Proximate mechanisms driving circadian control of neuroendocrine function: Lessons from the young and old

Abstract

Circadian rhythms impact a variety of behavioral and physiological functions contributing to longevity and successful reproduction. In their natural environments, individuals of a species are faced with a multitude of challenges and the coordination of internal processes and behavior with external pressures has been hypothesized to be an important target of natural selection. Several lines of evidence from cyanobacteria, Drosophila, and plants provide strong support for an important role of the circadian clock in survival and reproductive success. Similarly in mammals, disruptions in circadian function markedly impact reproduction and lifespan. The present review discusses research outlining the proximate and ultimate mechanisms responsible for the central and peripheral control of the reproductive axis. Because precise temporal coordination of the endocrine system is particularly crucial for reproduction by females, the present overview focuses on the role of circadian timing in this sex.

Introduction

In order to survive and successfully reproduce, animals in their natural environments must satisfy a variety of competing behavioral and physiological demands on a daily basis. Because selection pressures vary over the course of the day, and all requirements cannot be filled concomitantly, the most successful individuals of a species are those whose behavioral and physiological functions are organized temporally. As a result, organisms have evolved in ways that restrict behavioral activities such as foraging, defending territory, seeking and competing for mates, and sleeping, each to their own specific temporal niche. Not surprisingly, the internal central and peripheral physiological processes subserving these behaviors exhibit the same systematic orchestration. The harmony maintained among organismal events and the environment is coordinated by a clock in the brain, located in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus (Fig. 1). This overview examines the organization of the circadian system and its role in driving appropriate timing of endocrine-mediated imperatives. Via the cerebrospinal fluid (CSF) and the bloodstream, hormones are in a position to have widespread influence over the central nervous system (CNS) and periphery. Presumably, most endocrine mediators acting on the periphery have also been co-opted throughout evolutionary time to drive behaviors associated with systemic alterations. For example, the hormones responsible for ovulation (i.e., estrogen and progesterone) commonly trigger sexual motivation and associated species-specific behavior across taxa. As a result, the timing of endocrine function represents an ideal model system in which to explore the proximate and ultimate mechanisms underlying circadian organization.

Fig. 1.

The mammalian circadian clock is located in the suprachiasmatic nucleus (SCN) of the anterior hypothalamus. The SCN is situated at the base of the brain directly above the optic chiasm (oc) and surrounding the third ventricle (3V). The image depicts a low-power photomicrograph of the SCN of a Syrian hamster (Mesocricetus auratus) stained immunohistochemically for FOS protein as previously described (Gibson et al., 2008). Arrows point to the lateral borders of the SCN.

Evolution and adaptive significance of the circadian system

The pervasiveness of circadian clocks among prokaryotic and eukaryotic organisms living longer than a day has been used to argue for the adaptive significance of circadian function, although this hypothesis went largely untested until the late 1990s (Ouyang et al., 1998; Sharma, 2003; Yan et al., 2008). Although these studies are still in their infancy, several lines of evidence suggest a role for the circadian system in reproductive fitness. Interested readers are directed to several excellent reviews on this topic (Sharma, 2003; Johnson, 2005). One early study examined survivorship among free-living chipmunks in the Allegheny Mountains following destruction of the circadian clock versus a sham procedure. Loss of circadian function led to nighttime restlessness and increased susceptibility to terrestrial predators, suggesting an important advantage of fitness in avoiding predators (DeCoursey et al., 2000). These data must be interpreted cautiously as such naturalistic studies do not afford the control of variables possible in the laboratory. Although a correlation exists between clock disruption and nighttime restlessness, it is not possible to determine whether or not nighttime activity resulted in predation. Likewise, because the surgeries took place in the laboratory, and animals were released into their natural habitats after recovery, it is possible that re-establishment of territory contributed to mortality. Sham surgeries were conducted to control for this, but it is possible that the greater destruction of the brain in lesioned animals made re-acclimation to the environment more difficult. Although this study did not uncover increased mortality resulting from “natural” causes in arrhythmic animals, studies across species, including Drosophila melanogaster (Pittendrigh and Minis, 1972; Klarsfeld and Rouyer, 1998) and Syrian hamsters (Mesocricetus auratus) (Hurd and Ralph, 1998) demonstrated that adult longevity is increased when animals are housed in environments in which the period of the light:dark cycle matches the endogenous period of an individual's clock. Presumably, a circadian clock allows organisms to anticipate daily environmental change and synchronize internal processes accordingly. As more findings emerge in both laboratory and natural settings, a clearer role of the circadian clock in effective adaptation can be more firmly established.

In addition to alterations in lifespan, a functioning circadian clock can contribute to an organism's fitness by affecting reproductive output. In studies of Drosophila with mutations of core clock genes controlling circadian function at the cellular level, those animals with dysfunctional clocks exhibited a 40% reduction in the production of progeny relative to flies with functional clocks (Beaver et al., 2002). This reduction in reproductive fitness was the result of reduced output of eggs and an increase in the number of unfertilized eggs. These latter deficits were associated with abnormal reproductive function of males, with the authors noting reduced production and transport of sperm, presumably due to loss of clock gene function in the testes and seminal vesicles. Additionally, when clock-mutant males were mated with wild-type (WT) females, reductions in reproductive capacity remained, further indicating a role for cellular clocks in the fertility of males. Importantly, when clock function was rescued in male flies, reproductive abnormalities were abolished (Beaver et al., 2002). Analogous results were obtained in studies of the pitcher-plant mosquito, Wyeomyia smithii, in which incongruence between the free-running period of these insects and the light:dark cycle resulted in reductions in the fecundity of females (mean number of eggs in which eclosion was observed) (Emerson et al., 2008).

