Physiology and Endocrinology of the Ovarian Cycle in Macaques.

From the viewpoint of reproductive toxicology, female macaques provide excellent models for preclinical testing and safety assessment of female fertility. Currently, the predominant species in toxicity testing is the cynomolgus monkey ( Macaca fascicularis , synonymous with Macaca cynomolgus, Macaca irus , long-tailed macaque, or crab-eating macaque). Occasionally, the rhesus monkey ( Macaca mulatto ) and the (common) marmoset ( Callithrix jacchus ) are also used for reproductive toxicity evaluation. At Covance Laboratories GmbH, Muenster, Germany, over the past years, approximately 80% of toxicity studies have been conducted in the cynomolgus monkey model, approximately 15% in the marmoset model, and the remainder in the rhesus monkey model. The purpose of this article is (1) to briefly review the physiology and endocrinology of the primate ovarian cycle, (2) to provide practical guidance on how to assess and monitor ovarian cyclicity in toxicity studies, and (3) to describe new data on the effects of psychosocial and environmental enrichment on ovarian cyclicity in the context of toxicological studies. The latter is considered particularly relevant in light of the current changes in housing recommendations for nonhuman primates ( Council of Europe 2007 ). Special emphasis is given to the cynomolgus monkey and some comparative information is provided for other nonhuman primate species currently used in toxicology (i.e., rhesus monkey and marmoset). Animal experiments conducted at Covance Laboratories were performed in accordance with the German Animal Welfare Act. In terms of ovarian cycle characteristics and regulation, primates exhibit substantial differences from other mammalian species, in particular rodents or ruminants. Primates have a comparatively long life span of the corpus luteum—about two weeks or longer—irrespective of whether conception occurs. If a pregnancy is established, the corpus luteum has an extended duration of function and delayed luteal regression to permit implantation and the luteal-placental shift. Unlike in rodents, prolactin is not considered to play a decisive role during the luteal phase, and luteolysis does not involve a uterine signal. A wealth of studies is available to demonstrate that the ovarian cycle in various macaque species shows close similarities to that in women. The precise endocrine sequelae and control of the primate ovarian cycle were unravelled initially in rhesus monkeys ( Zeleznik and Pohl 2006 ), and insights from nonhuman primate research have yielded significant insights into women’s health ( Archer 2004 ; Cline et al. 2001 ; Kaplan 2004 ). Macaque models have been specifically useful for research on ovarian cycle control, endometriosis, anovulatory infertility, osteoporosis, stress-associated infertility, menopausal changes in physiology, and related metabolic disorders (diabetes, atherosclerosis). Overall, excellent reviews on the endocrinology and physiology of the primate ovarian cycle are available ( Gougeon 1996 ; Jabbour et al. 2006 ; Stouffer 2003 ; Zeleznik and Pohl 2006 ). A wealth of information and new insights into endocrine requirements and local control of the ovarian cycle were also gained from experimental studies in macaques using controlled ovarian (hyper)stimulation protocols and/or selective steroid/steroid receptor modulation ( Cline 2007 ; Macklon et al. 2006 ).

At approximately twenty weeks following birth, LH and FSH levels in circulation peak sharply followed by a decline toward prepubertal levels in macaques ( Mann et al. 2000 ; Plant 2006 ). This neonatal gonadotropin peak is reduced in females ovariectomized at 1 week of age compared with males orchidectomized at the same age. The onset of puberty is triggered by initiation of pulsatile GnRH release, with a corresponding increase in the tonic level of LH secretion, particularly at night ( Terasawa and Fernandez 2001 ; Wilson, Fisher, and Chikazawa 2004 ). Neither adrenalectomy nor pinealectomy alter the timing of the prepubertal/pubertal onset of gonadotropin release ( Plant, Ramaswamy, and Dipietro 2006 ). Activation of the G protein–coupled GPR54 receptor on GnRH neurons by metastin/ kisspeptin triggers GnRH release on onset of puberty ( Plant, Ramaswamy, and Dipietro 2006 ; Seminara et al. 2006 ). In the juvenile and prepubertal ovary, follicle development continues to the preantral and early antral follicle stage but does not lead to formation of preovulatory follicles. Instead, many primordial follicles are lost through atresia during these periods. Follicles in the prepubertal ovary are steroidogenically active as suggested by several-fold higher estradiol concentrations in the ovarian vein compared with circulating levels and by the fact that ovariectomy in prepubertal animals decreases circulating estradiol levels ( Winter et al. 1978 ). Menarche in female cynomolgus monkeys and rhesus monkeys occurs at an age of two to three years ( Honjo, Cho, and Terao 1984 ; Watanabe et al. 2006 ; Wilen and Naftolin 1976 ). The earliest onset of vaginal bleeding in our experience among cynomolgus monkeys was encountered in an approximately eighteen-month-old female. Female reproductive aging has been described for a variety of nonhuman primate species ( Nozaki, Mitsunaga, and Shimizu 