluntary exercise can each promote steroid secretion and can also alter SCN cellular clocks under certain circumstances, the results provoke the question of what might be the mechanism whereby these afferent cues act on the SCN. As we discuss next, the possibility that steroid receptors in the SCN are a mediating mechanism merits examination.

Androgen receptors (ARs) have been identified in the SCN of many species, including humans, mice, rats, ferrets, goats, and baboons (Handa et al., 1988; Wu et al., 1995; Kashon et al., 1996; Fernández-Guasti et al., 2000; Wang et al., 2007; Iwahana et al., 2008; Maejima et al., 2009; Jahan et al., 2015), pointing to the potentially broad generality of receptor-mediated mechanisms. Gonadectomy reduces AR expression in mice (Karatsoreos et al., 2007), and systematically administrated T restores AR expression in a concentration dependent manner (Butler et al., 2012). ARs in mice are located in gastrin-releasing peptide containing neurons lying in the core region of the SCN (Figure 5A). They express FOS after a light pulse indicating that they are retinorecipient. Importantly, gonadectomy reduces the FOS response after a phase-shifting light pulse, whereas androgen replacement restores the FOS response to that seen in intact animals. The results highlight the fact that retinal input and the androgenic steroid act on the same neurons within the SCN (Karatsoreos and Silver, 2007).

There is a sex difference in AR expression in the SCN with more robust expression in men vs. women and in male vs. female mice (Fernández-Guasti et al., 2000; Wang et al., 2007). This difference is attributable to adult circulating T as the sex difference disappears following castration of male mice and is reinstated with T administration (Iwahana et al., 2008). In females, administration of T by implants of silastic capsules in gonadectomized mice produces a male-like patterns of SCN AR expression. Finally, in the four core genotype mouse model (gonadal males with either XX or XY chromosomes and gonadal females with either XX or XY chromosomes) gonadal males have more AR than gonadal females, indicating that the sex difference in AR expression in the SCN is independent of sex chromosomes (Kuljis et al., 2013).

The functional significance of ARs in the SCN has been demonstrated by direct intracranial implants of T. Following gonadectomy, the free-running period of locomotor activity is lengthened in male mice and testosterone treatment reverses these effects (Daan et al., 1975). Importantly, unilateral intracranial T implants in gonadectomized mice reinstate both intact-male-typical locomotor activity and ipsilateral SCN AR expression (Model et al., 2015). The results indicate that androgens acting directly within the SCN are sufficient to support male-typical circadian locomotor rhythms. The mechanism of how a change in AR expression results in altered behavior remains unexplored.

ERα and/or ERβ receptor have been detected in the SCN in many species, including humans, rats, mice, plains vizcacha, sheep (Shughrue et al., 1997a,b; Hileman et al., 1999; Su et al., 2001; Kruijver and Swaab, 2002; Vida et al., 2008; Cao et al., 2014; Inserra et al., 2017). The expression of both ERα and ERβ has been delineated in mice (Figure 5B; Vida et al., 2008). In cultured SCN cells and astrocytes of neonatal rats ERα and ERβ are colocalized (Su et al., 2001). In young and adult, but not old rats, ERβ mRNA in the SCN has a diurnal rhythm, peaking around 03:00 h in animals held in a 14:10 LD cycle with lights on at 04:00 h (Wilson et al., 2002). As in the case of T, circulating estrogens affect receptor expression. A single injection of β-estradiol-3benzoate reduced ER-β expression in SCN of gonadectomized male and female mice, though ER-α was not affected by estrogen administration (Vida et al., 2008; Figure 5C). In both young and old OVX rats, ER-β mRNA expression was decreased in SCN after implantation of E2-bearing capsules (Shima et al., 2003; Yamaguchi and Yuri, 2014).

There are sex differences in SCN ER (Figure 5C). About 25.5% of cells in female mouse SCN and 16% in male show ERβ-ir neurons in the SCN respectively, approximately 2% of which co-express AVP (Vida et al., 2008). In humans, females have more ERα in the SCN than males (Figure 6 in Kruijver and Swaab, 2002). ERβ demonstrates five- to sixfold higher immunoreactivity than ERα. The quantification of ERα and ERβ also demonstrated a dimorphic pattern, with nuclear ERα significantly higher in women’s SCN.

Significant effects of estrogen on circadian period, measured by locomotor activity, are very well documented. For example, ovariectomy increases free-running period and E2 replacement shortens free-running period in hamsters, rats, and mice. The period-shortening effects of E2 can be mimicked by agonists of both the ERα and ERβ, with ERα agonists effective at lower doses (Joye and Evans, 2021).