The adaptive significance of circadian function has perhaps been best demonstrated using asexual strains of cyanobacteria, Synechococcus, which differ in free-running periods (Ouyang et al., 1998; Johnson and Golden, 1999; Woelfle et al., 2004). In these studies, strains of cyanobacteria with a circadian period closest to the light:dark cycle of the environment have greater reproductive success when examined in studies of competition (Ouyang et al., 1998). Strains whose clocks were disrupted were defeated by strains with a functional clock, but only when held in a light:dark cycle; any competitive advantage was lost under constant conditions (Woelfle et al., 2004). Similar results have been obtained in the plant, Arabidopsis thaliana, where plants with periods closest to the natural light:dark cycle contain more chlorophyll, fix more carbon, grow faster, and survive better (Green et al., 2002; Dodd et al., 2005a; Dodd et al., 2005b). Together, these findings indicate a selective advantage of having an internal circadian clock when living in a rhythmic environment, presumably by allowing organisms to anticipate environmental change and initiate appropriate functions in advance of the optimal time of day.

Mammalian reproduction as a model system

All of the studies investigating the evolutionary/adaptive significance of clock function to date have explored these questions by either using organisms whose clock is disrupted at the cellular level or whose master circadian clock has been destroyed. Our approach to investigating the significance of the circadian clock in reproductive success has been to examine the role of circadian mechanisms in the neural circuitry driving sexual motivation, ovulation, and maintenance of pregnancy in rodents. In most laboratory rodents (e.g., rats, Syrian hamsters, and mice), an intact circadian system is essential for females’ reproductive success (Kriegsfeld and Silver, 2006). Lesions of the SCN result in arrhythmicity in all behavioral and physiological parameters, including endocrine function (Moore and Eichler, 1972; Stephan and Zucker, 1972; Wiegand and Terasawa, 1982). With regard to endocrine function specifically, SCN lesions or mutation of “clock” genes driving rhythms at the cellular level result in marked deficits in ovulation and fecundity in most rodents (Nunez and Stephan, 1977; Wiegand and Terasawa, 1982; Miller et al., 2004). In mice, rats, and hamsters the circadian clock is essential for normal ovulation, whereas ovulation and sexual behavior in primates is under less stringent temporal control (Yamaji et al., 1971; Knobil, 1974).

Because ovulation and sexual behavior are tightly controlled by the circadian system in rodents in the laboratory, one can exploit this tractable model system to explore the importance of endogenous daily timing for reproductive success. Throughout the rodent ovulatory cycle, concentrations of sex steroids are controlled by a neuroendocrine cascade beginning with the secretion of hypothalamic gonadotropin-releasing hormone (GnRH) into the hypophyseal portal system. In turn, GnRH acts on the anterior pituitary to stimulate the synthesis and secretion of the gonadotropins, luteinizing hormone (LH) and follicle-stimulating hormone (FSH). LH and FSH then act on the gonads to regulate steroidogenesis and gametogensis, respectively. The activity of the reproductive axis is controlled through the action of negative feedback from sex steroids, with this mechanism maintaining LH at low concentrations throughout most of the ovulatory cycle. At the time of the LH surge, however, high concentrations of estradiol are required for the SCN to trigger ovulation (Legan and Karsch, 1975; Kriegsfeld and Silver, 2006). The mechanisms mediating both the positive and negative feedback of estradiol are complex and not fully understood, though evidence points to the importance of ERα receptors in mediating estradiol negative and positive feedback in regions upstream of GnRH, especially in the AVPV (Herbison et al., 1992; Dorling et al., 2003; Wintermantel et al., 2006a). Likewise, it is unclear whether a single system switches from negative to positive feedback, whether two independent systems (one positive and one negative) differentially dominate throughout the cycle, or a combination of the two mechanisms mediate the switch from negative to positive feedback, thereby allowing the LH surge. Below, we highlight the role of two hypothalamic RFamide (Arg-Phe-NH₂) peptides, kisspeptin and RFamide-related peptide (RFRP; also known as gonadotropin-inhibitory hormone or GnIH), in the temporal coordination of the activity of the HPG axis required for successful reproduction (Kriegsfeld, 2006; Kriegsfeld et al., 2006; Smith et al., 2006a; Adachi et al., 2007).

RFamide peptide control of the timing of the reproductive axis

Discovery and functions of kisspeptin and RFRP

Kisspeptin and RFRP have pronounced, but opposing, actions on the reproductive axis (Gottsch et al., 2004; Irwig et al., 2004; Navarro et al., 2005; Kriegsfeld, 2006; Johnson et al., 2007). Likewise, there is significant evidence that both systems are targeted by the SCN and are key regulators of ovulation (Gibson et al., 2008; Williams et al., 2008). Through coordinated regulation of these opposing peptidergic systems, the circadian clock is in a position to maintain the reproductive axis within optimal operating limits in males and allow for the LH surge in females.