1995 ). In that context, the rhesus monkey has been studied in particular detail. Overall, the processes and consequences of female reproductive aging display similarities between Old World monkeys and women (for a review, see Bellino and Wise 2003 ). This, for example, holds true for endocrine profiles, menstrual-cycle irregularities, fertility decline, dyslipidemia, body weight gain, and altered body composition. Hot flush–like phenomena were observed in rhesus monkeys ovariectomized for simulation of menopause ( Dierschke 1985 ). What appears different is the timing and duration of menopause and a more abrupt onset of cycle irregularities in monkeys ( Shideler et al. 2001 ) when compared with women. In rhesus monkeys, menopausal changes occur very late in life (e.g., third decade of life; Walker 1995 ; Shideler et al. 2001 ), and consequently, duration of menopause is shorter relative to life span in the rhesus monkey than in women. For the seasonal Japanese monkey ( Macaca fuscata ), menopause has been observed at approximately twenty-seven years of age ( Nozaki, Mitsunaga, and Shimizu 1995 ). Endocrinologically, menopausal monkeys have elevated gonadotropin levels, reduced estradiol levels, and unaltered or slightly reduced progesterone levels. Recent work using pushpull perfusion from stalk-median eminence in rhesus monkeys indicates that during menopause, GnRH secretion is increased ( Gore, Windsor-Engnell, and Terasawa 2004 ). Whether this is the consequence of reduced estradiol or progesterone feedback and/or inherent aging in GnRH-synthesizing hypothalamic neurons remains to be clarified. Histologically, ovarian senescence has recently been characterized in rhesus monkeys throughout a twenty-five-year period ( Nichols et al. 2005 ) and in aged cynomolgus monkeys ( Buse, Zöller, and van Esch, 2008 [this issue]). The proportion of primary follicles decreased from approximately 80% in one-year-old ovaries to approximately 55% in more than twenty-year-old ovaries in the rhesus monkey. Accordingly, total number of primary and antral follicles per ovarian section also decreased several-fold until the age of more than twenty years. In terms of reproduction, female monkeys lost fertility a couple of years prior to ovarian endocrine senescence, a feature qualitatively similar to women ( Nozaki, Mitsunaga, and Shimizu 1995 ). Unlike for the rhesus monkey, information on female reproductive aging in cynomolgus monkeys is limited. However, a more recent report identified naturally occurring menopause in cynomolgus monkeys ( Kavanagh, Williams, and Wagner 2005 ). In this study of sixteen animals, age estimates, based on dentition, were twenty-two to thirty-one years ( M = 29) for seven postmenopausal females and nine to twenty-one years ( M = 16) for nine premenopausal females. Postmenopausal animals had elevated FSH and reduced estradiol levels.

Menstrual cycles, cycle-related changes in hormone production, and the provision of a mature and fertilization-competent oocyte are governed by a complex but delicate interplay among hypothalamus, pituitary, and ovary. This brain-gonadal axis represents the core unit for the maintenance of the endocrine balance and fertility. Ovarian functions (i.e., production of steroid hormones and of ova) are entirely subject to regulation by endocrine factors derived from the brain. GnRH is secreted from the hypothalamus and stimulates the synthesis and release of the gonadotropic hormones, luteinizing hormone (LH), and follicle-stimulating hormone (FSH) from the pituitary gland. LH acts on ovarian theca cells and governs the production and secretion of androgens by these cells. Within the follicle, androgens act on the somatic granulosa cells and are converted by them into estradiol. FSH acts directly on the granulosa cells to stimulate follicular growth, and receptors for FSH have been identified only on this cell type. Estrogens exert a positive feedback effect on gonadotropin release prior to ovulation and provoke the ovulatory LH peak, whereas during the postovulatory luteal phase and in the early follicular phase, estrogens inhibit gonadotropin levels. During the luteal phase, LH stimulates progesterone production leading to further negative feedback effects on gonadotropin secretion and, along with estrogen, is believed to result in the luteal inhibition of early follicular development. Among primates, the decapeptide GnRH is predominantly synthesized in neurons of the mediobasal hypothalamus and the arcuate nucleus. Two forms of GnRH (GnRH-I and GnRH-II) have been described for primates that have receptors in the gonadotropic cells but are encoded by different genes ( Cheng and Leung 2005 ). GnRH is produced by successive cleavage stages from a longer precursor, called preproGnRH, transported along the axons to the median eminence and there released into portal blood. GnRH-I is involved in gonadotropin regulation. A second GnRH receptor gene was identified in marmosets (GnRH type II receptor), which is structurally and functionally distinct from the classical, type I receptor. The GnRH type II receptor, however, is not functional in the human and many other species, and its role is as yet unknown ( Millar 2005 ). Both GnRH-I and GnRH-II were shown to signal through the GnRH type I receptor. While GnRH-I regulates gonadotropins, GnRH-II