Systemic application of E2 alters the expression of Cry1 and Cry2 mRNA (Nakamura et al., 2001), affects FOS-IR (Gibbs et al., 1990; Peterfi et al., 2004) and Per2 mRNA (Nakamura et al., 2005), alters neurotransmitter and transporter activity (Hery et al., 1984; Krajnak et al., 2003), and changes responsiveness to Ach and 5-HT (Kow and Pfaff, 1984). Given that these studies involve systemic hormone application, they do not probe the site of action and may be due to afferent input from extra-SCN target sites (Mong et al., 2011; Hatcher et al., 2020).

Direct bath application of E2 to the SCN increase the spontaneous firing frequency and depolarized cell membrane of SCN neurons in whole-cell patch clamp recordings (Fatehi and Fatehi-Hassanabad, 2008). These responses were thought to be receptor mediated as all these effects were abolished by pretreatment with the E2 receptor antagonist, ICI 182780. However, in vitro exposure of SCN tissue from OVX rats exposed to high physiological levels of E2 did not significantly alter the period of per1-luc expression (Murphy et al., 2013). E2 treatment didn’t affect PER2::LUC either (Sellix et al., 2006; Nakamura et al., 2008). These results suggest that either E2 does not act directly on the SCN or that E2 does act on the SCN but in a manner not detected by this measure.

In men and women, PRs are found in SCN neurons, in both cellular nuclei and cytoplasm, with no significant sex differences (Figures 4C,D in Kruijver and Swaab, 2002). Little or no PR is reported in the SCN in rats, guinea pigs, rabbits and monkeys (Blaustein et al., 1988; Bethea et al., 1992; Goldsmith et al., 1997; Auger and De Vries, 2002; Caba et al., 2003; Murphy et al., 2013). Very high concentrations of P and P+E2 lengthened the period of Per1-luc in SCN explants, while physiological levels of the steroids were not effective (Blaustein et al., 1988; Bethea et al., 1992; Goldsmith et al., 1997; Auger and De Vries, 2002; Caba et al., 2003; Murphy et al., 2013), leading the authors to conclude that at physiological levels ovarian steroids do not directly affect the timing of Per1-luc expression in SCN neurons despite the presence of E2 and P4 receptors in SCN pacemaker neurons.

The two corticoid receptors, glucocorticoid receptors (GR) and mineralocorticoid receptors (MR), are functionally differentiated by their responses to circulating glucocorticoids, with MR mainly participating in the baseline activity of glucocorticoids and GR regulating acute responses to stress (Kloet and Reul, 1985; Spencer and Deak, 2017). CORT receptors in the SCN change markedly in the postnatal period. In neonatal rats, the expression of GR is high during the first week after birth. GR immunoactivity gradually declines, and by PND20, only low density of GR-ir was detected in this brain area, comparable to the expression level of adult rats (Rosenfeld et al., 1988, 1993; Ahima and Harlan, 1990; Cintra et al., 1994b). There is a correlation during development in rats between the rise of circulating CORT and the decrease of glucocorticoid receptors in the SCN (Henning, 1978; Rosenfeld et al., 1988).

In a widely cited study, it was reported that there is no significantly detectable GR mRNA in the SCN of adult mice (Balsalobre et al., 2000). The authors postulated that an SCN devoid of corticosteroids receptors provides a mechanism to resist the impacts of sudden hormonal changes in circadian rhythms resulting from stress. However, several studies report the presence of GR protein or mRNA in adult (Rosenfeld et al., 1988; Cintra et al., 1994a,b; Morimoto et al., 1996; Figure 6). Additionally, real-time polymerase chain reaction analysis indicates that GR is present at low levels and is upregulated after ADX (Su et al., 2015). The available evidence is sparse; in some studies of GR and MR distribution in adult brain, the SCN is not mentioned (Aronsson et al., 1988; Sutanto et al., 1988; Sousa et al., 1989), while in other work, GR is studied only in young animals (Yi et al., 1994).

As in the case of systemically administered estrogen, many studies point to corticoid effects on the SCN though direct vs. indirect sites of action cannot be determined.