Numerous studies across a variety of species indicate a striking stimulatory role for kisspeptin in the reproductive axis (de Roux et al., 2003; Funes et al., 2003; Seminara et al., 2003; Gottsch et al., 2004; Irwig et al., 2004; Navarro et al., 2004; Thompson et al., 2004; Shahab et al., 2005; Gottsch et al., 2006; Smith et al., 2006a; Smith, 2009). Early evidence for a stimulatory role of kisspeptin was revealed with the identification of GPR54, the orphan G-protein-coupled receptor for the kisspeptin ligand. In humans, a congenital mutation of the GPR54 gene is associated with hypogonadotropic hypogonadism (de Roux et al., 2003; Semple et al., 2005). In Gpr54 knockout (KO) mice, a similar phenotype to those patients with mutations in Gpr54 is observed (Funes et al., 2003), whereas kisspeptin administration to female rats advances vaginal opening and induces precocious puberty (Navarro et al., 2004). Kisspeptin mRNA and protein are localized to hypothalamic areas associated with reproductive function, including the anteroventral periventricular (AVPV) and arcuate (ARC) nuclei in mice, rats, and hamsters (Gottsch et al., 2004; Irwig et al., 2004; Kinoshita et al., 2005; Greives et al., 2007; Mason et al., 2007; Revel et al., 2008), the preoptic area (POA) and ARC in sheep (Franceschini et al., 2006; Smith et al., 2008a), and the infundibular nucleus in humans and monkeys (Rometo et al., 2007). Administration of kisspeptin stimulates GnRH activity and the secretion of gonadotropins across species (Dhillo et al., 2005; Messager et al., 2005; Navarro et al., 2005; Shahab et al., 2005; Dhillo et al., 2007). Kisspeptinergic neurons project to GnRH cells that express GPR54 (Irwig et al., 2004; Han et al., 2005; Kinoshita et al., 2005; Clarkson and Herbison, 2006; Smith et al., 2008a), suggesting a direct effect of kisspeptin on the GnRH system. GnRH antagonists block the stimulatory actions of kisspeptin on LH in vivo, suggesting that kisspeptin does not act at the level of the anterior pituitary (Gottsch et al., 2004; Greives et al., 2007; Irwig et al., 2004). However, it is conceivable that there are subtle effects of kisspeptin on secretion of LH by the pituitary as Kiss1 and Gpr54 have been localized to a subset of LHβ-immunoreactive (ir) cells (Richard et al., 2008) and kisspeptin can elicit LH secretion in primary pituitary cultures collected during the follicular phase of the ovulatory cycle (Smith et al., 2008b). In contrast to these positive results, other studies find no effect of kisspeptin on secretion of pituitary LH (Matsui et al., 2004). It is undeniable that kisspeptin, at least in part, modulates the reproductive axis by stimulating the release of GnRH, but determining whether or not kisspeptin also acts at the level of the pituitary requires further investigation.

In contrast to kisspeptin, RFRP has been shown to have rapid, pronounced inhibitory effects on the reproductive axis across species. RFRP was first identified in the avian brain and was named GnIH due to its suppressive actions on secretion of LH by the pituitary (Tsutsui et al., 2000; Satake et al., 2001; Bentley et al., 2003). In birds, GnIH inhibits synthesis (Ciccone et al., 2004) and secretion (Tsutsui et al., 2000; Osugi et al., 2004) of gonadotropin as well as gonadal development and maintenance (Ubuka et al., 2006). GnIH-ir cells project to GnRH neurons that express the receptor for this peptide (Bentley et al., 2003; Bentley et al., 2006; Ubuka et al., 2006; Ubuka et al., 2008), suggesting a direct effect on the GnRH system in addition to the pituitary. The mammalian ortholog is typically referred to as RFRP because there is equivocal evidence regarding its effects on the pituitary (see below). In rodents, RFRP is expressed principally in the dorsomedial hypothalamus (DMH), with direct projections to GnRH cells (Kriegsfeld et al., 2006; Johnson et al., 2007; Gibson et al., 2008). In vivo administration of RFRP markedly attenuates LH secretion and inhibits sexual behavior in rodents (Kriegsfeld et al., 2006; Johnson et al., 2007). In one recent study, application of RFRP to GnRH cells in brains slices from male and female mice decreased neural activity in a subset of cells (Ducret et al., 2009), further supporting a suppressive role for this peptide on GnRH. As suggested previously, it is unclear whether RFRP acts at the level of the pituitary in mammals. Suggestive evidence for this possibility comes from studies showing that the RFRP receptor (GPR147) is localized to rat and Syrian hamster pituitaries (Hinuma et al., 2000; Gibson et al., 2008), and RFRP-ir fibers have been reported to extend into the external layer of the median eminence in hamsters (Gibson et al., 2008). Additionally, recent studies of rats suggest that RFRP inhibits GnRH-elicited release of LH at the level of the pituitary (Murakami et al., 2008). In other studies using a different antiserum to RFRP, RFRP-ir was not observed in the median eminence of rat brain, and intraperitoneal injections of the retrograde tracer Fluorogold only labeled a small number of RFRP cell bodies (Rizwan et al., 2009), suggesting that most of these cells do not reach the hypophyseal portal system. However, intravenously administered RFRP rapidly (within 3 min) inhibited GnRH-induced release of LH (Rizwan et al., 2009) as also occurs in birds (Osugi et al., 2004). These data indicate the potential for RFRP to act on the release of pituitary gonadotropin. Whether or not endogenous RFRP acts on the pituitary in addition to GnRH cells in mammalian species requires further investigation to clarify whether discrepant findings represent interspecies differences or result from technical variation across studies.

Kisspeptin and RFRP in ovulatory control

The timing of the LH surge is tightly regulated by the circadian system in Syrian hamsters. In this species, the LH surge occurs 4 h prior to the onset of locomotor activity, with a cessation of the surge occurring 2 h later (Stetson, 1978). GnRH neurons exhibit robust increases in immediate-early gene expression one hour after the peak in the LH surge (Lee et al., 1990; Lee et al., 1992; Hoffman et al., 1993). Hamsters housed in a light:dark (LD) schedule exhibit one activity bout every 24 h. This activity is associated with symmetrical, bilateral activation of the SCN and GnRH system, which elicits an LH surge every 24 h in ovariectomized, estradiol-treated animals. When housed in constant light, some hamsters exhibit a splitting of behavior, with two daily activity bouts, each reflecting an anti-phase oscillation of the left and right sides of the bilateral SCN (de la Iglesia et al., 2000; Yan et al., 2005). Notably, behaviorally split animals exhibit two daily surges of LH (Swann and Turek, 1985) as well as asymmetrical activation of the GnRH system (de la Iglesia et al., 2000; de la Iglesia et al., 2003; Gibson et al., 2008). We examined the possibility that, in addition to driving the LH surge positively, the SCN concomitantly coordinates the removal of steroid-mediated RFRP inhibition of the gonadotropic axis to permit the surge (Fig. 2). The SCN forms close appositions with RFRP cells, suggesting the possibility for direct temporal control of RFRP activity (Gibson et al., 2008). During the time of the surge in LH, immediate-early gene expression is reduced in RFRP cells. Additionally, in split animals, activation of the RFRP system is asymmetrical. Importantly, this asymmetry is opposite to the state of the GnRH system (Gibson et al., 2008). Together, these findings suggest that the SCN acts to temporally balance the contribution of negative and positive inputs to the GnRH system, thereby permitting the LH surge and ultimately ovulation.