appears to be a neuromodulator and stimulates sexual behavior. GnRH is released from the hypothalamus in a pulsatile manner and this pulsatility is indispensable for intact pituitary gonadotropin secretion and, subsequently, ovarian function. Experimental manipulation provoking increased (including continuous exposure) or decreased frequency of GnRH pulses reduces or abolishes gonadotropin secretion and thus disturbs the ovarian cycle. It has become clear over the past years that the kisspeptin/GPR54 system is pivotal for the regulation of GnRH secretion ( Figure 1 ). Kisspeptin-expressing neurons are located in the anteroventral periventricular nucleus, in the periventricular nucleus, in the anterodorsal preoptic nucleus, and in the arcuate nucleus (AN). Outside the nervous system, the KISS1 gene is expressed in placenta, testis, pancreas, liver, and intestine ( Popa, Clifton, and Steiner 2008 ). Kisspeptin is sensitive to steroid levels within the circulation and mediates the negative feedback regulation of gonadotropin secretion. In fact, although androgens, estrogens, and progesterone suppress gonadotropin secretion through androgen receptor–, estrogen receptor-α–, and progesterone receptor–dependent mechanisms, respectively, none of these sex steroids affects GnRH secretion by direct action on GnRH neurons. On the contrary, kisspeptin neurons in the arcuate nucleus are direct targets of sex steroids in all species and should be viewed as the site of the negative feedback control of GnRH production. In addition, kisspeptin produced by the anteroventral periventricular nucleus, a sexually dimorphic nucleus rich in steroid-sensitive neurons in the female, may mediate the positive feedback effects of estrogen on GnRH secretion. Expression of KISS1 is also influenced by adiposity and satiety factors including leptins and Ghrelin ( Tena-Sempere 2008 ). The hypothalamic release of GnRH is modulated by a variety of factors including estradiol, progesterone, catecholamines, dopamine, and endorphins ( Figure 1 ). Norepinephrine, glutamate, and neuropeptide Y stimulate, whereas endogenous opioids, GABA, and corticotrophin-releasing hormone inhibit GnRH secretion ( Zeleznik and Pohl 2006 ). Ovarian steroid hormones can exert positive (e.g., during ovulation trigger) and negative feedback effects (e.g., during follicular and luteal phases) and can act at the hypothalamic and/or pituitary level. In the ovariectomized monkey, estradiol reduced hypothalamic pulsatile multiunit activity compatible with a direct hypothalamic effect ( O’Byrne and Knobil 1993 ). The inhibitory effects of progesterone might be mediated via an inhibitory factor ( Zeleznik and Pohl 2006 ). The GnRH-driven synthesis and secretion of gonadotropic hormones is modulated by ovarian steroid and protein factors ( Figure 2 ). During the follicular phase, increased estradiol production by the maturing follicles exerts a negative feedback action and decreases FSH secretion (and LH secretion) by lowering pulse amplitudes. Once serum estradiol concentration surpasses a certain threshold over a defined period, a positive feedback action of estradiol on FSH and LH release occurs, thus provoking the midcycle peak of gonadotropin secretion and inducing ovulation. This effect is assumed to involve actions at the hypothalamic level (increase of GnRH release) and at the pituitary level. It is assumed that the midcycle progesterone increase is also involved in this positive feedback activity since progesterone antagonists delay or even block ovulation, an effect that is reversible on provision of progesterone. On the other hand, work in rhesus monkeys suggests that progesterone has a direct beneficial effect on oocyte maturation/follicle survival but is unable to elicit ovulation in the absence of LH ( Borman et al. 2004 ). Substance P, a neurokinin, may be involved in the control of the preovulatory LH release since administration of a specific substance P antagonist acting via the NK1 receptor significantly lowered LH concentrations ( Kerdelhue et al. 2000 ). Following ovulation, estradiol and progesterone exert negative feedback actions on both gonadotropic hormones, ultimately resulting in restoration of preovulatory levels via reduction of pulse frequency rather than pulse amplitude. It seems that pulse amplitude and frequency are inversely related, yielding comparable LH output during the follicular and the luteal phases ( Veldhuis and Johnson 1990 ). In addition, a peptide-mediated feedback loop has been described for FSH that is mediated by the effects of inhibins/activins and follistatin ( Bilezikjian et al. 2006 ; Messinis 2006 ). Inhibin secretion from granulosa cells is stimulated by LH and FSH and in turn inhibits FSH secretion. Activin and follistatin are also involved in FSH feedback regulation but act more as local regulators than as endocrine factors. Within the pituitary, local and differential regulation of FSH versus LH synthesis might also involve pituitary-adenylase cyclase-activating peptide produced in folliculostellate cells ( Winters and Moore 2007 ). Peripheral feedback effects of inhibins and activins occur at the level of the pituitary. Activins selectively stimulate FSH secretion, and follistatin binds to activin and, presumably, determines and regulates activin-associated effects through this mechanism.