Glucocorticoids modulate light entrainment of a molecular clock in the SCN and also affect locomotor activities, indicating the stress hormone may participate in regulating the circadian timing system through the central clock (Sage et al., 2004; Wang et al., 2012). Systemic treatment with dexamethasone changed the level and/or rhythmicity of AVP and VIP mRNA expression in the SCN (Larsen et al., 1994). Consistent with this result, in post-mortem material from humans treated with glucocorticoids, the AVP mRNA was substantially lower than in patients who died of diseases where glucocorticoids was not administered (Liu et al., 2006), suggesting an effect of corticoid treatment. Systemic injections of dexamethasone upregulated Fos-ir and Jun-ir in the SCN (Briski et al., 1997). ADX upregulated cry2 mRNA expression in the rat SCN (Su et al., 2015) and abolished its oscillation. ADX also reduced the immunostaining of glial fibrillary acidic protein in the ventral SCN and CORT supplement reversed this effect (Maurel et al., 2000). Dexamethasone shifted the phase of PER2::LUC in fetal mouse SCN explants (Čečmanová et al., 2019) pointing to a direct effect of the steroid on the SCN in early development. In the adults, ADX had no impact on the phase of per1-Luc rhythms in the SCN (Wang et al., 2012). Exogenous administration of CORT induced higher AVP expression in the SCN only during the early morning indicating that stress hormones may feedback the SCN in a narrow time window (Larsen et al., 1994).

Low density immunostaining of MR (Ahima et al., 1991) and its mRNA expression (Su et al., 2015) has been detected in the adult rat SCN. MR mRNA level does not vary across day. Similar to effects on GR, ADX appears to elevate MR expression in the rat SCN. However, to our knowledge, there is no understanding of the function on the MR in the SCN.

The evidence for the existence of steroid receptors in the SCN is robust. The evidence clearly supports the conclusion that there are receptors of androgens estrogens and corticoids in the SCN. There is also evidence that circulating steroids have direct effects on SCN function. The hypothesis that emerges from this evidence is that steroids act on their cognate receptors in the brain clock and alter its sensitivity to afferent input, its oscillation and/or its output. Surprisingly little work has been done to explore mechanisms of action and function of SCN steroid receptors. Nevertheless, the foregoing review suggests that SCN steroid receptors have functions. Behavioral and physiological factors related to feeding, sexercise and stress each affect hormone secretion. Though few in number, there are studies indicating that each of these factors can alter SCN activity, either by altering sensitivity to light, or changing the timing or expression level of specific genes or proteins.

Consider the possibility that hormonal inputs to the SCN are specialized, and that they share some but not all properties with afferent photic input. The SCN clock responds to regularly recurring photic signals at specific times of the day. If the SCN’s responses to CORT have similar properties to those of SCN responses to light, then we must anticipate that the GR mediated responses will occur to regularly recurrent signals, and that the efficacy of those signals will be time of day dependent. Additionally, in the adult, the retinorecipient region of the SCN encompasses many more neurons than those bearing GR. As a consequence, it may require a substantial and recurrent hormonal cue for the few GR neurons to effectively alter the overall oscillation of the SCN. Importantly the time course of changes in receptor expression can be very rapid in response to a hormone. For example, systemically administered T can change receptor expression within 15 min (Freeman et al., 1995).

Cortisol or corticosterone is an especially attractive and most ignored candidate as an important SCN acting signal. This is the least noisy of the rhythms in hormones, a desirable feature in a timing mechanism. Plasma corticoid levels have a reliable high amplitude daily rhythm with a distinct peak with a 20-fold change as a function of time of day (Spencer and Deak, 2017). They provide a clear time stamp as they have a relatively narrow peak at the onset of the active phase in both diurnal and nocturnal species. Stress, feeding and exercise-mediated alterations of the daily pattern are readily discerned on this steady baseline. CORT blood levels rise within several minutes after stressor onset, typically reach peak levels within 30 min, and can return to pre-stress levels within 60 min after discrete stressor termination. These transients rest atop a stable baseline.

Corticoids acting on GR are known to be effective in entraining clock gene expression in peripheral and extra-SCN brain tissue. Interestingly while GR are thought to be ubiquitous throughout the body, they are considered virtually absent in the adult SCN (Balsalobre et al., 2000). In fact, there are GR receptors in the adult SCN, though they are sparsely distributed (Figure 6).

Given the sparse distribution and small numbers of GRs, these neurons are not likely to produce a response on the oscillator population as a whole, unless the cue is robust and repeats. There is a lack of experiments testing the possibility that the SCN responds to regularly recurring hormonal cues which might explain the function of these receptors in the SCN. One problem is that there are many potential responses that might be relevant. For example, a steroid might act by increasing or decreasing the amplitude of rhythms by changing the period of oscillations, altering sensitivity to phase shifts, lowering the peak or raising the trough of oscillations, changing the signal to noise ratio, altering phase relationships among oscillators, responding to phase-setting signals, etc. (reviewed in Silver and Balsam, 2010). And of course, it may be that one clock gene or protein is affected but another is not. For example, in studies of effects of daylength on the SCN, it appears that Per1 and Per2 play differential roles in photoperiodic responses (Yoshikawa et al., 2017).