Fig. 2.

Lateralization of GnRH and RFRP Activational Patterns in ‘Split’ Hamsters is Associated with the LH Surge. (A and B) Actograms of wheel-running activity in estradiol-implanted, ovariectomized hamsters kept under constant light conditions (LL). (A) Non-split (NS) hamsters were sacrificed (*) 1 h or 13 h prior to the onset of the activity bout. (B) Split hamsters were sacrificed (*) 1 h prior to the onset of one of the two activity bouts. (C and D) Photomicrographs of FOS activation in SCN and GnRH cells (insets) of NS and split hamsters. (E and F) Mean (±SEM) percentage of FOS-ir GnRH cells in non-split hamsters sacrificed 1 h (NS1) or 13 h (before the surge; NS13) prior to their activity bout and split hamsters sacrificed 1 h prior to one of their activity bouts. (G and H) Photomicrographs of FOS-ir RFRP cells in the DMH of NS and split hamsters. (I and J) Mean (±SEM) percentage of FOS-ir RFRP cells in NS1, NS13 and split hamsters. Asterisk indicates significantly different from all other values, P < 0.05. Reprinted with permission from (Gibson et al., 2008). Copyright: The Endocrine Society.

In addition to the direct stimulatory actions of the circadian system on GnRH described previously, the SCN may stimulate the LH surge indirectly through kisspeptin (Fig. 3). Early evidence suggested an important role for the AVPV in the initiation of the LH surge, although the cell phenotype mediating these actions remained elusive (Le et al., 1997; Wiegand et al., 1978; Le et al., 1999; Wintermantel et al., 2006b). In the AVPV, Kiss1 mRNA is downregulated following ovariectomy and upregulated by estradiol treatment, suggesting a role for this peptide in the positive feedback effects of estradiol. The opposite pattern is seen in the ARC of these animals (Smith et al., 2005). A majority of kisspeptin-ir cells co-express ERα, the receptor subtype implicated in estrogen positive feedback (Franceschini et al., 2006; Wintermantel et al., 2006b), and exhibit increases in FOS-ir around the time of the LH surge (Smith et al., 2006b; Clarkson et al., 2008). Gpr54- and Kiss1-null mice do not exhibit LH surges or FOS induction in GnRH cells (Clarkson et al., 2008). This study contrasts with previous work suggesting that Gpr54 mutant mice could initiate an LH surge (Dungan et al., 2007). The conflicting evidence for the necessity of GPR54 in the generation of the preovulatory LH surge may be a result of the mutant mouse models used by each group. Dungan and colleagues employed a retroviral insertion, with no loss in the coding sequence of Gpr54. This mutation might result in residual gene function and may explain why a subset of animals maintain the ability to generate an LH surge. Conversely, the Gpr54 mouse line used by Clarkson and colleagues had a complete knockout of Gpr54 (Clarkson et al., 2008). Administration of kisspeptin upregulates the release of GnRH and LH in sheep (Messager et al., 2005), thus stimulating ovulation in anestrus ewes (Caraty et al., 2007). In addition to direct actions of kisspeptin on GnRH cells (Irwig et al., 2004; Han et al., 2005; Kinoshita et al., 2005; Clarkson and Herbison, 2006; Smith et al., 2008a), kisspeptin may stimulate the release of GnRH via direct actions on GnRH terminals in mice (d'Anglemont de Tassigny et al., 2008). Taken together, these studies suggest an important role for kisspeptin as an integration point for circadian and estrogenic signals and its role as an essential stimulatory element in generation of the LH surge.

Fig. 3.

Integration of RFRP and kisspeptin in control of the LH surge. A proposed model incorporating a role for the RFRP and kisspeptin systems in regulation of the ovulatory circuitry. Black lines depict projections from the circadian system to GnRH neurons and to neurons containing estrogen receptors (de la Iglesia et al., 1995) as well as to the RFRP system (Gibson et al., 2008) and the AVPV, including the kisspeptin system (Watson et al., 1995; Williams et al., 2008) reported in rats, mice, and hamsters. Kisspeptin cells in the AVPV are active at the time of the LH surge (Irwig et al., 2004, Kauffman et al., 2007). Neurons containing ER-α in the preoptic area and elsewhere are known to project to the SCN (de la Iglesia et al., 1999) and to the vicinity of GnRH neurons (Simonian et al., 1999) and may play a role in mediating the circadian signal to GnRH neurons directly and/or indirectly. While ER-responsive cells have not been definitively shown to project specifically to GnRH neurons, the emergence and sexual dimorphism of kisspeptin cells and fibers that project to GnRH cell bodies provide compelling evidence for the direct connection between these two neural phenotypes (Clarkson and Herbison, 2006). Future studies are necessary to determine whether neurons in the AVPV project to GnRH cells or adjacent interneurons in the POA. Connections between the RFRP and GnRH systems (Kriegsfeld et al., 2006; Gibson et al., 2008,) may indicate a putative role for RFRP in modulating the negative-feedback effects of estrogen (Kriegsfeld et al., 2006), with SCN communication allowing for removal of negative feedback on the reproductive axis during the time of the LH surge. Connections between kisspeptin and GnRH cells indicate a possible role for the kisspeptin system in mediation of the GnRH system (Irwig et al., 2004). Dashed lines indicate indirect connections between systems that have not yet been explored. According to this model, RFRP cells respond to estradiol with inhibition of the HPG axis during most of the estrous cycle. At the time of the LH surge, however, the SCN signals the RFRP system to “ignore” estradiol input and remove the influences of negative feedback on the HPG axis while simultaneously signaling activation of the kisspeptin system. This allows for activation of the GnRH system directly or indirectly via the circadian system and/or through activation of the kisspeptin system.