It is important to recognize that some nonhuman primate species exhibit distinct seasonal variation in reproductive functions depending on the latitude of their habitat. Species residing in equatorial habitats and not exposed to widely varying seasons lack such clear-cut annual variation of reproductive functions. Marmosets reside in the canopy of tropical forests and do not show reproductive seasonality of fertility or breeding. Cynomolgus monkeys are sexually active and fertile throughout the entire year and are not considered to express distinct and significant reproductive seasonality. Under feral conditions, pregnant females were discovered throughout the entire year ( Kavanagh and Laursen 1984 ) with the incidence of birth being somewhat related to the rainy season. Menstrual cyclicity was unrelated to seasonal environment ( Dang 1977 ). Our own observations in seventy-eight cynomolgus monkeys housed individually ( Figure 8 and Figure 13 ; data for reference group) and twenty animals housed in ten pairs ( Figure 10 – Figure 12 ) for at least twelve months are concordant with the absence of ovarian seasonality in the cynomolgus monkey under laboratory conditions and under natural-light conditions. In contrast, rhesus monkeys display pronounced seasonality of reproductive activity for both sexes ( Ghosh and Sengupta 1992 ; Herndon et al. 1996 ). In this primate, reproductive functions are entirely shut off or severely diminished for approximately half of the year. In our latitudes, sexual activity and gonadal activity are present roughly throughout October until March, but this may also vary substantially for individual animals. It is crucial to consider that the annual rhythmicity of reproductive cycles can persist over years in captivity and under indoor artificial-light conditions ( Wickings and Nieschlag 1980 ). Hence, in these circumstances, seasonality of reproduction may be uncoupled from actual season, and eventually, an animal may display periodic reproductive activation that is uncoupled to outside season. During the out-of-season periods, reproductive hormone secretion and gonadal activity are at a complete halt. Therefore, the seasonal status of rhesus monkeys should be assessed carefully in general toxicity studies and obviously in special toxicity studies when reproductive parameters are a suspected target of the test article.

Recent developments in regulations governing the use of primates demand social housing whenever possible ( Council of Europe 2007 ). Although it is evident that nonhuman primates generally benefit from social housing ( Capitanio, Kyes, and Fairbanks 2006 ), systematic studies on the effects of social housing of cynomolgus monkeys in the context of reproductive toxicity evaluation are scarce. Earlier work in this species indicated that subordinate females have fewer ovulatory menstrual cycles and more cycles with deficient luteal-phase progesterone production ( Adams, Kaplan, and Koritnik 1985 ). In these studies, however, group housing involved a male along with several females, whereas during the conduct of toxicity studies, the females were housed in isosexual pairs or groups. This experimental settings raise the question about effects of social housing on the endocrinology and duration of the menstrual cycle. Also, based on an article on “menstrual synchrony and suppression” by McClintock (1971) , a series of studies were performed to evaluate the original hypothesis that close physical contact as well as pheromonal mechanisms may trigger the synchronization of human ovarian cycles. Comparative analysis of these studies yielded ambiguous results ( Wilson, Kiefhaber, and Gravel 1991 ) leading to the proposition that avoiding synchrony might rather be a strategy to prevent female-female competition. We performed a series of studies to evaluate the effects of social housing on female reproductive endocrinology and to clarify whether ovarian cycle synchronization occurs in the cynomolgus monkey model during pair housing and group housing. In one experiment, a newly formed group of eight previously single-housed animals was studied over fifteen ovarian cycles—that is, up to three cycles under single housing and twelve cycles following group formation. Cyclicity was assessed by daily vaginal smears and cycles were numbered consecutively. Data were compared to cycle data from seventy-eight animals housed individually during a comparable period. Blood samples were taken during cycle 1 (single housing) and during cycles 3, 5, 7, 9, and 11 (all group housing) on cycle days 1, 4, 7, 10, 11, 12, 13, 14, 15, 16, 18, 20, 22, 24, and 27 for estradiol and progesterone determination. Social rank was determined by food challenge test in regular intervals ( Rilling, Winslow, and Kilts 2004 ). Clear effects on variation in cycle length were encountered when animals were transferred from individual housing to group housing. During group housing, average cycle duration became prolonged from