Aging as a tool for understanding reproductive control

Reproductive senescence in females is defined as the inability to produce viable oocytes. Historically, it was generally accepted that the transition to reproductive senescence resulted from a depletion of ovarian follicles and that changes in the CNS components of the reproductive axis resulted from reductions in ovarian estrogen (Mandl and Shelton, 1959; Nelson et al., 1987; Richardson et al., 1987; Hansen et al., 2008). Whereas this mechanism drives the transition to menopause in primates, results from numerous studies suggest that aging causes changes in the brain that precede deficits at the level of the ovary in rodents (van der Schoot, 1976; Cooper et al., 1980; Sopelak and Butcher, 1982; Wise, 1982b; Felicio et al., 1983; Lloyd et al., 1994; Lu et al., 1994; Nelson et al., 1995). Thus, it is likely that the transition to reproductive senescence results from dysregulation of the reproductive axis at the levels of both the brain and the ovaries, at least in rodents (Rubin, 2000). Importantly, because ovarian follicles are the primary source of estrogen, many non-reproductive functions are also compromised following the transition to reproductive senescence, including: the regulation both of cardiovascular function and bone metabolism as well as protection from neurodegenerative diseases such as Alzheimer's and Parkinson's disease (Downs and Wise, 2009). By unraveling the contributions of the brain in the transition to reproductive senescence, insight into the general process of aging of the brain can be gained, as reproductive aging occurs in advance of many other pathological changes that tend to confound gerontological studies (Wise, 1999).

Peripheral control of reproductive aging

As suggested above, reduction in the number of ovarian follicles is associated with reproductive aging in rats (Mandl and Shelton, 1959; LaPolt et al., 1998), mice (Nelson et al., 1987) and humans (Hansen et al., 2008), and is considered to be the hallmark of reproductive senescence. These peripheral changes in ovarian function can have a marked impact on the HPG axis. For example, when follicular reserves are experimentally reduced in young rats, reproductive deficiencies that are normally seen in aged animals emerge, including deficits in secretion of gonadotropin and abnormal estrous cycles (Meredith and Butcher, 1985). Conversely, slowing the loss of ovarian follicles through chronic treatment with progesterone (LaPolt et al., 1998) or through dietary restriction (Nelson et al., 1985) can delay age-related anomalies in estrous cycles. Moreover, experimental reduction in follicular reserves reduces the magnitude of the preovulatory LH surge, ovulation rate and estrous cyclicity in both young and middle-aged rats, further suggesting that with advancing age ovarian decline impacts hypothalamo–pituitary function (Anzalone et al., 1998). Importantly, animals with experimental reduction in follicular pools have unaltered estradiol profiles throughout their cycles, suggesting that the reduction in follicular pool-size affects the central neuroendocrine response independent of changes in gonadal steroids.

Role of ovarian oscillators in reproductive decline

The molecular machinery that drives circadian rhythms in cells of the SCN also operates relatively ubiquitously throughout the brain and periphery (Hardin, 1994; Tosini and Menaker, 1996; Sun et al., 1997; Balsalobre et al., 1998; Zylka et al., 1998; Yamazaki et al., 2000). The intracellular clockwork consists of interacting transcriptional and translational feedback loops that operate with a period of about 24 h (for review see Okamura, 2007). Briefly, the transcription of three Per genes (1–3) and two Cry genes (1 and 2) is driven by the dimerization of the bHLH/PAS containing CLOCK and BMAL1 transcription factors (Gekakis et al., 1998; Takahata et al., 1998; Shearman et al., 2000). To complete the cycle, PER and CRY proteins translocate into the nucleus as multimeric complexes to discontinue CLOCK::BMAL-mediated transcription (Shearman et al., 2000). A number of cytoplasmic enzymes phosphorylate the PER and CRY proteins and play an important role in maintaining the precise timing of feedback loops (Lowrey et al., 2000). The SCN synchronizes independent oscillators in peripheral tissues, with loss of synchrony among independent cellular oscillators in the absence of SCN input (Welsh et al., 2004). Thus, the SCN acts as a “master clock” that synchronizes molecular rhythms in the periphery that, in turn, regulate the timing of a vast array of physiological and behavioral rhythms (Pando et al., 2002; Schibler and Sassone-Corsi, 2002; Guo et al., 2006).

The ovaries have been shown to express daily rhythms in the clock genes Per1 and Per2 (Fahrenkrug et al., 2006; Karman and Tischkau, 2006). The integrity of these local clocks may be important for normal ovarian function; genes contributing to the initiation of ovarian steroid production are rhythmic in the human ovary (Bao et al., 2003; Foster et al., 2005). Ovarian clock genes may also have a functional role in the synthesis of progesterone. In domestic chickens, the largest and most mature preovulatory follicle expresses diurnal changes in Per2 and Per3, whereas smaller follicles do not exhibit changes in clock-gene expression (Nakao et al., 2007). These changes are concomitant with increases in StAR, a steroidogenic acute regulatory protein that increases synthesis of progesterone (Nakao et al., 2007). Importantly, StAR genes contain E-Box enhancers (Christenson and Strauss, 2001), DNA elements that are direct targets of the molecular clock. Although correlational, these studies suggest a potential role for local timing mechanisms in ovulatory control. Although previous work has confirmed the existence of rhythmically expressed clock genes in the ovarian cells of rats, the role of murine ovarian circadian clock genes is largely speculative (Fahrenkrug et al., 2006). Whether or not alterations in ovarian clock genes contribute to the maintenance of ovarian function and/or reproductive decline with age remains an open question.

Central control of reproductive aging

Whereas alterations at the level of the ovary play an integral role in reproductive decline associated with advancing age, a number of lines of evidence suggest that central mechanisms also contribute to these changes. The timing of neural signals that control the release of gonadotropin exhibit marked changes in middle-aged women and non-human animals, and these modifications may accelerate the loss of follicles with advancing age (Nass et al., 1984; Wise, 1993; Hall et al., 2000; Gore et al., 2004). Transplanting the ovaries of old rats into young, ovariectomized rats restores follicular development and ovulation, indicating that aged ovaries maintain ovulatory function in some cases (Krohn, 1955; Peng and Huang, 1972). Likewise in young rats, electrolytic lesions of the medial preoptic area (POA), where GnRH neurons reside, result in irregular estrous cycling akin to that seen in older rats (Clemens and Bennett, 1977). Conversely, stimulation of the medial POA in old, acyclic rats results in an enhanced LH response (Wuttke and Meites, 1973). Together, these studies suggest that the brain contributes to aging of the reproductive axis and represents an important consideration for further investigation.

One caveat to consider when generalizing from rodent models to humans is that changes at the level of the brain may not be equivalent. In postmenopausal women, for example, gonadotropin levels are high from loss of negative feedback, whereas aged acyclic rats have normal levels of LH (Lu et al., 1994; Wise et al., 1999). Furthermore, follicular loss beginning in middle-aged women eventually leads to the complete depletion of ovarian follicles. In aged, acyclic rats, however, remaining follicles have been reported (Wise, 1999). These findings suggest that the contribution of hypothalamic decline versus the exhaustion of ovarian follicles to reproductive senescence likely differs between rodents and humans. Indeed, estrogen and progesterone administration in postmenopausal women can still generate or inhibit LH and FSH surges, depending on the timing and level of gonadal steroid treatment, suggesting the maintenance of positive and negative feedback in postmenopausal women (Gill et al., 2002; Ottowitz et al., 2008). Despite these differences, rodent models can provide important insight into human menopause. In both rats and humans, for example, a rise in concentration of FSH remains a hallmark of impending reproductive decline (DePaolo, 1987; Klein et al., 1996), though a rise in human FSH has been primarily linked to a decrease in inhibin B, the main ovarian-derived negative feedback peptide in humans, through depletion in follicle numbers (Burger, 1993; Burger et al., 2000). The pattern of LH secretion also changes in both perimenopausal humans and in rats transitioning into acyclicity. The duration of LH pulses increases and the frequency decreases in both premenopausal women and middle-aged, regularly cycling rats (Scarbrough and Wise, 1990; Matt et al., 1998). The ability of estradiol to induce LH surges becomes attenuated in perimenopausal women (van Look et al., 1977) and middle-aged rats (Wise, 1984). The studies by van Look and colleagues and Matt and colleagues, however, take place in women that are well in to the transition to menopause and therefore have a marked decrease in follicle number, which alters levels of ovarian hormones such as inhibin B and may result in changes in cyclicity that precede any alterations in the brain's responsiveness to positive feedback. Therefore, while changes in the brain's ability to control the timing of the LH surge and respond to the negative and positive feedback of circulating estradiol contribute to the desynchrony of gonadotropin secretion in rats transitioning into senescence, the impact of changes in positive and negative feedback on the transition to menopause in humans remains controversial (Hall, 2007). This underscores the idea that the origin of the dysregulation of the brain's control of the HPG axis differs between rats and humans. In rats, changes in the hypothalamus contribute to reproductive senescence, while alterations in gonadotropin secretion in humans transitioning to menopause are preceded by the elimination of negative feedback originating from follicle depletion. Although the mechanisms differ, the resulting impact on general HPG axis activity affects both aged rodents and primates. Recent studies indicate that the increase in circulating gonadotropins may be a result of hypertrophy of kisspeptin neurons in the infundibular nucleus (Rance, 2009). These results are almost identical to rodent studies of arcuate kisspeptin neurons after ovariectomy, and may constitute a neural compensatory mechanism to ovarian failure (Rance, 2009). Further research is needed to fully elucidate the neural mechanisms underlying HPG activity during the transition to menopause. Thus, using rodent models of hypothalamic decline in reproductive aging can lend insight into human menopause, as reliable endocrine markers for predicting the full transition to anovulation (including final menses) in humans have yet to be specified (Burger et al., 2008).

One of the first indicators of reproductive decline in rodents involves an attenuated and delayed LH surge on the day of proestrus (Wise, 1982a). This decreased LH response occurs prior to any changes in pituitary responsiveness to GnRH or reductions in GnRH cell numbers (Wise, 1982a; Lloyd et al., 1994; Krajnak et al., 2001), although a decrease in neuronal activation is seen in GnRH neurons in middle-aged, regularly cycling rats (Lloyd et al., 1994). It is likely that the decreased and delayed LH surge seen in middle-aged animals stems from an alteration in the timed activation pattern of GnRH neurons (Lloyd et al., 1994; Rubin and King, 1994; Krajnak et al., 2001). This decline in GnRH activation may reflect alterations in the strength and timing of the afferent inputs to this system originating from the circadian clock (Krajnak et al., 1998). Additionally, significant alterations in SCN rhythmicity are observed with advancing age and may contribute to deficits in the timing of GnRH neuronal activation at the time of the LH surge (Wise et al., 1988). Furthermore, unlike the condition in young animals, rhythms in vasoactive intestinal polypeptide (VIP) mRNA are not apparent in middle-aged female rodents (Krajnak et al., 1998). Suppression of VIP in the SCN through the administration of antisense oligonucleotides also leads to an aging-like attenuation of cyclic secretion of LH (Harney et al., 1996). Together, these findings suggest that changes in the timing and amplitude of the preovulatory LH surge may result from age-related changes in the SCN and in its neurochemical output.