thirty-one to forty-six days for about six months but subsequently returned to baseline duration from cycle 6 onward ( Figure 8 ). The prolongation of menstrual-cycle duration corresponded to a statistically significant absent or decreased preovulatory estradiol peak and loss of the normal luteal progesterone rise in cycle 3 ( Figure 9 ), whereas from cycle 6 onward, hormone patterns appeared normal. These observations have a significant bearing on the design of female reproductive toxicity studies, suggesting a need for either individual housing or use of pairs and groups with an established history of social contact for at least six months. In another experiment, twenty mature animals—already accustomed to each other—were pair housed (pairs were denoted as A/B, C/D, E/F, G/H, I/J, K/L, M/N, O/P, R/S, T/U) in a climate-controlled room over a period of approximately twelve months. Artificial lighting was controlled automatically to provide a cycle of twelve hours of light and twelve hours of dark exposure. During the observation period, the menstrual cycle was monitored using daily vaginal smears for menses detection. Average cycle duration across the year ranged between 29.3 and 31.3 days, and no evidence for seasonal variation of ovarian cycles was obtained (data not shown). Analysis of individual menstrual cycle patterns revealed no evidence for a menstrual-cycle synchrony in pair-housed (examples in Figure 10 – Figure 12 ) animals. For pair E/F ( Figure 10 ), cycle timing was different but very robust and not influenced by the cage mate. Pair C/D ( Figure 11 ) maintained a parallel cycle, whereas pair I/J maintained an irregular cyclicity ( Figure 12 ) irrespective of the cage mate. Generally, all pairs presented robust and regular cycle durations. It has been reported earlier that female proximity or social contact might stabilize ovarian cycles ( Wallis, King, and Roth-Meyer 1986 ). Our data set was not sufficiently large to ascertain such effect. During the group-housing experiment comprising eight animals described above and shown in Figure 8 and Figure 9 , ovarian cycles were studied throughout fifteen cycles. These data were replotted to evaluate whether seasonal or synchronization effects were present ( Figure 13 ). Neither seasonality nor synchronization of ovarian cycles could be detected in these experiments. Hence, there is no evidence for the synchronization of ovarian cycles in female cynomolgus monkeys housed in pairs or groups. In further experiments, cycle duration was studied throughout a six-month period in three pairs that had been in contact for more than twelve months and in two pairs that had not had contact previously. Artificial lighting was controlled automatically to provide a cycle of twelve hours of light and twelve hours of dark exposure. These animals were also subjected to daily vehicle administration during toxicity evaluation. Ovarian cycle duration remained constant in the three accustomed pairs ( Figure 14 ). In contrast, in the two pairs without previous contact, one animal each of the newly formed pairs presented with prolonged and irregular cycle ( Figure 15 ). These observations along with the findings of the first two experiments clearly demonstrate that for the study of ovarian cycle length and reproductive endocrinology in socially housed female cynomolgus monkeys, it is essential to consider the housing history and familiarity between the animals prior to pair or group formation.

Vaginal smears for menstruation are sufficient for identifying normally cycling female animals. For a special female fertility study, it is important to use only animals with a regular menstrual cycle—that is, cycle monitoring should be available for three to five consecutive cycles prior to recruiting the animals into a toxicity study. Only 30% to 50% of cycling females may present with a regular ovarian cycle. If it is decided to include endocrine measurements, frequent blood-sampling schedules (e.g., as described above) should be included both during predose, during test-item exposure, and during the recovery period. Typically, one observation and two treatment cycles are used for closely monitoring hormone levels, whereas inclusion of further sampling cycles might impose limitations on blood-volume availability. The use of weekly or monthly determinations of female reproductive hormones is not recommended, as these intervals might prove to be too long for making judgments about endocrine effects. At a minimum, estradiol and progesterone levels should be determined since these two steroids are the main regulators of the primate ovarian cycle. Preferably, LH and FSH should be included in the analysis. In our experience, it is not advisable to combine vaginal smears with weekly or monthly hormone determinations. Should vaginal smear analysis indicate ovarian irregularities, it is usually not possible to reconcile or interpret these alterations in conjunction with infrequent hormone data.