Role of central oscillators in reproductive decline

Perhaps the most noticeable alteration in behavior associated with advancing age is in the quality, quantity and phase of sleep (Dijk et al., 1999). Whether such age-related changes are the result of alterations at the level of the SCN or extra-SCN clocks (e.g., clocks in the reticular activating system) remains a topic of intense interest. Several lines of evidence suggest that age-related changes in the SCN (Sutin et al., 1993; Benloucif et al., 1997; Aujard et al., 2001) are responsible for the decline in behavioral and physiological rhythms seen in aged animals (Yamazaki et al., 2002), and that these changes may contribute to reductions in fertility. For example, the amplitude of electrical activity in SCN slices taken from aged rats is dampened relative to that seen in the SCN of young rats (Watanabe et al., 1995). Furthermore, the number of SCN neurons expressing arginine vasopressin (AVP) and vasoactive intestinal polypeptide (VIP), two key SCN neuropeptides, decreases in aged rats (Roozendaal et al., 1987; Chee et al., 1988). Because both of these neurochemicals have been implicated in the circadian control of ovulation (van der Beek et al., 1997; Kalsbeek and Buijs, 2002), the integrity of these neurons in the SCN may be critical for the maintenance of reproductive functioning with age.

Studies investigating age-related changes in the molecular components of the circadian clock have uncovered equivocal results. In rats, the rhythm of SCN Per1 and Per2 expression does not differ between young and old animals (Asai et al., 2001; Yamazaki et al., 2002). However, in mice, the day-night rhythm of SCN Per2 is blunted in aged animals (Weinert et al., 2001). Some findings suggest that reductions in clock-gene amplitude with advanced age may be due to changes in the synchrony of the internal clock with external time. The expression of Bmal1 in old hamsters, for example, is lower than in young hamsters during the subjective night, when Bmal1 expression is normally at its peak (Kolker et al., 2003). These subtle changes in Bmal1 may have marked consequences for circadian function, as Bmal1 is the only single clock-gene deletion that causes loss of rhythmic function (Bunger et al., 2000). Aging does not seem to affect other clock genes in the SCN, as Per1 and Per2 rhythms do not diminish in the aged SCN (Davidson et al., 2008) and mice with Clock mutations do not exhibit an increase in the effects of age on circadian rhythms over WT littermates (Kolker et al., 2004). Therefore, the reduction in the expression of Bmal1 in aged rats may principally account for the diminished molecular rhythms in the SCN.

As mentioned previously, the SCN is essential for ovulation in a number of rodent species and AVP may be an important output signal triggering the LH surge. Injections of AVP into estradiol-treated, SCN-lesioned rats induce an LH surge (Grace et al., 1999; Palm et al., 1999). Notably, rhythms in AVP expression are regulated by the core molecular clock through an E-BOX motif (Grace et al., 1999). Clock mutant mice have blunted AVP rhythms (Jin et al., 1999; Silver et al., 1999) that presumably impact ovulatory function negatively (Miller et al., 2006). Because AVP is regulated by the core molecular clockwork, age-related disruptions in SCN cellular function can clearly impact the output/timing of this peptide and, ultimately, ovulatory functioning. Interestingly, the diurnal expression of AVP mRNA within the SCN is maintained in aged animals (Krajnak et al., 1998), although it is possible that, with age, cells targeted by AVP, upstream of the GnRH system, become insensitive to this peptide. Neurons in the AVPV that play a critical role in the initiation of the LH surge (Le et al., 1997; Le et al., 1999; Terasawa et al., 1980) express V1a receptors (Ostrowski et al., 1994) and are synaptic targets of the SCN (Watson et al., 1995). Importantly, these cells are upregulated in the presence of estradiol in young rats but not middle-aged rats (Funabashi et al., 2000).

Timed coordination of RFamide peptides during aging

As described previously, the cellular activity of the inhibitory peptide, RFRP, appears to be orchestrated by the SCN in ways that remove negative input to the GnRH system during the LH surge (Gibson et al., 2008). Likewise, the kisspeptin system represents a potential key integration point for estrogenic and circadian signals necessary to stimulate the LH surge (Smith et al., 2006b; Williams et al., 2008). Because the precision in the coordination of these neuropeptides is important for ovulatory control in young animals, we examined whether changes in the expression or timing of RFRP and kisspeptin are associated with reproductive decline in middle-aged Syrian hamsters. Young (60–90 days of age) and middle-aged (12–13 months of age) animals were ovariectomized and implanted with Silastic capsules containing estradiol to remove potential effects of reduced gonadal steroids in middle-aged animals. Brains and serum samples were collected two, six, ten, and twelve hours after light onset (light:dark cycle = 14L:10D) to capture time points before, during, and after the LH surge. Brains were labeled immunofluorescently for RFRP and FOS or kisspeptin and FOS and serum samples were assayed for LH as previously described (Greives et al., 2007; Gibson et al., 2008).

Although the timing of the LH surge was unaffected in middle-aged females, the amplitude was significantly lower in middle-aged relative to young animals (P < 0.05; Fig. 4). In both young and middle-aged animals, the number of RFRP cells visualized with immunohistochemistry was greatest during the LH surge (ZT10; P < 0.05 in both cases) and the percentage of cells expressing FOS was lowest during the LH surge in both age groups (P < 0.05 in both cases; Fig. 5). In contrast, the number of RFRP cells labeled and the percentage of RFRP cells expressing FOS was unaffected by age (P > 0.05 in all cases; Fig. 5). It is unclear whether or not more pronounced changes in the RFRP system would have been observed in aged hamsters had the animals been left intact. However, as middle-aged females were cycling irregularly, or became anestrus, it was impossible to select appropriate time-points for comparisons without ovariectomy and replacement of estradiol.

Fig. 4.

The preovulatory LH surge is blunted in middle-aged hamsters. LH concentrations are shown relative to zeitgeber (light: dark cycle) time, with ZT0 = lights on and ZT14 = lights off. Concentrations of LH in the serum were measured in duplicate in a single RIA with reagents obtained from the National Institutes of Health (Bethesda, MD) as previously described. The antiserum was rLH-S-11, and the standard was rLH-RP3. The sensitivity was 0.01 ng/tube, and the intraassay coefficient of variation was 2.8% for the low pool and 8.4% for the high pool. The antisera were highly specific for the hormones measured, with low cross-reactivity with other hormones. *Significantly greater than all other groups, P < 0.05.

Fig. 5.

Number of, and percent of FOS-positive, RFRP-ir cells are associated with the timing of the LH surge. The mean (±SEM) number of RFRP-ir cells and percent (±SEM) of FOS-positive RFRP-ir cells do not differ in young and middle-aged animals at any time-point (P > 0.05 in all cases). Data are shown relative to zeitgeber (light: dark cycle) time, with ZT0 = lights on and ZT14 = lights off. RFRP cells and FOS were labeled immunohistochemically as previously described (Gibson et al., 2008). *Significantly greater (top) or less (bottom) than all other time-points in the same age group P < 0.05. **Significantly less than all other time-points from the same age group, excluding ZT6, P < 0.05.

Numbers of AVPV kisspeptin cells were reduced during the LH surge in young animals (P < 0.05) but not in middle-aged (P > 0.05) ones (Fig. 6). In both age groups, the percentage of cells expressing kisspeptin was increased at ZT10/ZT12 relative to time-points prior to the LH surge (P < 0.05 in all cases; Fig. 6). There were no age differences in the percentage of cells expressing FOS at any time-point (P > 0.05 in all cases; Fig. 6). As with RFRP, it is possible that changes in kisspeptin may have been observed had animals been examined under conditions of endogenous estrogen exposure, particularly given that the activity of kisspeptin is sensitive to estradiol (Smith et al., 2006b; Roa et al., 2008).

Fig. 6.

Number and percent of FOS-positive, kisspeptin-ir cells are associated with the timing of the LH surge. The mean (±SEM) number of kisspeptin-ir cells and percent (±SEM) of FOS-positive kisspeptin-ir cells do not differ in young and middle-aged animals at any time-point (P > 0.05 in all cases). Data are shown relative to zeitgeber (light: dark cycle) time, with ZT0 = lights on and ZT14 = lights off. Kisspeptin cells and FOS were labeled immunohistochemically as previously described (Williams et al., 2008; Greives et al., 2007). *Significantly less than all other groups, P < 0.05. **Significantly greater than values at ZT2 and ZT6, P < 0.05.

Given the relatively small differences in kisspeptin and RFRP expression in young and middle-aged animals, it is unlikely that alterations in these systems account entirely for the reduction in the amplitude of LH surge in older animals. Both in young and middle-aged animals, it appears that the circadian system is equally capable of removing the influence of RFRP inhibition at the time of the surge. In young animals, there is a reduction in kisspeptin cell numbers at the time of the surge, possibly due to increased transport and release in this age group relative to those in middle-aged animals. Likewise, we have recently shown that the GnRH system is differentially responsive to kisspeptin stimulation across the day, presumably due to local regulation by clock genes (Zhao and Kriegsfeld, 2009). Thus, it is also possible that middle-aged animals lose the ability to time the responsiveness of GnRH to kisspeptin to coincide with times of RFRP disinhibition and concomitant stimulation by estradiol.

In agreement with results from studies of the effects of estradiol on kisspeptin neurons in the arcuate of Syrian hamsters (Smith 2005; Franceschini 2006; Roa 2006), postmenopausal rhesus monkeys exhibit an increase in Kiss1 gene expression and cell size relative to reproductively competent animals. The authors speculated that this increase is due to decreased negative feedback control by ovarian steroids (Rometo et al., 2007; Kim et al., 2009). In humans, the number and size of kisspeptin mRNA neurons increases in postmenopausal women (Rometo et al., 2007). In primates, reproductive decline is primarily due to ovarian aging, whereas reproductive aging in rodents may be the result of alterations both in the brain and ovary (Yin and Gore, 2006). Results in Syrian hamsters indicate that kisspeptin expression is higher during the LH surge in middle-aged animals relative to young animals. Whether this increased expression pattern represents a reduction in release with continued mRNA turnover, or an increase in mRNA transcription/translation, requires further examination. However, because estrogen concentrations were clamped at equivalent values both in young and middle-aged animals, these changes are not the result of alterations in estrogen negative feedback as might occur in human/non-human primates. Instead, these findings suggest that kisspeptin regulation may change independently of decreasing ovarian steroids.

Conclusions and considerations

The control of reproduction by the circadian system represents an exciting opportunity for exploring the role of endogenous timing mechanisms in reproductive success. In females, the RFamide peptides, kisspeptin and RFRP, represent important intermediary systems in the circadian control of the preovulatory surge in LH. The SCN projects to both the kisspeptin and RFRP systems, and the activity of these neuronal populations is coordinated with the time of the LH surge. Presumably, activity of the kisspeptin system is upregulated at the time of the LH surge whereas that of the RFRP system is downregulated. In middle-aged animals, although significant suppression of the amplitude of the surge in LH is observed, there is little evidence that alterations in the RFRP system contribute to this deficit in Syrian hamsters. In contrast, changes in the positive drive by the kisspeptin system, independent of gonadal steroids, may impact reproductive aging. Additional exploration of the role of the circadian timing system in the neural circuits regulating the LH surge and ovulatory function is necessary to further understand how aging might impact timing mechanisms contributing to reproductive senescence.

Funding

National Institutes of Health Grant HD050470 (to L.J.K.) and NSF grant IOS-0641188 (G.E.B.).